The Robert H. Wurtz Award and Lecture in Systems Neuroscience recognizes advancements in neural research, including discoveries deciphering how brain circuits generate vision and guide behavior, and how their dysfunction contributes to visual disorders and disease. The award honors Bob Wurtz, PhD, whose pioneering vision research profoundly influenced the field of neuroscience. Recipients are chosen by a panel of esteemed scientists from the National Institutes of Health (NIH) and top universities.
Jazayeri, who is also an HHMI Investigator and a Professor and Director of Education in MIT’s Department of Brain and Cognitive Sciences, studies how the brain creates internal models of the external world. By combining sophisticated experiments in humans and nonhuman primates with computational modeling, his work has revealed fundamental principles by which the brain represents time, integrates information, and develops flexible strategies to guide behavior.
“His rigorous, integrative approach exemplifies systems-level neuroscience and illuminates how neural circuits transform sensory information into perception, decisions, and actions,” according to an announcement by the Foundation for the National Institutes of Health, who manages the award along with the National Eye Institute (NEI).
Jazayeri delivered his prize lecture, “Strategic problem solving in the primate brain” at the NEI on September 29, 2026.
As artificial intelligence becomes more and more powerful, its ability to handle human-like tasks is undeniable. Artificial neural networks can learn to recognize images, use language, or write computer code. Like the brain’s circuits, they reorganize themselves as they learn, adjusting the strengths of their connections to improve their performance.
The question for many neuroscientists is, what can these machine systems teach them about how the brain learns? And what kinds of models are likely to be the most informative?
In the October 9 issue of the journal PNAS, scientists at MIT’s McGovern Institute and K. Lisa Yang Integrative Computational Neuroscience Center argue that a crucial metric for the biological plausibility of an artificial neural network’s approach to learning is the network’s ability to organize itself, without the guiding hand of a developer. They show that it is possible for an artificial neural network to build its own learning machinery. With some simple instructions about how and when connections should change, a group of randomly connected units can reshape itself into an effective information-processing network, learning from the data it is trained on.
Biological learning
Qianli Liao is a K. Lisa Yang ICoN Center Fellow in Tomaso Poggio and Mark Harnett’s labs.
“Current AI technology is working very well, but it doesn’t necessarily use a learning algorithm that is used by the brain,” says Qianli Liao, a K. Lisa Yang ICoN Center Postdoctoral Fellow in Tomaso Poggio and Mark Harnett‘s labs, and co-first author of the PNAS paper with MIT research scientist Liu Ziyin and graduate student Yulu Gan. Liao explains that engineers design the rules about how connections change as an artificial neural network processes training data and receives feedback about its mistakes.
When it comes to training AI to do its job, the approach is powerful. But Liao and his colleagues caution that networks designed in this way may not offer much insight into biological learning. “The brain doesn’t have an external designer or architect,” Liao says. “The brain has to organize itself. And we need to know how simple rules are going to make the brain organize.”
The rules that reshape neural connections as an animal learns may be different in different parts of the brain, or they may change as the brain becomes more mature. Neuroscientists are working to understand how those rules work and how they might be disrupted by aging or disease—a vital step toward developing new therapies and prevention strategies. Liao points to the development of brain-computer interfaces, which aim to restore communication when the brain’s language abilities are lost or let paralyzed patients control neuroprosthetic limbs, as one potential application.
“None of these can work without knowing the rules behind neuron-neuron interactions,” he says. “That’s not achievable by just designing a circuit that works very well.” An artificial neural network’s value to neuroscience should instead be judged by the extent to which it can organize itself—not just its connections, but also the machinery that adjusts those connections.
Self-organized artificial learning
Working with Poggio and Harnett, Liao, Ziyin, and Gan set out to develop and validate a self-organizing computational network capable of learning. Liao explained that the network’s structure begins with multiple layers of randomly interconnected neuron-like units. Within each layer, every neuron-like component is subject to the same rules about how connections should be adjusted. “They just interact with each other using that rule, and they develop some type of behavior,” he says.
The team’s network implements the kind of simple rules thought to rewire brain circuits as humans and other animals learn from their experiences. In both the brain and the machine network, for example, “neurons that fire together wire together.” In other words, when two connected neurons are activated at the same time, the connection between them strengthens.
Above: A feedforward network (blue) and a feedback network (orange) are linked by cross-connections (purple), and line thickness shows connection strength. Using only local learning rules, both pathways reshape their connections at the same time until the feedback network sends useful training signals to the forward network. Graphic courtesy of the researchers.
The team showed that a self-organizing network governed by rules like these can learn to recognize patterns and classify images. They explain that in their model, learning is enabled by self-assembling motifs that emerge according to simple rules, then serve as building blocks for functional circuits. As their network learned, these motifs reorganized themselves using an optimization strategy similar to the approach built into most modern machine learning systems, called gradient descent, in which connections are gradually adjusted in response to errors in order to improve performance.
Most modern AI implements gradient descent through some form of the backpropagation algorithm, which neuroscientists say is unlikely to be used by the brain. The team compared its self-organizing network to networks designed to learn in this way and found that it performed at a similar level.
Connecting brains and machines
Poggio says finding a biological parallel to backpropagation could trigger significant progress in both AI and neuroscience.“If the synaptic motifs we propose are, in fact, used by the brain to do what backpropagation does for artificial neural networks, then we will have a strong connection at the level of fundamental mechanisms between artificial and natural networks—between machines and brains,” he says.
Liao says a self-organizing network does not necessarily learn the same way the brain does—but it’s a critical starting point for biological plausibility. “Self-organization is the first high-level framework,” he says. “Once we are good with self-organization as a measure, then we can talk about next thing.”
While the group’s main objective is better understanding the brain, Liao says that knowledge could ultimately improve machine learning systems too, enabling engineers to develop networks that perform better or in more human-like ways.
When people reach out during a mental health crisis, a top priority for counselors is identifying those with a high risk of suicide. The distressed person’s language holds critical clues, and a new tool developed by scientists at MIT’s McGovern Institute is designed to pick up on and rapidly evaluate those signals.
The language-processing tool was developed by Daniel Low, a former graduate student in Satra Ghosh’s Senseable Intelligence Group, who is now a research scientist at the Child Mind Institute, where he leads its AI, Risk, and Contemplative Science Lab, and is also a visiting scholar at Harvard University. It uses a custom-built list of words and phrases linked to 49 suicide risk factors, searching text for these and using them to estimate an individual’s risk.
Ghosh, Low, and colleagues report today in the Journal of Psychopathology and Clinical Science that their tool accurately predicts suicide risk from text conversations with crisis counselors. It is already helping to clarify which suicide risk factors matter most in times of crisis. With more validation, it could help with risk assessment in clinical settings and crisis-support situations.
