visual learning representations

Primed To Learn

Embark on Your Learning Journey

21 Effective Visual Learning Strategies To Engage Visual Learners

If you’re a teacher or a parent who’s ever wondered, “How can I make learning more engaging and effective for my students or children?”, then you’re in the right place. Visual learning strategies are powerful tools that can transform the way information is absorbed, retained, and recalled. They make the most of our brain’s ability to process visual information, which is inherently more interesting and memorable than plain text or spoken words. So, ready to explore these game-changing techniques with us? Let’s get started!

Visual Learning Strategies

Visual learning strategies can greatly benefit students by making complex concepts more accessible and engaging. Here’s a list of visual learning strategies:

1. Visual Aids

As a parent or teacher, one effective way of employing these strategies is by integrating diagrams, charts, or infographics into the learning process. For instance, let’s say you’re tasked with teaching a visual learner about the water cycle. Instead of relying solely on verbal explanations or text, consider using a detailed diagram of the water cycle. 

This visual aid can clearly illustrate each stage – from evaporation, to condensation, to precipitation, and collection – offering an easily comprehensible and memorable representation of the concept. This strategy not only caters to visual learners’ strengths, but also helps to foster a more engaging and interactive learning environment.

2. Graphic Organizers

Graphic Organizers are a potent visual learning strategy that can significantly aid in understanding and retaining complex information. Essentially, they’re visual displays teachers or parents can utilize to organize information in a manner that makes it easier for visual learners to grasp. 

For instance, let’s consider you’re helping a student understand the storyline of a novel. You could use a Story Map graphic organizer, which visually outlines the key elements of the story such as setting, characters, conflict, resolution, and plot events. This allows the student to see the relationships between different parts of the story, promoting a deeper understanding and recall. Thus, Graphic Organizers can turn a daunting task into an engaging, manageable, and visually stimulating learning experience.

3. Mind Maps

Mind maps are an extraordinary visual learning strategy that teachers and parents can effortlessly utilize to enhance a visual learner’s comprehension. By creating a central concept and branching out with related ideas, mind maps serve as an effective tool for brainstorming, note-taking, or summarizing a topic. 

For instance, you can create a mind map while teaching a history lesson. The central concept could be ‘World War II’, with branches sprouting to various key aspects like ‘Causes’, ‘Key Figures’, ‘Major Battles’, and ‘Consequences’. Each branch can further be divided into smaller branches, encapsulating all the details in a visually appealing and understandable format. 

Best Mind Mapping Tools For Learning

With a glance, students can grasp the overall structure of the topic, seeing how different pieces of information connect to each other. This approach resonates particularly well with visual learners, making learning an enjoyable and productive process.

4. Color Coding

Color coding is a powerful visual learning strategy that can significantly enhance information retention and understanding. For teachers and parents, this is a practical and efficient tool to help visual learners excel. For example, when helping a child learn mathematics, color coding can be used to differentiate between various mathematical operations. Multiplication problems could be highlighted in blue, division in green, addition in yellow, and subtraction in red. 

This way, the child can visually organize the information, making it easier to recognize and solve different types of problems. Using color coding as a visual learning strategy, you can effectively connect symbols and meanings, making learning more enticing and fun for visual learners.

5. Visual Timelines

Visual learning strategies are effective tools for enhancing comprehension and retention among visual learners. One particularly effective strategy is the use of “Visual Timelines”. Visual timelines provide a graphical representation of events in chronological order. For example, a teacher teaching a history lesson about World War II can use a visual timeline to plot key events, battles, and political shifts. This way, the students can easily understand the progression of events, their interconnections, and their relative significance. 

Similarly, a parent helping their child learn daily routines or understand concepts of time can create a visual timeline of a typical day. This could include images representing waking up, eating breakfast, going to school, doing homework, and sleeping. This visual reference aids in developing a clear and logical understanding of sequences and timeframes, fostering effective learning.

6. Interactive Whiteboards

Interactive Whiteboards (IWBs) are a powerful visual learning strategy that can significantly enhance comprehension and engagement for visual learners. They are essentially ‘touch-sensitive’ screens connected to a computer and projector, enabling the display of interactive content. For instance, a teacher or parent can use an IWB when teaching fractions. They could display pie charts or bars that can be manipulated to show different fractions.

The learner can physically interact with the content, changing the pie chart’s size or the bar’s length. This hands-on interaction, coupled with the visual representation, helps the learner understand the concept better. It’s not just about hearing the information; it’s about seeing it, touching it, and interacting with it. With IWBs, learning becomes a dynamic experience, fostering a deeper understanding and retention of the material.

7. Visual Summaries

Visual Summaries are an excellent strategy for visual learners, providing clear, easy-to-understand overviews of a topic. This method is great for processing complex information, breaking it down into digestible, visual chunks. As a teacher or parent, you can utilize this strategy to enhance comprehension and retention of knowledge.

For instance, let’s say you’re teaching a unit on the solar system. Instead of relying solely on text-based materials, you could create a Visual Summary. This tool could include illustrations of the planets, their orbits, and other significant features, each labeled with important facts. By doing this, you’re offering a visual learning aid that helps students or your child to understand, remember, and recall the information more effectively.

8. Videos and Animations

As a teacher or parent, one powerful visual learning strategy at your disposal is the use of videos and animations. These dynamic tools bring concepts to life in a way that textbooks cannot, making complex information more digestible. For instance, if you’re teaching a child about the water cycle, a simple animation can illustrate each step—evaporation, condensation, precipitation, and collection—in an engaging, easy-to-understand manner. The child can see the process unfolding, helping them to grasp and remember the concept more effectively. This approach not only enhances comprehension but also fosters an enjoyable learning experience.

9. Illustrated Stories

Illustrated Stories are a powerful visual learning strategy that can be capitalized on by both teachers and parents to enhance comprehension and retention for visual learners. This approach involves using graphic elements, such as pictures or animations, to accompany and illustrate the narrative of a story. The idea is to leverage the visual learner’s innate ability to process and remember information presented visually. 

For example, let’s say a teacher is introducing a new topic – “The Life Cycle of a Butterfly.” Instead of simply describing the stages, the teacher can present an illustrated storybook that vividly depicts each stage in a butterfly’s life cycle. This not only makes the lesson more engaging but also enables the visual learner to form a mental picture of the process, aiding in long-term retention of the information. The use of Illustrated Stories can be a fun and effective addition to visual learning strategies.

10. Virtual Field Trips

Virtual Field Trips are a fantastic visual learning strategy that can significantly enhance the learning experience, especially for visual learners. These online journeys allow students to explore different locations, cultures, or events from the comfort of their classroom or home, delivering a vibrant and immersive learning experience that textbooks might struggle to provide. This method is both interactive and visually stimulating, facilitating a higher level of engagement and understanding.

For instance, a teacher might utilize a Virtual Field Trip to the Smithsonian National Museum of Natural History during a lesson on dinosaurs. The students can virtually navigate through the museum, exploring the exhibits, and closely observing the creature’s skeletons. This experience, paired with a guided discussion or follow-up activities, can help reinforce the lesson in a way that’s unforgettable and meaningful for visual learners.

What is Virtual Learning? 10 Best Practices to Implement

11. Conceptual Models

Conceptual models are a powerful tool in visual learning strategies. They enable you to depict complex concepts or processes graphically, making them more comprehensible and engaging for visual learners. For instance, a teacher teaching the solar system can create a 3D model displaying the planets and their relative positions and sizes. 

This hands-on, visual strategy allows students to grasp the concept of the solar system in a more tangible and memorable way than text alone. Similarly, parents can use conceptual models at home to explain day-to-day processes. For example, a simple model of a plant can be used to teach children about photosynthesis, turning an abstract concept into a relatable visual experience.

12. Visual Note-Taking

Visual note-taking can be an effective strategy for aiding visual learners in their educational journey. It’s a method that allows learners to represent their thoughts and ideas in a dynamic, visual way, which can significantly enhance their understanding and retention of information. For instance, a teacher or a parent implementing this strategy could encourage a student to draw a diagram or sketch to represent the life cycle of a butterfly when studying biology. 

This exercise not only stimulates visual cognition but also makes the learning process more enjoyable and memorable for the student. Remember, the goal of visual note-taking isn’t to create a piece of art, but rather to create a personalized visual understanding of the information.

13. Visual Vocabulary

Visual Vocabulary is a compelling strategy that can enable visual learners to comprehend and remember new words or concepts more efficiently. It involves associating words with relevant images, symbols, or diagrams to create a visual context. For instance, a teacher teaching the concept of photosynthesis to her students could use a diagram depicting how plants take in carbon dioxide and sunlight to produce glucose and oxygen. 

Similarly, a parent helping a child learn new vocabulary could draw a picture of an ‘apple’ while teaching the word ‘apple’. This association helps the child to remember the word and its meaning for a longer time. The Visual Vocabulary strategy capitalizes on the visual learner’s innate ability to remember and understand visual cues, making learning more engaging and effective.

14. Interactive Diagrams

Interactive diagrams are an excellent visual learning strategy that you can utilize either as a teacher or a parent to enhance the learning experience of visual learners. These diagrams facilitate the understanding of complex concepts by visually depicting the relationships and processes involved. 

For instance, consider teaching the solar system. Instead of relying solely on verbal or textual descriptions, you can use an interactive diagram of the solar system. This diagram can allow learners to click on each planet to reveal information about its size, composition, and its distance from the sun. This not only aids in retaining information but also stimulates curiosity and encourages exploration. The visual representation of the solar system can help visual learners grasp the concept more effectively as they can ‘see’ the information, rather than just read or listen to it.

15. Conceptual Art Projects

Conceptual Art Projects can be an effective visual learning strategy for visual learners, providing a hands-on approach to understanding complex concepts. For instance, let’s consider a teacher or parent trying to explain the concept of the Solar System. Instead of relying solely on textual or oral descriptions, they could initiate a project where the child is involved in creating a 3D model of the Solar System. 

This hands-on project not only allows the child to visually connect with the concept, but also enables them to comprehend the relative sizes and distances of the planets in a more concrete manner. This helps to reinforce the child’s understanding and retention of the subject matter, making learning an enjoyable and lasting experience.

16. Storyboarding

Storyboarding is a powerful visual learning strategy that you, as a teacher or parent, can utilize to enrich the learning experience for visual learners. This method involves creating a visual sequence of events, like a comic strip, to depict a story or process. It aids in comprehension and retention by allowing the learner to visualize the information, thus making abstract concepts more concrete.

For instance, suppose you’re teaching the process of photosynthesis to a child. Instead of solely relying on textual information, you could draw a storyboard illustrating the stages of photosynthesis. The first panel might show a tree absorbing sunlight, the second could depict water and carbon dioxide being absorbed through the roots and leaves, the third would show the production of glucose and oxygen, and so on. This visual representation can make the complex process easier to understand and remember, reinforcing the learning outcome.

17. Comparative Charts

Comparative charts are a fantastic visual learning strategy that can be effectively used by teachers and parents to boost a visual learner’s understanding. These charts allow learners to see comparisons and contrasts between different concepts clearly, making the information more digestible and memorable. 

Let’s consider an example: if a teacher is trying to make students understand the differences and similarities between two historical events, a comparative chart could be an excellent tool. The teacher can list the events vertically down the left side of the chart, with categories for comparison (like cause, impact, key figures) along the top. The corresponding cells can then be filled with the relevant information. As a result, students can visually compare and contrast the two events, aiding their understanding and retention. This approach simplifies complex information and enhances learning for visual learners.

18. Digital Collages

Digital Collages constitute an effective visual learning strategy that can be harnessed by teachers and parents alike to enhance the learning experience of visual learners. They offer a creative avenue to compile and represent information, ideas, or concepts in a visually appealing and comprehensive manner. 

For instance, let’s consider a history lesson on the American Revolution. A teacher or parent can create a Digital Collage that includes key figures, maps, battlefields, and significant events of the period. They can also add brief descriptions or captions to the images. This would not only aid in visualizing the historical events but also in constructing mental links between the different elements. Thus, Digital Collages, by amalgamating text and visuals, can greatly facilitate the learning process for visual learners, making it more engaging and effective.

19. Visual Quizzes

Visual quizzes can be a game changer in the world of visual learning strategies. As a teacher or a parent, you have the opportunity to utilize visual quizzes to enhance understanding and retention for visual learners. Here’s how it works. Let’s say you’re teaching your students or children about the animal kingdom.

Rather than relying solely on verbal or textual descriptions, you could create a visual quiz. For this, compile a set of images showcasing different animals, and ask them to identify which ones are mammals, which are reptiles, and so on. By doing this, you’re allowing them to associate visual elements with the concepts they’re learning, which can boost their memory retention and make learning a fun and interactive experience.

20. Art Integration

Art integration serves as an exceptional visual learning strategy, especially for visual learners, as it emphasizes the use of images, diagrams, and other visual aids to facilitate understanding. For instance, a teacher or parent might choose to integrate art into a history lesson by creating a time-period-specific collage. This could involve gathering pictures, symbols, or drawings that resonate with the era being studied, arranging them in chronological order on a large piece of paper. 

This visual representation not only offers learners a comprehensive overview of the historical period but also allows them to connect more deeply with the subject matter. The tangible, visual nature of the collage fosters a richer learning experience, catering perfectly to the needs of visual learners.

21. Science Labs and Demonstrations

Science labs and demonstrations are a highly effective visual learning strategy that can immensely benefit visual learners. These hands-on activities provide clear, visual representations of scientific concepts, making abstract ideas more tangible and easier to understand. For instance, imagine a teacher or parent demonstrating the reaction between baking soda and vinegar. 

This experiment isn’t just fun and engaging; it visually illustrates the concept of chemical reactions. The learner can see the vinegar (an acid) react with the baking soda (a base) to produce a new substance, carbon dioxide (the bubbles). This visual demonstration brings the science to life, aiding comprehension and making learning a more enjoyable experience for visual learners.

Visual learning strategies are powerful tools that teachers and parents can harness to empower visual learners. As you’ve seen, these methods can turn challenging concepts into memorable visuals, enhancing understanding and engagement. Remember, it’s all about making learning visible, tangible, and interactive. So, don’t be afraid to experiment and to incorporate charts, diagrams, mind maps, or even virtual reality into your teaching methods. Make learning a vibrant, visual journey. After all, for your visual learners, seeing truly is believing.

Frequently Asked Questions

How do visual learners learn best.

As a visual learner, you thrive when information is presented in a way that you can see. Graphs, charts, infographics, timelines, animated videos, or pictorial flashcards tend to work best for you. You probably find it easy to remember information from movies or presentations where visual aids were incorporated extensively. Mind maps are another effective tool for you, helping to visualize the connections between different pieces of information. So, if you’re revising for an exam or seeking to understand a complex theory, try translating that information into a diagram or flowchart. Remember, your strength lies in ‘seeing’ information.

Why is visual learning the best learning style?

Visual learning strategies can be incredibly effective because they cater to a fundamental way that many people process information. As a visual learner, you’re likely to find that information makes the most sense when you can see it. This is because our brains naturally tend to absorb and recall visual information better than auditory or text-based data. It’s like painting a picture in your mind – the colors, shapes, and patterns all contribute to a memorable image that’s easy to recall when you need it. 

What do visual learners struggle with?

Visual learners, while having the advantage of learning quickly through images, diagrams, and other visual aids, often struggle with auditory instructions and long passages of written information. They may find lectures or discussion-based classes difficult to follow, as their strength lies in seeing and visualizing rather than hearing or reading. Additionally, they may struggle with complex concepts that are presented without accompanying visual aids, such as charts or diagrams.

What is the best material for visual learners?

Infographics and diagrams often serve as the most effective materials for visual learners. These types of content allow you, as a visual learner, to quickly grasp complex information and new concepts. Infographics are beneficial because they break down data into a visually appealing and digestible format. Diagrams, on the other hand, help you visualize the structure of an idea or process, making it easier to remember and understand. So, when it comes to visual learning strategies, incorporating infographics and diagrams into your study routine could significantly boost your comprehension and retention of information.

You might also like:

• 15 Surprising Benefits of Multisensory Learning
• 10 Innovative Blended Learning Strategies to Transform Your Classroom

• Open access
• Published: 19 July 2015

The role of visual representations in scientific practices: from conceptual understanding and knowledge generation to ‘seeing’ how science works

• Maria Evagorou 1 ,
• Sibel Erduran 2 &
• Terhi Mäntylä 3  

International Journal of STEM Education volume  2 , Article number:  11 ( 2015 ) Cite this article

76k Accesses

78 Citations

13 Altmetric

Metrics details

The use of visual representations (i.e., photographs, diagrams, models) has been part of science, and their use makes it possible for scientists to interact with and represent complex phenomena, not observable in other ways. Despite a wealth of research in science education on visual representations, the emphasis of such research has mainly been on the conceptual understanding when using visual representations and less on visual representations as epistemic objects. In this paper, we argue that by positioning visual representations as epistemic objects of scientific practices, science education can bring a renewed focus on how visualization contributes to knowledge formation in science from the learners’ perspective.

This is a theoretical paper, and in order to argue about the role of visualization, we first present a case study, that of the discovery of the structure of DNA that highlights the epistemic components of visual information in science. The second case study focuses on Faraday’s use of the lines of magnetic force. Faraday is known of his exploratory, creative, and yet systemic way of experimenting, and the visual reasoning leading to theoretical development was an inherent part of the experimentation. Third, we trace a contemporary account from science focusing on the experimental practices and how reproducibility of experimental procedures can be reinforced through video data.

Conclusions

Our conclusions suggest that in teaching science, the emphasis in visualization should shift from cognitive understanding—using the products of science to understand the content—to engaging in the processes of visualization. Furthermore, we suggest that is it essential to design curriculum materials and learning environments that create a social and epistemic context and invite students to engage in the practice of visualization as evidence, reasoning, experimental procedure, or a means of communication and reflect on these practices. Implications for teacher education include the need for teacher professional development programs to problematize the use of visual representations as epistemic objects that are part of scientific practices.

During the last decades, research and reform documents in science education across the world have been calling for an emphasis not only on the content but also on the processes of science (Bybee 2014 ; Eurydice 2012 ; Duschl and Bybee 2014 ; Osborne 2014 ; Schwartz et al. 2012 ), in order to make science accessible to the students and enable them to understand the epistemic foundation of science. Scientific practices, part of the process of science, are the cognitive and discursive activities that are targeted in science education to develop epistemic understanding and appreciation of the nature of science (Duschl et al. 2008 ) and have been the emphasis of recent reform documents in science education across the world (Achieve 2013 ; Eurydice 2012 ). With the term scientific practices, we refer to the processes that take place during scientific discoveries and include among others: asking questions, developing and using models, engaging in arguments, and constructing and communicating explanations (National Research Council 2012 ). The emphasis on scientific practices aims to move the teaching of science from knowledge to the understanding of the processes and the epistemic aspects of science. Additionally, by placing an emphasis on engaging students in scientific practices, we aim to help students acquire scientific knowledge in meaningful contexts that resemble the reality of scientific discoveries.

Despite a wealth of research in science education on visual representations, the emphasis of such research has mainly been on the conceptual understanding when using visual representations and less on visual representations as epistemic objects. In this paper, we argue that by positioning visual representations as epistemic objects, science education can bring a renewed focus on how visualization contributes to knowledge formation in science from the learners’ perspective. Specifically, the use of visual representations (i.e., photographs, diagrams, tables, charts) has been part of science and over the years has evolved with the new technologies (i.e., from drawings to advanced digital images and three dimensional models). Visualization makes it possible for scientists to interact with complex phenomena (Richards 2003 ), and they might convey important evidence not observable in other ways. Visual representations as a tool to support cognitive understanding in science have been studied extensively (i.e., Gilbert 2010 ; Wu and Shah 2004 ). Studies in science education have explored the use of images in science textbooks (i.e., Dimopoulos et al. 2003 ; Bungum 2008 ), students’ representations or models when doing science (i.e., Gilbert et al. 2008 ; Dori et al. 2003 ; Lehrer and Schauble 2012 ; Schwarz et al. 2009 ), and students’ images of science and scientists (i.e., Chambers 1983 ). Therefore, studies in the field of science education have been using the term visualization as “the formation of an internal representation from an external representation” (Gilbert et al. 2008 , p. 4) or as a tool for conceptual understanding for students.

In this paper, we do not refer to visualization as mental image, model, or presentation only (Gilbert et al. 2008 ; Philips et al. 2010 ) but instead focus on visual representations or visualization as epistemic objects. Specifically, we refer to visualization as a process for knowledge production and growth in science. In this respect, modeling is an aspect of visualization, but what we are focusing on with visualization is not on the use of model as a tool for cognitive understanding (Gilbert 2010 ; Wu and Shah 2004 ) but the on the process of modeling as a scientific practice which includes the construction and use of models, the use of other representations, the communication in the groups with the use of the visual representation, and the appreciation of the difficulties that the science phase in this process. Therefore, the purpose of this paper is to present through the history of science how visualization can be considered not only as a cognitive tool in science education but also as an epistemic object that can potentially support students to understand aspects of the nature of science.

Scientific practices and science education

According to the New Generation Science Standards (Achieve 2013 ), scientific practices refer to: asking questions and defining problems; developing and using models; planning and carrying out investigations; analyzing and interpreting data; using mathematical and computational thinking; constructing explanations and designing solutions; engaging in argument from evidence; and obtaining, evaluating, and communicating information. A significant aspect of scientific practices is that science learning is more than just about learning facts, concepts, theories, and laws. A fuller appreciation of science necessitates the understanding of the science relative to its epistemological grounding and the process that are involved in the production of knowledge (Hogan and Maglienti 2001 ; Wickman 2004 ).

The New Generation Science Standards is, among other changes, shifting away from science inquiry and towards the inclusion of scientific practices (Duschl and Bybee 2014 ; Osborne 2014 ). By comparing the abilities to do scientific inquiry (National Research Council 2000 ) with the set of scientific practices, it is evident that the latter is about engaging in the processes of doing science and experiencing in that way science in a more authentic way. Engaging in scientific practices according to Osborne ( 2014 ) “presents a more authentic picture of the endeavor that is science” (p.183) and also helps the students to develop a deeper understanding of the epistemic aspects of science. Furthermore, as Bybee ( 2014 ) argues, by engaging students in scientific practices, we involve them in an understanding of the nature of science and an understanding on the nature of scientific knowledge.

Science as a practice and scientific practices as a term emerged by the philosopher of science, Kuhn (Osborne 2014 ), refers to the processes in which the scientists engage during knowledge production and communication. The work that is followed by historians, philosophers, and sociologists of science (Latour 2011 ; Longino 2002 ; Nersessian 2008 ) revealed the scientific practices in which the scientists engage in and include among others theory development and specific ways of talking, modeling, and communicating the outcomes of science.

Visualization as an epistemic object

Schematic, pictorial symbols in the design of scientific instruments and analysis of the perceptual and functional information that is being stored in those images have been areas of investigation in philosophy of scientific experimentation (Gooding et al. 1993 ). The nature of visual perception, the relationship between thought and vision, and the role of reproducibility as a norm for experimental research form a central aspect of this domain of research in philosophy of science. For instance, Rothbart ( 1997 ) has argued that visualizations are commonplace in the theoretical sciences even if every scientific theory may not be defined by visualized models.

Visual representations (i.e., photographs, diagrams, tables, charts, models) have been used in science over the years to enable scientists to interact with complex phenomena (Richards 2003 ) and might convey important evidence not observable in other ways (Barber et al. 2006 ). Some authors (e.g., Ruivenkamp and Rip 2010 ) have argued that visualization is as a core activity of some scientific communities of practice (e.g., nanotechnology) while others (e.g., Lynch and Edgerton 1988 ) have differentiated the role of particular visualization techniques (e.g., of digital image processing in astronomy). Visualization in science includes the complex process through which scientists develop or produce imagery, schemes, and graphical representation, and therefore, what is of importance in this process is not only the result but also the methodology employed by the scientists, namely, how this result was produced. Visual representations in science may refer to objects that are believed to have some kind of material or physical existence but equally might refer to purely mental, conceptual, and abstract constructs (Pauwels 2006 ). More specifically, visual representations can be found for: (a) phenomena that are not observable with the eye (i.e., microscopic or macroscopic); (b) phenomena that do not exist as visual representations but can be translated as such (i.e., sound); and (c) in experimental settings to provide visual data representations (i.e., graphs presenting velocity of moving objects). Additionally, since science is not only about replicating reality but also about making it more understandable to people (either to the public or other scientists), visual representations are not only about reproducing the nature but also about: (a) functioning in helping solving a problem, (b) filling gaps in our knowledge, and (c) facilitating knowledge building or transfer (Lynch 2006 ).

Using or developing visual representations in the scientific practice can range from a straightforward to a complicated situation. More specifically, scientists can observe a phenomenon (i.e., mitosis) and represent it visually using a picture or diagram, which is quite straightforward. But they can also use a variety of complicated techniques (i.e., crystallography in the case of DNA studies) that are either available or need to be developed or refined in order to acquire the visual information that can be used in the process of theory development (i.e., Latour and Woolgar 1979 ). Furthermore, some visual representations need decoding, and the scientists need to learn how to read these images (i.e., radiologists); therefore, using visual representations in the process of science requires learning a new language that is specific to the medium/methods that is used (i.e., understanding an X-ray picture is different from understanding an MRI scan) and then communicating that language to other scientists and the public.

There are much intent and purposes of visual representations in scientific practices, as for example to make a diagnosis, compare, describe, and preserve for future study, verify and explore new territory, generate new data (Pauwels 2006 ), or present new methodologies. According to Latour and Woolgar ( 1979 ) and Knorr Cetina ( 1999 ), visual representations can be used either as primary data (i.e., image from a microscope). or can be used to help in concept development (i.e., models of DNA used by Watson and Crick), to uncover relationships and to make the abstract more concrete (graphs of sound waves). Therefore, visual representations and visual practices, in all forms, are an important aspect of the scientific practices in developing, clarifying, and transmitting scientific knowledge (Pauwels 2006 ).

Methods and Results: Merging Visualization and scientific practices in science

In this paper, we present three case studies that embody the working practices of scientists in an effort to present visualization as a scientific practice and present our argument about how visualization is a complex process that could include among others modeling and use of representation but is not only limited to that. The first case study explores the role of visualization in the construction of knowledge about the structure of DNA, using visuals as evidence. The second case study focuses on Faraday’s use of the lines of magnetic force and the visual reasoning leading to the theoretical development that was an inherent part of the experimentation. The third case study focuses on the current practices of scientists in the context of a peer-reviewed journal called the Journal of Visualized Experiments where the methodology is communicated through videotaped procedures. The three case studies represent the research interests of the three authors of this paper and were chosen to present how visualization as a practice can be involved in all stages of doing science, from hypothesizing and evaluating evidence (case study 1) to experimenting and reasoning (case study 2) to communicating the findings and methodology with the research community (case study 3), and represent in this way the three functions of visualization as presented by Lynch ( 2006 ). Furthermore, the last case study showcases how the development of visualization technologies has contributed to the communication of findings and methodologies in science and present in that way an aspect of current scientific practices. In all three cases, our approach is guided by the observation that the visual information is an integral part of scientific practices at the least and furthermore that they are particularly central in the scientific practices of science.

