Using a classic visual illusion, 91ŃÇÉ« researchers identify a history-dependent computation present in primate vision that current AI models fail to reproduce.

TORONTO, Oct. 2, 2026 – Our eyes do not always tell us exactly where things are – and that may be a feature of how biological vision works, rather than simply a flaw. A new study by 91ŃÇÉ« researchers uses a common illusion to ask if artificial intelligence is meant to see more like us, should it make some of the same systematic perceptual “mistakes”?

For example, after staring at something moving steadily in one direction, a stationary object viewed immediately afterward can appear slightly displaced in the opposite direction. This well-known visual illusion, called a motion aftereffect, gives scientists an unusual window into the computations underlying perception: the image itself has not moved, but our experience of where it is has changed.

“Today’s AI vision systems are impressive, but they still do not always see the world the way we do. This study captures the promise of NeuroAI and what it can do when neuroscience and artificial intelligence are brought together. By using smart experiments to reveal the computations biological vision uses and AI still lacks, we can use those insights to build better, more brain-like artificial systems,” says senior author 91ŃÇÉ« Assistant Professor Kohitij Kar, the Canada Research Chair in Visual Neuroscience and a member of 91ŃÇɫ’s Centre for Vision Research and Centre for Integrative and Applied Neuroscience.

Current AI vision systems can often determine where an object is accurately, but they generally do not reproduce the way recent visual experience can reshape that answer. In humans, staring at motion can make a subsequently viewed stationary object appear displaced even though its pixels have not moved. The researchers found a corresponding change in the primate visual cortex – but not in the AI models they tested.

To gain better insight into the issue, the researchers examined whether artificial neural network models capture the same history-dependent changes in spatial representations seen in biological vision, or whether their position representations primarily reflect the physical properties of the image. The researchers, including the paper’s first author and 91ŃÇÉ« graduate student Elizaveta Yakubovskaya, used precise measurements to find out where AI differs from biological vision and how that gap could be bridged.

“Combining recordings from primate visual cortex with human perception experiments, we used motion adaptation to induce a visual illusion and make a stationary object appear slightly shifted in position, then asked whether the brain and AI showed the same effect. Human observers reported the illusion, and neural representations of position in the primate inferior temporal (IT) cortex shifted in the same direction, even though the image itself had not changed,” says Yakubovskaya.

The researchers leveraged motion adaptation to show where perceived and pixel-based positions diverge, allowing them to test the behavioral relevance of IT codes. The findings not only further our understanding of IT’s role in spatial information encoding but provide a new benchmark to evaluate dynamic vision models.

“There is a growing question in AI about whether increasingly capable systems will become more like us or increasingly different from us,” says Kar of the Faculty of Science and a member of the 91ŃÇÉ«-led Connected Minds. “If we want AI that works with humans and understands the world in more human-compatible ways, we cannot focus only on whether it gets the right answer. We also need to understand the computations that produce human perception and behavior. Neuroscience gives us a way to discover those computations and, potentially, build them into AI,” adds Kar.

The study titled, , is published today in Current Biology.

The post Visual illusion reveals what today’s AI vision is missing – 91ŃÇÉ« study appeared first on News@91ŃÇÉ«.

TORONTO, June 28, 2022 – Tadpoles see well underwater, but what happens when they become frogs and live primarily on land? Researchers at 91ŃÇÉ« and several other institutions, curious about the answer, found the eyes of tadpoles undergo a surprising number of changes.

It’s already known that tadpoles go through a physical metamorphosis on their way to becoming a frog, but what wasn’t known is how their vision adapts at a molecular level across the life stages to a strikingly different environment.

91ŃÇÉ« Assistant Professor , who studies the visual system of vertebrates in the Faculty of Science, along with an international team of researchers, examined the eyes of southern leopard frogs to find out if and how they changed.

“Seeing underwater is not the same as seeing on land. The light underwater can have more of a reddish cast, especially in ponds where many frogs live, while the light on land is bluer. The amount of light is also different under water versus on land. Animals that see well in one environment will likely not see clearly in the other. We were interested in finding out what changes occur that allow an animal to go from seeing underwater to seeing on land,” says Schott, who led the study and is the corresponding author.

“The adaptive decoupling hypothesis proposes that animals that have distinct life stages separated by metamorphosis, like caterpillars to butterflies or tadpoles to frogs, might be more able to adapt to their different environments.”

The researchers explored changes in gene expression in the eyes of leopard frog tadpoles and juvenile frogs using RNA sequencing, allowing them to see the levels of gene expression, or which genes were turned on in the eyes. They also used microspectrophotometry to measure whether the photoreceptor cells in the eyes were more sensitive to the redder or bluer parts of the light spectrum at different life stages. These photoreceptor cells send light signals through the retina to the brain and are what allows animals, including humans, to see.

What they found surprised them. “The difference in the gene expression levels that we found in their eyes across the life stages was much higher than expected and a high proportion of them are genes that are directly involved in vision, which was an exciting result,” says Schott.