Identifying key risk factors
Suicide attempts are notoriously difficult to predict. Dozens of risk factors have been linked to suicide, and even trained clinicians struggle to identify who will make an attempt among those who have some form of suicidal ideation. Among the factors that can make suicidal thoughts and behaviors more likely are certain psychiatric symptoms and disorders, like depression, borderline personality disorder, and post-traumatic stress disorder, as well as environmental and social stressors, like poverty, incarceration, discrimination, and loneliness.
“You see all these 50 risk factors, and they’re all interacting in ways we don’t really understand,” Low says. “Many different pathways could lead to someone feeling they want to escape their internal pain,” he says—and it’s challenging to know whose path will lead to a suicide attempt or death.
Ghosh and Low wanted to understand which risk factors counselors and clinicians should most look out for during a mental health crisis. To do that, they partnered with the Crisis Text Line, a nonprofit whose trained volunteers provide confidential text-based support to people in distress.
Crisis Text Line, a global mental health nonprofit that provides free, 24/7, confidential mental health support for people in need, provided specialized training and controlled access to this restricted dataset. The researchers analyzed de-identified texts from approximately 16,000 conversations with Crisis Text Line’s volunteer crisis counselors. Based on Crisis Text Line’s assessments, those conversations were grouped into three different risk levels: non-suicidal, suicidal ideation without imminent risk, and imminent risk. It was this imminent risk group—those with a plan for suicide or have an intent to die within the next 48 hours—that the researchers most wanted to understand.
“We wanted to know what type of symptoms predict the highest suicide risk,” Low says. This question has been studied before, he says—but typically through epidemiological surveys that ask a person to recall their symptoms and experiences, often after their mental health crisis has passed. In contrast, he says, “Crisis Text Line gives us an opportunity to assess many different symptoms and potential risk factors as people are having the crises.”
Reading between the lines
Before analyzing the crisis line texts, the research team built a suicide-risk lexicon. They turned to artificial intelligence to generate a preliminary list of words and phrases tied to established suicide risk factors, including factors associated with suicidal ideation, suicide attempt, and suicide death. Then they manually reviewed and curated that list. Their final lexicon includes about 60 words or phrases for each of 49 risk factors, with the relevance of each one confirmed by expert clinicians.
Then they trained a machine learning model to search the crisis conversations for words and phrases in their lexicon and use these to predict suicide risk. Because the lexicon links each word or phrase to a specific risk factor, they could use this data to determine which risk factors are most closely tied to imminent risk among people in crisis.
What they found was consistent with patterns found in previous research, though not always intuitive. For example, depression is a well-known risk factor for suicidal ideation, but their model found that mentions of lethal means and substance use were more likely to be expressed by the highest-risk group than depressed mood or fatigue. Expressions of active suicidal ideation and self-injury were also strong predictors. Intermediate predictors included anxiety, PTSD, and emotional pain.
The predictive model assigns a weight to each risk factor based on its contribution to risk. For example, mentions of lethal means for suicide, like “cut” or “pills,” are weighed heavily, whereas terms related to hopelessness, like “don’t know what to do” or “hopeless,” contribute to a lesser degree. After training their model, the team found they could use it to accurately predict risk severity in new conversations the model had not previously seen.
One limitation of lexicons, the researchers note, is that they do not consider the context of terms, and they can miss terms that are similar to those in the lexicon, but not explicitly included. Large language models have reasoning abilities, and Low and colleagues have developed ways of using large language models to detect suicide risk in other projects. However, they say they often use their lexicon in parallel to guarantee flagging certain terms, as well as to maintain data privacy.
Low stresses that while the team used the power of a large language model to develop its lexicon, its prediction model is a simpler, “lightweight” model. Unlike large language models, which require massive computational power, it can be run easily on a personal computer, reducing both cost and privacy concerns. Just as importantly, it is interpretable: Rather than merely generating a risk estimate like some deep learning models can do more effectively, it tells users how it got there. Words of concern can be flagged so users understand the basis for each assessment and act on that information. They are working on similar explainability approaches with large language models.
That’s critical, because the stakes are so high. “This is such a complex space that having a human in the loop is, I think, going to be critical for a long, long time,” says Ghosh, who is a Senior Research Scientist and Director of the Open Data in Neuroscience Initiative at the MIT McGovern Institute. Likewise, the researchers add that any predictive model must be thoroughly validated before clinical use, and might need to be continually refined to keep up with changes in language use or target populations.
Because a reliable lexicon opens doors to new ways of understanding mental health, Ghosh and Low are widely sharing not just their suicide risk lexicon, but also the software package they developed to build it. Researchers can use that tool to efficiently build lexicons for other mental health conditions. Meanwhile, Low says, the suicide risk lexicon is already being used to explore how text data from a variety of sources, from social media to electronic health records, might help researchers and clinicians better estimate risk.
Longtime supporters Patricia and James Poitras ’63 have launched a fellowship program for MIT graduate students and postdoctoral researchers studying major mental illness, expanding their philanthropy to directly support early-career scientists. The commitment establishes 50 two-year fellowships through the Poitras Center for Psychiatric Disorders Research at the McGovern Institute. Five fellowships will be awarded every year for the next decade, creating a long-term talent pipeline focused specifically on psychiatric disorders.
The $10M gift is the latest in a series of philanthropic investments from the Poitras family to strengthen MIT’s capacity to address the growing burden of severe depression and anxiety, bipolar disorder, schizophrenia, and other complex psychiatric conditions. “Pat and Jim have remained steadfast in their decades-long commitment to bold research that can transform mental wellness,” says Robert Desimone, director of the McGovern Institute and head of the Poitras Center. “Their remarkable support of rising talent in the MIT ecosystem is yet another emblem of their commitment to that cause.”
A philanthropic legacy
Many recent mental health discoveries emerging from MIT—from molecular tools that can rewrite DNA to AI-powered technologies that can calculate a person’s risk for developing mental illness—were hard to imagine two decades ago. Yet, Patricia and James Poitras envisioned a future where enigmatic mental health conditions could be solved. They understood this future would require not just research, but a fundamental reimagining of how psychiatric research itself is conducted.
In 2007, inspired by meetings with leadership at the McGovern Institute, the Poitras Family gifted $20 million to launch the Poitras Center for Psychiatric Disorders Research. By bridging the fields of basic neuroscience, clinical psychiatry, and molecular biology, the center sought to establish a unified blueprint for understanding how psychiatric disorders hijack the mind at every level, from molecular mechanisms to whole brain systems, and guide the creation of novel therapies to better treat them.
Patricia (right) and James Poitras ‘63 (left).
Since the center’s establishment, additional investments by the Poitras family have supported research ranging from genome engineering to cognitive neuroscience. These investments have empowered scientists across disciplines to pursue innovative research questions, including how ketamine acts on synaptic communication and why schizophrenia distorts inner speech and reasoning. These efforts have ushered in major breakthroughs in mental health: an AI-powered calculator for predicting bipolar disorder risk in adolescents, molecular carriers that precisely deliver therapies throughout the body, and strategies that use patients’ brain activity patterns to match them with optimal treatments, among other advances.