Case study 1: use visual representations as evidence in the discovery of DNA

The focus of the first case study is the discovery of the structure of DNA. The DNA was first isolated in 1869 by Friedrich Miescher, and by the late 1940s, it was known that it contained phosphate, sugar, and four nitrogen-containing chemical bases. However, no one had figured the structure of the DNA until Watson and Crick presented their model of DNA in 1953. Other than the social aspects of the discovery of the DNA, another important aspect was the role of visual evidence that led to knowledge development in the area. More specifically, by studying the personal accounts of Watson ( 1968 ) and Crick ( 1988 ) about the discovery of the structure of the DNA, the following main ideas regarding the role of visual representations in the production of knowledge can be identified: (a) The use of visual representations was an important part of knowledge growth and was often dependent upon the discovery of new technologies (i.e., better microscopes or better techniques in crystallography that would provide better visual representations as evidence of the helical structure of the DNA); and (b) Models (three-dimensional) were used as a way to represent the visual images (X-ray images) and connect them to the evidence provided by other sources to see whether the theory can be supported. Therefore, the model of DNA was built based on the combination of visual evidence and experimental data.

An example showcasing the importance of visual representations in the process of knowledge production in this case is provided by Watson, in his book The Double Helix (1968):

…since the middle of the summer Rosy [Rosalind Franklin] had had evidence for a new three-dimensional form of DNA. It occurred when the DNA 2molecules were surrounded by a large amount of water. When I asked what the pattern was like, Maurice went into the adjacent room to pick up a print of the new form they called the “B” structure. The instant I saw the picture, my mouth fell open and my pulse began to race. The pattern was unbelievably simpler than those previously obtained (A form). Moreover, the black cross of reflections which dominated the picture could arise only from a helical structure. With the A form the argument for the helix was never straightforward, and considerable ambiguity existed as to exactly which type of helical symmetry was present. With the B form however, mere inspection of its X-ray picture gave several of the vital helical parameters. (p. 167-169)

As suggested by Watson’s personal account of the discovery of the DNA, the photo taken by Rosalind Franklin (Fig.  1 ) convinced him that the DNA molecule must consist of two chains arranged in a paired helix, which resembles a spiral staircase or ladder, and on March 7, 1953, Watson and Crick finished and presented their model of the structure of DNA (Watson and Berry 2004 ; Watson 1968 ) which was based on the visual information provided by the X-ray image and their knowledge of chemistry.

X-ray chrystallography of DNA

In analyzing the visualization practice in this case study, we observe the following instances that highlight how the visual information played a role:

Asking questions and defining problems: The real world in the model of science can at some points only be observed through visual representations or representations, i.e., if we are using DNA as an example, the structure of DNA was only observable through the crystallography images produced by Rosalind Franklin in the laboratory. There was no other way to observe the structure of DNA, therefore the real world.

Analyzing and interpreting data: The images that resulted from crystallography as well as their interpretations served as the data for the scientists studying the structure of DNA.

Experimenting: The data in the form of visual information were used to predict the possible structure of the DNA.

Modeling: Based on the prediction, an actual three-dimensional model was prepared by Watson and Crick. The first model did not fit with the real world (refuted by Rosalind Franklin and her research group from King’s College) and Watson and Crick had to go through the same process again to find better visual evidence (better crystallography images) and create an improved visual model.

Example excerpts from Watson’s biography provide further evidence for how visualization practices were applied in the context of the discovery of DNA (Table  1 ).

In summary, by examining the history of the discovery of DNA, we showcased how visual data is used as scientific evidence in science, identifying in that way an aspect of the nature of science that is still unexplored in the history of science and an aspect that has been ignored in the teaching of science. Visual representations are used in many ways: as images, as models, as evidence to support or rebut a model, and as interpretations of reality.

Case study 2: applying visual reasoning in knowledge production, the example of the lines of magnetic force

The focus of this case study is on Faraday’s use of the lines of magnetic force. Faraday is known of his exploratory, creative, and yet systemic way of experimenting, and the visual reasoning leading to theoretical development was an inherent part of this experimentation (Gooding 2006 ). Faraday’s articles or notebooks do not include mathematical formulations; instead, they include images and illustrations from experimental devices and setups to the recapping of his theoretical ideas (Nersessian 2008 ). According to Gooding ( 2006 ), “Faraday’s visual method was designed not to copy apparent features of the world, but to analyse and replicate them” (2006, p. 46).

The lines of force played a central role in Faraday’s research on electricity and magnetism and in the development of his “field theory” (Faraday 1852a ; Nersessian 1984 ). Before Faraday, the experiments with iron filings around magnets were known and the term “magnetic curves” was used for the iron filing patterns and also for the geometrical constructs derived from the mathematical theory of magnetism (Gooding et al. 1993 ). However, Faraday used the lines of force for explaining his experimental observations and in constructing the theory of forces in magnetism and electricity. Examples of Faraday’s different illustrations of lines of magnetic force are given in Fig.  2 . Faraday gave the following experiment-based definition for the lines of magnetic forces:

a Iron filing pattern in case of bar magnet drawn by Faraday (Faraday 1852b , Plate IX, p. 158, Fig. 1), b Faraday’s drawing of lines of magnetic force in case of cylinder magnet, where the experimental procedure, knife blade showing the direction of lines, is combined into drawing (Faraday, 1855, vol. 1, plate 1)

A line of magnetic force may be defined as that line which is described by a very small magnetic needle, when it is so moved in either direction correspondent to its length, that the needle is constantly a tangent to the line of motion; or it is that line along which, if a transverse wire be moved in either direction, there is no tendency to the formation of any current in the wire, whilst if moved in any other direction there is such a tendency; or it is that line which coincides with the direction of the magnecrystallic axis of a crystal of bismuth, which is carried in either direction along it. The direction of these lines about and amongst magnets and electric currents, is easily represented and understood, in a general manner, by the ordinary use of iron filings. (Faraday 1852a , p. 25 (3071))

The definition describes the connection between the experiments and the visual representation of the results. Initially, the lines of force were just geometric representations, but later, Faraday treated them as physical objects (Nersessian 1984 ; Pocovi and Finlay 2002 ):

I have sometimes used the term lines of force so vaguely, as to leave the reader doubtful whether I intended it as a merely representative idea of the forces, or as the description of the path along which the power was continuously exerted. … wherever the expression line of force is taken simply to represent the disposition of forces, it shall have the fullness of that meaning; but that wherever it may seem to represent the idea of the physical mode of transmission of the force, it expresses in that respect the opinion to which I incline at present. The opinion may be erroneous, and yet all that relates or refers to the disposition of the force will remain the same. (Faraday, 1852a , p. 55-56 (3075))

He also felt that the lines of force had greater explanatory power than the dominant theory of action-at-a-distance:

Now it appears to me that these lines may be employed with great advantage to represent nature, condition, direction and comparative amount of the magnetic forces; and that in many cases they have, to the physical reasoned at least, a superiority over that method which represents the forces as concentrated in centres of action… (Faraday, 1852a , p. 26 (3074))

For giving some insight to Faraday’s visual reasoning as an epistemic practice, the following examples of Faraday’s studies of the lines of magnetic force (Faraday 1852a , 1852b ) are presented:

(a) Asking questions and defining problems: The iron filing patterns formed the empirical basis for the visual model: 2D visualization of lines of magnetic force as presented in Fig.  2 . According to Faraday, these iron filing patterns were suitable for illustrating the direction and form of the magnetic lines of force (emphasis added):

It must be well understood that these forms give no indication by their appearance of the relative strength of the magnetic force at different places, inasmuch as the appearance of the lines depends greatly upon the quantity of filings and the amount of tapping; but the direction and forms of these lines are well given, and these indicate, in a considerable degree, the direction in which the forces increase and diminish . (Faraday 1852b , p.158 (3237))

Despite being static and two dimensional on paper, the lines of magnetic force were dynamical (Nersessian 1992 , 2008 ) and three dimensional for Faraday (see Fig.  2 b). For instance, Faraday described the lines of force “expanding”, “bending,” and “being cut” (Nersessian 1992 ). In Fig.  2 b, Faraday has summarized his experiment (bar magnet and knife blade) and its results (lines of force) in one picture.

(b) Analyzing and interpreting data: The model was so powerful for Faraday that he ended up thinking them as physical objects (e.g., Nersessian 1984 ), i.e., making interpretations of the way forces act. Of course, he made a lot of experiments for showing the physical existence of the lines of force, but he did not succeed in it (Nersessian 1984 ). The following quote illuminates Faraday’s use of the lines of force in different situations:

The study of these lines has, at different times, been greatly influential in leading me to various results, which I think prove their utility as well as fertility. Thus, the law of magneto-electric induction; the earth’s inductive action; the relation of magnetism and light; diamagnetic action and its law, and magnetocrystallic action, are the cases of this kind… (Faraday 1852a , p. 55 (3174))

(c) Experimenting: In Faraday's case, he used a lot of exploratory experiments; in case of lines of magnetic force, he used, e.g., iron filings, magnetic needles, or current carrying wires (see the quote above). The magnetic field is not directly observable and the representation of lines of force was a visual model, which includes the direction, form, and magnitude of field.

(d) Modeling: There is no denying that the lines of magnetic force are visual by nature. Faraday’s views of lines of force developed gradually during the years, and he applied and developed them in different contexts such as electromagnetic, electrostatic, and magnetic induction (Nersessian 1984 ). An example of Faraday’s explanation of the effect of the wire b’s position to experiment is given in Fig.  3 . In Fig.  3 , few magnetic lines of force are drawn, and in the quote below, Faraday is explaining the effect using these magnetic lines of force (emphasis added):

Picture of an experiment with different arrangements of wires ( a , b’ , b” ), magnet, and galvanometer. Note the lines of force drawn around the magnet. (Faraday 1852a , p. 34)

It will be evident by inspection of Fig. 3 , that, however the wires are carried away, the general result will, according to the assumed principles of action, be the same; for if a be the axial wire, and b’, b”, b”’ the equatorial wire, represented in three different positions, whatever magnetic lines of force pass across the latter wire in one position, will also pass it in the other, or in any other position which can be given to it. The distance of the wire at the place of intersection with the lines of force, has been shown, by the experiments (3093.), to be unimportant. (Faraday 1852a , p. 34 (3099))

In summary, by examining the history of Faraday’s use of lines of force, we showed how visual imagery and reasoning played an important part in Faraday’s construction and representation of his “field theory”. As Gooding has stated, “many of Faraday’s sketches are far more that depictions of observation, they are tools for reasoning with and about phenomena” (2006, p. 59).

Case study 3: visualizing scientific methods, the case of a journal

The focus of the third case study is the Journal of Visualized Experiments (JoVE) , a peer-reviewed publication indexed in PubMed. The journal devoted to the publication of biological, medical, chemical, and physical research in a video format. The journal describes its history as follows:

JoVE was established as a new tool in life science publication and communication, with participation of scientists from leading research institutions. JoVE takes advantage of video technology to capture and transmit the multiple facets and intricacies of life science research. Visualization greatly facilitates the understanding and efficient reproduction of both basic and complex experimental techniques, thereby addressing two of the biggest challenges faced by today's life science research community: i) low transparency and poor reproducibility of biological experiments and ii) time and labor-intensive nature of learning new experimental techniques. ( http://www.jove.com/ )

By examining the journal content, we generate a set of categories that can be considered as indicators of relevance and significance in terms of epistemic practices of science that have relevance for science education. For example, the quote above illustrates how scientists view some norms of scientific practice including the norms of “transparency” and “reproducibility” of experimental methods and results, and how the visual format of the journal facilitates the implementation of these norms. “Reproducibility” can be considered as an epistemic criterion that sits at the heart of what counts as an experimental procedure in science:

Investigating what should be reproducible and by whom leads to different types of experimental reproducibility, which can be observed to play different roles in experimental practice. A successful application of the strategy of reproducing an experiment is an achievement that may depend on certain isiosyncratic aspects of a local situation. Yet a purely local experiment that cannot be carried out by other experimenters and in other experimental contexts will, in the end be unproductive in science. (Sarkar and Pfeifer 2006 , p.270)

We now turn to an article on “Elevated Plus Maze for Mice” that is available for free on the journal website ( http://www.jove.com/video/1088/elevated-plus-maze-for-mice ). The purpose of this experiment was to investigate anxiety levels in mice through behavioral analysis. The journal article consists of a 9-min video accompanied by text. The video illustrates the handling of the mice in soundproof location with less light, worksheets with characteristics of mice, computer software, apparatus, resources, setting up the computer software, and the video recording of mouse behavior on the computer. The authors describe the apparatus that is used in the experiment and state how procedural differences exist between research groups that lead to difficulties in the interpretation of results:

The apparatus consists of open arms and closed arms, crossed in the middle perpendicularly to each other, and a center area. Mice are given access to all of the arms and are allowed to move freely between them. The number of entries into the open arms and the time spent in the open arms are used as indices of open space-induced anxiety in mice. Unfortunately, the procedural differences that exist between laboratories make it difficult to duplicate and compare results among laboratories.

The authors’ emphasis on the particularity of procedural context echoes in the observations of some philosophers of science:

It is not just the knowledge of experimental objects and phenomena but also their actual existence and occurrence that prove to be dependent on specific, productive interventions by the experimenters” (Sarkar and Pfeifer 2006 , pp. 270-271)

The inclusion of a video of the experimental procedure specifies what the apparatus looks like (Fig.  4 ) and how the behavior of the mice is captured through video recording that feeds into a computer (Fig.  5 ). Subsequently, a computer software which captures different variables such as the distance traveled, the number of entries, and the time spent on each arm of the apparatus. Here, there is visual information at different levels of representation ranging from reconfiguration of raw video data to representations that analyze the data around the variables in question (Fig.  6 ). The practice of levels of visual representations is not particular to the biological sciences. For instance, they are commonplace in nanotechnological practices:

Visual illustration of apparatus

Video processing of experimental set-up

Computer software for video input and variable recording

In the visualization processes, instruments are needed that can register the nanoscale and provide raw data, which needs to be transformed into images. Some Imaging Techniques have software incorporated already where this transformation automatically takes place, providing raw images. Raw data must be translated through the use of Graphic Software and software is also used for the further manipulation of images to highlight what is of interest to capture the (inferred) phenomena -- and to capture the reader. There are two levels of choice: Scientists have to choose which imaging technique and embedded software to use for the job at hand, and they will then have to follow the structure of the software. Within such software, there are explicit choices for the scientists, e.g. about colour coding, and ways of sharpening images. (Ruivenkamp and Rip 2010 , pp.14–15)

On the text that accompanies the video, the authors highlight the role of visualization in their experiment:

Visualization of the protocol will promote better understanding of the details of the entire experimental procedure, allowing for standardization of the protocols used in different laboratories and comparisons of the behavioral phenotypes of various strains of mutant mice assessed using this test.

The software that takes the video data and transforms it into various representations allows the researchers to collect data on mouse behavior more reliably. For instance, the distance traveled across the arms of the apparatus or the time spent on each arm would have been difficult to observe and record precisely. A further aspect to note is how the visualization of the experiment facilitates control of bias. The authors illustrate how the olfactory bias between experimental procedures carried on mice in sequence is avoided by cleaning the equipment.

Our discussion highlights the role of visualization in science, particularly with respect to presenting visualization as part of the scientific practices. We have used case studies from the history of science highlighting a scientist’s account of how visualization played a role in the discovery of DNA and the magnetic field and from a contemporary illustration of a science journal’s practices in incorporating visualization as a way to communicate new findings and methodologies. Our implicit aim in drawing from these case studies was the need to align science education with scientific practices, particularly in terms of how visual representations, stable or dynamic, can engage students in the processes of science and not only to be used as tools for cognitive development in science. Our approach was guided by the notion of “knowledge-as-practice” as advanced by Knorr Cetina ( 1999 ) who studied scientists and characterized their knowledge as practice, a characterization which shifts focus away from ideas inside scientists’ minds to practices that are cultural and deeply contextualized within fields of science. She suggests that people working together can be examined as epistemic cultures whose collective knowledge exists as practice.

It is important to stress, however, that visual representations are not used in isolation, but are supported by other types of evidence as well, or other theories (i.e., in order to understand the helical form of DNA, or the structure, chemistry knowledge was needed). More importantly, this finding can also have implications when teaching science as argument (e.g., Erduran and Jimenez-Aleixandre 2008 ), since the verbal evidence used in the science classroom to maintain an argument could be supported by visual evidence (either a model, representation, image, graph, etc.). For example, in a group of students discussing the outcomes of an introduced species in an ecosystem, pictures of the species and the ecosystem over time, and videos showing the changes in the ecosystem, and the special characteristics of the different species could serve as visual evidence to help the students support their arguments (Evagorou et al. 2012 ). Therefore, an important implication for the teaching of science is the use of visual representations as evidence in the science curriculum as part of knowledge production. Even though studies in the area of science education have focused on the use of models and modeling as a way to support students in the learning of science (Dori et al. 2003 ; Lehrer and Schauble 2012 ; Mendonça and Justi 2013 ; Papaevripidou et al. 2007 ) or on the use of images (i.e., Korfiatis et al. 2003 ), with the term using visuals as evidence, we refer to the collection of all forms of visuals and the processes involved.

Another aspect that was identified through the case studies is that of the visual reasoning (an integral part of Faraday’s investigations). Both the verbalization and visualization were part of the process of generating new knowledge (Gooding 2006 ). Even today, most of the textbooks use the lines of force (or just field lines) as a geometrical representation of field, and the number of field lines is connected to the quantity of flux. Often, the textbooks use the same kind of visual imagery than in what is used by scientists. However, when using images, only certain aspects or features of the phenomena or data are captured or highlighted, and often in tacit ways. Especially in textbooks, the process of producing the image is not presented and instead only the product—image—is left. This could easily lead to an idea of images (i.e., photos, graphs, visual model) being just representations of knowledge and, in the worse case, misinterpreted representations of knowledge as the results of Pocovi and Finlay ( 2002 ) in case of electric field lines show. In order to avoid this, the teachers should be able to explain how the images are produced (what features of phenomena or data the images captures, on what ground the features are chosen to that image, and what features are omitted); in this way, the role of visualization in knowledge production can be made “visible” to students by engaging them in the process of visualization.

The implication of these norms for science teaching and learning is numerous. The classroom contexts can model the generation, sharing and evaluation of evidence, and experimental procedures carried out by students, thereby promoting not only some contemporary cultural norms in scientific practice but also enabling the learning of criteria, standards, and heuristics that scientists use in making decisions on scientific methods. As we have demonstrated with the three case studies, visual representations are part of the process of knowledge growth and communication in science, as demonstrated with two examples from the history of science and an example from current scientific practices. Additionally, visual information, especially with the use of technology is a part of students’ everyday lives. Therefore, we suggest making use of students’ knowledge and technological skills (i.e., how to produce their own videos showing their experimental method or how to identify or provide appropriate visual evidence for a given topic), in order to teach them the aspects of the nature of science that are often neglected both in the history of science and the design of curriculum. Specifically, what we suggest in this paper is that students should actively engage in visualization processes in order to appreciate the diverse nature of doing science and engage in authentic scientific practices.

However, as a word of caution, we need to distinguish the products and processes involved in visualization practices in science:

If one considers scientific representations and the ways in which they can foster or thwart our understanding, it is clear that a mere object approach, which would devote all attention to the representation as a free-standing product of scientific labor, is inadequate. What is needed is a process approach: each visual representation should be linked with its context of production (Pauwels 2006 , p.21).

The aforementioned suggests that the emphasis in visualization should shift from cognitive understanding—using the products of science to understand the content—to engaging in the processes of visualization. Therefore, an implication for the teaching of science includes designing curriculum materials and learning environments that create a social and epistemic context and invite students to engage in the practice of visualization as evidence, reasoning, experimental procedure, or a means of communication (as presented in the three case studies) and reflect on these practices (Ryu et al. 2015 ).

Finally, a question that arises from including visualization in science education, as well as from including scientific practices in science education is whether teachers themselves are prepared to include them as part of their teaching (Bybee 2014 ). Teacher preparation programs and teacher education have been critiqued, studied, and rethought since the time they emerged (Cochran-Smith 2004 ). Despite the years of history in teacher training and teacher education, the debate about initial teacher training and its content still pertains in our community and in policy circles (Cochran-Smith 2004 ; Conway et al. 2009 ). In the last decades, the debate has shifted from a behavioral view of learning and teaching to a learning problem—focusing on that way not only on teachers’ knowledge, skills, and beliefs but also on making the connection of the aforementioned with how and if pupils learn (Cochran-Smith 2004 ). The Science Education in Europe report recommended that “Good quality teachers, with up-to-date knowledge and skills, are the foundation of any system of formal science education” (Osborne and Dillon 2008 , p.9).

However, questions such as what should be the emphasis on pre-service and in-service science teacher training, especially with the new emphasis on scientific practices, still remain unanswered. As Bybee ( 2014 ) argues, starting from the new emphasis on scientific practices in the NGSS, we should consider teacher preparation programs “that would provide undergraduates opportunities to learn the science content and practices in contexts that would be aligned with their future work as teachers” (p.218). Therefore, engaging pre- and in-service teachers in visualization as a scientific practice should be one of the purposes of teacher preparation programs.

Achieve. (2013). The next generation science standards (pp. 1–3). Retrieved from http://www.nextgenscience.org/ .

Google Scholar  

Barber, J, Pearson, D, & Cervetti, G. (2006). Seeds of science/roots of reading . California: The Regents of the University of California.

Bungum, B. (2008). Images of physics: an explorative study of the changing character of visual images in Norwegian physics textbooks. NorDiNa, 4 (2), 132–141.

Bybee, RW. (2014). NGSS and the next generation of science teachers. Journal of Science Teacher Education, 25 (2), 211–221. doi: 10.1007/s10972-014-9381-4 .

Article   Google Scholar  

Chambers, D. (1983). Stereotypic images of the scientist: the draw-a-scientist test. Science Education, 67 (2), 255–265.

Cochran-Smith, M. (2004). The problem of teacher education. Journal of Teacher Education, 55 (4), 295–299. doi: 10.1177/0022487104268057 .

Conway, PF, Murphy, R, & Rath, A. (2009). Learning to teach and its implications for the continuum of teacher education: a nine-country cross-national study .

Crick, F. (1988). What a mad pursuit . USA: Basic Books.

Dimopoulos, K, Koulaidis, V, & Sklaveniti, S. (2003). Towards an analysis of visual images in school science textbooks and press articles about science and technology. Research in Science Education, 33 , 189–216.

Dori, YJ, Tal, RT, & Tsaushu, M. (2003). Teaching biotechnology through case studies—can we improve higher order thinking skills of nonscience majors? Science Education, 87 (6), 767–793. doi: 10.1002/sce.10081 .

Duschl, RA, & Bybee, RW. (2014). Planning and carrying out investigations: an entry to learning and to teacher professional development around NGSS science and engineering practices. International Journal of STEM Education, 1 (1), 12. doi: 10.1186/s40594-014-0012-6 .

Duschl, R., Schweingruber, H. A., & Shouse, A. (2008). Taking science to school . Washington DC: National Academies Press.

Erduran, S, & Jimenez-Aleixandre, MP (Eds.). (2008). Argumentation in science education: perspectives from classroom-based research . Dordrecht: Springer.

Eurydice. (2012). Developing key competencies at school in Europe: challenges and opportunities for policy – 2011/12 (pp. 1–72).

Evagorou, M, Jimenez-Aleixandre, MP, & Osborne, J. (2012). “Should we kill the grey squirrels?” A study exploring students’ justifications and decision-making. International Journal of Science Education, 34 (3), 401–428. doi: 10.1080/09500693.2011.619211 .

Faraday, M. (1852a). Experimental researches in electricity. – Twenty-eighth series. Philosophical Transactions of the Royal Society of London, 142 , 25–56.

Faraday, M. (1852b). Experimental researches in electricity. – Twenty-ninth series. Philosophical Transactions of the Royal Society of London, 142 , 137–159.

Gilbert, JK. (2010). The role of visual representations in the learning and teaching of science: an introduction (pp. 1–19).

Gilbert, J., Reiner, M. & Nakhleh, M. (2008). Visualization: theory and practice in science education . Dordrecht, The Netherlands: Springer.

Gooding, D. (2006). From phenomenology to field theory: Faraday’s visual reasoning. Perspectives on Science, 14 (1), 40–65.

Gooding, D, Pinch, T, & Schaffer, S (Eds.). (1993). The uses of experiment: studies in the natural sciences . Cambridge: Cambridge University Press.

Hogan, K, & Maglienti, M. (2001). Comparing the epistemological underpinnings of students’ and scientists’ reasoning about conclusions. Journal of Research in Science Teaching, 38 (6), 663–687.

Knorr Cetina, K. (1999). Epistemic cultures: how the sciences make knowledge . Cambridge: Harvard University Press.

Korfiatis, KJ, Stamou, AG, & Paraskevopoulos, S. (2003). Images of nature in Greek primary school textbooks. Science Education, 88 (1), 72–89. doi: 10.1002/sce.10133 .

Latour, B. (2011). Visualisation and cognition: drawing things together (pp. 1–32).

Latour, B, & Woolgar, S. (1979). Laboratory life: the construction of scientific facts . Princeton: Princeton University Press.

Lehrer, R, & Schauble, L. (2012). Seeding evolutionary thinking by engaging children in modeling its foundations. Science Education, 96 (4), 701–724. doi: 10.1002/sce.20475 .

Longino, H. E. (2002). The fate of knowledge . Princeton: Princeton University Press.

Lynch, M. (2006). The production of scientific images: vision and re-vision in the history, philosophy, and sociology of science. In L Pauwels (Ed.), Visual cultures of science: rethinking representational practices in knowledge building and science communication (pp. 26–40). Lebanon, NH: Darthmouth College Press.

Lynch, M. & S. Y. Edgerton Jr. (1988). ‘Aesthetic and digital image processing representational craft in contemporary astronomy’, in G. Fyfe & J. Law (eds), Picturing Power; Visual Depictions and Social Relations (London, Routledge): 184 – 220.

Mendonça, PCC, & Justi, R. (2013). An instrument for analyzing arguments produced in modeling-based chemistry lessons. Journal of Research in Science Teaching, 51 (2), 192–218. doi: 10.1002/tea.21133 .

National Research Council (2000). Inquiry and the national science education standards . Washington DC: National Academies Press.

National Research Council (2012). A framework for K-12 science education . Washington DC: National Academies Press.