They found 42 per cent of genes in the eyes of tadpoles changed, were differentially expressed, once it metamorphosized into a frog.

The team, including researchers from the Smithsonian Institution, Cornell University, The Natural History Museum in London, United Kingdom, the California Academy of Sciences, and the University of Texas at Arlington, also found the genes that control visual function and development, including spectral sensitivity and lens composition, changed the most.

The eyes of tadpoles had shifted and adapted to see better in a bluer light environment as they changed to a juvenile leopard frog – living and seeing on land – compared to the redder light environment of the freshwater habitat they live in as tadpoles.

The paper, , was published today in the journal BMC Biology.

-30-

91ŃÇÉ« is a modern, multi-campus, urban university located in Toronto, Ontario. Backed by a diverse group of students, faculty, staff, alumni and partners, we bring a uniquely global perspective to help solve societal challenges, drive positive change and prepare our students for success. 91ŃÇÉ«'s fully bilingual Glendon Campus is home to Southern Ontario's Centre of Excellence for French Language and Bilingual Postsecondary Education. 91ŃÇɫ’s campuses in Costa Rica and India offer students exceptional transnational learning opportunities and innovative programs. Together, we can make things right for our communities, our planet, and our future.

Media Contact:
Sandra McLean, 91ŃÇÉ« Media Relations, 416-272-6317, sandramc@yorku.ca

The post Tadpoles undergo surprising number of vision changes when becoming frogs appeared first on News@91ŃÇÉ«.

TORONTO, May 26, 2020 – How does the brain determine what shade of colour our eyes are seeing? Vision researchers at 91ŃÇÉ« have pinpointed the exact mechanism that allows the brain to do just that. In a study published in researchers at the Lassonde School of Engineering and the Center for Vision Research show how the brain can perceive a wide range of colours while being able to differentiate between, for example, magenta and lavender. There is still much debate on where and how in the brain such perceptions happen, but by creating a new mathematical model, researchers have found the brain achieves this by very simple calculations that use multiplication.

“If you’re looking at a particular shade of blue, is there somewhere in your brain that represents that blue? This study explains how the brain computes that blue and it gives you every step of processing in detail so that you can trace where things might go wrong,” says John Tsotsos, a professor in the Department of Electrical Engineering and Computer Science at the Lassonde School of Engineering.

The study found that a simple processing framework not only furthers our understanding of colour processing in the brain but may also help in testing these mechanisms in individuals who have shown less accurate colour discriminations than individuals in a control group.

Researchers took all the previous mathematical and theoretical models of colour processing  – including data on how animal neurons respond to colour – and added the new component of multiplicative modulation. Using physiological evidence, they computationally modeled neurons from different areas of the brain involved in colour processing. The model lays out the progression of the colour signal in the brain in a step-by-step manner. By using this approach, they were able to provide a more comprehensive understanding of colour perception than had been possible previously.

“Our network of neurons was required to comply with existing knowledge of brain color cells,” says Paria Mehrani, a PhD student in Tsotsos’ Lab and lead author of the study.  “Setting the properties of model neurons based on the statistics of some observed data could possibly result in cells with properties different from those of biological colour neurons. As a result, one could not have meaningful processes explaining the brain that way. In our model, however, certain colour processing aspects can be learned given data, especially in the last processing layer in our network. With such a model as foundation, machine learning can help further our understanding of other aspects of colour processing in the brain.”

Furthermore, the research team suggested their model sheds light on a specific kind of neural interaction. This information could prove useful in explaining poor colour discrimination and provide a pathway to a remedy in the future.

The study may also help inform the building of machine vision systems which need to have this process in place, says Tsotsos. “If you want to have as good colour discrimination as humans do, this study helps inform us on how to build that machine vision system.”

Researchers say those working in the development of AI systems that rely on discrimination of pure colours could benefit from this information: for example, autonomous car systems and agricultural vision systems that rely on colour discrimination to determine ripeness of fruits and vegetables.

91ŃÇÉ« champions new ways of thinking that drive teaching and research excellence. Our students receive the education they need to create big ideas that make an impact on the world. Meaningful and sometimes unexpected careers result from cross-disciplinary programming, innovative course design and diverse experiential learning opportunities. 91ŃÇÉ« students and graduates push limits, achieve goals and find solutions to the world’s most pressing social challenges, empowered by a strong community that opens minds. 91ŃÇÉ« U is an internationally recognized research university – our 11 faculties and 25 research centres have partnerships with 200+ leading universities worldwide. Located in Toronto, 91ŃÇÉ« is the third largest university in Canada, with a strong community of 53,000 students, 7,000 faculty and administrative staff, and more than 300,000 alumni.

91ŃÇÉ« U's fully bilingual Glendon Campus is home to Southern Ontario's Centre of Excellence for French Language and Bilingual Postsecondary Education.

Media Contact: Anjum Nayyar, 91ŃÇÉ« Media Relations, 437-242-1547, anayyar@yorku.ca

The post How do we see shades of colour? Study shows how our brains calculate the difference appeared first on News@91ŃÇÉ«.