Expanding support for early career scientists
The Poitras family’s latest gift invests directly in the PhD students and postdoctoral fellows who will carry the field of psychiatric research forward. It comes at a time when federal funding has grown especially precarious. “To make the greatest impact on the world’s mental health, we recognize that we must not only support transformational research but also the young people driving its progress,” says James Poitras, who is also chair of the McGovern Institute’s board.
Five McGovern Institute researchers have been selected as the inaugural cohort of Poitras Center Fellows and Graduate Scholars. Their projects span multiple areas in brain research and could reveal a suite of new ways to heal the mind.
Amrita Lamba | Postdoctoral fellow, Saxe lab Drawing on her expertise in social neuroscience and game theory, Lamba will examine whether a two-player economic game could be used to distinguish generalized anxiety from social anxiety — a critical step towards personalized treatments for disorder subtypes.
Hoonwon Lee | Postdoctoral fellow, Jasanoff lab
As a molecular biologist who studies memory formation in the brain, Lee examines how fear memories develop and persist across brain networks, connecting what’s happening at the cellular level to larger brain-wide patterns. His findings may lead to new targets for post-traumatic stress disorder.”
Karen Pang | Graduate student, Anikeeva lab
Pang studies how the gut communicates with the brain to influence anxiety and depression, developing new tools to understand this connection and ultimately create treatments that address both digestive and mental health problems together.
Smriti Saini | Graduate student, Gabrieli lab
Saini is using brain imaging and computational techniques to identify which people with severe social anxiety will benefit from cognitive behavioral therapy, so doctors can personalize treatment decisions rather than relying only on symptoms. “.
Linghua Zhang | Postdoctoral fellow, Wang lab
Exercise is known to boost mood, but how it reshapes the mind is unclear. Zhang will study the circuit-level mechanisms behind the brain benefits of physical activity.
The new gift extends the Poitras family’s support of mental health research at MIT to over $100 million. It also marks another step toward a bold vision years in the making. “Serious brain disorders profoundly affect patients, families, and caregivers,” says Patricia Poitras. “We believe that investing in the next generation of researchers will accelerate powerful discoveries that lead to life-changing treatments and better future for countless patients and families.”
The next application window for Poitras Center fellowships will open in May 2027.
The tools of artificial intelligence (AI) are enabling McGovern researchers to ask new kinds of questions and solve more complex problems — and that’s leading to new insights into how the brain works and how mental illness might be more effectively managed.
“I can’t think of any facet [of brain research] where there isn’t an AI-based tool or technology that helps us,” says McGovern Institute Senior Research Scientist Satrajit Ghosh. Ghosh thinks today’s AI technology is particularly powerful in the early stages of research, when researchers are figuring out which questions are important and how to answer them. Historically, this has meant deep reading of the scientific literature — but existing knowledge can be overlooked when people don’t know where to look for it.
“The entire scientific workflow — from ideation to research to execution to dissemination — now has AI tools and technologies that can help.”
– Satra Ghosh
Large language models like Gemini or ChatGPT can pull information and ideas from sources across the internet, connecting users with resources they might not consider accessing on their own. “We get caught in our tracks and grooves that are hard to step outside of,” Ghosh says. “There are still biases in these tools, but I think overall they’re in a much more flexible space than humans typically are.”
McGovern Investigator Feng Zhang adds that by accelerating this vital phase of doing science, AI tools are broadening the scope of what an individual lab can explore. “We’ve never been short on good questions — what’s new is that we can chase them the moment they occur to us, and follow them as far as the data goes,” says Zhang, who is the James and Patricia Poitras Professor of Neuroscience at MIT.
Testing ideas faster
Once a plan is in place, researchers can often go from idea to execution much faster with the help of AI. McGovern Investigators Nancy Kanwisher and Josh McDermott say they have begun using AI models to pilot experiments, refining their approach before working with human study participants. They even use AI to generate images or sounds that their subjects will hear or see during their studies, designing stimuli to uncover specific aspects of the brain’s response.
The models they use are far more reliable representations of human perception than researchers were able to build just a few years ago. And with major advances in AI’s ability to generate computer code, models that took months or years to piece together can now be generated in days.
McDermott, who is the Uncas (1923) and Helen Whitaker Professor of Brain and Cognitive Sciences at MIT, uses an array of models to investigate different aspects of auditory processing in the brain. His team has models that recognize speech, models that localize sounds, and models that detect changes in pitch. “I fantasized about doing these things when I was a grad student, and it just wasn’t possible,” he says. “We were always trying to build working models of things, and it never worked. Now it’s really possible, and it’s just opening huge numbers of doors.”
Using a computational model of the auditory system, Josh McDermott has shown how the brain can selectively focus attention on one voice among many in a noisy environment — shedding light on a longstanding neuroscientific phenomenon known as the “cocktail party problem.” Photo: Steph Stevens
For McDermott, one priority is finding ways to develop better devices to help people hear. Computational models can facilitate that by enabling experimentation on a scale that is not possible with human subjects, allowing researchers to explore thousands of different conditions or rapidly test the consequences of modifying a hearing aid algorithm in different ways.
AI models also allow researchers to explore conditions they can’t test in the real world. McDermott’s lab, for example, has studied how hearing might work in a world where sounds behave differently, helping reveal which aspects of perception were shaped by the environment in which our ancestors evolved. And Kanwisher says models help overcome the limitations of experiments with humans, where the tools for monitoring and manipulating brain activity are imprecise and researchers have no control over their subjects’ life experiences.
That’s particularly valuable for studying cognitive functions that are uniquely human. “If you want to study things like language, or mathematical cognition, or theory of mind, it’s pretty hard to do in a mouse or even a monkey,” says McGovern Investigator and Associate Professor of Brain and Cognitive Sciences Evelina Fedorenko, who studies how the brain creates and understands language. AI models, on the other hand, can be trained to mimic these abilities. And what scientists like Fedorenko are learning is that models that behave similarly to humans often process information in ways that are similar to the human brain.
“If you’re in the business of studying sensory systems, like I am, it’s totally changed the way that we do science.” – Josh McDermott
Kanwisher, the Walter A. Rosenblith Professor of Cognitive Neuroscience at MIT, and her colleagues have used AI to explore how and why brains have networks specialized for the processing of certain kinds of objects. Her team and others have tasked AI models with processing visual information, training them with an assortment of images without further information or instruction. Remarkably, such models develop specific responses for certain categories of images: faces, places, bodies, and words — all of which Kanwisher’s team has shown are preferentially processed by specific regions in the human brain.
Kanwisher says that was surprising, because cognitive scientists have reasoned that human brains likely evolved these selective regions to optimize the way perceptual information is later used by the brain. But that doesn’t explain why AI responds to these categories with the same selectivity.