Nersessian, NJ. (1984). Faraday to Einstein: constructing meaning in scientific theories . Dordrecht: Martinus Nijhoff Publishers.

Book   Google Scholar  

Nersessian, NJ. (1992). How do scientists think? Capturing the dynamics of conceptual change in science. In RN Giere (Ed.), Cognitive Models of Science (pp. 3–45). Minneapolis: University of Minnesota Press.

Nersessian, NJ. (2008). Creating scientific concepts . Cambridge: The MIT Press.

Osborne, J. (2014). Teaching scientific practices: meeting the challenge of change. Journal of Science Teacher Education, 25 (2), 177–196. doi: 10.1007/s10972-014-9384-1 .

Osborne, J. & Dillon, J. (2008). Science education in Europe: critical reflections . London: Nuffield Foundation.

Papaevripidou, M, Constantinou, CP, & Zacharia, ZC. (2007). Modeling complex marine ecosystems: an investigation of two teaching approaches with fifth graders. Journal of Computer Assisted Learning, 23 (2), 145–157. doi: 10.1111/j.1365-2729.2006.00217.x .

Pauwels, L. (2006). A theoretical framework for assessing visual representational practices in knowledge building and science communications. In L Pauwels (Ed.), Visual cultures of science: rethinking representational practices in knowledge building and science communication (pp. 1–25). Lebanon, NH: Darthmouth College Press.

Philips, L., Norris, S. & McNab, J. (2010). Visualization in mathematics, reading and science education . Dordrecht, The Netherlands: Springer.

Pocovi, MC, & Finlay, F. (2002). Lines of force: Faraday’s and students’ views. Science & Education, 11 , 459–474.

Richards, A. (2003). Argument and authority in the visual representations of science. Technical Communication Quarterly, 12 (2), 183–206. doi: 10.1207/s15427625tcq1202_3 .

Rothbart, D. (1997). Explaining the growth of scientific knowledge: metaphors, models and meaning . Lewiston, NY: Mellen Press.

Ruivenkamp, M, & Rip, A. (2010). Visualizing the invisible nanoscale study: visualization practices in nanotechnology community of practice. Science Studies, 23 (1), 3–36.

Ryu, S, Han, Y, & Paik, S-H. (2015). Understanding co-development of conceptual and epistemic understanding through modeling practices with mobile internet. Journal of Science Education and Technology, 24 (2-3), 330–355. doi: 10.1007/s10956-014-9545-1 .

Sarkar, S, & Pfeifer, J. (2006). The philosophy of science, chapter on experimentation (Vol. 1, A-M). New York: Taylor & Francis.

Schwartz, RS, Lederman, NG, & Abd-el-Khalick, F. (2012). A series of misrepresentations: a response to Allchin’s whole approach to assessing nature of science understandings. Science Education, 96 (4), 685–692. doi: 10.1002/sce.21013 .

Schwarz, CV, Reiser, BJ, Davis, EA, Kenyon, L, Achér, A, Fortus, D, et al. (2009). Developing a learning progression for scientific modeling: making scientific modeling accessible and meaningful for learners. Journal of Research in Science Teaching, 46 (6), 632–654. doi: 10.1002/tea.20311 .

Watson, J. (1968). The Double Helix: a personal account of the discovery of the structure of DNA . New York: Scribner.

Watson, J, & Berry, A. (2004). DNA: the secret of life . New York: Alfred A. Knopf.

Wickman, PO. (2004). The practical epistemologies of the classroom: a study of laboratory work. Science Education, 88 , 325–344.

Wu, HK, & Shah, P. (2004). Exploring visuospatial thinking in chemistry learning. Science Education, 88 (3), 465–492. doi: 10.1002/sce.10126 .

Download references

Acknowledgements

The authors would like to acknowledge all reviewers for their valuable comments that have helped us improve the manuscript.

Author information

Authors and affiliations.

University of Nicosia, 46, Makedonitissa Avenue, Egkomi, 1700, Nicosia, Cyprus

Maria Evagorou

University of Limerick, Limerick, Ireland

Sibel Erduran

University of Tampere, Tampere, Finland

Terhi Mäntylä

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Maria Evagorou .

Additional information

Competing interests.

The authors declare that they have no competing interests.

Authors’ contributions

ME carried out the introductory literature review, the analysis of the first case study, and drafted the manuscript. SE carried out the analysis of the third case study and contributed towards the “Conclusions” section of the manuscript. TM carried out the second case study. All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0 ), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Cite this article.

Evagorou, M., Erduran, S. & Mäntylä, T. The role of visual representations in scientific practices: from conceptual understanding and knowledge generation to ‘seeing’ how science works. IJ STEM Ed 2 , 11 (2015). https://doi.org/10.1186/s40594-015-0024-x

Download citation

Received : 29 September 2014

Accepted : 16 May 2015

Published : 19 July 2015

DOI : https://doi.org/10.1186/s40594-015-0024-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Visual representations
• Epistemic practices
• Science learning

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• HHS Author Manuscripts

Visual Perceptual Learning and Models

Barbara dosher.

1 Department of Cognitive Sciences, Institute for Mathematical Behavioral Sciences, and Center for Neurobiology and Behavior, University of California, Irvine, Irvine, California 92617-5100;

Zhong-Lin Lu

2 Department of Psychology, Center for Cognitive and Brain Science, and Center for Cognitive and Behavioral Brain Imaging, The Ohio State University, Columbus, Ohio, 43210;

Visual perceptual learning through practice or training can significantly improve performance on visual tasks. Originally seen as a manifestation of plasticity in the primary visual cortex, perceptual learning is more readily understood as improvements in the function of brain networks that integrate processes including sensory representations, decision, attention, and reward and balance plasticity with system stability. This review considers the primary phenomena of perceptual learning, and theories of perceptual learning and its effect on signal and noise in visual processing and decision. Models, especially computational models, play a key role in behavioral and physiological evaluation of the mechanisms of perceptual learning, and for understanding, predicting, and optimizing human perceptual processes, learning, and performance. Performance improvements resulting from reweighting or readout of sensory inputs to decision provide a strong theoretical framework for interpreting perceptual learning and transfer that may prove useful in optimizing learning in real world applications.

INTRODUCTION

Visual perceptual learning is the improvement in visual task performance with practice or training ( Sagi 2011 ). It reflects learning and plasticity in the visual system and a network of other brain substrates of behavior. Substantial improvements in visual task performance can occur in adults whose cortical organization and function are developmentally mature, where the architecture of the visual system—absent major injury and reorganization—is relatively stable ( Wandell & Smirnakis 2009 ). Although research has heavily focused on the development of perceptual expertise in adults ( Lu et al. 2011 ), visual perceptual learning contributes to functional improvements during development ( Atkinson et al. 1977 , Gibson 1969 ), can improve visual performance during aging ( DeLoss et al. 2015 ), and plays an important function in visual rehabilitation ( Lu et al. 2016 ). Finally, the state of the perceptual system depends on experience, and cannot be fully understood without understanding its plasticity.

Human perceptual processes are a necessary gateway to experience and integral to planning and executing behavior. The visual system is a complex processing engine, with many areas and modules that coordinate a complex flow of perceptual information ( Van Essen et al. 1992 ). The visual system, like many sensory systems, seems to have evolved to support processing of the important stimulus cues in the environment ( Geisler 2008 ). However, even after millions of years of evolution, it continues to improve through development and through experience. Human visual functions improve considerably during a developmental period starting in infancy. For normal adults, performance in many visual tasks may be far from optimal and can be improved with the right kind of training or practice ( Lu et al. 2011 ).

Observations of perceptual learning date back to early in the study of perception. William James describes improvements in performance with practice in his chapter on discrimination and comparison, citing work by Volkmann and Fechner on improvements in two-point discrimination on skin from the mid-1800s ( James 1890 ). Stratton studied the role of experience with prism distortions of visual input in the late 1800s ( Stratton 1897 ). Perceptual learning and its role in visual development was studied by E. Gibson starting in the 1950s ( Gibson 1969 ). There were many observations of expertise in naturalistic tasks such as wool sorting, wine tasting etc. More controlled laboratory studies of perceptual learning refocused in the late 1980s, and many features of perceptual learning, its specificity to stimuli and tasks, and its mechanisms have been discovered during the last nearly three decades.

OBSERVING PERCEPTUAL LEARNING

Learning in perceptual tasks.

Perceptual learning has been documented in virtually all tasks and sensory modalities—although performance in a few tasks is relatively unchanged by practice, perhaps because they are so common in everyday life. A number of reviews have considered this now-extensive body of work, including ( Adini et al. 2004 , Ahissar & Hochstein 2004 , Fahle & Poggio 2002 , Fine & Jacobs 2002 , Lu et al. 2011 , Vogels 2010 , Watanabe & Sasaki 2015 ).

Perceptual learning occurs for different kinds of tasks at different levels of visual analysis ( figure 1 ). It improves detection or discrimination for single features such as orientation ( Dosher & Lu 1998 , Dosher & Lu 1999 , Schoups et al. 1995 , Vogels & Orban 1985 ), spatial frequency ( Bennett & Westheimer 1991 , Fiorentini & Berardi 1981 ), phase ( Dosher et al. 2010 , Fiorentini & Berardi 1980 ), contrast ( Adini et al. 2004 , Dorais & Sagi 1997 , Sowden et al. 2002 ), color ( Casey & Sowden 2012 , Özgen & Davies 2002 , Thurston & Dobkins 2007 ), acuity ( Bennett & Westheimer 1991 , Westheimer 2001 ), and hyper-acuity ( Crist et al. 1997 , McKee & Westheimer 1978 , Poggio et al. 1992 ). It improves pattern discrimination in tasks involving compound stimuli ( Fiorentini & Berardi 1980 , Fiorentini & Berardi 1981 ), textures ( Ahissar & Hochstein 1993 , Karni & Sagi 1991 ), depth ( Fendick & Westheimer 1983 , Ramachandran & Braddick 1973 ), and motion ( Ball & Sekuler 1982 , Lu et al. 2006 , Lu et al. 2005 , Matthews & Welch 1997 , Watanabe et al. 2002 ). And it improves identification of objects and natural scenes such as faces and entities ( Gauthier et al. 1998 , Gold et al. 1999 ), shapes and objects ( Kourtzi et al. 2005 , Nazir & O’Regan 1990 ), and biological motion ( Grossman et al. 2004 , Jastorff et al. 2006 ). In many of these cases, the literature is extensive; we have often provided earlier references.

Visual perceptual learning improves task performance measured in different ways: improving percent correct discrimination (a), contrast thresholds (b), or feature difference threshold differences (c) (hypothetical data and generating exponential learning curves). (d) Difference thresholds (arcsec) from a line offset hyperacuity task for vertical, horizontal, and oblique layouts (data from McKee & Westheimer, 1981), fitted exponential learning curves added. Reproduced with permission from B. Dosher and Z.L. Lu.

Perceptual learning can powerfully affect performance, improving accuracy from near chance to more than 90% correct in many two-choice tasks ( Ball & Sekuler 1982 , Fiorentini & Berardi 1980 , Poggio et al. 1992 ). For this reason, any characterization of visual functions based on testing will depend upon the level of expertise. Finally, training effects are sufficiently large that they can contribute to performance in practical domains. On the other hand, there are cases where perceptual learning does not alter performance ( Sagi 2011 ); failure to improve often involves already practiced judgments such as training at fovea with prototypical features in easy viewing conditions ( Lu & Dosher 2004 ), or when potentially conflicting judgments are intermixed or roved ( Herzog et al. 2012 , Yu et al. 2004 ).

Visual perceptual learning has been used to improve visual performance in visual conditions, including amblyopia ( Levi & Li 2009 , Li et al. 2013 , Xi et al. 2014 ), myopia ( Durrie & McMinn 2007 , Yan et al. 2015 ), aging ( DeLoss et al. 2015 ), presbyopia ( Polat et al. 2012 ), low vision ( Liu et al. 2007 , Yu et al. 2010 ), cortical blindness ( Huxlin et al. 2009 , Kasten & Sabel 1995 , Nelles et al. 2001 ), and rehabilitation after surgical interventions ( Huber et al. 2015 , Kalia et al. 2014 ). It has also been applied in education ( Kellman & Massey 2013 , Merzenich et al. 1996 , Strong et al. 2011 ) and training of visual expertise ( Deveau et al. 2014 , Gauthier et al. 1998 , Sowden et al. 2002 ).

Perceptual learning often involves thousands of trials of practice over days or weeks ( Dosher & Lu 1998 ), although sometimes a few exposures of easy stimuli accelerate learning ( Liu et al. 2012 , Rubin et al. 1997 ), and in some domains initial learning occurs within a few dozens trials ( Ramachandran & Braddick 1973 ). Some studies find that REM sleep was critical for perceptual learning in texture discrimination ( Karni & Sagi 1991 , Mednick et al. 2003 ). Training effects can persist for periods up to years ( Karni & Sagi 1993 ). In sum, perceptual learning is a major phenomenon of adult plasticity with important theoretical and practical implications.

Specificity and Transfer

The specificity of visual perceptual learning ( Karni & Sagi 1991 ) is one of its hallmark characteristics ( figure 2 ). Specificity—in which learned improvements are lost when the stimuli or task is altered—has been reported for orientation, spatial frequency, motion direction, pattern, and even (significantly) location in the visual field ( Ball & Sekuler 1982 , Dosher & Lu 1999 , Fahle & Edelman 1993 , Fiorentini & Berardi 1980 , Karni & Sagi 1991 , Schoups et al. 1995 ). For example, training to detect a small patch of differently oriented lines in the lower right quadrant does not fully transfer to texture processing in other quadrants ( Karni & Sagi 1991 ). The specificity of trained improvements to a portion of the visual field were considered especially salient, leading some to infer that perceptual learning reflects plasticity in early visual cortex, V1, which has small retinotopic receptive fields.

Learning in a training task can express varying benefits for a transfer task. (a) Hypothetical learning curves showing full specificity, partial specificity, and full transfer; (b) corresponding patterns in bar graphs that often summarize these results; and (c) data from an experiment training texture detection in different quadrants of the visual field that shows significant specificity to retinal location and partial transfer (data from Karni & Sagi, 1999, figure 1 ). (a)-(c) reproduced with permission from B. Dosher and Z.L. Lu. (d) with permission from XXXX

While the literature emphasized the specificity of visual perceptual learning, often it is a graded phenomenon with some specificity and some transfer of trained improvements to other stimuli, tasks, or locations ( Dosher & Lu 2009 ). Generalization is more important in practical applications, such as the development of expertise or remediation, where benefits beyond the training conditions are valuable. We now know that the relative specificity versus transfer of training can depend on several factors, including the processing level of the trained task ( Fine & Jacobs 2002 ), the task difficulty ( Ahissar & Hochstein 1993 , Liu 1999 ), precision of the transfer task ( Jeter et al. 2009 ), the extent of training ( Jeter et al. 2010 ), the state of adaptation induced by training ( Censor et al. 2006 ), and the exact training and transfer procedure ( Hung & Seitz 2014 , Xiao et al. 2008 ). More demanding tasks tend to experience more specificity.

Feedback and Reward

Feedback occurs when the learning environment provides information about the quality or value of behavior. The form of feedback can differ, and these differences can affect the magnitude and speed of perceptual learning. Training may use full feedback about the desired response on each trial, occur on a subset of trial, be misleading, or feedback may be unavailable ( Dosher & Lu 2009 ). Almost all experimental investigations of perceptual learning involve two-alternative choices (e.g., left/right, same different) and trial-by-trial accuracy feedback. However, learning can occur with only block feedback ( Herzog & Fahle 1997 , Shibata et al. 2009 ) or without any feedback if performance before training is high enough ( Fahle & Edelman 1993 , Herzog & Fahle 1997 , Liu et al. 2010 , McKee & Westheimer 1978 , Petrov et al. 2006 ). If performance accuracy before training is low, observers may not learn without feedback ( Liu et al. 2010 , Rubin et al. 1997 ), or the rate of learning increases with feedback ( Crist et al. 1997 ). Reverse or random feedback can prevent learning ( Aberg & Herzog 2012 , Herzog & Fahle 1997 , Herzog & Fahle 1999 ), while exaggerated (positive) block feedback can change its rate ( Shibata et al. 2009 ). In some unusual demonstrations, feedback in the absence of a stimulus can alter performance ( Choi & Watanabe 2012 , Shibata et al. 2012 ). In short, learning can occur in the absence of feedback in certain situations, yet feedback is important when the task is difficult and initial performance is poor. Trial-by-trial feedback is more effective than block feedback, and inaccurate feedback can disrupt learning.

Physical rewards can result in perceptual learning in the absence of verbal instructions ( Seitz et al. 2009 ). Yet explicit rewards, and especially the systematic effects of the magnitude of rewards on perceptual learning, are just beginning to be investigated ( Zhang et al. 2016 ). A model with the right learning rule has the potential to systematize and predict these varied effects of feedback and reward.

Selection by Task and Attention

Perceptual learning balances stability with plasticity in part through selectivity. Real world sensory stimulation is rich, containing many potential cues for guiding behavior. Yet generally only task relevant stimuli, features, or locations participate in learning ( Ahissar & Hochstein 1993 , Fahle & Morgan 1996 , Shiu & Pashler 1992 ). However, task-irrelevant learning sometimes occurs for extraneous stimuli appearing in temporal proximity to a training stimulus if they are subliminal ( Gutnisky et al. 2009 , Watanabe et al. 2002 ).

In addition to task-relevance, attention may select what is learned. Although there are claims that attention to a stimulus is required for learning ( Ahissar & Hochstein 2004 , Dolan et al. 1997 , Gilbert et al. 2001 ), only a few studies have explicitly manipulated attention and evaluated learning in attended and unattended conditions ( Mukai et al. 2011 , Xu et al. 2012 ). Conversely, the functional importance of attention in determining task performance can be reduced through extensive perceptual training ( Dosher et al. 2010 ). Selection by task or by attention selectively reweights only those stimulus representations or attended features that are relevant to the trained task.

MODELS OF PERCEPTUAL LEARNING

Even the simplest task of detecting or discriminating a perceptual stimulus involves a network of processes or brain regions supporting sensory processing, decision, action selection, top-down task relevance, attention, and processing of rewards or feedback. Each of these sensory and cognitive processes may be engaged during natural behavioral episodes and in each experimental trial. Learning alters processing, generally to improve performance. Although early explanations focused on plasticity in the sensory cortices, perceptual learning must engage multiple processes, levels, and brain areas ( Kourtzi 2010 , Kourtzi et al. 2005 ).

Maintaining stability in the face of plasticity, the plasticity-stability dilemma, constrains how the system learns perceptual tasks ( Dosher & Lu 2009 ). Plasticity of visual system is normally considered an advantage associated with performance improvements, yet must be balanced with maintaining stability in standard visual functions. Too much plasticity could result in catastrophic forgetting of one task or set of stimuli by training on another ( French 1999 , Grossberg 1987 ), and an inability to optimize several tasks simultaneously.

Several conceptual frameworks for perceptual learning have been proposed, including the primary visual cortical plasticity theory ( Karni & Sagi 1991 ), the reverse hierarchy theory ( Ahissar & Hochstein 1993 , Ahissar & Hochstein 2004 ), the reweighting model of perceptual learning ( Dosher & Lu 1998 ), and the dual plasticity model ( Watanabe & Sasaki 2015 ). Representation enhancement theories of perceptual learning identify changes in early visual areas such as V1 as the substrate. Selective reweighting theories of perceptual learning promote stability by improving readout from sensory representations that remain largely unchanged ( Dosher & Lu 1998 , Mollon & Danilova 1996 ). Although task-dependent reweighting alters the inputs (representations) at later stages of processing, stable early sensory representations (e.g. V1) could contribute to maintaining performance in previously learned tasks. And, if sensory representations are altered, further reweighting would be necessary to optimize readout or decoding of the (new) neurosensory evidence ( Dosher & Lu 2009 ). Recent theoretical overviews cite both forms of plasticity, along with attention and reward ( Watanabe & Sasaki 2015 ).

Models can play a powerful role in testing these theories of learning and plasticity and account for complex patterns in the empirical literature. A complete model includes modules for sensory representation, decision, and learning, and possibly attention, reward, and feedback. Another important component is noise, or variability, in the internal responses of the system. The accuracy of performance depends critically on extracting signal from the internal noise in the system responses. A computational model predicts behavioral performance by taking the stimuli as inputs and specifying the computations carried out in each module. It can be as abstract as a set of simple computations or it can mimic the architecture of brain areas and the behavior of neural populations.

The Learning Rule

Learning rules, core to neural network theories of learning, have also been central to models of perceptual learning. In artificial neural networks, the learning rules change the weights from input units representing the stimuli to decision or response units. Network learning theories distinguish purely supervised, purely unsupervised, and hybrid or semi-supervised learning ( Barlow 1989 , Jordan & Rumelhart 1992 , Reed & Marks 1998 ). Supervised learning requires a teaching signal that specifies full information about the desired responses, while unsupervised learning does not. In semi-supervised learning, full information may be provided on a subset of cases or about accuracy but not the nature of the error. In hybrid learning, the rule may be modified by feedback, attention, or reward.

Empirical findings on feedback can inform learning rules in models. That perceptual learning can occur in the absence of feedback, yet can sometimes benefit from or require feedback implies a hybrid of supervised and unsupervised learning ( Dosher & Lu 2009 , Herzog & Fahle 1998 ). One plausible rule is augmented Hebbian reweighting ( Petrov et al. 2005 , Petrov et al. 2006 ), which combined unsupervised Hebbian learning with guidance from feedback and bias. Other models, motivated by physiological concepts of reward and reward prediction error, use reweighting through reinforcement learning, a form of weak supervised learning ( Law & Gold 2009 ).

Computational Models of Perceptual Learning

Essentially all current computational models of perceptual learning are reweighting models. They specify a task domain, including stimuli and desired responses, the network architecture, the decision, and how learning rules change the weights between representations and decision. Computational models have been developed for hyperacuity ( Herzog & Fahle 1998 , Huang et al. 2012 , Poggio et al. 1992 , Sotiropoulos et al. 2011 , Weiss et al. 1993 , Zhaoping et al. 2003 ), orientation discrimination ( Petrov et al. 2005 , Petrov et al. 2006 , Teich & Qian 2003 ), tilt ( Jacobs 2009 ), motion direction discrimination ( Law & Gold 2009 , Vaina et al. 1995 ), and contrast discrimination ( Adini et al. 2004 ) tasks.

For example, the early hyper basis function ( HBF ) network model learned a hyperacuity task in which observers judged lines as offset either top-line left or right ( Poggio et al. 1992 ). Its feed forward architecture included a stimulus input layer, a representation layer for localization, and a single-unit output layer (left/right). Modeling studies ( Weiss et al. 1993 ) identified self-supervised learning rules and added internal noise to match predictions to behavioral data. A related model learned global motion direction judgments from input motion vectors for individual dot motions using integrated motion direction templates in an intermediate layer to generate left/right decisions ( Vaina et al. 1995 ). In all these computational models, learning alters the decoding of the activity in a stable stimulus input layer, possibly through an intermediate representation layer, to the output decision layer. They use unsupervised, or self-supervised, learning rules to account for learning without feedback or reward, and may incorporate supervision to account for feedback ( Herzog & Fahle 1999 ). The input representations are highly simplified. And in general they were tested against simple empirical data such as a learning curve or pattern of specificity ( Weiss et al. 1993 ).

The augmented Hebbian reweighting model ( AHRM ) ( Petrov et al. 2005 , Petrov et al. 2006 ) is a full model of perceptual learning. It uses a sensory representation module that mimics the spatial-frequency and orientation tuned responses of early visual cortices like V1, including nonlinearities ( Carandini et al. 1997 ) and internal noise, compatible with observer models of signal and noise in perception (see SIGNAL AND NOISE , below). An output or decision unit weights evidence from activations in the sensory representation to make a decision. Another input corrects for bias in recent responses and a teaching signal augments unsupervised Hebbian learning when trial-by-trial feedback is available (see figure 3 ). Learning occurs through reweighting, changing the weights on stimulus evidence to make a decision. This yields an improved weight structure for the task after training or practice. Parameters specifying the sensory representations, such as orientation or spatial frequency bandwidths, are set a priori from physiology; nonlinearities are estimated once and then held constant. Internal noise and learning rate parameters are varied to fit model simulated data to the behavioral data. The model takes stimulus images as inputs to a decision, and learns on each trial, reprising experiments exactly.

A reweighting framework for visual perceptual learning takes images as input, processes them in task-relevant representations, makes decisions based on the weighted sum of the relevant normalized noisy representation activations, and learns by changing the weights through unsupervised and supervised learning algorithms. As in the observer models, performance depends on normalization and internal noise. Reproduced with permission from B. Dosher and Z.L. Lu.

The AHRM accounts for many perceptual learning phenomena, including improvements in contrast threshold in multiple levels of external noise (visual ‘snow’) with training ( Lu et al. 2010 ), more efficient pretraining with low than high external noise conditions ( Lu et al. 2010 ), the importance of feedback in learning tasks with low but not high initial performance ( Liu et al. 2010 ), how to improve learning by including high-accuracy training trails ( Liu et al. 2012 , Petrov et al. 2005 ), and how false feedback induces response biases ( Liu et al. 2014 ).

Other similar models use representation modules coded for orientation ( Sotiropoulos et al. 2011 ), location ( Huang et al. 2012 ) or motion ( Lu et al. 2010 ). The original AHRM spatial-frequency/orientation module combined with an adaptive precision pooling decision module was used for perceptual learning in tilt judgments ( Jacobs 2009 ). A model developed for monkey behavior learns by reweighting evidence from a motion representation using a reinforcement-learning rule ( Law & Gold 2009 ).

Modeling Specificity and Transfer

Reweighting models have also been used to explain specificity and transfer in perceptual learning. They predict transfer if the learned weight structure of the training task and the (semi-) optimal weight structure for the transfer task are compatible, and predict specificity otherwise ( Dosher & Lu 2009 ). A few studies have modeled transfer within a single retinal location: The HBF model predicted performance for stimuli differing in line spacing or length ( Weiss et al. 1993 ). A related orientation basis function model simulated specificity and transfer for different hyperacuity tasks ( Sotiropoulos et al. 2011 ). Specificity and transfer to other stimuli at the same location was predicted by the compatibility of weight structures for the AHRM ( Petrov et al. 2005 ).