“The AI models don’t use face images to engage social cognition or place images to engage navigation,” she says. “They don’t do a thing with them, and yet they pull out the same categories as important.” That doesn’t rule out previous ideas about human brains — but, Kanwisher says, “It’s proof that you can have these shockingly similar things for very different reasons.” Her group is now exploring how the models’ training experience shapes the development of selective regions — something that cannot be done in humans.
Building better tools
AI is also helping McGovern scientists build physical tools, both to allow researchers to study the brain in new ways and to get better treatments to patients.
“Molecular syringes” (above), engineered by the AI tool AlphaFold in the Zhang lab, can deliver therapeutic cargo into specific cells in the body. These nanosyringes may lead to safer and more effective treatments for a variety of conditions, including cancer. Image: Joseph Kreitz, Feng Zhang
In Feng Zhang and Ed Boyden’s labs, researchers are using an AI tool called AlphaFold to design proteins with specific properties. Zhang’s lab is searching for proteins that can deliver therapies to specific brain cells, while Boyden’s team is developing tools for monitoring signaling in the brain.
AlphaFold predicts the structure of biomolecules, such as how a sequence of amino acids will fold into a three-dimensional protein. This eliminates much of the trial-and-error stage of protein engineering, speeding scientists’ ability to find molecules that interact with biological systems in the ways that they want.
Boyden, who is the Y. Eva Tan Professor in Neurotechnology at MIT, envisions another powerful AI tool: one that predicts how a brain works from its structure. The idea, he says, is to use machine learning to associate detailed brain maps with physiology and circuit function.
Building such a model would require vast amounts of data, but Boyden says the tools and tissue samples needed for large-scale structural analyses are available.
He envisions a kind of “compiler” built from molecular maps of brain cells and tissues — one that could help neuroscientists understand circuit function and eventually predict the effects of medications and other therapies.
Decoding data
Once an experiment is complete, data analysis begins — and AI is enabling scientists to find insights in datasets that, until recently, were too large and complex to make sense of.
AI tools have been invaluable, for instance, in creating detailed wiring diagrams, or connectomes, which many neuroscientists see as critical guides for brain research. That work, which involves tracing neurons across millions of microscope images, relies heavily on computer vision.
Without AI, piecing together the circuitry of an entire brain would take lifetimes. Remarkably, scientists now have access to a fully mapped connectome for the brain of a fruit fly, whose neurons make more than 50 million connections with one another.
A consortium of scientists, including McGovern Investigator Sven Dorkenwald, created a complete connectome of the fly brain (above) using computer vision algorithms and hundreds of human proofreaders. It is the largest, most complete neuronal map ever produced by scientists. Image: FlyWire Consortium, Sven Dorkenwald
McGovern Investigator Tomaso Poggio and his team are using AI to interact with that connectome by running simulations of circuit function, testing for a motion-detecting circuit that Poggio first proposed in theoretical work he did in the 1980s. Until recently, Poggio, who is the Eugene McDermott Professor Emeritus of Brain and Cognitive Sciences at MIT, had no way to find it.
Personalizing mental health
Other McGovern researchers are deploying AI to find ways to improve the diagnosis and treatment of brain disorders. They are finding that AI can pick up on signals that can be difficult for clinicians to detect — sometimes years before diagnosis.
Working with Satra Ghosh, John Gabrieli’s team found that a machine learning model could predict from clinical data which children and teenagers would later be diagnosed with bipolar disorder. With other collaborators, Ghosh has found that AI can recognize autism-related behaviors in parents’ videos of their toddlers, and he trained an AI model to detect changes in voice and speech patterns associated with conditions like Alzheimer’s and Parkinson’s disease. With further development, tools like these could facilitate earlier diagnosis and intervention for people who need it.
Researchers in Gabrieli’s lab think AI can also enable a more personalized approach to treating mental illness. It’s difficult to predict who will respond to a particular medication or therapy, leaving doctors and patients with a trial-and-error approach to treatment. Gabrieli’s team is betting that machine learning algorithms can find patterns hidden within complex clinical datasets that can enable more informed treatment decisions.
By merging machine-learning approaches with clinical and neuroimaging data, Gun Ahn, a graduate student in John Gabrieli’s lab, is pioneering precision-based methods that could help heal mental illness more quickly and effectively than current strategies.
In a recent success, Gun Ahn, a graduate student on Gabrieli’s team, followed up on a study in which the group found that a daily mindfulness practice meaningfully reduced anxiety for about 50 percent of autistic adults. No single factor separated those who benefitted from those who did not, so the team used machine learning to analyze demographic data and patient questionnaires. Their model was able to predict, with 80 percent accuracy, who responded to the practice.
This finding could help patients decide whether mindfulness is worth trying, particularly if it can one day be paired with predictions about alternative treatments. Gabrieli’s lab is taking a similar approach to predict patients’ response to medications for attention-deficit/hyperactivity disorder, hoping to help people bypass ineffective options and quickly find what works for them.
Tailoring interventions to a patient’s personal biology and experience has long been a goal in psychiatry — but the knowledge clinicians need to achieve this has been elusive. “For about 10 to 15 years, people have been talking about the incredible value that precision psychiatry would have,” says Gabrieli, who is the Grover Hermann Professor of Health Sciences and Technology at MIT. “This is, to me, the most promising scientific approach that could be widely usable.”
AI could even help clinicians move beyond formal diagnoses to understanding and treating people more holistically. Ahn imagines a model trained broadly, with data from a wide swath of patients.
Moving the field forward
Science isn’t done once the data has been analyzed: sharing knowledge is essential, too. Poggio, for example, integrated AI into the writing of a new book, “Brains Minds Machines: The Mystery of Human Intelligence, The Enigmas of the Artificial,” even crediting Gemini and ChatGPT as coauthors. He says the tools helped him organize his thinking and offload tedious tasks.
Looking ahead, most McGovern scientists say AI technologies are advancing so rapidly, their future impacts on neuroscience are difficult to predict. Could AI one day replace human researchers altogether? Some think it’s possible — at least in some areas.
“The scary prospect is that AI is going to put humans out of business. We’re definitely not close to that now, but nobody can predict the future,” says McDermott. “Physically doing experiments may be something that continues to be the domain of humans. But deciding what experiments to run and what they mean, and analyzing all the data? It’s unclear whether humans are still going to be the most efficient way to do that.”
Others are more skeptical. “It’s hard for AI to generate truly creative new ideas,” Boyden says. “I haven’t seen that happen yet.”
For now at least, today’s tools need oversight and guidance. To get the most out of them, users need experience and expertise. “They still have hallucinations from time to time,” Poggio warns. “One has to be careful.”
Nonetheless, today’s tools are tremendously empowering. They’re not just getting more powerful; they’re also getting more accessible. “These models are becoming exciting because they’re becoming usable by everyone, not just specialists,” Zhang says. “And when more people can do more, the whole field moves faster.”
Consciousness transfer, the hypothetical process of moving a human mind to a digital medium, has been a cornerstone of science fiction movies and books for decades. But as AI advances, questions about whether human consciousness could be transferred to a machine have moved from science fiction to serious academic debate.
The human brain is widely considered the most complex structure in the known universe, with tens of billions of neurons and more than 100 trillion points of connection. The theoretical process of uploading the mind, or “whole brain emulation,” would require mapping all of these circuits and understanding their unique function – a feat that many neuroscientists say exceeds current technologies and computing power. Mapping even the poppy seed-sized brain of a fly required an unprecedented collaboration of hundreds of scientists, various AI tools, and a global community of human proofreaders six years to complete. Researchers estimate it will take another 10 to 15 years to do the same in the mouse.
These technical hurdles aside, consciousness transfer also raises profound philosophical questions about identity, mortality, and what it means to exist. Even if we could upload our minds to a machine, should we?
To explore these questions, we asked members of the McGovern community to weigh in on the debate.
Jeffrey Brown II
Jeffrey Brown II is an EECS PhD student in Edward Boyden’s lab at the McGovern Institute. He builds AI agents to help construct a complete and accurate wiring diagram of the brain.“
Working on this problem can be extremely stimulative for neuroscience because, if you approach it seriously, you will have to deal with the some of the field’s most challenging problems.” – Jeffrey Brown
Jeffrey Brown II is an EECS graduate student using AI tools in the Boyden lab to construct a complete wiring diagram of the brain.
No one one really knows how the brain works, says Brown, and it’s unclear how much we need to know about the human brain in order to build an AI model of it.
“We don’t even have a list of all the circuits in the human brain,” says Brown, adding that he hopes to scale this up as part of his doctoral thesis in the Boyden lab. “We know many pieces of the brain, but we don’t really know how they all interact across massive scales of space and time,” he explains.
Some researchers believe functional data suffices; others insist on biophysical accuracy; still others use different approaches entirely. There’s no scientific consensus on what “good enough” actually means, Brown says.
Brown emphasizes that serious work in this space could be tremendously valuable for neuroscience, pushing researchers to grapple with some of the field’s most fundamental challenges. “On the other hand, if you don’t do the science carefully, I think people could hurt themselves, most likely by ceding their health (physical and mental) or agency to building a digital model that is not really “them.””
Davy Deng
Davy Deng is a Harvard-MIT HST PhD student in Ed Boyden’s lab at the McGovern Institute. His research goal is to build a digital brain using high-resolution multimodal data from the nervous systems of small animals.
“The most scientifically grounded and promising path toward whole brain emulation involves studying the nervous system of a small worm.” – Davy Deng
Davy Deng is a Harvard-MIT HST PhD student in Ed Boyden’s developing technologies toward reconstruction of the mind in machines.
In his recent TEDxMIT talk, Deng highlights a striking geographic divide: “When I mention whole brain emulation in Silicon Valley, people lean forward… But if I mention this in Boston, I tend to get some suspicious looks.”
This hype and skepticism has clouded real progress, he says, and a rigorous scientific foundation is needed to make headway. He argues that C. elegans, a 1-millimeter transparent worm, offers a unique opportunity to do this. The complete wiring diagram of the worm is well established, yet scientists have been unsuccessful in emulating its brain activity in a computer.
“This tells us that a wiring diagram alone does not contain sufficient information to simulate a brain,” he says.
His solution is to incorporate two additional layers of information – the chemical signatures of each connection and whole-organism imaging during behavior – creating what he calls the “ground truth” necessary for brain emulation.
Ila Fiete
Ila Fiete is an associate investigator at the McGovern Institute, a professor of brain and cognitive sciences, and the director of the K. Lisa Yang ICoN Center at MIT.
“I do think consciousness transfer could be possible!” – Ila Fiete
McGovern Institute Associate Investigator Ila Fiete.
Initial proofs of concept based on the connectome show some promise, Fiete says. (Nature, Sept 2024; Nature, Oct 2024; NeurIPS 2025) These studies combine supervised learning with existing circuit knowledge to infer the function of individual neurons.
“The challenge will be to show whether it is possible to do this for the whole brain, where we do not have knowledge of what most of the circuits exactly do, or how they do it.”
Rather than declaring consciousness transfer inherently good or bad, Fiete recognizes that there are profound philosophical concerns that deserve careful attention from ethicists.
“I think once people have an idea, it’s impossible to stop it from being pursued,” she says. “That’s a pathology of our species stemming from our curiosity and drive for improvement.”
“Fortunately, I do think and hope that we are capable of building guardrails around how these things are done.”
“Given our current level of knowledge, it is hard to imagine how this could be done.
But never say never!” – Alan Jasanoff
Even if we mapped the billions of neurons and their interactions in the human brain, says Jasanoff, we still can’t easily explain the “phenomenal” aspect of consciousness: why we feel and experience things the way we do.
McGovern Institute Associate Investigator Alan Jasanoff.
“This aspect of consciousness doesn’t seem reducible to more fundamental characteristics the way that biological phenomena like disease, digestion, or even life itself seem to be,” he says.
“And if phenomenal consciousness were somehow recreated, consider what kind of existence awaits,” Jasanoff reflects.
“What would give that life satisfaction or a sense of purpose? The virtual consciousness might achieve freedom from biological suffering and death, but how could it escape virtual equivalents of those in the machine environment?”
Qiyao (Catherine) Liang
Catherine Liang is an EECS PhD student in Ila Fiete’s lab at the McGovern Institute. She studies how intelligence emerges in brains and artificial systems.
“Consciousness uploading could be brilliant in principle, but terrible if developed or used without a clear understanding of its consequence.” – Catherine Liang
Catherine Liang is an EECS PhD student in the Fiete lab studying how intelligence emerges in biological and artificial systems.
“It may one day be possible to create models that imitate a person’s memories, personality, and behavior closely enough to function as a digital descendant,” says Liang. “Even whole-brain emulation may become a reality if consciousness depends mainly on the brain’s functional organization.”
“But that is assuming that consciousness is an emergent property of the physical brain, and precisely what level of details we need to emulate is the open question,” she says.
Consciousness uploading could theoretically overcome aging, disease, and biological limitations, says Liang, but it poses catastrophic risks if the process merely copies consciousness while the original person dies.
Profound ethical concerns would emerge around access, consent, ownership, and the legal rights of digital minds, she cautions.
Caitlin Lienkaemper
Caitlin Lienkaemper is the Swartz Foundation Postdoctoral Fellow for Theory in Neuroscience in Ila Fiete’s lab where she uses mathematical tools to study how the brain encodes and processes information.