Specificity to retinal location implies dependence on local retinotopic neural representations and raises the question—how can learning be transferred to a separate set of local neural representations? The integrated reweighting theory (IRT) uses a hierarchical multi-location architecture consisting of location-specific representations and a location-invariant representation ( Dosher et al. 2013 ) to predict transfer over retinal locations. It ( figure 4 ) explains why the same stimuli and task in a new location shows partial transfer, while the same task in the same location with new stimuli shows nearly full specificity. Learned weights for the location-invariant representations are valid when the same stimuli and task occur in other retinal locations. Training in the same location is specific if the stimuli or task are switched because the optimal weight structures compete. This framework also accounts for why different tasks can be sometimes be trained in different locations ( Liu et al. 2014 ) and how related tasks in different locations interact. A modified version of the IRT also predicts transfer of training to new locations in a double-training experiment that practiced two different tasks in two locations ( Talluri et al. 2015 ), a transfer phenomenon attributed by some to higher order inference ( Xiao et al. 2008 ).

A framework for predicting transfers across and interactions between learning in multiple retinal locations based on a hierarchical architecture of sensory representations including relatively location invariant representations. Perceptual learning in one retinal location trains weights between both location-specific and location-invariant representations and decision. Transfer reflects compatibility of optimized weight structures, while specificity reflects independence or incompatibility of optimized weight structures for the training and transfer tasks. Reproduced with permission from B. Dosher and Z.L. Lu.

Future Model Development

Current computational models are necessarily simplified, yet have the potential to be elaborated in many ways. Different stimuli and tasks may require new stimulus representation modules, new forms of classification or decision, and/or more complex system architectures. Top-down attention could amplify activity in sensory representations; reward may require modified learning rules; and feedback between or reweighting within representation modules may be required to account for new phenomena. The impact of correlated and uncorrelated noise in population networks ( Bejjanki et al. 2011 ) could be modeled. As the physiological substrates for perception and visual perceptual learning in different tasks are discovered, computational models may increasingly mimic these processes. Still, existing reweighting models, despite their spare and simplified form, have provided a strong basis for systematizing and understanding many phenomena in visual perceptual learning and visual expertise

SIGNAL AND NOISE

Performance improvements and improved signal and noise.

The human brain devotes specialized brain regions to the processing of visual stimuli and visual decisions ( Ungerleider et al. 1998 ). Analysis by visual cortex is complex, likened to powerful deep learning network models ( Yamins et al. 2014 ). Visual functions and behavior are not, however, perfect. Accurate performance is limited by the signal to noise ratio in the evidence leading to behavior ( Lu & Dosher 2008 ). Perceptual learning, then, improves the signal to noise ratio in perceptual processing ( Dosher & Lu 1998 ).

Each act of visual detection, discrimination, or identification is limited by the ability to extract the appropriate signal and by intrinsic noise in the system and extrinsic noise in the stimuli. Internal or intrinsic noise is the stochastic variability in the neural responses to a stimulus arising at every stage of processing, while extrinsic noise is variability in the stimuli. The signal to noise ratio determines the accuracy and response time of an identification or choice behavior. In neural encoding, a driving stimulus leads to a noisy pattern of firing across populations of neurons that must be decoded. Limiting noise should be an explicit factor in models, whether they are signal detection models of behavior ( Green & Swets 1966 ), computational models of perceptual learning ( Dosher et al. 2013 ), or descriptive or computational models of neural responses ( Goris et al. 2014 ). When perceptual learning improves the signal to noise ratio, this could occur through improvements in the extraction of the signal or reduction of noise, or both. Behavioral and neural methods can be used to analyze changes in signal and noise, and can help to identify the mechanisms of visual perceptual learning and can guide the development of models of perceptual learning. These mechanisms are considered next.

Mechanisms: Psychophysics

Discovering how perceptual learning changes the signal and noise processing in the perceptual system from behavioral evidence requires a model of the observer and systematic psychophysical experiments. Observer models inspired by properties of the visual system systematize the behavior of observers in different testing circumstances by characterizing signal and noise processing in perception. These models initially characterized the observer as a single-channel linear system with an additive noise source (linear amplifier model) ( Pelli 1981 ); they have since been successively elaborated to account for data and now include multiple noise sources, nonlinearity and gain control (perceptual template model ( Lu & Dosher 2008 ), or incorporate multiple sensory channels each limited by noise ( Dosher & Lu 1998 , Hou et al. 2014 ). These observer models of the sensory representations and decision can be used to characterize changes due to perceptual learning, and could be further elaborated to include learning rules and other factors.

External noise paradigms can be used to specify observer models, sometimes augmented with double-pass paradigms ( Lu & Dosher 1998 , Saarinen & Levi 1995 ). To estimate nonlinearities and other parameters that predict the observer’s performance in all signal and external noise conditions, detection or discrimination thresholds are measured at three or more performance levels, e.g., 65%, 75% and 85% correct in several conditions with different levels of external noise added to the signal stimulus ( Lu & Dosher 1999 ). Double pass paradigms, repeating the identical stimuli, including the exact target and external noise sample, can add constraints on observer models and parameters, especially in the ratio of internal to external noise ( Burgess & Colborne 1988 ). These experiments are also used to measure how perceptual learning changes signal and noise processes to improve behavioral performance by specifying which components of the observer model have been changed during learning.

The perceptual template model (PTM) has been the most widely used observer model in perceptual learning ( figure 5 ). The model includes a perceptual template (channel) that extracts evidence, nonlinearity in transduction, internal noises, and a decision stage to predict behavior. Three mechanisms can improve performance: improved filtering or external noise exclusion by tuning the template; enhancement of the target stimulus or equivalently reduction in internal additive noise; or change in nonlinearity/reduction in internal multiplicative noise ( Dosher & Lu 1999 ). Each learning mechanisms has a signature pattern of changes in empirical thresholds at different levels of external noise and accuracy (see figure 3 ). Mechanisms are inferred by comparing threshold versus external noise contrast (TvC) functions early and late in learning and estimating how learning changes the model parameters for that task, i.e., template quality, amount of internal noise. The model and estimated parameters together can then predict the behavior for many different stimuli, and with simple elaborations, for tasks of different precision ( Dosher & Lu 1999 , Lu & Dosher 2008 ).

Perceptual template model (PTM) of the observer and signature changes in contrast threshold versus external noise contrast (TvC) functions for three mechanisms of perceptual learning. The observer model includes (from left to right) a template tuned to the target stimulus, point nonlinearity, multiplicative noise, additive noise, and a decision template. Three mechanisms of perceptual learning correspond to stimulus enhancement (or internal additive noise reduction), external noise exclusion (filtering), and changes in multiplicative internal noise or nonlinearity)—or mixtures of these. Modified from Dosher & Lu (1999) , figure 3 ). Reproduced with permission from B. Dosher and Z.L. Lu.

Perceptual learning has now been studied using TvC curves in several tasks. In most cases, perceptual learning simultaneously improved external noise exclusion—by improving the perceptual template—and reduced internal additive noise ( Dosher & Lu 1999 , Gold et al. 1999 ). The magnitudes of the two improvements can be decoupled. Sometimes, pure patterns of improved template ( Lu & Dosher 2004 ) or internal additive noise reduction ( Dosher & Lu 2006 ) occur. Changes in nonlinearity and/or internal multiplicative noise have not been observed. The different mix of mechanisms of perceptual learning revealed by TvC experiments is consistent with reweighting models of learning (see below) ( Bejjanki et al. 2011 , Dosher & Lu 1998 , Lu et al. 2010 ).

Mechanisms: Physiology

Physiological substrates and mechanisms of perceptual learning are investigated by measuring the amplitude, tuning curves, absolute variability, variability relative to the mean response (fano-factor), and topography of neural response in visual areas (i.e., V1, V4, MT, LIP, etc.) before and after perceptual training—or by fMRI brain imaging. Neural responses can be measured either while actively performing the task, which engages task-induced strategy and attention and top-down modification of neural responses, or under various passive fixation controls or anesthesia (in monkey), which reflect more persistent plasticity of neural responses ( Schoups et al. 2001 ). Population neural response models can estimate the magnitude of behavioral improvement that could be accounted for by changes in neural responses in a given brain area ( Ghose et al. 2002 , Raiguel et al. 2006 , Schoups et al. 2001 , Yang & Maunsell 2004 ).

The substrates of perceptual learning are more complex than the early claims of plasticity in V1 ( Karni & Sagi 1991 ). Under passive recording, subtle changes in V1 tuning curves can occur in neurons tuned near a trained orientation ( Schoups et al. 2001 ), while others ( Ghose et al. 2002 ) find essentially no change in V1 or V2 neurons. Even when changes in the slope of the tuning functions in V1 or V2 occur, they are estimated to account for perhaps only 1/10 th of the observed behavioral threshold reduction. Perceptual learning increased the response and narrowed the tuning in a small subset of V4 neurons tuned near the trained orientations under passive viewing ( Raiguel et al. 2006 , Yang & Maunsell 2004 ), but these changes still fell an order of magnitude short of accounting for behavioral improvements. In contrast, changes in V4 neuron responses measured while actively performing the task, decoded by separate optimal Bayes classifiers before and after training, came closer to explaining behavior ( Adab & Vogels 2011 ). And, improved V1 responses during active contour detection following training reflect feedback from V4 and higher areas later in the response interval ( Gilbert & Li 2012 ). Correspondingly, learning altered responses in LIP, but not in earlier MT during a dot-motion direction task ( Law & Gold 2008 ). In other tasks, higher-order areas receiving visual inputs are part of the learning circuits observed during object identification, including IT ( Kobatake et al. 1998 , Logothetis et al. 1995 ), lateral PFC ( Rainer & Miller 2000 ), and decision circuitry ( Kourtzi 2010 ). In addition, perceptual learning can change the brain areas used in a task: Training fine depth judgments inoculated coarse depth judgments from MT deactivation by injection ( Chowdhury & DeAngelis 2008 ). Because the brain is an interconnected network, changes observed in one cortical area need not be the only or even the primary locus of learned plasticity ( Wandell & Smirnakis 2009 ).

Finally, training may affect other neural properties, such as reducing the correlations between neurons to yield more independence across neural responses. Such a reduction was found between pairs of dorsal MST neurons in a fine motion task ( Gu et al. 2011 ). One computational study ( Bejjanki et al. 2011 ) confirmed that reweighting neuronal responses from one brain area to the next (e.g., LGN to V1) could account for the TvC improvements observed in Dosher and Lu (1998) . Reweighting reduced correlations between neurons at the later area, while only slightly changing the amplitude or tuning of individual neurons. This study suggests that the amplitude and tuning of individual neurons is less informative than the pattern and correlation of neuronal population responses measured in simultaneous recordings. It also found that TvC functions estimated from neuronal responses could be a robust measure of the relevant population properties.

Broader networks engaged during perceptual learning have been evaluated using fMRI in humans ( Furmanski et al. 2004 , Jehee et al. 2012 , Kahnt et al. 2011 , Kourtzi 2010 , Kuai et al. 2013 , Schwartz et al. 2002 , Shibata et al. 2012 , Zhang & Kourtzi 2010 ). Training-related changes in primary sensory areas sometimes occur; but there are almost always also changes in many other areas. Multi-voxel pattern classifiers tuned to decode the voxel activity before and after perceptual learning is one method for connecting brain activity to behavior. While a number of design factors and the role of top-down attention and decision complicate interpretation in the current literature, imaging could ultimately identify the brain networks, from sensory areas to decision, attention, and expectation, involved in learning and performing different perceptual tasks.

OPTIMIZATION

Investigations of perceptual learning have contributed to scientific theories of the perceptual system and its plasticity. Another goal is to maximize the magnitude of perceptual learning and transfer through development of optimized training protocols. A training protocol includes selection of the training stimuli, the task(s) used to train, the number and sequence of practice trials, the use of feedback, or reward, etc. Identifying the best training methods is a problem that can be formulated in terms of mathematical optimization ( Lu et al. 2016 ). First, an objective function must specify the goals and grade the outcome; this might include several factors such as the magnitude and efficiency of learning and the desired transfer characteristics, and their relative importance. Next, a generative model predicts the outcome(s) for all protocols within the domain of possible training manipulations. A search algorithm identifies the one or several candidate protocols to maximize the objective function. Optimizing training by intuition and experimentation is prohibitive. A robust computational model can be uniquely valuable in optimization—replacing expensive and time consuming empirical exploration with simulation. Empirical testing may still be required to estimate parameters of the generative model and potentially to improve the model itself. And experimental tests must ultimately validate the best training protocol.

SUMMARY, CHALLENGES, AND FUTURE DIRECTIONS

Since the earliest research in psychology and neuroscience, perceptual learning has been recognized for its important role in perception and performance. Since its resurgence in the late 1980s when it was recognized as a key demonstration of visual plasticity in adults, it has become an important topic of study in the visual sciences. We extract four principles of perceptual learning from this extensive literature:

• Perceptual learning occurs within a complex set of brain networks and may occur at multiple levels.
• Learned plasticity must be balanced by stability in order to optimize the behavior over many tasks and environmental contexts.
• Reweighting evidence from one level to another of representation or within levels is a major and perhaps the major form of perceptual learning.
• Perceptual learning improves the signal to noise ratio limiting human performance either by enhancing signal or reducing noise.

There are many exciting areas for future research. One is to characterize the interaction between perceptual learning and lifespan development. Perceptual learning may have powerful influences on visual function during early childhood ( Atkinson et al. 1977 ), and could aid in maintaining function during normal aging ( DeLoss et al. 2015 ). Both domains are only beginning to be mined, perhaps due to ethical considerations and the potential for unintended consequences, especially in early development.

The development of more sophisticated methods of measuring learning and transfer is another area for future research. The act of measuring visual functions can change the state of the system, analogous to the observer effect in physics. For example, transfer/specificity measures often compare post-training performance to pre-training baselines, yet measuring pre-training baselines causes learning that must be estimated. The magnitude and speed of learned improvements depend upon the mixture of stimuli used while measuring snapshots of performance. Developing better performance measures and understanding the pros and cons of different procedures and how to model them would benefit future applications.

Computational models have already played a powerful role in systematizing a broad range of empirical phenomena and generated new and testable predictions that can guide future basic and translational research in perceptual learning. The convergence between models at the level of learning architectures and learning networks with brain architecture and with cellular physiology could extend our understanding of perception and performance and underlying brain substrates. Multi-modal models that relate computations to behavior on one hand, and measures of brain activity on the other may reveal new principles of learning and plasticity and an improved ability to interpret brain function. Models can also play a key role in optimizing training protocols.

Research in perceptual learning has an increasingly recognized potential for important translation into systems for training expertise or improving remediation and recovery. One of the key challenges is how to improve generalizability of training that otherwise may show undesirable specificity. One recent direction goes beyond simplified laboratory tasks to more complex and realistic tasks and training systems. There have been widespread claims that video games and other digital training apps can enhance learning and generalization, and improve broad visual functions ( Green & Bavelier 2015 ). The roles of reward, pacing, and task variation in these games and apps all deserve investigation that could improve not just translation but reveal new insights in basic science of vision. Finally, as a practical matter, translation to the clinical, enrichment, or entertainment marketplaces requires a more sophisticated understanding of the regulatory environment, a refined approach to optimization, and a move towards testing that achieves the standards of clinical trials ( Lu et al. 2016 ).