“If I only existed as thoughts on a computer, I don’t think I’d be alive in a meaningful sense; I am my body as much as I am my thoughts.” – Caitlin Lienkaemper
Caitlin Lienkaemper is the Swartz Foundation Postdoctoral Fellow for Theory in Neuroscience in Ila Fiete’s lab.
While Lienkaemper acknowledges that a highly detailed simulation of a human brain could theoretically be conscious, she remains skeptical—in part because there’s no scientific consensus on how to measure consciousness in the first place.
“We don’t have a fully fleshed out theory of consciousness that would give us reason to trust any measurement,” she explains. “And even if we did, it’s unclear whether predictions valid for human brains would hold for machines.”
Beyond the technical barriers, though, Lienkaemper questions whether consciousness transfer is even desirable. She sees the pursuit as a “secular repackaging” of the concept of an immortal soul—and worries it fundamentally misses what makes life meaningful.
“I don’t think my thoughts are the only important thing about me,” she says. “I’m my body as much as I am my thoughts, and I value my connection to the natural world and to other people. Investing our energy in digital immortality rather than preserving Earth’s ecosystems seems both impractical and selfish.”
On June 9, the McGovern Institute convened leaders in science, biotechnology, and patient advocacy for RareNet 2026, a first-of-its-kind symposium aimed at dismantling the barriers between laboratory discovery and life-changing treatments for rare brain disorders.
Over 300 million people worldwide live with rare disorders—most affecting the brain and nervous system. Yet the vast majority lack an approved therapy. The Rare Brain Disorders Nexus (RareNet) was established at the McGovern Institute in 2025 by MIT alums Ana Méndez ’91 and Rajeev Jayavant ’86, (EE ’88, SM ’88) to address this need. Led by Guoping Feng, the James W. (1963) and Patricia T. Poitras Professor of Neuroscience at MIT, RareNet draws together expertise from the MIT community and beyond to expedite the path from lab to clinic.
MIT President Sally Kornbluth set the tone at the inaugural RareNet symposium, thanking founders Méndez and Jayavant for helping MIT focus on this important challenge: “I look forward to watching RareNet dissolve needless barriers, accelerate timelines, and bring new hope to millions of patients and their families for whom hope is long overdue.”
MIT President Sally Kornbluth (third from left) with RareNet founders Rajeev Jayavant (far left) and Ana Méndez (center) at the June 9 symposium at the McGovern Institute. Also pictured are the founders’ son Neal (second from left), RareNet Director Guoping Feng, RareNet Scientific Advisor Xian Gao, and McGovern Institute Director Robert Desimone. Photo: Steph Stevens
RareNet’s collaborative vision came into focus at the symposium, where more than a dozen leading neuroscientists, biotech innovators, and patient advocates shared the podium. The scientific program spanned the full translational pathway, from fundamental discovery to clinical development.
Feng Zhang (McGovern Institute, MIT; HHMI; Broad Institute), Katherine High (RhyGaze AG; Rockefeller University), Kiran Musunuru (University of Pennsylvania), and Timothy Yu (Boston Children’s Hospital; Harvard Medical School) discussed emerging genetic medicines–including genome editing, gene therapy, and individualized therapeutic strategies–and the challenges involved in bringing them safely to patients.
Kevin Bender (University of California San Francisco), Christopher Walsh (Boston Children’s Hospital; Harvard Medical School), Sonia Vallabh (Broad Institute; MGH; Harvard Medical School), and Joseph Buxbaum (Icahn School of Medicine at Mount Sinai) explored how insights into disease mechanisms, human genetics, and patient-derived data are advancing research in neurodevelopmental disorders, autism, and prion disease.
Representatives from the Sturge-Weber and FOXP1 communities–including Karen Ball, Matt Shirley, and Samit Dasgupta–demonstrated how patient foundations can help define research priorities, build essential resources, and drive promising discoveries toward meaningful treatments.
Also among the day’s speakers was Monica Coenraads, who transformed her child’s rare disease into a powerful research initiative. When her daughter Chelsea was diagnosed with Rett syndrome in 1998 at age two, Coenraads faced an uncertain future. Today, as founder and CEO of the Rett Syndrome Research Trust, she is helping to rewrite that story for other families. In a compelling talk with John Sinnamon, Director of Research at RSRT, Coenraads shared both the scientific breakthroughs reshaping Rett syndrome treatment and the deeply personal journey that sparked it all.
“Monica and John’s talks capture the vital connection between patients and families, cutting-edge research, and translational innovation,” says RareNet Executive Director Xian Gao. “It’s a powerful reminder that behind every research breakthrough is a human story demanding progress.”
Within the brain, neurons compute by generating electrical impulses. These signals travel throughout neurons, which are in turn connected in vast networks that control brain functions such as sensory perception, memory formation, and control of movement.
McGovern Institute Investigator Edward Boyden. Photo: Justin Knight
In an advance that could help neuroscientists map those neural networks, leading to a better understanding of how neural activity underlies behavior and other brain functions, MIT engineers have invented a new microscope that can image electrical activity in neurons distributed across the brain of an entire organism, the experimental model Danio rerio (zebrafish).
Using a microscope that they adapted for fast, high-volumetric rate imaging, the researchers were able to track electrical activity across the brain on the scale of milliseconds. This method revealed patterns of neural activity from neurons throughout the brain that were activated in response to ultraviolet light.
“All of the parts of the brain are connected together, so if you want to truly understand the brain, you have to understand how all the neurons work together as an emergent whole,” says Ed Boyden, the Y. Eva Tan Professor in Neurotechnology at MIT; a professor of biological engineering, media arts and sciences, and brain and cognitive sciences; and a member of MIT’s McGovern Institute for Brain Research, Yang Tan Collective, and the Koch Institute for Integrative Cancer Research.
Boyden is the senior author of the study, which appears today in Nature Methods. Former J. Douglas Tan Postdoctoral Fellow Zeguan Wang ’PhD 24 and former MIT research scientist Jie Zhang are the lead authors of the paper. Other authors include former MIT postdoc Panagiotis Symvoulidis, Picower Institute research scientist Wei Guo, graduate students Davy Deng and Lige Zhang, Koch Institute research scientist Adam Amsterdam, Picower Institute research scientist Takato Honda, Boston College undergraduate Steven Roche, and Matthew Wilson, the Sherman Fairchild Professor of Neuroscience at MIT and a member of the Picower Institute.
High-speed imaging
One technique often used to measure neuron activity in the brain is calcium imaging. Calcium flows into neurons after they fire an electrical impulse, so measuring calcium levels in the cells can serve as a proxy for neural activity. However, this type of imaging isn’t fast enough to capture single spikes of activity.
“Calcium imaging inherently is very slow, so you’re talking about imaging activity on the order of seconds or even minutes. Typically that is too slow for us to be able to see a lot of these high-speed neural activities,” Zhang says. “Neurons compute using electrical activity, so with voltage imaging, you can get direct observation of that.”