• Aberg KC, Herzog MH. 2012. Different types of feedback change decision criterion and sensitivity differently in perceptual learning . Journal of vision 12 : 3–3 [ PubMed ] [ Google Scholar ]
• Adab HZ, Vogels R. 2011. Practicing coarse orientation discrimination improves orientation signals in macaque cortical area v4 . Current biology 21 : 1661–66 [ PubMed ] [ Google Scholar ]
• Adini Y, Wilkonsky A, Haspel R, Tsodyks M, Sagi D. 2004. Perceptual learning in contrast discrimination: The effect of contrast uncertainty . Journal of Vision 4 : 2–2 [ PubMed ] [ Google Scholar ]
• Ahissar M, Hochstein S. 1993. Attentional control of early perceptual learning . Proceedings of the National Academy of Sciences 90 : 5718–22 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Ahissar M, Hochstein S. 2004. The reverse hierarchy theory of visual perceptual learning . Trends in cognitive sciences 8 : 457–64 [ PubMed ] [ Google Scholar ]
• Atkinson J, Braddick O, Moar K. 1977. Development of contrast sensitivity over the first 3 months of life in the human infant . Vision Research 17 : 1037–44 [ PubMed ] [ Google Scholar ]
• Ball K, Sekuler R. 1982. A specific and enduring improvement in visual motion discrimination . Science 218 : 697–98 [ PubMed ] [ Google Scholar ]
• Barlow HB. 1989. Unsupervised learning . Neural computation 1 : 295–311 [ Google Scholar ]
• Bejjanki VR, Beck JM, Lu Z-L, Pouget A. 2011. Perceptual learning as improved probabilistic inference in early sensory areas . Nature neuroscience 14 : 642–48 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Bennett RG, Westheimer G. 1991. The effect of training on visual alignment discrimination and grating resolution . Perception & Psychophysics 49 : 541–46 [ PubMed ] [ Google Scholar ]
• Burgess A, Colborne B. 1988. Visual signal detection. IV. Observer inconsistency . JOSA A5 : 617–27 [ PubMed ] [ Google Scholar ]
• Carandini M, Heeger DJ, Movshon JA. 1997. Linearity and normalization in simple cells of the macaque primary visual cortex . The Journal of Neuroscience 17 : 8621–44 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Casey MC, Sowden PT. 2012. Modeling learned categorical perception in human vision . Neural Networks 33 : 114–26 [ PubMed ] [ Google Scholar ]
• Censor N, Karni A, Sagi D. 2006. A link between perceptual learning, adaptation and sleep . Vision research 46 : 4071–74 [ PubMed ] [ Google Scholar ]
• Choi H, Watanabe T. 2012. Perceptual learning solely induced by feedback . Vision research 61 : 77–82 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Chowdhury SA, DeAngelis GC. 2008. Fine discrimination training alters the causal contribution of macaque area MT to depth perception . Neuron 60 : 367–77 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Crist RE, Kapadia MK, Westheimer G, Gilbert CD. 1997. Perceptual learning of spatial localization: specificity for orientation, position, and context . Journal of neurophysiology 78 : 2889–94 [ PubMed ] [ Google Scholar ]
• DeLoss DJ, Watanabe T, Andersen GJ. 2015. Improving Vision Among Older Adults Behavioral Training to Improve Sight . Psychological science 26 : 456–66 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Deveau J, Ozer DJ, Seitz AR. 2014. Improved vision and on-field performance in baseball through perceptual learning . Current Biology 24 : R146–R47 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Dolan R, Fink G, Rolls E, Booth M, Holmes A, et al. 1997. How the brain learns to see objects and faces in an impoverished context . Nature 389 : 596–99 [ PubMed ] [ Google Scholar ]
• Dorais A, Sagi D. 1997. Contrast masking effects change with practice . Vision Research 37 : 1725–33 [ PubMed ] [ Google Scholar ]
• Dosher BA, Han S, Lu Z-L. 2010. Perceptual learning and attention: Reduction of object attention limitations with practice . Vision research 50 : 402–15 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Dosher BA, Jeter P, Liu J, Lu Z-L. 2013. An integrated reweighting theory of perceptual learning . Proceedings of the National Academy of Sciences 110 : 13678–83 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Dosher BA, Lu Z-L. 1998. Perceptual learning reflects external noise filtering and internal noise reduction through channel reweighting . Proceedings of the National Academy of Sciences 95 : 13988–93 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Dosher BA, Lu Z-L. 1999. Mechanisms of perceptual learning . Vision research 39 : 3197–221 [ PubMed ] [ Google Scholar ]
• Dosher BA, Lu Z-L. 2006. Level and mechanisms of perceptual learning: Learning first-order luminance and second-order texture objects . Vision Research 46 : 1996–2007 [ PubMed ] [ Google Scholar ]
• Dosher BA, Lu Z-L. 2009. Hebbian reweighting on stable representations in perceptual learning . Learning & Perception 1 : 37–58 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Durrie D, McMinn PS. 2007. Computer-based primary visual cortex training for treatment of low myopia and early presbyopia . Trans Am Ophthalmol Soc 105 : 132–8 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Fahle M, Edelman S. 1993. Long-term learning in vernier acuity: Effects of stimulus orientation, range and of feedback . Vision research 33 : 397–412 [ PubMed ] [ Google Scholar ]
• Fahle M, Morgan M. 1996. No transfer of perceptual learning between similar stimuli in the same retinal position . Current Biology 6 : 292–97 [ PubMed ] [ Google Scholar ]
• Fahle M, Poggio T. 2002. Perceptual learning: MIT Press
• Fendick M, Westheimer G. 1983. Effects of practice and the separation of test targets on foveal and peripheral stereoacuity . Vision research 23 : 145–50 [ PubMed ] [ Google Scholar ]
• Fine I, Jacobs RA. 2002. Comparing perceptual learning across tasks: A review . Journal of vision 2 : 5–5 [ PubMed ] [ Google Scholar ]
• Fiorentini A, Berardi N. 1980. Perceptual learning specific for orientation and spatial frequency . Nature [ PubMed ] [ Google Scholar ]
• Fiorentini A, Berardi N. 1981. Learning in grating waveform discrimination: Specificity for orientation and spatial frequency . Vision research 21 : 1149–58 [ PubMed ] [ Google Scholar ]
• French RM. 1999. Catastrophic forgetting in connectionist networks . Trends in cognitive sciences 3 : 128–35 [ PubMed ] [ Google Scholar ]
• Furmanski CS, Schluppeck D, Engel SA. 2004. Learning strengthens the response of primary visual cortex to simple patterns . Current Biology 14 : 573–78 [ PubMed ] [ Google Scholar ]
• Gauthier I, Williams P, Tarr MJ, Tanaka J. 1998. Training ‘greeble’experts: a framework for studying expert object recognition processes . Vision research 38 : 2401–28 [ PubMed ] [ Google Scholar ]
• Geisler WS. 2008. Visual perception and the statistical properties of natural scenes . Annu. Rev. Psychol 59 : 167–92 [ PubMed ] [ Google Scholar ]
• Ghose GM, Yang T, Maunsell JH. 2002. Physiological correlates of perceptual learning in monkey V1 and V2 . Journal of neurophysiology 87 : 1867–88 [ PubMed ] [ Google Scholar ]
• Gibson EJ. 1969. Principles of perceptual learning and development .
• Gilbert CD, Li W. 2012. Adult visual cortical plasticity . Neuron 75 : 250–64 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Gilbert CD, Sigman M, Crist RE. 2001. The neural basis of perceptual learning . Neuron 31 : 681–97 [ PubMed ] [ Google Scholar ]
• Gold J, Bennett P, Sekuler A. 1999. Signal but not noise changes with perceptual learning . Nature 402 : 176–78 [ PubMed ] [ Google Scholar ]
• Goris RL, Movshon JA, Simoncelli EP. 2014. Partitioning neuronal variability . Nature neuroscience 17 : 858–65 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Green CS, Bavelier D. 2015. Action video game training for cognitive enhancement . Current Opinion in Behavioral Sciences 4 : 103–08 [ Google Scholar ]
• Green DM, Swets JA. 1966. Signal detection theory and psychophysics .
• Grossberg S 1987. Competitive learning: From interactive activation to adaptive resonance . Cognitive science 11 : 23–63 [ Google Scholar ]
• Grossman ED, Blake R, Kim C-Y. 2004. Learning to see biological motion: brain activity parallels behavior . Journal of Cognitive Neuroscience 16 : 1669–79 [ PubMed ] [ Google Scholar ]
• Gu Y, Liu S, Fetsch CR, Yang Y, Fok S, et al. 2011. Perceptual learning reduces interneuronal correlations in macaque visual cortex . Neuron 71 : 750–61 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Gutnisky DA, Hansen BJ, Iliescu BF, Dragoi V. 2009. Attention alters visual plasticity during exposure-based learning . Current Biology 19 : 555–60 [ PubMed ] [ Google Scholar ]
• Herzog MH, Aberg KC, Frémaux N, Gerstner W, Sprekeler H. 2012. Perceptual learning, roving and the unsupervised bias . Vision research 61 : 95–99 [ PubMed ] [ Google Scholar ]
• Herzog MH, Fahle M. 1997. The role of feedback in learning a vernier discrimination task . Vision research 37 : 2133–41 [ PubMed ] [ Google Scholar ]
• Herzog MH, Fahle M. 1998. Modeling perceptual learning: Difficulties and how they can be overcome . Biological cybernetics 78 : 107–17 [ PubMed ] [ Google Scholar ]
• Herzog MH, Fahle M. 1999. Effects of biased feedback on learning and deciding in a vernier discrimination task . Vision Research 39 : 4232–43 [ PubMed ] [ Google Scholar ]
• Hou F, Lu Z-L, Huang C-B. 2014. The external noise normalized gain profile of spatial vision . Journal of vision 14 : 9–9 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Huang C-B, Lu Z-L, Dosher BA. 2012. Co-learning analysis of two perceptual learning tasks with identical input stimuli supports the reweighting hypothesis . Vision research 61 : 25–32 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Huber E, Webster JM, Brewer AA, MacLeod DI, Wandell BA, et al. 2015. A lack of experience-dependent plasticity after more than a decade of recovered sight . Psychological science : 0956797614563957 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Hung S-C, Seitz AR. 2014. Prolonged training at threshold promotes robust retinotopic specificity in perceptual learning . The Journal of Neuroscience 34 : 8423–31 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Huxlin KR, Martin T, Kelly K, Riley M, Friedman DI, et al. 2009. Perceptual relearning of complex visual motion after V1 damage in humans . The Journal of Neuroscience 29 : 3981–91 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Jacobs RA. 2009. Adaptive precision pooling of model neuron activities predicts the efficiency of human visual learning . Journal of Vision 9 : 22–22 [ PubMed ] [ Google Scholar ]
• James W. The Principles of. Psychology. 1890; 2 [ Google Scholar ]
• Jastorff J, Kourtzi Z, Giese MA. 2006. Learning to discriminate complex movements: Biological versus artificial trajectories . Journal of vision 6 : 3–3 [ PubMed ] [ Google Scholar ]
• Jehee JF, Ling S, Swisher JD, van Bergen RS, Tong F. 2012. Perceptual learning selectively refines orientation representations in early visual cortex . The Journal of neuroscience 32 : 16747–53 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Jeter PE, Dosher BA, Liu S-H, Lu Z-L. 2010. Specificity of perceptual learning increases with increased training . Vision research 50 : 1928–40 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Jeter PE, Dosher BA, Petrov A, Lu Z-L. 2009. Task precision at transfer determines specificity of perceptual learning . Journal of Vision 9 : 1–1 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Jordan MI, Rumelhart DE. 1992. Forward models: Supervised learning with a distal teacher . Cognitive science 16 : 307–54 [ Google Scholar ]
• Kahnt T, Grueschow M, Speck O, Haynes J-D. 2011. Perceptual learning and decision-making in human medial frontal cortex . Neuron 70 : 549–59 [ PubMed ] [ Google Scholar ]
• Kalia A, Lesmes LA, Dorr M, Gandhi T, Chatterjee G, et al. 2014. Development of pattern vision following early and extended blindness . Proceedings of the National Academy of Sciences 111 : 2035–39 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Karni A, Sagi D. 1991. Where practice makes perfect in texture discrimination: evidence for primary visual cortex plasticity . Proceedings of the National Academy of Sciences 88 : 4966–70 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Karni A, Sagi D. 1993. The time course of learning a visual skill . Nature 365 : 250–52 [ PubMed ] [ Google Scholar ]
• Kasten E, Sabel BA. 1995. Visual field enlargement after computer training in brain-damaged patients with homonymous deficits: an open pilot trial . Restorative neurology and neuroscience 8 : 113–27 [ PubMed ] [ Google Scholar ]
• Kellman PJ, Massey CM. 2013. Perceptual learning, cognition, and expertise . The psychology of learning and motivation 58 : 117–65 [ Google Scholar ]
• Kobatake E, Wang G, Tanaka K. 1998. Effects of shape-discrimination training on the selectivity of inferotemporal cells in adult monkeys . Journal of Neurophysiology 80 : 324–30 [ PubMed ] [ Google Scholar ]
• Kourtzi Z 2010. Visual learning for perceptual and categorical decisions in the human brain . Vision research 50 : 433–40 [ PubMed ] [ Google Scholar ]
• Kourtzi Z, Betts LR, Sarkheil P, Welchman AE. 2005. Distributed neural plasticity for shape learning in the human visual cortex . PLoS Biol 3 : e204. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Kuai S-G, Levi D, Kourtzi Z. 2013. Learning optimizes decision templates in the human visual cortex . Current Biology 23 : 1799–804 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Law C-T, Gold JI. 2008. Neural correlates of perceptual learning in a sensory-motor, but not a sensory, cortical area . Nature neuroscience 11 : 505–13 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Law C-T, Gold JI. 2009. Reinforcement learning can account for associative and perceptual learning on a visual-decision task . Nature neuroscience 12 : 655–63 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Levi DM, Li RW. 2009. Perceptual learning as a potential treatment for amblyopia: a mini-review . Vision research 49 : 2535–49 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Li J, Thompson B, Deng D, Chan LY, Yu M, Hess RF. 2013. Dichoptic training enables the adult amblyopic brain to learn . Current Biology 23 : R308–R09 [ PubMed ] [ Google Scholar ]
• Liu J, Dosher B, Lu Z-L. 2014. Modeling trial by trial and block feedback in perceptual learning . Vision research 99 : 46–56 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Liu J, Lu Z-L, Dosher BA. 2010. Augmented Hebbian reweighting: Interactions between feedback and training accuracy in perceptual learning . Journal of vision 10 : 29–29 [ PubMed ] [ Google Scholar ]
• Liu J, Lu Z-L, Dosher BA. 2012. Mixed training at high and low accuracy levels leads to perceptual learning without feedback . Vision research 61 : 15–24 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Liu L, Kuyk T, Fuhr P. 2007. Visual search training in subjects with severe to profound low vision . Vision research 47 : 2627–36 [ PubMed ] [ Google Scholar ]
• Liu Z 1999. Perceptual learning in motion discrimination that generalizes across motion directions . Proceedings of the National Academy of Sciences 96 : 14085–87 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Logothetis NK, Pauls J, Poggio T. 1995. Shape representation in the inferior temporal cortex of monkeys . Current Biology 5 : 552–63 [ PubMed ] [ Google Scholar ]
• Lu Z-L, Chu W, Dosher BA. 2006. Perceptual learning of motion direction discrimination in fovea: Separable mechanisms . Vision Research 46 : 2315–27 [ PubMed ] [ Google Scholar ]
• Lu Z-L, Chu W, Dosher BA, Lee S. 2005. Independent perceptual learning in monocular and binocular motion systems . Proceedings of the National Academy of Sciences of the United States of America 102 : 5624–29 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Lu Z-L, Dosher BA. 1998. External noise distinguishes attention mechanisms . Vision research 38 : 1183–98 [ PubMed ] [ Google Scholar ]
• Lu Z-L, Dosher BA. 1999. Characterizing human perceptual inefficiencies with equivalent internal noise . JOSA A16 : 764–78 [ PubMed ] [ Google Scholar ]
• Lu Z-L, Dosher BA. 2004. Perceptual learning retunes the perceptual template in foveal orientation identification . Journal of Vision 4 : 5–5 [ PubMed ] [ Google Scholar ]
• Lu Z-L, Dosher BA. 2008. Characterizing observers using external noise and observer models: assessing internal representations with external noise . Psychological review 115 : 44. [ PubMed ] [ Google Scholar ]
• Lu Z-L, Hua T, Huang C-B, Zhou Y, Dosher BA. 2011. Visual perceptual learning . Neurobiology of learning and memory 95 : 145–51 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Lu Z-L, Lin Z, Dosher BA. 2016. Translating Perceptual Learning from the Laboratory to Applications . Trends in cognitive sciences [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Lu Z-L, Liu J, Dosher BA. 2010. Modeling mechanisms of perceptual learning with augmented Hebbian re-weighting . Vision research 50 : 375–90 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Matthews N, Welch L. 1997. Velocity-dependent improvements in single-dot direction discrimination . Perception & psychophysics 59 : 60–72 [ PubMed ] [ Google Scholar ]
• McKee SP, Westheimer G. 1978. Improvement in vernier acuity with practice . Perception & psychophysics 24 : 258–62 [ PubMed ] [ Google Scholar ]
• Mednick S, Nakayama K, Stickgold R. 2003. Sleep-dependent learning: a nap is as good as a night . Nature neuroscience 6 : 697–98 [ PubMed ] [ Google Scholar ]
• Merzenich MM, Jenkins WM, Johnston P, Schreiner C. 1996. Temporal processing deficits of language-learning impaired children ameliorated by training . Science 271 : 77. [ PubMed ] [ Google Scholar ]
• Mollon JD, Danilova MV. 1996. Three remarks on perceptual learning . Spatial vision 10 : 51–58 [ PubMed ] [ Google Scholar ]
• Mukai I, Bahadur K, Kesavabhotla K, Ungerleider LG. 2011. Exogenous and endogenous attention during perceptual learning differentially affect post-training target thresholds . Journal of vision 11 : 25–25 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Nazir TA, O’Regan JK. 1990. Some results on translation invariance in the human visual system . Spatial vision 5 : 81–100 [ PubMed ] [ Google Scholar ]
• Nelles G, Esser J, Eckstein A, Tiede A, Gerhard H, Diener HC. 2001. Compensatory visual field training for patients with hemianopia after stroke . Neuroscience letters 306 : 189–92 [ PubMed ] [ Google Scholar ]
• Özgen E, Davies IR. 2002. Acquisition of categorical color perception: a perceptual learning approach to the linguistic relativity hypothesis . Journal of Experimental Psychology: General 131 : 477. [ PubMed ] [ Google Scholar ]
• Pelli DG. 1981. Effects of visual noise . University of Cambridge . [ Google Scholar ]
• Petrov AA, Dosher BA, Lu Z-L. 2005. The dynamics of perceptual learning: an incremental reweighting model . Psychological review 112 : 715. [ PubMed ] [ Google Scholar ]
• Petrov AA, Dosher BA, Lu Z-L. 2006. Perceptual learning without feedback in non-stationary contexts: Data and model . Vision research 46 : 3177–97 [ PubMed ] [ Google Scholar ]
• Poggio T, Edelman S, Fahle M. 1992. Learning of visual modules from examples: A framework for understanding adaptive visual performance . CVGIP: Image Understanding 56 : 22–30 [ Google Scholar ]
• Polat U, Schor C, Tong J-L, Zomet A, Lev M, et al. 2012. Training the brain to overcome the effect of aging on the human eye . Scientific reports 2 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Raiguel S, Vogels R, Mysore SG, Orban GA. 2006. Learning to see the difference specifically alters the most informative V4 neurons . The Journal of neuroscience 26 : 6589–602 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Rainer G, Miller EK. 2000. Effects of visual experience on the representation of objects in the prefrontal cortex . Neuron 27 : 179–89 [ PubMed ] [ Google Scholar ]
• Ramachandran V, Braddick O. 1973. Orientation-specific learning in stereopsis . Perception 2 : 371–76 [ PubMed ] [ Google Scholar ]
• Reed RD, Marks RJ. 1998. Neural smithing: supervised learning in feedforward artificial neural networks: Mit Press
• Rubin N, Nakayama K, Shapley R. 1997. Abrupt learning and retinal size specificity in illusory-contour perception . Current Biology 7 : 461–67 [ PubMed ] [ Google Scholar ]
• Saarinen J, Levi DM. 1995. Perceptual learning in vernier acuity: what is learned? Vision research 35 : 519–27 [ PubMed ] [ Google Scholar ]
• Sagi D 2011. Perceptual learning in vision research . Vision research 51 : 1552–66 [ PubMed ] [ Google Scholar ]
• Schoups A, Vogels R, Qian N, Orban G. 2001. Practising orientation identification improves orientation coding in V1 neurons . Nature 412 : 549–53 [ PubMed ] [ Google Scholar ]
• Schoups AA, Vogels R, Orban GA. 1995. Human perceptual learning in identifying the oblique orientation: retinotopy, orientation specificity and monocularity . The Journal of physiology 483 : 797. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Schwartz S, Maquet P, Frith C. 2002. Neural correlates of perceptual learning: a functional MRI study of visual texture discrimination . Proceedings of the National Academy of Sciences 99 : 17137–42 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Seitz AR, Kim D, Watanabe T. 2009. Rewards evoke learning of unconsciously processed visual stimuli in adult humans . Neuron 61 : 700–07 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Shibata K, Chang L-H, Kim D, Náñez JE Sr, Kamitani Y, et al. 2012. Decoding reveals plasticity in V3A as a result of motion perceptual learning . PLoS One 7 : e44003. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Shibata K, Yamagishi N, Ishii S, Kawato M. 2009. Boosting perceptual learning by fake feedback . Vision research 49 : 2574–85 [ PubMed ] [ Google Scholar ]
• Shiu L-P, Pashler H. 1992. Improvement in line orientation discrimination is retinally local but dependent on cognitive set . Perception & psychophysics 52 : 582–88 [ PubMed ] [ Google Scholar ]
• Sotiropoulos G, Seitz AR, Seriès P. 2011. Perceptual learning in visual hyperacuity: A reweighting model . Vision research 51 : 585–99 [ PubMed ] [ Google Scholar ]
• Sowden PT, Rose D, Davies IR. 2002. Perceptual learning of luminance contrast detection: Specific for spatial frequency and retinal location but not orientation . Vision research 42 : 1249–58 [ PubMed ] [ Google Scholar ]
• Stratton GM. 1897. Vision without inversion of the retinal image . Psychological review 4 : 341 [ Google Scholar ]
• Strong GK, Torgerson CJ, Torgerson D, Hulme C. 2011. A systematic meta‐analytic review of evidence for the effectiveness of the ‘Fast ForWord’language intervention program . Journal of Child Psychology and Psychiatry 52 : 224–35 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Talluri BC, Hung S-C, Seitz AR, Seriès P. 2015. Confidence-based integrated reweighting model of task-difficulty explains location-based specificity in perceptual learning . Journal of vision 15 : 17–17 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Teich AF, Qian N. 2003. Learning and adaptation in a recurrent model of V1 orientation selectivity . Journal of Neurophysiology 89 : 2086–100 [ PubMed ] [ Google Scholar ]
• Thurston C, Dobkins K. 2007. Stimulus-specific perceptual learning for chromatic, but not luminance, contrast detection . Journal of Vision 7 : 469–69 [ Google Scholar ]
• Ungerleider LG, Courtney SM, Haxby JV. 1998. A neural system for human visual working memory . Proceedings of the National Academy of Sciences 95 : 883–90 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Vaina LM, Sundareswaran V, Harris JG. 1995. Learning to ignore: Psychophysics and computational modeling of fast learning of direction in noisy motion stimuli . Cognitive Brain Research 2 : 155–63 [ PubMed ] [ Google Scholar ]
• Van Essen DC, Anderson CH, Felleman DJ. 1992. Information processing in the primate visual system: an integrated systems perspective . Science 255 : 419. [ PubMed ] [ Google Scholar ]
• Vogels R 2010. Mechanisms of visual perceptual learning in macaque visual cortex . Topics in Cognitive Science 2 : 239–50 [ PubMed ] [ Google Scholar ]
• Vogels R, Orban GA. 1985. The effect of practice on the oblique effect in line orientation judgments . Vision research 25 : 1679–87 [ PubMed ] [ Google Scholar ]
• Wandell BA, Smirnakis SM. 2009. Plasticity and stability of visual field maps in adult primary visual cortex . Nature Reviews Neuroscience 10 : 873–84 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Watanabe T, Náñez JE, Koyama S, Mukai I, Liederman J, Sasaki Y. 2002. Greater plasticity in lower-level than higher-level visual motion processing in a passive perceptual learning task . Nature neuroscience 5 : 1003–09 [ PubMed ] [ Google Scholar ]
• Watanabe T, Sasaki Y. 2015. Perceptual learning: toward a comprehensive theory . Annual review of psychology 66 : 197 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Weiss Y, Edelman S, Fahle M. 1993. Models of perceptual learning in vernier hyperacuity . Neural Computation 5 : 695–718 [ Google Scholar ]
• Westheimer G 2001. Is peripheral visual acuity susceptible to perceptual learning in the adult? Vision Research 41 : 47–52 [ PubMed ] [ Google Scholar ]
• Xi J, Jia W-L, Feng L-X, Lu Z-L, Huang C-B. 2014. Perceptual learning improves stereoacuity in amblyopiaperceptual learning and restoration of stereoacuity . Investigative ophthalmology & visual science 55 : 2384–91 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Xiao L-Q, Zhang J-Y, Wang R, Klein SA, Levi DM, Yu C. 2008. Complete transfer of perceptual learning across retinal locations enabled by double training . Current Biology 18 : 1922–26 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Xu JP, He ZJ, Ooi TL. 2012. Push–pull training reduces foveal sensory eye dominance within the early visual channels . Vision research 61 : 48–59 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Yamins DL, Hong H, Cadieu CF, Solomon EA, Seibert D, DiCarlo JJ. 2014. Performance-optimized hierarchical models predict neural responses in higher visual cortex . Proceedings of the National Academy of Sciences 111 : 8619–24 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Yan F-F, Zhou J, Zhao W, Li M, Xi J, et al. 2015. Perceptual learning improves neural processing in myopic vision . Journal of vision 15 : 12–12 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Yang T, Maunsell JH. 2004. The effect of perceptual learning on neuronal responses in monkey visual area V4 . The Journal of Neuroscience 24 : 1617–26 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Yu C, Klein SA, Levi DM. 2004. Perceptual learning in contrast discrimination and the (minimal) role of context . Journal of Vision 4 : 4–4 [ PubMed ] [ Google Scholar ]
• Yu D, Cheung S-H, Legge GE, Chung ST. 2010. Reading speed in the peripheral visual field of older adults: Does it benefit from perceptual learning? Vision research 50 : 860–69 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Zhang J, Kourtzi Z. 2010. Learning-dependent plasticity with and without training in the human brain . Proceedings of the National Academy of Sciences 107 : 13503–08 [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Zhang P, Hou F, Xi J, Zhang M-y, He Q, et al. 2016. Reward enhances perceptual learning and transfer . In Vision Sciences Society , pp. 54.4113 [ Google Scholar ]
• Zhaoping L, Herzog MH, Dayan P. 2003. Nonlinear ideal observation and recurrent preprocessing in perceptual learning . Network: Computation in Neural Systems 14 : 233–47 [ PubMed ] [ Google Scholar ]

• Our Mission

The Power of Visualization in Math

Creating visual representations for math students can open up understanding. We have resources you can use in class tomorrow.

When do you know it’s time to try something different in your math lesson?

For me, I knew the moment I read this word problem to my fifth-grade summer school students: “On average, the sun’s energy density reaching Earth’s upper atmosphere is 1,350 watts per square meter. Assume the incident, monochromatic light has a wavelength of 800 nanometers (each photon has an energy of 2.48 × 10 -19 joules at this wavelength). How many photons are incident on the Earth’s upper atmosphere in one second?”

My students couldn’t get past the language, the sizes of the different numbers, or the science concepts addressed in the question. In short, I had effectively shut them down, and I needed a new approach to bring them back to their learning. So I started drawing on the whiteboard and created something with a little whimsy, a cartoon photon asking how much energy a photon has.

Immediately, students started yelling out, “2.48 × 10 -19 joules,” and they could even cite the text where they had learned the information. I knew I was on to something, so the next thing I drew was a series of boxes with our friend the photon.

If all of the photons in the image below were to hit in one second, how much energy is represented in the drawing?

Students realized that we were just adding up all the individual energy from each photon and then quickly realized that this was multiplication. And then they knew that the question we were trying to answer was just figuring out the number of photons, and since we knew the total energy in one second, we could compute the number of photons by division.

The point being, we reached a place where my students were able to process the learning. The power of the visual representation made all the difference for these students, and being able to sequence through the problem using the visual supports completely changed the interactions they were having with the problem.

If you’re like me, you’re thinking, “So the visual representations worked with this problem, but what about other types of problems? Surely there isn’t a visual model for every problem!”

The power of this moment, the change in the learning environment, and the excitement of my fifth graders as they could not only understand but explain to others what the problem was about convinced me it was worth the effort to pursue visualization and try to answer these questions: Is there a process to unlock visualizations in math? And are there resources already available to help make mathematics visual?

I realized that the first step in unlocking visualization as a scaffold for students was to change the kind of question I was asking myself. A powerful question to start with is: “How might I represent this learning target in a visual way?” This reframing opens a world of possible representations that we might not otherwise have considered. Thinking about many possible visual representations is the first step in creating a good one for students.

The Progressions published in tandem with the Common Core State Standards for mathematics are one resource for finding specific visual models based on grade level and standard. In my fifth-grade example, what I constructed was a sequenced process to develop a tape diagram—a type of visual model that uses rectangles to represent the parts of a ratio. I didn’t realize it, but to unlock my thinking I had to commit to finding a way to represent the problem in a visual way. Asking yourself a very simple series of questions leads you down a variety of learning paths, and primes you for the next step in the sequence—finding the right resources to complete your visualization journey.

Posing the question of visualization readies your brain to identify the right tool for the desired learning target and your students. That is, you’ll more readily know when you’ve identified the right tool for the job for your students. There are many, many resources available to help make this process even easier, and I’ve created a matrix of clickable tools, articles, and resources .

The process to visualize your math instruction is summarized at the top of my Visualizing Math graphic; below that is a mix of visualization strategies and resources you can use tomorrow in your classroom.

Our job as educators is to set a stage that maximizes the amount of learning done by our students, and teaching students mathematics in this visual way provides a powerful pathway for us to do our job well. The process of visualizing mathematics tests your abilities at first, and you’ll find that it makes both you and your students learn.

Learning Through Visuals

Visual imagery in the classroom.

Posted July 20, 2012

• Why Education Is Important
• Take our ADHD Test
• Find a Child Therapist

A large body of research indicates that visual cues help us to better retrieve and remember information. The research outcomes on visual learning make complete sense when you consider that our brain is mainly an image processor (much of our sensory cortex is devoted to vision), not a word processor. In fact, the part of the brain used to process words is quite small in comparison to the part that processes visual images.

Words are abstract and rather difficult for the brain to retain, whereas visuals are concrete and, as such, more easily remembered. To illustrate, think about your past school days of having to learn a set of new vocabulary words each week. Now, think back to the first kiss you had or your high school prom date. Most probably, you had to put forth great effort to remember the vocabulary words. In contrast, when you were actually having your first kiss or your prom date, I bet you weren’t trying to commit them to memory . Yet, you can quickly and effortlessly visualize these experiences (now, even years later). You can thank your brain’s amazing visual processor for your ability to easily remember life experiences. Your brain memorized these events for you automatically and without you even realizing what it was doing.

There are countless studies that have confirmed the power of visual imagery in learning. For instance, one study asked students to remember many groups of three words each, such as dog, bike, and street. Students who tried to remember the words by repeating them over and over again did poorly on recall. In comparison, students who made the effort to make visual associations with the three words, such as imagining a dog riding a bike down the street, had significantly better recall.

Various types of visuals can be effective learning tools: photos, illustrations, icons, symbols, sketches, figures, and concept maps, to name only a few. Consider how memorable the visual graphics are in logos, for example. You recognize the brand by seeing the visual graphic, even before reading the name of the brand. This type of visual can be so effective that earlier this year Starbucks simplified their logo by dropping their printed name and keeping only the graphic image of the popularly referred to mermaid (technically, it’s a siren). I think we can safely assume that Starbucks Corporation must be keenly aware of how our brains have automatically and effortlessly committed their graphic image to memory.

So powerful is visual learning that I embrace it in my teaching and writing. Each page in the psychology textbooks I coauthor has been individually formatted to maximize visual learning. Each lecture slide I use in class is presented in a way to make the most of visual learning. I believe the right visuals can help make abstract and difficult concepts more tangible and welcoming, as well as make learning more effective and long lasting. This is why I scrutinize every visual I use in my writing and teaching to make sure it is paired with content in a clear, meaningful manner.

Based upon research outcomes, the effective use of visuals can decrease learning time, improve comprehension, enhance retrieval, and increase retention. In addition, the many testimonials I hear from my students and readers weigh heavily in my mind as support for the benefits of learning through visuals. I hear it often and still I can’t hear it enough times . . . by retrieving a visual cue presented on the pages of a book or on the slides of a lecture presentation, a learner is able to accurately retrieve the content associated with the visual.

McDaniel, M. A., & Einstein, G. O. (1986). Bizarre imagery as an effective memory aid: The importance of distinctiveness. Journal of Experimental Psychology: Learning, Memory, and Cognition , 12(1), 54-65.

Meier, D. (2000). The accelerated learning handbook. NY: McGraw-Hill.

Patton, W. W. (1991). Opening students’ eyes: Visual learning theory in the Socratic classroom. Law and Psychology Review, 15, 1-18.

Schacter, D.L. (1966). Searching for memory. NY: Basic Books.

Verdi, M. P., Johnson, J. T., Stock, W. A., Kulhavy, R. W., Whitman-Ahern, P. (1997). Organized spatial displays and texts: Effects of presentation order and display type on learning outcomes. Journal of Experimental Education , 65, 303-317.

Haig Kouyoumdjian, Ph.D. , is a clinical psychologist and coauthor of Introduction to Psychology , 9th ed. and the innovative Discovery Series: Introduction to Psychology.

• Find a Therapist
• Find a Treatment Center
• Find a Psychiatrist
• Find a Support Group
• Find Online Therapy
• United States
• Brooklyn, NY
• Chicago, IL
• Houston, TX
• Los Angeles, CA
• New York, NY
• Portland, OR
• San Diego, CA
• San Francisco, CA
• Seattle, WA
• Washington, DC
• Asperger's
• Bipolar Disorder
• Chronic Pain
• Eating Disorders
• Passive Aggression
• Personality
• Goal Setting
• Positive Psychology
• Stopping Smoking
• Low Sexual Desire
• Relationships
• Child Development
• Self Tests NEW
• Therapy Center
• Diagnosis Dictionary
• Types of Therapy

Sticking up for yourself is no easy task. But there are concrete skills you can use to hone your assertiveness and advocate for yourself.

• Emotional Intelligence
• Gaslighting
• Affective Forecasting
• Neuroscience

Visual Learning: 10 Examples, Definition, Pros & Cons

Dave Cornell (PhD)

Dr. Cornell has worked in education for more than 20 years. His work has involved designing teacher certification for Trinity College in London and in-service training for state governments in the United States. He has trained kindergarten teachers in 8 countries and helped businessmen and women open baby centers and kindergartens in 3 countries.

Learn about our Editorial Process

Chris Drew (PhD)

This article was peer-reviewed and edited by Chris Drew (PhD). The review process on Helpful Professor involves having a PhD level expert fact check, edit, and contribute to articles. Reviewers ensure all content reflects expert academic consensus and is backed up with reference to academic studies. Dr. Drew has published over 20 academic articles in scholarly journals. He is the former editor of the Journal of Learning Development in Higher Education and holds a PhD in Education from ACU.

Visual learning refers to the process of coming to understand information by seeing it – often, represented in graphs or films.

Teachers that utilize visual learning strategies present information in various visual formats such as: flowcharts, diagrams, videos, simulations, graphs, cartoons, coloring books, PPT slide shows, posters, movies, games, and flash cards.

Human beings are very visually-oriented creatures. Our visual system is central to many aspects of our lives. We can see the centrality of visual stimuli in the arts in the form of theatre and film, paintings and sculptures.

It plays a central role in our daily lives as we wear clothes and put on make-up to enhance our visual aesthetic. Fashion and beauty industries exist in every country and tally billions of dollars in sales a year.

However, despite the importance of visual stimuli, in educational contexts visual learning may not be suitable for all students. Because every student is different, visual learning may be effective for some, but not others.

Visual Learning as a Learning Style

Visual learning is the processing of visually presented information. A visual learning style, on the other hand, refers to times when visual learning is an individual’s preferred method of learning.

Whereas some students may be especially capable of visual learning, others may prefer to learn through other means, such as through text or auditory processing.

Others may prefer to have something to touch and manipulate.

This has led scholars to devise the concept of learning styles (see Pritchard, 2017). Each student has a different way of learning. Such scholars argue that teachers should utilize a range of instructional approaches that present information in a range of formats.

Over the years, a plethora of theoretical frameworks regarding learning styles has developed, with visual learning being a common category.

For instance, Neil Fleming’s VARK model (Fleming & Baume, 2006) contains four learning modalities : visual, auditory, reading and writing, and kinesthetic ( similar to tactile learning ).

Glossary Term: Visual Literacy

Visual literacy is a slightly different concept. It refers to a skill or the ability to decipher and create visually presented information.

Avgerinou and Pettersson (2011) point out the difficulty scholars have had in agreeing upon a definition of visual literacy. However, the one provided by Heinich et al. (1982) seems sufficient, despite the fact that it was offered last century:

“Visual literacy is the learned ability to interpret visual messages accurately and to create such messages.  Thus, interpretation and creation in visual literacy can be said to parallel reading and writing in print literacy” (p. 62).

Visual Learning Examples

• Concept Maps: A concept map is a way to graphically organize information that can enhance a student’s understanding of how different ideas are interconnected. Each concept is displayed as a circle, and students draw lines to other concepts/circles that are related in some way.  
• Data Animations: Large amounts of complex data can be presented in animation form. For example, explaining the economic growth and decline of various countries across decades can be demonstrated by animating the placement of each country’s economic rank year-over-year.
• PowerPoint Slides: Creating a PPT presentation that includes various charts and images can help convey meaning that cannot be accomplished through text alone.
• Gamification: Adding game elements to academic concepts generates student engagement and allows students to have a non-academic experience with academic concepts.
• Minecraft Education Edition: The Education Edition of Minecraft is a great way for students to learn programming skills and about academic subjects by creating their own visual stories.
• Dioramas: A diorama gives students a chance to create their own 3-D displays pertaining to academic subjects. For example, students can learn about animals and their habitats by constructing a scene in a shoebox.
• Interactive Smartboards: The interactive smartboard can display interactive charts, demonstrate complex principles in chemistry and physics, and even give preschoolers a chance to get out of their seats and touch the correct phoneme displayed on the board.
• Computer Simulations: It’s one thing to hear a lecture on the synaptic gap and neurotransmission. It’s quite another to see the process depicted in a sophisticated computer simulation.       
• Video Production: Students can learn about a key historical event by producing their own micro-play on video. The performance aspect is also visual and the end result is a student-designed video that depicts the crucial moments and characters of an important historical happening.
• Flowcharts: Complicated processes can be explained through a verbal explanation, but having a visual representation will be much more effective. Seeing each step sequentially helps students understand the big picture while at the same time seeing how each step is connected.

Strengths and Weaknesses of Visual Learning

1. strength: explaining the complex.

Very complex processes, such as those in physics, chemistry, and medicine, can be more easily understood through a visual format.

Well-done computer animations can show the dynamics of a complex process that simply cannot be discerned so thoroughly if presented through a verbal or text format.

2. Strength: Availability of Resources

Visual learning resources can be found within a few seconds on the internet. An image or video search will generate an incredible number of graphs, images, and videos which a teacher can easily download and incorporate into instruction.

3. Strength: Increases Student Engagement

Students today live in a very visual world. Short videos on social media and sites such as YouTube are viewed by students every day.

When in the classroom, listening or reading about academic concepts can lead to a lack of interest among students. However, presenting the same information in a visual format can pique interest and therefore increase student engagement.

4. Strength: Convenience

Visual learning resources are usually in a digital format. That means students can view the material just about anywhere, as long as they have their phone with them.

This convenience expands the opportunities for students to engage in learning. They no longer have to be seated at a desk to learn.

5. Strength: Efficiency

Visual learning is very efficient. For example, a lot of information can be presented in a short video lasting just a couple of minutes. However, to read and digest the same amount of information presented in text may consume many pages in a book.

Reading all of those pages may take three or four times longer than the same content presented in a video.

6. Weakness: Requires Equipment

When we think of the classroom, we usually envision a room well-equipped with video projectors and screens and teachers with laptops and laser pointers.

Unfortunately, that is a distorted perception of what exists in most of the world. A vast majority of classrooms around the globe are simply not equipped with the necessary hardware to capitalize on the value of visual learning material.

7. Weakness: Requires Less Thinking

Some visual learning activities, certainly not all, are passive experiences. For example, watching a video is a passive experience. The student simply needs to keep their eyes on the screen and let the information enter their mind.

This is a quite different cognitive process than needing to focus on a lecture and processing the meaning of each word spoken.

One is a passive cognitive process, while the other requires thinking.

8. Weakness: Can Create Edutainment Expectations

Because today’s students are so immersed in videos that are eye-catching and exciting to watch, it can create the expectation that education should be entertaining. This is not only unrealistic, but also may not be in the student’s best interest.

Learning to endure educational experiences that are not always pleasurable can help students develop self-discipline.

Disengaging from a learning experience simply because it is not entertaining denies students an opportunity for personal growth and the opportunity for them to develop higher-order thinking .

Case Study: Visual Learning in Ed. Teach

Applications of technology to improve classroom instruction has steadily increased as software has become more user friendly.

Numerous commercial products are available that can enhance students’ understanding of academic concepts, generate interest in technology, and improve higher-order thinking skills such as logical reasoning and problem-solving.

Many of those products capitalize on visual learning.

For example, Rodger et al. (2009) demonstrated the use of Alice to design lessons in math, language arts, and social studies. The program allows students to create their own interactive games, animations, and videos.

Scratch is a media tool that allows students to program their own interactive stories and games, which helps students build computational thinking and programming skills (Brennan & Resnick, 2012; Wilson et al., 2009). 

Kodu Game Lab is a 3-D visual programming platform that can enhance creativity and problem-solving skills (Stolee & Fristoe, 2011).

Hero et al. (2015) used MIT App to spark student interest in programming by enabling students to design their own Android-based apps and games.

These kinds of technology platforms, which utilize visual learning, can produce numerous educational benefits.

Visual learning is learning by seeing. Information is presented in a visual format such as a video, graph, or computer animation.

Although many students can benefit from visually presented information, not all will. Some students are more motivated to learn through auditory or textual channels, so they prefer to listen or read.

Recognizing that students differ in how they prefer to learn has led to the notion of learning styles. This is the idea that each student has a preferred way of learning and that therefore, teachers should design instructional strategies that suit various learning styles in a process called differentiation .

While visual learning has many advantages in terms of explaining complex processes and capturing student attention, there are also some disadvantages.

Most classrooms in the world are not equipped for visual learning. A reliance on visual learning can create the expectation in students that learning is passive and/or should be entertaining.

In other aspects, some visual learning formats can involve less active cognitive processing and fail to exercise a valuable mental skill known as thinking .

Avgerinou, M. D., & Pettersson, R. (2011). Toward a cohesive theory of visual literacy. Journal of Visual Literacy , 30 (2), 1-19.

Brennan, K., & Resnick, M. (2012, April). New frameworks for studying and assessing the development of computational thinking. In Proceedings of the 2012 annual meeting of the American educational research association, Vancouver, Canada (Vol. 1, p. 25).

Coffield F., Moseley D., Hall E., Ecclestone K. (2004). Learning styles and pedagogy in post-16 learning. A systematic and critical review . London: Learning and Skills Research Centre.

Fleming, N., & Baume, D. (2006). Learning Styles Again: VARKing up the right tree! Educational Developments , 7 (4), 4.

Heinich, R., Molenda, M., & Russell, J. D. (1982). Instructional media and the new technologies of instruction . New York: Macmillan.

Herro, D., McCune-Gardner, C., & Boyer, M. D. (2015). Perceptions of coding with MIT App Inventor: Pathways for their future. Journal for Computing Teachers .

Pritchard, A. (2017). Ways of learning: Learning theories for the classroom . London: Routledge.

Raiyn, J. (2016). The Role of Visual Learning in Improving Students’ High-Order Thinking Skills. Journal of Education and Practice , 7 (24), 115-121.

Rodger, S. H., Hayes, J., Lezin, G., Qin, H., Nelson, D., Tucker, R., … & Slater, D. (2009, March). Engaging middle school teachers and students with alice in a diverse set of subjects. In Proceedings of the 40th ACM technical symposium on Computer science education (pp. 271-275).

Stolee, K. T., & Fristoe, T. (2011, March). Expressing computer science concepts through Kodu game lab. In Proceedings of the 42nd ACM technical symposium on Computer science education (pp. 99-104).

Wilson, A., Hainey, T., & Connolly, T. (2012, October). Evaluation of computer games developed by primary school children to gauge understanding of programming concepts. In European Conference on Games Based Learning (p. 549). Academic Conferences International Limited.

• Dave Cornell (PhD) https://helpfulprofessor.com/author/dave-cornell-phd/ 23 Achieved Status Examples
• Dave Cornell (PhD) https://helpfulprofessor.com/author/dave-cornell-phd/ 25 Defense Mechanisms Examples
• Dave Cornell (PhD) https://helpfulprofessor.com/author/dave-cornell-phd/ 15 Theory of Planned Behavior Examples
• Dave Cornell (PhD) https://helpfulprofessor.com/author/dave-cornell-phd/ 18 Adaptive Behavior Examples

• Chris Drew (PhD) https://helpfulprofessor.com/author/chris-drew-phd/ 23 Achieved Status Examples
• Chris Drew (PhD) https://helpfulprofessor.com/author/chris-drew-phd/ 15 Ableism Examples
• Chris Drew (PhD) https://helpfulprofessor.com/author/chris-drew-phd/ 25 Defense Mechanisms Examples
• Chris Drew (PhD) https://helpfulprofessor.com/author/chris-drew-phd/ 15 Theory of Planned Behavior Examples

Leave a Comment Cancel Reply

Your email address will not be published. Required fields are marked *

Initial Thoughts

Perspectives & resources, what is high-quality mathematics instruction and why is it important.

• Page 1: The Importance of High-Quality Mathematics Instruction
• Page 2: A Standards-Based Mathematics Curriculum
• Page 3: Evidence-Based Mathematics Practices

What evidence-based mathematics practices can teachers employ?

• Page 4: Explicit, Systematic Instruction

Page 5: Visual Representations

• Page 6: Schema Instruction
• Page 7: Metacognitive Strategies
• Page 8: Effective Classroom Practices
• Page 9: References & Additional Resources
• Page 10: Credits

Research Shows

• Students who use accurate visual representations are six times more likely to correctly solve mathematics problems than are students who do not use them. However, students who use inaccurate visual representations are less likely to correctly solve mathematics problems than those who do not use visual representations at all. (Boonen, van Wesel, Jolles, & van der Schoot, 2014)
• Students with a learning disability (LD) often do not create accurate visual representations or use them strategically to solve problems. Teaching students to systematically use a visual representation to solve word problems has led to substantial improvements in math achievement for students with learning disabilities. (van Garderen, Scheuermann, & Jackson, 2012; van Garderen, Scheuermann, & Poch, 2014)
• Students who use visual representations to solve word problems are more likely to solve the problems accurately. This was equally true for students who had LD, were low-achieving, or were average-achieving. (Krawec, 2014)

Visual representations are flexible; they can be used across grade levels and types of math problems. They can be used by teachers to teach mathematics facts and by students to learn mathematics content. Visual representations can take a number of forms. Click on the links below to view some of the visual representations most commonly used by teachers and students.

How does this practice align?

High-leverage practice (hlp).

• HLP15 : Provide scaffolded supports

CCSSM: Standards for Mathematical Practice

• MP1 : Make sense of problems and persevere in solving them.

Number Lines

Definition : A straight line that shows the order of and the relation between numbers.

Common Uses : addition, subtraction, counting

Strip Diagrams

Definition : A bar divided into rectangles that accurately represent quantities noted in the problem.

Common Uses : addition, fractions, proportions, ratios

Definition : Simple drawings of concrete or real items (e.g., marbles, trucks).

Common Uses : counting, addition, subtraction, multiplication, division

Graphs/Charts

Definition : Drawings that depict information using lines, shapes, and colors.

Common Uses : comparing numbers, statistics, ratios, algebra

Graphic Organizers

Definition : Visual that assists students in remembering and organizing information, as well as depicting the relationships between ideas (e.g., word webs, tables, Venn diagrams).

Common Uses : algebra, geometry

Before they can solve problems, however, students must first know what type of visual representation to create and use for a given mathematics problem. Some students—specifically, high-achieving students, gifted students—do this automatically, whereas others need to be explicitly taught how. This is especially the case for students who struggle with mathematics and those with mathematics learning disabilities. Without explicit, systematic instruction on how to create and use visual representations, these students often create visual representations that are disorganized or contain incorrect or partial information. Consider the examples below.

Elementary Example

Mrs. Aldridge ask her first-grade students to add 2 + 4 by drawing dots.

Notice that Talia gets the correct answer. However, because Colby draws his dots in haphazard fashion, he fails to count all of them and consequently arrives at the wrong solution.

High School Example

Mr. Huang asks his students to solve the following word problem:

The flagpole needs to be replaced. The school would like to replace it with the same size pole. When Juan stands 11 feet from the base of the pole, the angle of elevation from Juan’s feet to the top of the pole is 70 degrees. How tall is the pole?

Compare the drawings below created by Brody and Zoe to represent this problem. Notice that Brody drew an accurate representation and applied the correct strategy. In contrast, Zoe drew a picture with partially correct information. The 11 is in the correct place, but the 70° is not. As a result of her inaccurate representation, Zoe is unable to move forward and solve the problem. However, given an accurate representation developed by someone else, Zoe is more likely to solve the problem correctly.

Manipulatives

Some students will not be able to grasp mathematics skills and concepts using only the types of visual representations noted in the table above. Very young children and students who struggle with mathematics often require different types of visual representations known as manipulatives. These concrete, hands-on materials and objects—for example, an abacus or coins—help students to represent the mathematical idea they are trying to learn or the problem they are attempting to solve. Manipulatives can help students develop a conceptual understanding of mathematical topics. (For the purpose of this module, the term concrete objects refers to manipulatives and the term visual representations refers to schematic diagrams.)

It is important that the teacher make explicit the connection between the concrete object and the abstract concept being taught. The goal is for the student to eventually understand the concepts and procedures without the use of manipulatives. For secondary students who struggle with mathematics, teachers should show the abstract along with the concrete or visual representation and explicitly make the connection between them.

A move from concrete objects or visual representations to using abstract equations can be difficult for some students. One strategy teachers can use to help students systematically transition among concrete objects, visual representations, and abstract equations is the Concrete-Representational-Abstract (CRA) framework.

If you would like to learn more about this framework, click here.

Concrete-Representational-Abstract Framework

• Concrete —Students interact and manipulate three-dimensional objects, for example algebra tiles or other algebra manipulatives with representations of variables and units.
• Representational — Students use two-dimensional drawings to represent problems. These pictures may be presented to them by the teacher, or through the curriculum used in the class, or students may draw their own representation of the problem.
• Abstract — Students solve problems with numbers, symbols, and words without any concrete or representational assistance.

CRA is effective across all age levels and can assist students in learning concepts, procedures, and applications. When implementing each component, teachers should use explicit, systematic instruction and continually monitor student work to assess their understanding, asking them questions about their thinking and providing clarification as needed. Concrete and representational activities must reflect the actual process of solving the problem so that students are able to generalize the process to solve an abstract equation. The illustration below highlights each of these components.

For Your Information

One promising practice for moving secondary students with mathematics difficulties or disabilities from the use of manipulatives and visual representations to the abstract equation quickly is the CRA-I strategy . In this modified version of CRA, the teacher simultaneously presents the content using concrete objects, visual representations of the concrete objects, and the abstract equation. Studies have shown that this framework is effective for teaching algebra to this population of students (Strickland & Maccini, 2012; Strickland & Maccini, 2013; Strickland, 2017).

Kim Paulsen discusses the benefits of manipulatives and a number of things to keep in mind when using them (time: 2:35).

Kim Paulsen, EdD Associate Professor, Special Education Vanderbilt University

View Transcript

Transcript: Kim Paulsen, EdD

Manipulatives are a great way of helping kids understand conceptually. The use of manipulatives really helps students see that conceptually, and it clicks a little more with them. Some of the things, though, that we need to remember when we’re using manipulatives is that it is important to give students a little bit of free time when you’re using a new manipulative so that they can just explore with them. We need to have specific rules for how to use manipulatives, that they aren’t toys, that they really are learning materials, and how students pick them up, how they put them away, the right time to use them, and making sure that they’re not distracters while we’re actually doing the presentation part of the lesson. One of the important things is that we don’t want students to memorize the algorithm or the procedures while they’re using the manipulatives. It really is just to help them understand conceptually. That doesn’t mean that kids are automatically going to understand conceptually or be able to make that bridge between using the concrete manipulatives into them being able to solve the problems. For some kids, it is difficult to use the manipulatives. That’s not how they learn, and so we don’t want to force kids to have to use manipulatives if it’s not something that is helpful for them. So we have to remember that manipulatives are one way to think about teaching math.

I think part of the reason that some teachers don’t use them is because it takes a lot of time, it takes a lot of organization, and they also feel that students get too reliant on using manipulatives. One way to think about using manipulatives is that you do it a couple of lessons when you’re teaching a new concept, and then take those away so that students are able to do just the computation part of it. It is true we can’t walk around life with manipulatives in our hands. And I think one of the other reasons that a lot of schools or teachers don’t use manipulatives is because they’re very expensive. And so it’s very helpful if all of the teachers in the school can pool resources and have a manipulative room where teachers can go check out manipulatives so that it’s not so expensive. Teachers have to know how to use them, and that takes a lot of practice.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Published: 09 August 2024

Visual interpretability of bioimaging deep learning models

• Oded Rotem 1 &
• Assaf Zaritsky   ORCID: orcid.org/0000-0002-1477-5478 1  

Nature Methods volume  21 ,  pages 1394–1397 ( 2024 ) Cite this article

562 Accesses

1 Citations

1 Altmetric

Metrics details

• Image processing
• Machine learning

The success of deep learning in analyzing bioimages comes at the expense of biologically meaningful interpretations. We review the state of the art of explainable artificial intelligence (XAI) in bioimaging and discuss its potential in hypothesis generation and data-driven discovery.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 12 print issues and online access

251,40 € per year

only 20,95 € per issue

Buy this article

• Purchase on Springer Link
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Zaritsky, A. et al. Cell Syst. 12 , 733–747.e6 (2021).

Article   CAS   PubMed   PubMed Central   Google Scholar  

DeGrave, A. J., Cai, Z. R., Janizek, J. D., Daneshjou, R. & Lee, S. I. Nat. Biomed. Eng. https://doi.org/10.1038/s41551-023-01160-9 (2023).

Article   PubMed   Google Scholar  

Lamiable, A. et al. Nat. Commun. 14 , 6386 (2023).

Rotem, O. et al. Preprint at bioRxiv https://doi.org/10.1101/2023.11.15.566968 . (2023).

Kobayashi, H., Cheveralls, K. C., Leonetti, M. D. & Royer, L. A. Nat. Methods 19 , 995–1003 (2022).

Razdaibiedina, A. et al. Mol. Syst. Biol. 20 , 521–548 (2024).

Article   PubMed   PubMed Central   Google Scholar  

Lundberg, S. & Lee, S.-I. NIPS’17: Proc. 31st International Conference on Neural Information Processing Systems 4768–4777 (ACM, 2017).

Soelistyo, C. J. et al. Nat. Mach. Intell. 4 , 636–644 (2022).

Article   Google Scholar  

Yamamoto, T., Cockburn, K., Greco, V. & Kawaguchi, K. PLOS Comput. Biol. 18 , e1010477 (2022).

Schmitt, M. S. et al. Cell 187 , 481–494.e24 (2024).

Article   CAS   PubMed   Google Scholar  

Doron, M. et al. Preprint at bioRxiv https://doi.org/10.1101/2023.06.16.545359 (2023).

LaChance, J., Suh, K., Clausen, J. & Cohen, D. J. PLOS Comput. Biol. 18 , e1009293 (2022).

Article   CAS   Google Scholar  

Yang, K. D. et al. PLOS Comput. Biol. 16 , e1007828 (2020).

Eckstein, N. et al. Cell 187 , 2574–2594 (2024).

Soelistyo, C. J. & Lowe, A. R. Preprint at arXiv https://doi.org/10.48550/arXiv.2402.03115 (2024).

Download references

Acknowledgements

This research was supported by the Israeli Council for Higher Education (CHE) via the Data Science Research Center, Ben-Gurion University of the Negev, Israel (to AZ), and by the Rosetree trust (to AZ). We thank Nadav Rappoport, Meghan Driscoll, Orit Kliper-Gross and Kevin Dean for critically reading this Comment.

Author information

Authors and affiliations.

Department of Software and Information Systems Engineering, Ben-Gurion University of the Negev, Be’er-Sheva, Israel

Oded Rotem & Assaf Zaritsky

You can also search for this author in PubMed   Google Scholar

Contributions

O.R. and A.Z. conceived and wrote this comment.

Corresponding author

Correspondence to Assaf Zaritsky .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Rotem, O., Zaritsky, A. Visual interpretability of bioimaging deep learning models. Nat Methods 21 , 1394–1397 (2024). https://doi.org/10.1038/s41592-024-02322-6

Download citation

Published : 09 August 2024

Issue Date : August 2024

DOI : https://doi.org/10.1038/s41592-024-02322-6

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Embedding ai in biology.

Nature Methods (2024)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Original Article
• Open access
• Published: 07 December 2016

Creating visual explanations improves learning

• Eliza Bobek   ORCID: orcid.org/0000-0003-2380-3108 1 &
• Barbara Tversky 2  

Cognitive Research: Principles and Implications volume  1 , Article number:  27 ( 2016 ) Cite this article

21k Accesses

80 Citations

49 Altmetric

Metrics details

Many topics in science are notoriously difficult for students to learn. Mechanisms and processes outside student experience present particular challenges. While instruction typically involves visualizations, students usually explain in words. Because visual explanations can show parts and processes of complex systems directly, creating them should have benefits beyond creating verbal explanations. We compared learning from creating visual or verbal explanations for two STEM domains, a mechanical system (bicycle pump) and a chemical system (bonding). Both kinds of explanations were analyzed for content and learning assess by a post-test. For the mechanical system, creating a visual explanation increased understanding particularly for participants of low spatial ability. For the chemical system, creating both visual and verbal explanations improved learning without new teaching. Creating a visual explanation was superior and benefitted participants of both high and low spatial ability. Visual explanations often included crucial yet invisible features. The greater effectiveness of visual explanations appears attributable to the checks they provide for completeness and coherence as well as to their roles as platforms for inference. The benefits should generalize to other domains like the social sciences, history, and archeology where important information can be visualized. Together, the findings provide support for the use of learner-generated visual explanations as a powerful learning tool.

Significance

Uncovering cognitive principles for effective teaching and learning is a central application of cognitive psychology. Here we show: (1) creating explanations of STEM phenomena improves learning without additional teaching; and (2) creating visual explanations is superior to creating verbal ones. There are several notable differences between visual and verbal explanations; visual explanations map thought more directly than words and provide checks for completeness and coherence as well as a platform for inference, notably from structure to process. Extensions of the technique to other domains should be possible. Creating visual explanations is likely to enhance students’ spatial thinking skills, skills that are increasingly needed in the contemporary and future world.

Dynamic systems such as those in science and engineering, but also in history, politics, and other domains, are notoriously difficult to learn (e.g. Chi, DeLeeuw, Chiu, & Lavancher, 1994 ; Hmelo-Silver & Pfeffer, 2004 ; Johnstone, 1991 ; Perkins & Grotzer, 2005 ). Mechanisms, processes, and behavior of complex systems present particular challenges. Learners must master not only the individual components of the system or process (structure) but also the interactions and mechanisms (function), which may be complex and frequently invisible. If the phenomena are macroscopic, sub-microscopic, or abstract, there is an additional level of difficulty. Although the teaching of STEM phenomena typically relies on visualizations, such as pictures, graphs, and diagrams, learning is typically revealed in words, both spoken and written. Visualizations have many advantages over verbal explanations for teaching; can creating visual explanations promote learning?

Learning from visual representations in STEM

Given the inherent challenges in teaching and learning complex or invisible processes in science, educators have developed ways of representing these processes to enable and enhance student understanding. External visual representations, including diagrams, photographs, illustrations, flow charts, and graphs, are often used in science to both illustrate and explain concepts (e.g., Hegarty, Carpenter, & Just, 1990 ; Mayer, 1989 ). Visualizations can directly represent many structural and behavioral properties. They also help to draw inferences (Larkin & Simon, 1987 ), find routes in maps (Levine, 1982 ), spot trends in graphs (Kessell & Tversky, 2011 ; Zacks & Tversky, 1999 ), imagine traffic flow or seasonal changes in light from architectural sketches (e.g. Tversky & Suwa, 2009 ), and determine the consequences of movements of gears and pulleys in mechanical systems (e.g. Hegarty & Just, 1993 ; Hegarty, Kriz, & Cate, 2003 ). The use of visual elements such as arrows is another benefit to learning with visualizations. Arrows are widely produced and comprehended as representing a range of kinds of forces as well as changes over time (e.g. Heiser & Tversky, 2002 ; Tversky, Heiser, MacKenzie, Lozano, & Morrison, 2007 ). Visualizations are thus readily able to depict the parts and configurations of systems; presenting the same content via language may be more difficult. Although words can describe spatial properties, because the correspondences of meaning to language are purely symbolic, comprehension and construction of mental representations from descriptions is far more effortful and error prone (e.g. Glenberg & Langston, 1992 ; Hegarty & Just, 1993 ; Larkin & Simon, 1987 ; Mayer, 1989 ). Given the differences in how visual and verbal information is processed, how learners draw inferences and construct understanding in these two modes warrants further investigation.

Benefits of generating explanations

Learner-generated explanations of scientific phenomena may be an important learning strategy to consider beyond the utility of learning from a provided external visualization. Explanations convey information about concepts or processes with the goal of making clear and comprehensible an idea or set of ideas. Explanations may involve a variety of elements, such as the use of examples and analogies (Roscoe & Chi, 2007 ). When explaining something new, learners may have to think carefully about the relationships between elements in the process and prioritize the multitude of information available to them. Generating explanations may require learners to reorganize their mental models by allowing them to make and refine connections between and among elements and concepts. Explaining may also help learners metacognitively address their own knowledge gaps and misconceptions.

Many studies have shown that learning is enhanced when students are actively engaged in creative, generative activities (e.g. Chi, 2009 ; Hall, Bailey, & Tillman, 1997 ). Generative activities have been shown to benefit comprehension of domains involving invisible components, including electric circuits (Johnson & Mayer, 2010 ) and the chemistry of detergents (Schwamborn, Mayer, Thillmann, Leopold, & Leutner, 2010 ). Wittrock’s ( 1990 ) generative theory stresses the importance of learners actively constructing and developing relationships. Generative activities require learners to select information and choose how to integrate and represent the information in a unified way. When learners make connections between pieces of information, knowledge, and experience, by generating headings, summaries, pictures, and analogies, deeper understanding develops.

The information learners draw upon to construct their explanations is likely important. For example, Ainsworth and Loizou ( 2003 ) found that asking participants to self-explain with a diagram resulted in greater learning than self-explaining from text. How might learners explain with physical mechanisms or materials with multi-modal information?

Generating visual explanations

Learner-generated visualizations have been explored in several domains. Gobert and Clement ( 1999 ) investigated the effectiveness of student-generated diagrams versus student-generated summaries on understanding plate tectonics after reading an expository text. Students who generated diagrams scored significantly higher on a post-test measuring spatial and causal/dynamic content, even though the diagrams contained less domain-related information. Hall et al. ( 1997 ) showed that learners who generated their own illustrations from text performed equally as well as learners provided with text and illustrations. Both groups outperformed learners only provided with text. In a study concerning the law of conservation of energy, participants who generated drawings scored higher on a post-test than participants who wrote their own narrative of the process (Edens & Potter, 2003 ). In addition, the quality and number of concept units present in the drawing/science log correlated with performance on the post-test. Van Meter ( 2001 ) found that drawing while reading a text about Newton’s Laws was more effective than answering prompts in writing.