To enable direct imaging of voltage, researchers have developed proteins called genetically encoded voltage indicators — fluorescent proteins that can be genetically expressed in neurons. When a neuron fires an impulse, the protein fluoresces, which can be detected with a fluorescence microscope.
“All of the parts of the brain are connected together, so if you want to truly understand the brain, you have to understand how all the neurons work together as an emergent whole.” – Ed Boyden
In previous work, researchers have used these proteins to image small populations of neurons, usually focusing on one localized part of the brain. Until now, there hasn’t been a way to image a large volume, such as the entire brain, with the millisecond-scale resolution needed to see electrical impulses from individual neurons.
To achieve that, the MIT team decided to modify a commonly used microscope known as a light sheet microscope. This type of microscope uses a sheet of laser light to illuminate a thin slice of a sample. By imaging many layers in sequence, this technique can generate 3D images of a large volume. However, with previous microscopes, the scanning of an entire volume would take too long to be able to capture neuronal impulses across the volume at single cell resolution.
“Different groups of neurons that are distributed across the brain coordinate together at millisecond timescales to generate a lot of behaviors and brain computations,” Wang says. “To understand the principles, we need the technology to observe their activity at the same time, across the whole brain, so we are not missing any important participant neurons.”
To make the imaging process fast enough to image millisecond-scale activity, the researchers increased the image acquisition speed of the microscope’s camera, and they also boosted the scanning speed of the microscope using a technique called remote refocusing.
Using this approach, the researchers showed that they could scan the entire zebrafish brain 200 times per second, or once every five milliseconds.
Mapping brain activity
To test the new microscope, the researchers engineered neurons in larval zebrafish to express a voltage indicator called Positron2-Kv. Although they had hoped that the indicator would end up in every neuron, it produced signals in neurons distributed throughout the brain, with about one quarter of the neurons exhibiting acceptable signals. This was enough, however, to observe patterns of activity across the brain. The researchers imaged the brain as the fish were resting, and they were able to observe single voltage spikes from neurons, as well as rapid bursts of spikes.
Additionally, this technique revealed patterns in how the brain is activated following a stimulus such as ultraviolet light. Immediately following the stimulus, activity was seen in the optic tectum, which receives and processes visual input from the retina. This activity propagated from one side of a part of the brain called the tectum to the other. Stimulus-independent activity also occurred in sequences across sets of neurons in the cerebellum and hindbrain.
The new microscope scans the entire zebrafish brain 200 times per second, revealing brain-wide patterns of electrical activity. Each flashing spot represents an active neuron. Image: Zeguan Wang
The researchers now hope to increase the percentage of neurons that they can image across the brain, as well as the microscope’s speed and resolution. They are also working on expanding the use of this technique to other experimental models, including mice.
This approach, they say, could offer neuroscientists a new way to generate hypotheses about what happens in the brain when it engages in specific behaviors, or about how brain activity is linked to states of mind such as daydreaming.
“A big question is simply to understand how neurons work together as a network. And this might be the first time that you could do that, because you can image the voltage of neurons distributed throughout the network,” Boyden says.
The research was funded by the National Institutes of Health, the BRAIN Initiative, the Picower Institute Innovation Fund, K. Lisa Yang, Ashar Aziz, the K. Lisa Yang and Hock E. Tan Center for Molecular Therapeutics in Neuroscience at MIT, the Hock E. Tan and K. Lisa Yang Center for Autism Research, the Alana Down Syndrome Center, John Doerr, Jed McCaleb, James Fickel, and the Howard Hughes Medical Institute.
McGovern Investigator Robert Horvitz shared the 2002 Nobel Prize in Medicine with colleagues Sydney Brenner and John Sulston for discoveries that helped explain how genes regulate programmed cell death and organ development. Photo: AP Images/Aynsley Floyd
A functional nervous system depends on the cooperation of many kinds of cells. So as developing organisms build their nervous systems, their neurons must take on different forms and functions to fulfill their designated roles. That carefully orchestrated process gives rise to thousands of different cell types in the human brain.
In the tiny worm known as C. elegans, the nervous system is far simpler, comprising a mere 118 classes of neurons.
At MIT, scientists in H. Robert Horvitz’s lab are studying the worms to learn about how nervous systems develop. Horvitz is the David H. Koch Professor of Biology at MIT, an Investigator at the McGovern Institute for Brain Research at MIT, and an Investigator at the Howard Hughes Medical Institute. His team has just discovered that a protein complex called cohesin, which helps shape the three-dimensional structure of the genome in both worms and humans, is critical for establishing some neurons’ identities as development unfolds.
The findings, reported July 31, 2026, in the journal Science Advances, could help scientists find a way to treat a rare developmental disorder called Cornelia de Lange syndrome, which is caused by mutations that interrupt the cohesin complex.
Model organism
Postdoctoral researcher Dongyeop Lee explains that C. elegans is a powerful model for studying neurodevelopment not just because its nervous system has been comprehensively mapped, but also because of the ease and speed with which scientists can study the function of its genes.
Dongyeop Lee is a postdoctoral researcher in the Horvitz lab and the first author of the Science Advances paper.
Because many of the worm’s genes have been retained through evolution, findings from studies of C. elegans often reveal important aspects of human biology. The current study began with worms that, because of a genetic mutation, make too many neurons of a certain type.
Adrenergic neurons, named for the kind of neurotransmitter they use to communicate with other neurons, are vital for enabling worms to respond to both their environment and their own internal state. Normally, C. elegans has just two pairs of adrenergic neurons: two RIM neurons and two RIC neurons. But the worms Lee studied had extras of both.
Takashi Hirose, a former member of the Horvitz lab, first observed this change in 2007.
Lee later continued the study and discovered that worms carrying a mutation in a gene called coh-1 have extra adrenergic neurons. The coh-1 gene encodes one part of the cohesin complex.
When Lee tested other mutations that disrupt cohesin, he found the same effect: Worms without fully functional cohesin had too many RIM neurons and too many RIC neurons.
Molecular switch
With a series of experiments designed to tease apart how cohesin impacts neurons’ identities, Lee discovered that cohesin cooperates with a gene-regulating protein called EOR-1 (known in humans as PLZF) to direct some neurons to develop into neurons that communicate with the inhibitory neurotransmitter GABA.
By reorganizing the structure of the genome, cohesin can change the way gene regulators like EOR-1 interact with DNA. Lee’s experiments showed that when either cohesin or EOR-1 couldn’t do its job, cells that should have become GABA-producing neurons become adrenergic neurons instead.
“What we found is that there are two alternative possible fates of certain neurons, and cohesin acts as a molecular switch that decides one of the possible neuronal fates,” Lee explains. “This means the structure of genomic DNA in the nucleus is important for neuronal fate determination.”