One aspect to explore is whether visual and verbal productions contain different types of information. Learning advantages for the generation of visualizations could be attributed to learners’ translating across modalities, from a verbal format into a visual format. Translating verbal information from the text into a visual explanation may promote deeper processing of the material and more complete and comprehensive mental models (Craik & Lockhart, 1972 ). Ainsworth and Iacovides ( 2005 ) addressed this issue by asking two groups of learners to self-explain while learning about the circulatory system of the human body. Learners given diagrams were asked to self-explain in writing and learners given text were asked to explain using a diagram. The results showed no overall differences in learning outcomes, however the learners provided text included significantly more information in their diagrams than the other group. Aleven and Koedinger ( 2002 ) argue that explanations are most helpful if they can integrate visual and verbal information. Translating across modalities may serve this purpose, although translating is not necessarily an easy task (Ainsworth, Bibby, & Wood, 2002 ).

It is important to remember that not all studies have found advantages to generating explanations. Wilkin ( 1997 ) found that directions to self-explain using a diagram hindered understanding in examples in physical motion when students were presented with text and instructed to draw a diagram. She argues that the diagrams encouraged learners to connect familiar but unrelated knowledge. In particular, “low benefit learners” in her study inappropriately used spatial adjacency and location to connect parts of diagrams, instead of the particular properties of those parts. Wilkin argues that these learners are novices and that experts may not make the same mistake since they have the skills to analyze features of a diagram according to their relevant properties. She also argues that the benefits of self-explaining are highest when the learning activity is constrained so that learners are limited in their possible interpretations. Other studies that have not found a learning advantage from generating drawings have in common an absence of support for the learner (Alesandrini, 1981 ; Leutner, Leopold, & Sumfleth, 2009 ). Another mediating factor may be the learner’s spatial ability.

The role of spatial ability

Spatial thinking involves objects, their size, location, shape, their relation to one another, and how and where they move through space. How then, might learners with different levels of spatial ability gain structural and functional understanding in science and how might this ability affect the utility of learner-generated visual explanations? Several lines of research have sought to explore the role of spatial ability in learning science. Kozhevnikov, Hegarty, and Mayer ( 2002 ) found that low spatial ability participants interpreted graphs as pictures, whereas high spatial ability participants were able to construct more schematic images and manipulate them spatially. Hegarty and Just ( 1993 ) found that the ability to mentally animate mechanical systems correlated with spatial ability, but not verbal ability. In their study, low spatial ability participants made more errors in movement verification tasks. Leutner et al. ( 2009 ) found no effect of spatial ability on the effectiveness of drawing compared to mentally imagining text content. Mayer and Sims ( 1994 ) found that spatial ability played a role in participants’ ability to integrate visual and verbal information presented in an animation. The authors argue that their results can be interpreted within the context of dual-coding theory. They suggest that low spatial ability participants must devote large amounts of cognitive effort into building a visual representation of the system. High spatial ability participants, on the other hand, are more able to allocate sufficient cognitive resources to building referential connections between visual and verbal information.

Benefits of testing

Although not presented that way, creating an explanation could be regarded as a form of testing. Considerable research has documented positive effects of testing on learning. Presumably taking a test requires retrieving and sometimes integrating the learned material and those processes can augment learning without additional teaching or study (e.g. Roediger & Karpicke, 2006 ; Roediger, Putnam, & Smith, 2011 ; Wheeler & Roediger, 1992 ). Hausmann and Vanlehn ( 2007 ) addressed the possibility that generating explanations is beneficial because learners merely spend more time with the content material than learners who are not required to generate an explanation. In their study, they compared the effects of using instructions to self-explain with instructions to merely paraphrase physics (electrodynamics) material. Attending to provided explanations by paraphrasing was not as effective as generating explanations as evidenced by retention scores on an exam 29 days after the experiment and transfer scores within and across domains. Their study concludes, “the important variable for learning was the process of producing an explanation” (p. 423). Thus, we expect benefits from creating either kind of explanation but for the reasons outlined previously, we expect larger benefits from creating visual explanations.

Present experiments

This study set out to answer a number of related questions about the role of learner-generated explanations in learning and understanding of invisible processes. (1) Do students learn more when they generate visual or verbal explanations? We anticipate that learning will be greater with the creation of visual explanations, as they encourage completeness and the integration of structure and function. (2) Does the inclusion of structural and functional information correlate with learning as measured by a post-test? We predict that including greater counts of information, particularly invisible and functional information, will positively correlate with higher post-test scores. (3) Does spatial ability predict the inclusion of structural and functional information in explanations, and does spatial ability predict post-test scores? We predict that high spatial ability participants will include more information in their explanations, and will score higher on post-tests.

Experiment 1

The first experiment examines the effects of creating visual or verbal explanations on the comprehension of a bicycle tire pump’s operation in participants with low and high spatial ability. Although the pump itself is not invisible, the components crucial to its function, notably the inlet and outlet valves, and the movement of air, are located inside the pump. It was predicted that visual explanations would include more information than verbal explanations, particularly structural information, since their construction encourages completeness and the production of a whole mechanical system. It was also predicted that functional information would be biased towards a verbal format, since much of the function of the pump is hidden and difficult to express in pictures. Finally, it was predicted that high spatial ability participants would be able to produce more complete explanations and would thus also demonstrate better performance on the post-test. Explanations were coded for structural and functional content, essential features, invisible features, arrows, and multiple steps.

Participants

Participants were 127 (59 female) seventh and eighth grade students, aged 12–14 years, enrolled in an independent school in New York City. The school’s student body is 70% white, 30% other ethnicities. Approximately 25% of the student body receives financial aid. The sample consisted of three class sections of seventh grade students and three class sections of eighth grade students. Both seventh and eighth grade classes were integrated science (earth, life, and physical sciences) and students were not grouped according to ability in any section. Written parental consent was obtained by means of signed informed consent forms. Each participant was randomly assigned to one of two conditions within each class. There were 64 participants in the visual condition explained the bicycle pump’s function by drawing and 63 participants explained the pump’s function by writing.

The materials consisted of a 12-inch Spalding bicycle pump, a blank 8.5 × 11 in. sheet of paper, and a post-test (Additional file 1 ). The pump’s chamber and hose were made of clear plastic; the handle and piston were black plastic. The parts of the pump (e.g. inlet valve, piston) were labeled.

Spatial ability was assessed using the Vandenberg and Kuse ( 1978 ) mental rotation test (MRT). The MRT is a 20-item test in which two-dimensional drawings of three-dimensional objects are compared. Each item consists of one “target” drawing and four drawings that are to be compared to the target. Two of the four drawings are rotated versions of the target drawing and the other two are not. The task is to identify the two rotated versions of the target. A score was determined by assigning one point to each question if both of the correct rotated versions were chosen. The maximum score was 20 points.

The post-test consisted of 16 true/false questions printed on a single sheet of paper measuring 8.5 × 11 in. Half of the questions related to the structure of the pump and the other half related to its function. The questions were adapted from Heiser and Tversky ( 2002 ) in order to be clear and comprehensible for this age group.

The experiment was conducted over the course of two non-consecutive days during the normal school day and during regularly scheduled class time. On the first day, participants completed the MRT as a whole-class activity. After completing an untimed practice test, they were given 3 min for each of the two parts of the MRT. On the second day, occurring between two and four days after completing the MRT, participants were individually asked to study an actual bicycle tire pump and were then asked to generate explanations of its function. The participants were tested individually in a quiet room away from the rest of the class. In addition to the pump, each participant was one instruction sheet and one blank sheet of paper for their explanations. The post-test was given upon completion of the explanation. The instruction sheet was read aloud to participants and they were instructed to read along. The first set of instructions was as follows: “A bicycle pump is a mechanical device that pumps air into bicycle tires. First, take this bicycle pump and try to understand how it works. Spend as much time as you need to understand the pump.” The next set of instructions differed for participants in each condition. The instructions for the visual condition were as follows: “Then, we would like you to draw your own diagram or set of diagrams that explain how the bike pump works. Draw your explanation so that someone else who has not seen the pump could understand the bike pump from your explanation. Don’t worry about the artistic quality of the diagrams; in fact, if something is hard for you to draw, you can explain what you would draw. What’s important is that the explanation should be primarily visual, in a diagram or diagrams.” The instructions for the verbal condition were as follows: “Then, we would like you to write an explanation of how the bike pump works. Write your explanation so that someone else who has not seen the pump could understand the bike pump from your explanation.” All participants then received these instructions: “You may not use the pump while you create your explanations. Please return it to me when you are ready to begin your explanation. When you are finished with the explanation, you will hand in your explanation to me and I will then give you 16 true/false questions about the bike pump. You will not be able to look at your explanation while you complete the questions.” Study and test were untimed. All students finished within the 45-min class period.

• Spatial ability

The mean score on the MRT was 10.56, with a median of 11. Boys scored significantly higher (M = 13.5, SD = 4.4) than girls (M = 8.8, SD = 4.5), F(1, 126) = 19.07, p  < 0.01, a typical finding (Voyer, Voyer, & Bryden, 1995 ). Participants were split into high or low spatial ability by the median. Low and high spatial ability participants were equally distributed in the visual and verbal groups.

Learning outcomes

It was predicted that high spatial ability participants would be better able to mentally animate the bicycle pump system and therefore score higher on the post-test and that post-test scores would be higher for those who created visual explanations. Table  1 shows the scores on the post-test by condition and spatial ability. A two-way factorial ANOVA revealed marginally significant main effect of spatial ability F(1, 124) = 3.680, p  = 0.06, with high spatial ability participants scoring higher on the post-test. There was also a significant interaction between spatial ability and explanation type F(1, 124) = 4.094, p  < 0.01, see Fig.  1 . Creating a visual explanation of the bicycle pump selectively helped low spatial participants.

Scores on the post-test by condition and spatial ability

Coding explanations

Explanations (see Fig.  2 ) were coded for structural and functional content, essential features, invisible features, arrows, and multiple steps. A subset of the explanations (20%) was coded by the first author and another researcher using the same coding system as a guide. The agreement between scores was above 90% for all measures. Disagreements were resolved through discussion. The first author then scored the remaining explanations.

Examples of visual and verbal explanations of the bicycle pump

Coding for structure and function

A maximum score of 12 points was awarded for the inclusion and labeling of six structural components: chamber, piston, inlet valve, outlet valve, handle, and hose. For the visual explanations, 1 point was given for a component drawn correctly and 1 additional point if the component was labeled correctly. For verbal explanations, sentences were divided into propositions, the smallest unit of meaning in a sentence. Descriptions of structural location e.g. “at the end of the piston is the inlet valve,” or of features of the components, e.g. the shape of a part, counted as structural components. Information was coded as functional if it depicted (typically with an arrow) or described the function/movement of an individual part, or the way multiple parts interact. No explanation contained more than ten functional units.

Visual explanations contained significantly more structural components (M = 6.05, SD = 2.76) than verbal explanations (M = 4.27, SD = 1.54), F(1, 126) = 20.53, p  < 0.05. The number of functional components did not differ between visual and verbal explanations as displayed in Figs.  3 and 4 . Many visual explanations (67%) contained verbal components; the structural and functional information in explanations was coded as depictive or descriptive. Structural and functional information were equally likely to be expressed in words or pictures in visual explanations. It was predicted that explanations created by high spatial participants would include more functional information. However, there were no significant differences found between low spatial (M = 5.15, SD = 2.21) and high spatial (M = 4.62, SD = 2.16) participants in the number of structural units or between low spatial (M = 3.83, SD = 2.51) and high spatial (M = 4.10, SD = 2.13) participants in the number of functional units.

Average number of structural and functional components in visual and verbal explanations

Visual and verbal explanations of chemical bonding

Coding of essential features

To further establish a relationship between the explanations generated and outcomes on the post-test, explanations were also coded for the inclusion of information essential to its function according to a 4-point scale (adapted from Hall et al., 1997 ). One point was given if both the inlet and the outlet valve were clearly present in the drawing or described in writing, 1 point was given if the piston inserted into the chamber was shown or described to be airtight, and 1 point was given for each of the two valves if they were shown or described to be opening/closing in the correct direction.

Visual explanations contained significantly more essential information (M = 1.78, SD = 1.0) than verbal explanations (M = 1.20, SD = 1.21), F(1, 126) = 7.63, p  < 0.05. Inclusion of essential features correlated positively with post-test scores, r = 0.197, p  < 0.05).

Coding arrows and multiple steps

For the visual explanations, three uses of arrows were coded and tallied: labeling a part or action, showing motion, or indicating sequence. Analysis of visual explanations revealed that 87% contained arrows. No significant differences were found between low and high spatial participants’ use of arrows to label and no signification correlations were found between the use of arrows and learning outcomes measured on the post-test.

The explanations were coded for the number of discrete steps used to explain the process of using the bike pump. The number of steps used by participants ranged from one to six. Participants whose explanations, whether verbal or visual, contained multiple steps scored significantly higher (M = 0.76, SD = 0.18) on the post-test than participants whose explanations consisted of a single step (M = 0.67, SD = 0.19), F(1, 126) = 5.02, p  < 0.05.

Coding invisible features

The bicycle tire pump, like many mechanical devices, contains several structural features that are hidden or invisible and must be inferred from the function of the pump. For the bicycle pump the invisible features are the inlet and outlet valves and the three phases of movement of air, entering the pump, moving through the pump, exiting the pump. Each feature received 1 point for a total of 5 possible points.

The mean score for the inclusion of invisible features was 3.26, SD = 1.25. The data were analyzed using linear regression and revealed that the total score for invisible parts significantly predicted scores on the post-test, F(1, 118) = 3.80, p  = 0.05.

In the first experiment, students learned the workings of a bicycle pump from interacting with an actual pump and creating a visual or verbal explanation of its function. Understanding the functionality of a bike pump depends on the actions and consequences of parts that are not visible. Overall, the results provide support for the use of learner-generated visual explanations in developing understanding of a new scientific system. The results show that low spatial ability participants were able to learn as successfully as high spatial ability participants when they first generated an explanation in a visual format.

Visual explanations may have led to greater understanding for a number of reasons. As discussed previously, visual explanations encourage completeness. They force learners to decide on the size, shape, and location of parts/objects. Understanding the “hidden” function of the invisible parts is key to understanding the function of the entire system and requires an understanding of how both the visible and invisible parts interact. The visual format may have been able to elicit components and concepts that are invisible and difficult to integrate into the formation of a mental model. The results show that including more of the essential features and showing multiple steps correlated with superior test performance. Understanding the bicycle pump requires understanding how all of these components are connected through movement, force, and function. Many (67%) of the visual explanations also contained written components to accompany their explanation. Arguably, some types of information may be difficult to depict visually and verbal language has many possibilities that allow for specificity. The inclusion of text as a complement to visual explanations may be key to the success of learner-generated explanations and the development of understanding.

A limitation of this experiment is that participants were not provided with detailed instructions for completing their explanations. In addition, this experiment does not fully clarify the role of spatial ability, since high spatial participants in the visual and verbal groups demonstrated equivalent knowledge of the pump on the post-test. One possibility is that the interaction with the bicycle pump prior to generating explanations was a sufficient learning experience for the high spatial participants. Other researchers (e.g. Flick, 1993 ) have shown that hands-on interactive experiences can be effective learning situations. High spatial ability participants may be better able to imagine the movement and function of a system (e.g. Hegarty, 1992 ).

Experiment 1 examined learning a mechanical system with invisible (hidden) parts. Participants were introduced to the system by being able to interact with an actual bicycle pump. While we did not assess participants’ prior knowledge of the pump with a pre-test, participants were randomly assigned to each condition. The findings have promising implications for teaching. Creating visual explanations should be an effective way to improve performance, especially in low spatial students. Instructors can guide the creation of visual explanations toward the features that augment learning. For example, students can be encouraged to show every step and action and to focus on the essential parts, even if invisible. The coding system shows that visual explanations can be objectively evaluated to provide feedback on students’ understanding. The utility of visual explanations may differ for scientific phenomena that are more abstract, or contain elements that are invisible due to their scale. Experiment 2 addresses this possibility by examining a sub-microscopic area of science: chemical bonding.

Experiment 2

In this experiment, we examine visual and verbal explanations in an area of chemistry: ionic and covalent bonding. Chemistry is often regarded as a difficult subject; one of the essential or inherent features of chemistry which presents difficulty is the interplay between the macroscopic, sub-microscopic, and representational levels (e.g. Bradley & Brand, 1985 ; Johnstone, 1991 ; Taber, 1997 ). In chemical bonding, invisible components engage in complex processes whose scale makes them impossible to observe. Chemists routinely use visual representations to investigate relationships and move between the observable, physical level and the invisible particulate level (Kozma, Chin, Russell, & Marx, 2002 ). Generating explanations in a visual format may be a particularly useful learning tool for this domain.

For this topic, we expect that creating a visual rather than verbal explanation will aid students of both high and low spatial abilities. Visual explanations demand completeness; they were predicted to include more information than verbal explanations, particularly structural information. The inclusion of functional information should lead to better performance on the post-test since understanding how and why atoms bond is crucial to understanding the process. Participants with high spatial ability may be better able to explain function since the sub-microscopic nature of bonding requires mentally imagining invisible particles and how they interact. This experiment also asks whether creating an explanation per se can increase learning in the absence of additional teaching by administering two post-tests of knowledge, one immediately following instruction but before creating an explanation and one after creating an explanation. The scores on this immediate post-test were used to confirm that the visual and verbal groups were equivalent prior to the generation of explanations. Explanations were coded for structural and functional information, arrows, specific examples, and multiple representations. Do the acts of selecting, integrating, and explaining knowledge serve learning even in the absence of further study or teaching?

Participants were 126 (58 female) eighth grade students, aged 13–14 years, with written parental consent and enrolled in the same independent school described in Experiment 1. None of the students previously participated in Experiment 1. As in Experiment 1, randomization occurred within-class, with participants assigned to either the visual or verbal explanation condition.

The materials consisted of the MRT (same as Experiment 1), a video lesson on chemical bonding, two versions of the instructions, the immediate post-test, the delayed post-test, and a blank page for the explanations. All paper materials were typed on 8.5 × 11 in. sheets of paper. Both immediate and delayed post-tests consisted of seven multiple-choice items and three free-response items. The video lesson on chemical bonding consisted of a video that was 13 min 22 s. The video began with a brief review of atoms and their structure and introduced the idea that atoms combine to form molecules. Next, the lesson showed that location in the periodic table reveals the behavior and reactivity of atoms, in particular the gain, loss, or sharing of electrons. Examples of atoms, their valence shell structure, stability, charges, transfer and sharing of electrons, and the formation of ionic, covalent, and polar covalent bonds were discussed. The example of NaCl (table salt) was used to illustrate ionic bonding and the examples of O 2 and H 2 O (water) were used to illustrate covalent bonding. Information was presented verbally, accompanied by drawings, written notes of keywords and terms, and a color-coded periodic table.

On the first of three non-consecutive school days, participants completed the MRT as a whole-class activity. On the second day (occurring between two and three days after completing the MRT), participants viewed the recorded lesson on chemical bonding. They were instructed to pay close attention to the material but were not allowed to take notes. Immediately following the video, participants had 20 min to complete the immediate post-test; all finished within this time frame. On the third day (occurring on the next school day after viewing the video and completing the immediate post-test), the participants were randomly assigned to either the visual or verbal explanation condition. The typed instructions were given to participants along with a blank 8.5 × 11 in. sheet of paper for their explanations. The instructions differed for each condition. For the visual condition, the instructions were as follows: “You have just finished learning about chemical bonding. On the next piece of paper, draw an explanation of how atoms bond and how ionic and covalent bonds differ. Draw your explanation so that another student your age who has never studied this topic will be able to understand it. Be as clear and complete as possible, and remember to use pictures/diagrams only. After you complete your explanation, you will be asked to answer a series of questions about bonding.”

For the verbal condition the instructions were: “You have just finished learning about chemical bonding. On the next piece of paper, write an explanation of how atoms bond and how ionic and covalent bonds differ. Write your explanation so that another student your age who has never studied this topic will be able to understand it. Be as clear and complete as possible. After you complete your explanation, you will be asked to answer a series of questions about bonding.”

Participants were instructed to read the instructions carefully before beginning the task. The participants completed their explanations as a whole-class activity. Participants were given unlimited time to complete their explanations. Upon completion of their explanations, participants were asked to complete the ten-question delayed post-test (comparable to but different from the first) and were given a maximum of 20 min to do so. All participants completed their explanations as well as the post-test during the 45-min class period.

The mean score on the MRT was 10.39, with a median of 11. Boys (M = 12.5, SD = 4.8) scored significantly higher than girls (M = 8.0, SD = 4.0), F(1, 125) = 24.49, p  < 0.01. Participants were split into low and high spatial ability based on the median.

The maximum score for both the immediate and delayed post-test was 10 points. A repeated measures ANOVA showed that the difference between the immediate post-test scores (M = 4.63, SD = 0.469) and delayed post-test scores (M = 7.04, SD = 0.299) was statistically significant F(1, 125) = 18.501, p  < 0.05). Without any further instruction, scores increased following the generation of a visual or verbal explanation. Both groups improved significantly; those who created visual explanations (M = 8.22, SD = 0.208), F(1, 125) = 51.24, p  < 0.01, Cohen’s d  = 1.27 as well as those who created verbal explanations (M = 6.31, SD = 0.273), F(1,125) = 15.796, p  < 0.05, Cohen’s d  = 0.71. As seen in Fig.  5 , participants who generated visual explanations (M = 0.822, SD = 0.208) scored considerably higher on the delayed post-test than participants who generated verbal explanations (M = 0.631, SD = 0.273), F(1, 125) = 19.707, p  < 0.01, Cohen’s d  = 0.88. In addition, high spatial participants (M = 0.824, SD = 0.273) scored significantly higher than low spatial participants (M = 0.636, SD = 0.207), F(1, 125) = 19.94, p  < 0.01, Cohen’s d  = 0.87. The results of the test of the interaction between group and spatial ability was not significant.

Scores on the post-tests by explanation type and spatial ability

Explanations were coded for structural and functional content, arrows, specific examples, and multiple representations. A subset of the explanations (20%) was coded by both the first author and a middle school science teacher with expertise in Chemistry. Both scorers used the same coding system as a guide. The percentage of agreement between scores was above 90 for all measures. The first author then scored the remainder of the explanations. As evident from Fig.  4 , the visual explanations were individual inventions; they neither resembled each other nor those used in teaching. Most contained language, especially labels and symbolic language such as NaCl.

Structure, function, and modality

Visual and verbal explanations were coded for depicting or describing structural and functional components. The structural components included the following: the correct number of valence electrons, the correct charges of atoms, the bonds between non-metals for covalent molecules and between a metal and non-metal for ionic molecules, the crystalline structure of ionic molecules, and that covalent bonds were individual molecules. The functional components included the following: transfer of electrons in ionic bonds, sharing of electrons in covalent bonds, attraction between ions of opposite charge, bonding resulting in atoms with neutral charge and stable electron shell configurations, and outcome of bonding shows molecules with overall neutral charge. The presence of each component was awarded 1 point; the maximum possible points was 5 for structural and 5 for functional information. The modality, visual or verbal, of each component was also coded; if the information was given in both formats, both were coded.

As displayed in Fig.  6 , visual explanations contained a significantly greater number of structural components (M = 2.81, SD = 1.56) than verbal explanations (M = 1.30, SD = 1.54), F(1, 125) = 13.69, p  < 0.05. There were no differences between verbal and visual explanations in the number of functional components. Structural information was more likely to be depicted (M = 3.38, SD = 1.49) than described (M = 0.429, SD = 1.03), F(1, 62) = 21.49, p  < 0.05, but functional information was equally likely to be depicted (M = 1.86, SD = 1.10) or described (M = 1.71, SD = 1.87).

Functional information expressed verbally in the visual explanations significantly predicted scores on the post-test, F(1, 62) = 21.603, p  < 0.01, while functional information in verbal explanations did not. The inclusion of structural information did not significantly predict test scores. As seen Fig.  7 , explanations created by high spatial participants contained significantly more functional components, F(1, 125) = 7.13, p  < 0.05, but there were no ability differences in the amount of structural information created by high spatial participants in either visual or verbal explanations.

Average number of structural and functional components created by low and high spatial ability learners

Ninety-two percent of visual explanations contained arrows. Arrows were used to indicate motion as well as to label. The use of arrows was positively correlated with scores on the post-test, r = 0.293, p  < 0.05. There were no significant differences in the use of arrows between low and high spatial participants.

Specific examples

Explanations were coded for the use of specific examples, such as NaCl, to illustrate ionic bonding and CO 2 and O 2 to illustrate covalent bonding. High spatial participants (M = 1.6, SD = 0.69) used specific examples in their verbal and visual explanations more often than low spatial participants (M = 1.07, SD = 0.79), a marginally significant effect F(1, 125) = 3.65, p  = 0.06. Visual and verbal explanations did not differ in the presence of specific examples. The inclusion of a specific example was positively correlated with delayed test scores, r = 0.555, p  < 0.05.

Use of multiple representations

Many of the explanations (65%) contained multiple representations of bonding. For example, ionic bonding and its properties can be represented at the level of individual atoms or at the level of many atoms bonded together in a crystalline compound. The representations that were coded were as follows: symbolic (e.g. NaCl), atomic (showing structure of atom(s), and macroscopic (visible). Participants who created visual explanations generated significantly more (M =1.79, SD = 1.20) than those who created verbal explanations (M = 1.33, SD = 0.48), F (125) = 6.03, p  < 0.05. However, the use of multiple representations did not significantly correlate with delayed post-test scores on the delayed post-test.

Metaphoric explanations

Although there were too few examples to be included in the statistical analyses, some participants in the visual group created explanations that used metaphors and/or analogies to illustrate the differences between the types of bonding. Figure  4 shows examples of metaphoric explanations. In one example, two stick figures are used to show “transfer” and “sharing” of an object between people. In another, two sharks are used to represent sodium and chlorine, and the transfer of fish instead of electrons.

In the second experiment, students were introduced to chemical bonding, a more abstract and complex set of phenomena than the bicycle pump used in the first experiment. Students were tested immediately after instruction. The following day, half the students created visual explanations and half created verbal explanations. Following creation of the explanations, students were tested again, with different questions. Performance was considerably higher as a consequence of creating either explanation despite the absence of new teaching. Generating an explanation in this way could be regarded as a test of learning. Seen this way, the results echo and amplify previous research showing the advantages of testing over study (e.g. Roediger et al., 2011 ; Roediger & Karpicke, 2006 ; Wheeler & Roediger, 1992 ). Specifically, creating an explanation requires selecting the crucial information, integrating it temporally and causally, and expressing it clearly, processes that seem to augment learning and understanding without additional teaching. Importantly, creating a visual explanation gave an extra boost to learning outcomes over and above the gains provided by creating a verbal explanation. This is most likely due to the directness of mapping complex systems to a visual-spatial format, a format that can also provide a natural check for completeness and coherence as well as a platform for inference. In the case of this more abstract and complex material, generating a visual explanation benefited both low spatial and high spatial participants even if it did not bring low spatial participants up to the level of high spatial participants as for the bicycle pump.