Disease connection
Lee adds that extra adrenergic neurons were not the only abnormality he observed in worms with cohesin mutations. Cohesin is important for shaping cells and tissues throughout the body. “The mutants have severe developmental defects,” Lee says. “They grow slowly. They don’t move well, and they also have defects in reproduction.”
Notably, the problems Lee saw in the worms echo aspects of Cornelia de Lange syndrome, a rare genetic disorder that impacts physical, cognitive, and behavioral development. Cornelia de Lange syndrome can be caused by mutations in cohesin genes, and Lee says that the discovery of how cohesin mutations affect worm development and behavior opens new opportunities to study the disease and search for potential therapeutic targets in C. elegans.
The Horvitz lab already has some promising leads. Taking advantage of the quick genetic screens that are possible in worms, Lee has found additional mutations that can counteract impaired cohesin, improving the health of worms with cohesin mutations. The team is now working to identify the genes where these suppressor mutations occur, so they can investigate whether they might make good therapeutic targets in humans.
Meanwhile, the team is also exploring a potential role for cohesin in shaping the fates of other neuron types, as well as searching broadly for additional molecules that work with cohesin to guide development. “We expect we have opened up a new biology,” Lee says. “This paper is just the beginning.”
The wiring and rewiring of the brain never ends. Neural pathways are constantly being reshaped as we interact with the world and learn new things. At MIT’s McGovern Institute and York University in Toronto, scientists are combining detailed analysis of brain activity with computational modeling to better understand that change.
McGovern Institute postdoctoral fellow Lynn Sörensen, Investigator James DiCarlo, and York University assistant professor Kohitij Kar worked together to compare what happened when monkeys and an artificial neural network with brain-like architecture were trained to visually identify the same objects. As the model’s performance improved, it reorganized itself in ways that closely paralleled changes the team detected in the brains of monkeys.
Their work, reported today in the journalNature Communications, shows how changes in visual processing support animals’ ability to learn to discriminate new kinds of objects. By modeling these changes, the researchers hope to better predict how training reshapes perception, which could one day inform educational strategies for a wide range of learners.
Subtle changes
Learning about a new object calls on many parts of the brain. Visual-processing areas work together to make sense of information taken in through the eyes, then communicate with other brain areas to give the visual information meaning and guide behavior. Multiple parts of this system likely change during learning, and the research team wanted a clearer understanding of how that change is distributed.
Neuroscientists have debated how much change occurs in the brain’s visual-processing areas when an animal learns to recognize new objects. Some suspected that visual-processing pathways remain largely unchanged during learning to avoid broadly disrupting visual perception, but others have reported changes in activity within dedicated visual-processing areas with this kind of learning in humans and other primates.
To take a closer look, the team focused on neural activity in a key component of the brain’s visual object-processing network, the inferior temporal (IT) cortex. By the time visual information reaches the IT cortex, key object features are clearly represented—so much so that it’s possible to “decode” what object a monkey is seeing and even predict what errors it’s likely to make in identifying it, simply by analyzing patterns of neural activity there.
The team recorded neural activity in the IT cortex from two groups of monkeys as the animals looked at and identified images of objects. Some of the monkeys were untrained, so the images they saw had little meaning to them. Others had already learned to identify similar objects, so they could usually discriminate between elephants, chairs, and other select objects, even when those objects were presented at different sizes, from different angles, or against different backgrounds than the ones they had seen before.
The broad pattern of activity in the IT cortex was largely similar in trained and untrained monkeys, suggesting that learning had not dramatically rewritten this high-level visual representation. Still, the group found subtle but reliable differences in the way neurons in the IT cortex responded to images in monkeys that had learned to recognize the kinds of objects they were shown, compared to the untrained monkeys.
Modeling learning
The group turned to computational models to investigate how those modest changes might contribute to learning. Sörensen trained a suite of artificial neural networks whose internal components had been mapped to monkey IT cortex to identify the same categories of objects the monkeys had seen. The models were designed to learn using gradient descent, meaning they continually improved their accuracy by adjusting their parameters in response to errors.
Only some of the primate-like models showed learning behavior that matched that of the monkeys. In those that did, the IT-like stage changed in ways that resembled the learning-related changes the researchers had observed in the IT cortex of trained monkeys.
While gradient descent is commonly used to train artificial intelligence, it is generally considered biologically implausible as a direct model of how the brain learns. The researchers say the strong match in learning effects between the monkeys and their model demonstrates that these kinds of artificial neural networks can offer insights into biological learning at a useful level of abstraction, even if the brain does not learn in the same way.
“This shows that you can actually build in silico versions of future experiments,” Sörensen says. “I think that gives us this playground of asking ‘what if’ questions—and potentially predicting new things that go beyond the experimenter’s intuition.”
Most of the changes that allowed for learning in the model occurred outside of the IT cortex. “This tells us that there is a lot between the area we recorded from and the final behavioral readout that needs to change during this process,” Kar says. He adds that the team’s model will be useful as researchers look more deeply into how downstream brain areas contribute to learning.
The researchers stress that their study allowed more granular measurements of brain activity than would be possible in humans, and because monkeys’ brains are organized similarly to our own, their experiments have direct relevance to human learning. They say understanding the impact of plasticity in monkeys’ IT cortex could help researchers design new learning strategies for humans.
“Our prior conceptual working model of you—or a monkey—learning new objects was that your brain makes changes to synaptic connections that are largely downstream of your visual system, so you don’t destroy your visual system,” says DiCarlo, who is also the Peter de Florez Professor of Brain and Cognitive Sciences and Director of the MIT Siegel Family Quest for Intelligence. “You wouldn’t want your whole visual system to become an elephant detector [just because you’ve learned to identify an elephant]. But this study went beyond that to say actually, when you learn ‘elephant,’ your IT does change a little bit to make it a little more relevant to elephants.”
That likely has consequences for recognizing other visual features, too. Subtle changes in the IT cortex that support elephant recognition might also make you better at identifying things other than elephants, DiCarlo says. Likewise, the same changes might make it a little harder to identify something else.
These kinds of consequences may be difficult to predict intuitively, but become obvious with computational modeling. For instance, the team’s models revealed that after learning to recognize new objects, the IT cortex contained more information about objects’ locations. By providing insights like these, models could aid the design of more effective training strategies for visual tasks, including for people with altered sensory processing, who may learn from visual information in atypical ways.
The human brain is widely considered the most complex structure in the known universe, with tens of billions of neurons and more than 100 trillion points of connection. The theoretical process of uploading the mind, or “whole brain emulation,” would require mapping all of these circuits and understanding their unique function – a feat that many neuroscientists say exceeds current technologies and computing power. Mapping even the poppy seed-sized brain of a fly required an unprecedented collaboration of hundreds of scientists, various AI tools, and a global community of human proofreaders six years to complete. Researchers estimate it will take another 10 to 15 years to do the same in the mouse.