Participants high in spatial ability not only scored better, they also generated better explanations, including more of the information that predicted learning. Their explanations contained more functional information and more specific examples. Their visual explanations also contained more functional information.

As in Experiment 1, qualities of the explanations predicted learning outcomes. Including more arrows, typically used to indicate function, predicted delayed test scores as did articulating more functional information in words in visual explanations. Including more specific examples in both types of explanation also improved learning outcomes. These are all indications of deeper understanding of the processes, primarily expressed in the visual explanations. As before, these findings provide ways that educators can guide students to craft better visual explanations and augment learning.

General discussion

Two experiments examined how learner-generated explanations, particularly visual explanations, can be used to increase understanding in scientific domains, notably those that contain “invisible” components. It was proposed that visual explanations would be more effective than verbal explanations because they encourage completeness and coherence, are more explicit, and are typically multimodal. These two experiments differ meaningfully from previous studies in that the information selected for drawing was not taken from a written text, but from a physical object (bicycle pump) and a class lesson with multiple representations (chemical bonding).

The results show that creating an explanation of a STEM phenomenon benefits learning, even when the explanations are created after learning and in the absence of new instruction. These gains in performance in the absence of teaching bear similarities to recent research showing gains in learning from testing in the absence of new instruction (e.g. Roediger et al., 2011 ; Roediger & Karpicke, 2006 ; Wheeler & Roediger, 1992 ). Many researchers have argued that the retrieval of information required during testing strengthens or enhances the retrieval process itself. Formulating explanations may be an especially effective form of testing for post-instruction learning. Creating an explanation of a complex system requires the retrieval of critical information and then the integration of that information into a coherent and plausible account. Other factors, such as the timing of the creation of the explanations, and whether feedback is provided to students, should help clarify the benefits of generating explanations and how they may be seen as a form of testing. There may even be additional benefits to learners, including increasing their engagement and motivation in school, and increasing their communication and reasoning skills (Ainsworth, Prain, & Tytler, 2011 ). Formulating a visual explanation draws upon students’ creativity and imagination as they actively create their own product.

As in previous research, students with high spatial ability both produced better explanations and performed better on tests of learning (e.g. Uttal et al., 2013 ). The visual explanations of high spatial students contained more information and more of the information that predicts learning outcomes. For the workings of a bicycle pump, creating a visual as opposed to verbal explanation had little impact on students of high spatial ability but brought students of lower spatial ability up to the level of students with high spatial abilities. For the more difficult set of concepts, chemical bonding, creating a visual explanation led to much larger gains than creating a verbal one for students both high and low in spatial ability. It is likely a mistake to assume that how and high spatial learners will remain that way; there is evidence that spatial ability develops with experience (Baenninger & Newcombe, 1989 ). It is possible that low spatial learners need more support in constructing explanations that require imagining the movement and manipulation of objects in space. Students learned the function of the bike pump by examining an actual pump and learned bonding through a video presentation. Future work to investigate methods of presenting material to students may also help to clarify the utility of generating explanations.

Creating visual explanations had greater benefits than those accruing from creating verbal ones. Surely some of the effectiveness of visual explanations is because they represent and communicate more directly than language. Elements of a complex system can be depicted and arrayed spatially to reflect actual or metaphoric spatial configurations of the system parts. They also allow, indeed, encourage, the use of well-honed spatial inferences to substitute for and support abstract inferences (e.g. Larkin & Simon, 1987 ; Tversky, 2011 ). As noted, visual explanations provide checks for completeness and coherence, that is, verification that all the necessary elements of the system are represented and that they work together properly to produce the outcomes of the processes. Visual explanations also provide a concrete reference for making and checking inferences about the behavior, causality, and function of the system. Thus, creating a visual explanation facilitates the selection and integration of information underlying learning even more than creating a verbal explanation.

Creating visual explanations appears to be an underused method of supporting and evaluating students’ understanding of dynamic processes. Two obstacles to using visual explanations in classrooms seem to be developing guidelines for creating visual explanations and developing objective scoring systems for evaluating them. The present findings give insights into both. Creating a complete and coherent visual explanation entails selecting the essential components and linking them by behavior, process, or causality. This structure and organization is familiar from recipes or construction sets: first the ingredients or parts, then the sequence of actions. It is also the ingredients of theater or stories: the players and their actions. In fact, the creation of visual explanations can be practiced on these more familiar cases and then applied to new ones in other domains. Deconstructing and reconstructing knowledge and information in these ways has more generality than visual explanations: these techniques of analysis serve thought and provide skills and tools that underlie creative thought. Next, we have shown that objective scoring systems can be devised, beginning with separating the information into structure and function, then further decomposing the structure into the central parts or actors and the function into the qualities of the sequence of actions and their consequences. Assessing students’ prior knowledge and misconceptions can also easily be accomplished by having students create explanations at different times in a unit of study. Teachers can see how their students’ ideas change and if students can apply their understanding by analyzing visual explanations as a culminating activity.

Creating visual explanations of a range of phenomena should be an effective way to augment students’ spatial thinking skills, thereby increasing the effectiveness of these explanations as spatial ability increases. The proverbial reading, writing, and arithmetic are routinely regarded as the basic curriculum of school learning and teaching. Spatial skills are not typically taught in schools, but should be: these skills can be learned and are essential to functioning in the contemporary and future world (see Uttal et al., 2013 ). In our lives, both daily and professional, we need to understand the maps, charts, diagrams, and graphs that appear in the media and public places, with our apps and appliances, in forms we complete, in equipment we operate. In particular, spatial thinking underlies the skills needed for professional and amateur understanding in STEM fields and knowledge and understanding STEM concepts is increasingly required in what have not been regarded as STEM fields, notably the largest employers, business, and service.

This research has shown that creating visual explanations has clear benefits to students, both specific and potentially general. There are also benefits to teachers, specifically, revealing misunderstandings and gaps in knowledge. Visualizations could be used by teachers as a formative assessment tool to guide further instructional activities and scoring rubrics could allow for the identification of specific misconceptions. The bottom line is clear. Creating a visual explanation is an excellent way to learn and master complex systems.

Ainsworth, S. E., Bibby, P. A., & Wood, D. J. (2002). Examining the effects of different multiple representational systems in learning primary mathematics. Journal of the Learning Sciences, 11 (1), 25–62.

Article   Google Scholar  

Ainsworth, S. E., & Iacovides, I. (2005). Learning by constructing self-explanation diagrams. Paper presented at the 11th Biennial Conference of European Association for Resarch on Learning and Instruction, Nicosia, Cyprus.

Ainsworth, S. E., & Loizou, A. T. (2003). The effects of self-explaining when learning with text or diagrams. Cognitive Science, 27 (4), 669–681.

Ainsworth, S., Prain, V., & Tytler, R. (2011). Drawing to learn in science. Science, 26 , 1096–1097.

Alesandrini, K. L. (1981). Pictorial-verbal and analytic-holistic learning strategies in science learning. Journal of Educational Psychology, 73 , 358–368.

Aleven, V. & Koedinger, K. R. (2002). An effective metacognitive strategy: learning by doing and explaining with a computer-based cognitive tutor. Cognitive Science , 26 , 147–179.

Baenninger, M., & Newcombe, N. (1989). The role of experience in spatial test performance: A meta-analysis. Sex Roles, 20 (5–6), 327–344.

Bradley, J. D., & Brand, M. (1985). Stamping out misconceptions. Journal of Chemical Education, 62 (4), 318.

Chi, M. T. (2009). Active-Constructive-Interactive: A conceptual framework for differentiating learning activities. Topics in Cognitive Science, 1 , 73–105.

Article   PubMed   Google Scholar  

Chi, M. T. H., DeLeeuw, N., Chiu, M., & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18 , 439–477.

Google Scholar  

Craik, F., & Lockhart, R. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11 , 671–684.

Edens, K. M., & Potter, E. (2003). Using descriptive drawings as a conceptual change strategy in elementary science. School Science and Mathematics, 103 (3), 135–144.

Flick, L. B. (1993). The meanings of hands-on science. Journal of Science Teacher Education, 4 , 1–8.

Glenberg, A. M., & Langston, W. E. (1992). Comprehension of illustrated text: Pictures help to build mental models. Journal of Memory and Language, 31 , 129–151.

Gobert, J. D., & Clement, J. J. (1999). Effects of student-generated diagrams versus student-generated summaries on conceptual understanding of causal and dynamic knowledge in plate tectonics. Journal of Research in Science Teaching, 36 , 39–53.

Hall, V. C., Bailey, J., & Tillman, C. (1997). Can student-generated illustrations be worth ten thousand words? Journal of Educational Psychology, 89 (4), 677–681.

Hausmann, R. G. M., & Vanlehn, K. (2007). Explaining self-explaining: A contrast between content and generation. In R. Luckin, K. R. Koedinger, & J. Greer (Eds.), Artificial intelligence in education: Building technology rich learning contexts that work (pp. 417–424). Amsterdam: Ios Press.

Hegarty, M. (1992). Mental animation: Inferring motion from static displays of mechanical systems. Journal of Experimental Psychology: Learning, Memory & Cognition, 18 , 1084–1102.

Hegarty, M., Carpenter, P.A., & Just, M.A. (1990). Diagrams in the comprehension of scientific text. In R. Barr, M.S. Kamil, P. Mosenthal, & P. D. Pearson (Eds.), Handbook of reading research (pp. 641–669). New York: Longman.

Hegarty, M., & Just, M. A. (1993). Constructing mental models of machines from text and diagrams. Journal of Memory and Language, 32 , 717–742.

Hegarty, M., Kriz, S., & Cate, C. (2003). The roles of mental animations and external animations in understanding mechanical systems. Cognition & Instruction, 21 (4), 325–360.

Heiser, J., & Tversky, B. (2002). Diagrams and descriptions in acquiring complex systems. Proceedings of the Cognitive Science Society . Hillsdale: Erlbaum.

Hmelo-Silver, C., & Pfeffer, M. G. (2004). Comparing expert and novice understanding of a complex system from the perspective of structures, behaviors, and functions. Cognitive Science, 28 , 127–138.

Johnson, C. I., & Mayer, R. E. (2010). Applying the self-explanation principle to multimedia learning in a computer-based game-like environment. Computers in Human Behavior, 26 , 1246–1252.

Johnstone, A. H. (1991). Why is science difficult to learn? Things are seldom what they seem. Journal of Chemical Education, 61 (10), 847–849.

Kessell, A. M., & Tversky, B. (2011). Visualizing space, time, and agents: Production, performance, and preference. Cognitive Processing, 12 , 43–52.

Kozhevnikov, M., Hegarty, M., & Mayer, R. (2002). Revising the Visualizer–Verbalizer Dimension: Evidence for Two Types of Visualizers. Cognition & Instruction, 20 , 37–77.

Kozma, R., Chin, E., Russell, J., & Marx, N. (2002). The roles of representations and tools in the chemistry laboratory and their implication for chemistry learning. Journal of the Learning Sciences, 9 (2), 105–143.

Larkin, J., & Simon, H. (1987). Why a diagram is (sometimes) worth ten thousand words. Cognitive Science, 11 , 65–100.

Leutner, D., Leopold, C., & Sumfleth, E. (2009). Cognitive load and science text comprehension: Effects of drawing and mentally imagining text content. Computers in Human Behavior, 25 , 284–289.

Levine, M. (1982). You-are-here maps: Psychological considerations. Environment and Behavior, 14 , 221–237.

Mayer, R. E. (1989). Systematic thinking fostered by illustrations in scientific text. Journal of Educational Psychology, 81 , 240–246.

Mayer, R. E., & Sims, V. K. (1994). For whom is a picture worth a thousand words? Extensions of a dual-coding theory of multimedia learning. Journal of Educational Psychology, 86 (3), 389–401.

Perkins, D. N., & Grotzer, T. A. (2005). Dimensions of causal understanding: The role of complex causal models in students’ understanding of science. Studies in Science Education, 41 , 117–166.

Roediger, H. L., & Karpicke, J. D. (2006). Test enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17 , 249–255.

Roediger, H. L., Putnam, A. L., & Smith, M. A. (2011). Ten benefits of testing and their applications to educational practice. In B. H. Ross (Ed.), The psychology of learning and motivation (pp. 1–36). New York: Elsevier.

Roscoe, R. D., & Chi, M. T. H. (2007). Understanding tutor learning: Knowledge-building and knowledge-telling in peer tutors’ explanations and questions. Review of Educational Research, 77 , 534–574.

Schwamborn, A., Mayer, R. E., Thillmann, H., Leopold, C., & Leutner, D. (2010). Drawing as a generative activity and drawing as a prognostic activity. Journal of Educational Psychology, 102 , 872–879.

Taber, K. S. (1997). Student understanding of ionic bonding: Molecular versus electrostatic framework? School Science Review, 78 (285), 85–95.

Tversky, B. (2011). Visualizing thought. Topics in Cognitive Science, 3 , 499–535.

Tversky, B., Heiser, J., MacKenzie, R., Lozano, S., & Morrison, J. B. (2007). Enriching animations. In R. Lowe & W. Schnotz (Eds.), Learning with animation: Research implications for design (pp. 263–285). New York: Cambridge University Press.

Tversky, B., & Suwa, M. (2009). Thinking with sketches. In A. B. Markman & K. L. Wood (Eds.), Tools for innovation (pp. 75–84). Oxford: Oxford University Press.

Chapter   Google Scholar  

Uttal, D. H., Meadow, N. G., Tipton, E., Hand, L. L., Alden, A. R., Warren, C., et al. (2013). The malleability of spatial skills: A meta-analysis of training studies. Psychological Bulletin, 139 , 352–402.

Van Meter, P. (2001). Drawing construction as a strategy for learning from text. Journal of Educational Psychology, 93 (1), 129–140.

Vandenberg, S. G., & Kuse, A. R. (1978). Mental rotations: A group test of three-dimensional spatial visualization. Perceptual Motor Skills, 47 , 599–604.

Voyer, D., Voyer, S., & Bryden, M. P. (1995). Magnitude of sex differences in spatial abilities: A meta-analysis and consideration of critical variables. Psychological Bulletin, 117 , 250–270.

Wheeler, M. A., & Roediger, H. L. (1992). Disparate effects of repeated testing: Reconciling Ballard’s (1913) and Bartlett’s (1932) results. Psychological Science, 3 , 240–245.

Wilkin, J. (1997). Learning from explanations: Diagrams can “inhibit” the self-explanation effect. In M. Anderson (Ed.), Reasoning with diagrammatic representations II . Menlo Park: AAAI Press.

Wittrock, M. C. (1990). Generative processes of comprehension. Educational Psychologist, 24 , 345–376.

Zacks, J., & Tversky, B. (1999). Bars and lines: A study of graphic communication. Memory and Cognition, 27 , 1073–1079.

Download references

Acknowledgments

The authors are indebted to the Varieties of Understanding Project at Fordham University and The John Templeton Foundation and to the following National Science Foundation grants for facilitating the research and/or preparing the manuscript: National Science Foundation NSF CHS-1513841, HHC 0905417, IIS-0725223, IIS-0855995, and REC 0440103. We are grateful to James E. Corter for his helpful suggestions and to Felice Frankel for her inspiration. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the funders. Please address correspondence to Barbara Tversky at the Columbia Teachers College, 525 W. 120th St., New York, NY 10025, USA. Email: [email protected].

Authors’ contributions

This research was part of EB’s doctoral dissertation under the advisement of BT. Both authors contributed to the design, analysis, and drafting of the manuscript. Both authors read and approved the final manuscript.

Competing interests

The author declares that they have no competing interests.

Author information

Authors and affiliations.

University of Massachusetts Lowell, Lowell, MA, USA

Eliza Bobek

Stanford University, Columbia University Teachers College, New York, NY, USA

Barbara Tversky

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Eliza Bobek .

Additional file

Additional file 1:.

Post-tests. (DOC 44 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Cite this article.

Bobek, E., Tversky, B. Creating visual explanations improves learning. Cogn. Research 1 , 27 (2016). https://doi.org/10.1186/s41235-016-0031-6

Download citation

Received : 12 January 2016

Accepted : 15 November 2016

Published : 07 December 2016

DOI : https://doi.org/10.1186/s41235-016-0031-6

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Visual communication
• Dynamic system
• Diagrammatic reasoning
• Complex system

Pretrained Visual Representations in Reinforcement Learning

• Conference: Towards Autonomous Robotic Systems: 25th Annual Conference, TAROS 2024
• At: London, Brunel University, UK
• This person is not on ResearchGate, or hasn't claimed this research yet.

• University of Lincoln

Abstract and Figures

Discover the world's research

• 25+ million members
• 160+ million publication pages
• 2.3+ billion citations
• Siddharth Karamcheti
• Annie S. Chen
• Percy Liang
• P Bojanowski
• A Dosovitskiy
• A Kolesnikov
• D Weissenborn
• T Unterthiner
• Adv Neural Inform Process Syst
• A Escontrela
• K Hartikainen
• I Kostrikov
• T P Lillicrap
• 17359-17371
• A Rajeswaran
• S Purushwalkam
• K Schmeckpeper
• K Daniilidis
• Recruit researchers
• Join for free
• Login Email Tip: Most researchers use their institutional email address as their ResearchGate login Password Forgot password? Keep me logged in Log in or Continue with Google Welcome back! Please log in. Email · Hint Tip: Most researchers use their institutional email address as their ResearchGate login Password Forgot password? Keep me logged in Log in or Continue with Google No account? Sign up

• Machine Learning
• Español – América Latina
• Português – Brasil
• Tiếng Việt
• ML Concepts
• Crash Course

Neural networks: Nodes and hidden layers

To build a neural network that learns nonlinearities , begin with the following familiar model structure: a linear model of the form $y' = b + w_1x_1 + w_2x_2 + w_3x_3$.

We can visualize this equation as shown below, where $x_1$, $x_2$, and $x_3$ are our three input nodes (in blue), and $y'$ is our output node (in green).

In the model above, the weight and bias values have been randomly initialized. Perform the following tasks to familiarize yourself with the interface and explore the linear model. You can ignore the Activation Function dropdown for now; we'll discuss this topic later on in the module.

Click the Play (▶️) button above the network to calculate the value of the output node for the input values $x_1 = 1.00$, $x_2 = 2.00$, and $x_3 = 3.00$.

Click the second node in the input layer , and increase the value from 2.00 to 2.50. Note that the value of the output node changes. Select the output nodes (in green) and review the Calculations panel to see how the output value was calculated.

Click the output node (in green) to see the weight ($w_1$, $w_2$, $w_3$) and bias ($b$) parameter values. Decrease the weight value for $w_3$ (again, note that the value of the output node and the calculations below have changed). Then, increase the bias value. Review how these changes have affected the model output.

Adding layers to the network

Note that when you adjusted the weight and bias values of the network in Exercise 1 , that didn't change the overall mathematical relationship between input and output. Our model is still a linear model.

But what if we add another layer to the network, in between the input layer and the output layer? In neural network terminology, additional layers between the input layer and the output layer are called hidden layers , and the nodes in these layers are called neurons .

The value of each neuron in the hidden layer is calculated the same way as the output of a linear model: take the sum of the product of each of its inputs (the neurons in the previous network layer) and a unique weight parameter, plus the bias. Similarly, the neurons in the next layer (here, the output layer) are calculated using the hidden layer's neuron values as inputs.

This new hidden layer allows our model to recombine the input data using another set of parameters. Can this help our model learn nonlinear relationships?

We've added a hidden layer containing four neurons to the model.

Click the Play (▶️) button above the network to calculate the value of the four hidden-layer nodes and the output node for the input values $x_1 = 1.00$, $x_2 = 2.00$, and $x_3 = 3.00$.

Then explore the model, and use it to answer the following questions.

Try modifying the model parameters, and observe the effect on the hidden-layer node values and the output value (you can review the Calculations panel below to see how these values were calculated).

Can this model learn nonlinearities?

If you click on each of the nodes in the hidden layer and review the calculations below, you'll see that all of them are linear (comprising multiplication and addition operations).

If you then click on the output node and review the calculation below, you'll see that this calculation is also linear. Linear calculations performed on the output of linear calculations are also linear, which means this model cannot learn nonlinearities.

Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License , and code samples are licensed under the Apache 2.0 License . For details, see the Google Developers Site Policies . Java is a registered trademark of Oracle and/or its affiliates.

Last updated 2024-08-16 UTC.

Retraction Note: Learning visual representations with optimum-path forest and its applications to Barrett’s esophagus and adenocarcinoma diagnosis

• Retraction Note
• Published: 14 August 2024

Cite this article

• Luis A. de Souza Jr. 1 ,
• Luis C. S. Afonso 1 ,
• Alanna Ebigbo 2 ,
• Andreas Probst 2 ,
• Helmut Messmann 2 ,
• Robert Mendel 3 ,
• Christian Hook 3 ,
• Christoph Palm 3 &
• João P. Papa   ORCID: orcid.org/0000-0002-6494-7514 4  

47 Accesses

Explore all metrics

The Original Article was published on 08 January 2019

Avoid common mistakes on your manuscript.

Retraction Note to: Neural Computing and Applications (2019) 32:759-775 https://doi.org/10.1007/s00521-018-03982-0

The Editor-in-Chief and the publisher have retracted this article. The article was submitted to be part of a guest-edited issue. An investigation by the publisher found a number of articles, including this one, with a number of concerns, including but not limited to compromised editorial handling and peer review process, inappropriate or irrelevant references or not being in scope of the journal or guest-edited issue. Based on the investigation's findings the Editor-in-Chief therefore no longer has confidence in the results and conclusions of this article.

João P. Papa disagrees with this retraction. Luis A. de Souza Jr., Luis C. S. Afonso, Andreas Probst, Helmut Messmann, Christian Hook, and Christoph Palm have not responded to correspondence regarding this retraction. The Publisher has not been able to obtain a current email address for authors Alanna Ebigbo and Robert Mendel.

Author information

Authors and affiliations.

Department of Computing, Federal University of São Carlos - UFScar, São Carlos, Brazil

Luis A. de Souza Jr. & Luis C. S. Afonso

Medizinische Klinik III, Klinikum Augsburg, Augsburg, Germany

Alanna Ebigbo, Andreas Probst & Helmut Messmann

Regensburg Medical Image Computing (ReMIC), Ostbayerische Technische Hochschule Regensburg - OTH Regensburg, Regensburg, Germany

Robert Mendel, Christian Hook & Christoph Palm

Department of Computing, São Paulo State University - UNESP, Bauru, Brazil

João P. Papa

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to João P. Papa .

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

About this article

de Souza, L.A., Afonso, L.C.S., Ebigbo, A. et al. Retraction Note: Learning visual representations with optimum-path forest and its applications to Barrett’s esophagus and adenocarcinoma diagnosis. Neural Comput & Applic (2024). https://doi.org/10.1007/s00521-024-10343-7

Download citation

Published : 14 August 2024

DOI : https://doi.org/10.1007/s00521-024-10343-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Find a journal
• Publish with us
• Track your research

Grab your spot at the free arXiv Accessibility Forum

Help | Advanced Search

Computer Science > Computer Vision and Pattern Recognition

Title: whitening consistently improves self-supervised learning.

Abstract: Self-supervised learning (SSL) has been shown to be a powerful approach for learning visual representations. In this study, we propose incorporating ZCA whitening as the final layer of the encoder in self-supervised learning to enhance the quality of learned features by normalizing and decorrelating them. Although whitening has been utilized in SSL in previous works, its potential to universally improve any SSL model has not been explored. We demonstrate that adding whitening as the last layer of SSL pretrained encoders is independent of the self-supervised learning method and encoder architecture, thus it improves performance for a wide range of SSL methods across multiple encoder architectures and datasets. Our experiments show that whitening is capable of improving linear and k-NN probing accuracy by 1-5%. Additionally, we propose metrics that allow for a comprehensive analysis of the learned features, provide insights into the quality of the representations and help identify collapse patterns.

Submission history

Access paper:.

• HTML (experimental)
• Other Formats

References & Citations

• Google Scholar
• Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

A Biologically Inspired Attention Model for Neural Signal Analysis

• Find this author on Google Scholar
• Find this author on PubMed
• Search for this author on this site
• ORCID record for Nicolas Skatchkovsky
• For correspondence: [email protected]
• ORCID record for Natalia Glazman
• ORCID record for Sadra Sadeh
• ORCID record for Florencia Iacaruso
• Info/History
• Preview PDF

Understanding how the brain represents sensory information and triggers behavioural responses is a fundamental goal in neuroscience. Recent advances in neuronal recording techniques aim to progress towards this milestone, yet the resulting high dimensional responses are challenging to interpret and link to relevant variables. In this work, we introduce SPARKS, a model capable of generating low dimensional latent representations of high dimensional neural recordings. SPARKS adapts the self-attention mechanism of large language models to extract information from the timing of single spikes and the sequence in which neurons fire using Hebbian learning. Trained with a criterion inspired by predictive coding to enforce temporal coherence, our model produces interpretable embeddings that are robust across animals and sessions. Behavioural recordings can be used to inform the latent representations of the neural data, and we demonstrate state-of-the-art predictive capabilities across diverse electrophysiology and calcium imaging datasets from the motor, visual and entorhinal cortices. We also show how SPARKS can be applied to large neuronal networks by revealing the temporal evolution of visual information encoding across the hierarchy of the visual cortex. Overall, by integrating biological mechanisms into a machine learning model, we provide a powerful tool to study large-scale network dynamics. SPARKS' capacity to generalize across animals and behavioural states suggests it is capable of estimating the internal latent generative model of the world in animals, paving the way towards a foundation model for neuroscience.

Competing Interest Statement

The authors have declared no competing interest.

View the discussion thread.

Thank you for your interest in spreading the word about bioRxiv.

NOTE: Your email address is requested solely to identify you as the sender of this article.

Citation Manager Formats

• EndNote (tagged)
• EndNote 8 (xml)
• RefWorks Tagged
• Ref Manager
• Tweet Widget
• Facebook Like
• Google Plus One

Subject Area

• Neuroscience
• Animal Behavior and Cognition (5526)
• Biochemistry (12568)
• Bioengineering (9436)
• Bioinformatics (30819)
• Biophysics (15849)
• Cancer Biology (12919)
• Cell Biology (18522)
• Clinical Trials (138)
• Developmental Biology (10000)
• Ecology (14969)
• Epidemiology (2067)
• Evolutionary Biology (19151)
• Genetics (12735)
• Genomics (17541)
• Immunology (12680)
• Microbiology (29726)
• Molecular Biology (12369)
• Neuroscience (64725)
• Paleontology (479)
• Pathology (2002)
• Pharmacology and Toxicology (3455)
• Physiology (5330)
• Plant Biology (11091)
• Scientific Communication and Education (1728)
• Synthetic Biology (3063)
• Systems Biology (7685)
• Zoology (1729)