Unraveling the Neural Circuits of Cognition and Disease with Advanced Imaging Techniques
Paula Albeda
Dr. Aaron Kuan is an Assistant Professor in the Department of Neuroscience at Yale School of Medicine and a researcher at the Wu Tsai Institute. His work aims to unravel how brain-wide neural circuits give rise to cognition and contribute to neurological disorders. His lab, which was established in August 2023, applies advanced circuit-mapping techniques to try to bridge the gap between neural connectivity and function.
As a postdoctoral fellow at Harvard Medical School under the mentorship of Dr. Wei-Chung Lee and Dr. Christopher Harvey, he developed novel electron microscopy and X-ray imaging techniques which have allowed for high-resolution neural circuits mapping and more insight into the neural basis of decision-making. He earned his PhD in Applied Physics at Harvard University.
Through his research and interdisciplinary perspective, Dr. Kuan continues to employ state-of-the-art imaging techniques to decode the complex wiring of the brain and its implications for cognition and disease.
Research & Scientific Contributions
1. Your background is in applied physics, yet your research is deeply rooted in neuroscience. How has this interdisciplinary approach shaped the way you investigate the brain?
My training in the physical sciences encourages me to think about the limits of current technologies as opportunities, rather than barriers. I like to ask, what are the fundamental physical limits to the resolution and speed with which we can map the brain? Can we design new technologies that vastly expand our capabilities, opening new paths for understanding brain function?
2. What inspired you to transition from physics to neuroscience?
While I was a graduate student in physics, I became fascinated with electron microscopy. The ability to see the nanoscale world felt amazing and surreal. I was introduced to neuroscience research by my wife Sue, who was also a graduate student at the time. I was fascinated to learn that brain function relies fundamentally on nanoscale neural circuitry, and that emerging 3D electron microscopy techniques would be crucial in the quest to map and understand the brain.
3. Your lab focuses on applying circuit-mapping techniques at brain-wide scales. What are some key questions you hope to answer in the coming years?
Brain wiring has been studied at different scales for many decades, but we have had limited means to bridge these scales. For example, we can measure whole-brain connectivity non-invasively in humans with MRI, but with very coarse resolution. This is like looking at “Google Maps” but zoomed all the way out so that you only see the highways. You know how to get from Chicago to New York but you have no idea how to get to the grocery store. On the other hand, with high resolution microscopy techniques like electron microscopy, we can resolve each and every neuronal connection, but only for local circuits. This is like a local town map where you can find the ice cream shop but have no idea how to get to another city. With current technologies, we don’t have ways of relating these disparate scales, so it’s hard to understand holistically how the brain functions. In my lab, we’re developing imaging technologies that can bridge the top-down and bottom-up views of the brain. I think this is a key unanswered question in neuroscience: how does long-range wiring interact with local connectivity to give rise to brain function?
4. You’ve developed new electron microscopy and X-ray imaging techniques. Could you walk us through how these innovations improve circuit mapping?
Electron microscopy has been the “gold standard” for high resolution imaging for decades. Many of the pictures you see in textbooks showing cellular structures or synapses in detail are electron micrographs (a picture taken by the electron microscope). Generally, each such picture only captures an extremely tiny field of view, so we need to slice brain samples into ultra-thin sections (~40 nm, about 1000 times thinner than a human hair), then stitch together millions of individual images from these sections to form a contiguous 3D volume. Our automated electron microscopy systems streamline this process, so that we can take pictures rapidly enough to build up a volume large enough to contain neural circuits. In some sense, these systems are like the little car that drives around and takes pictures for the “Google Maps” 3D view. It can capture the important fine details, but it takes an incredible amount of work to build up the whole map.
X-ray imaging is perhaps more elegant in the sense that it can penetrate thick samples and reveal the interior structure of samples without cutting them into sections. This is why they are commonly used for medical and dental imaging. However, X-ray imaging currently typically can’t get to quite as high of a resolution as electron microscopy. Recently, we’ve demonstrated that new X-ray imaging techniques performed at synchrotrons (a type of particle accelerator that produces intense and coherent X-ray beams) can indeed reach levels of detail useful for mapping brain wiring. In our “Google Maps” analogy, X-ray techniques are a bit like satellite imaging – they can very efficiently map moderately sized features, but not the smallest details (yet).
5. What are some of the biggest technical challenges in mapping neural circuits at such a large scale?
Brain circuits are really unforgiving because the wiring can be extremely thin (~100 nm) but traverse extremely long distances (~10 cm in the human brain). This means that the accuracy of tracing needs to be extremely high. Any mistakes can cut off large portions of these axons, which could significantly affect analysis of connectivity. Thus, for mapping neural circuits, we have to aspire to extremely high standards at every step of the pipeline, including sample preparation, image quality, image processing, data analysis, and proofreading.
6. Decision-making is a complex process that involves multiple brain regions. What have your studies revealed about how neural circuits encode and execute decisions?
In our recent work, we focused on micro-circuits within the Posterior Parietal Cortex (PPC), a cortical brain region known to be involved in decision-making. Because decision-making indeed involves many brain regions, it is still unclear exactly what role the PPC plays. We found an interesting pattern of connectivity (circuit motif) where inhibitory neurons in PPC mediate competition between neurons that are active on right or left trials. Using circuit modeling, we showed that this inhibitory motif pushes the network towards decisively-right or decisively-left activity patterns, which improves the accuracy of decisions. This discovery is interesting because it showed that inhibitory neurons play a much more specific role than previously appreciated, and may be involved in the specialization of cortical regions to particular tasks such as visual processing, decision-making, language, etc.
7. How do you see your work contributing to our understanding of cognition and neurological diseases?
Our work aims to identify basic and fundamental principles of how neural circuits give rise to neuron activity and cognition. We use these types of knowledge when we look at an electronic circuit diagram. We can understand how the overall function is built up from simple circuit elements and connections, and accurately simulate its function. When something goes wrong, we can test different elements and isolate where the malfunction originates. The goal is to be able to do similar things with brain circuits – understand cognition as an emergent property of basic circuit motifs, accurately simulate brain circuits based on detailed circuit maps, and understand the cause of neurological diseases in order to support therapeutics.
Other Questions
8. If you could map the neural circuits of any cognitive function with unlimited resources, what would it be and why
Given unlimited resources, I would try to study something that isn’t commonly studied in systems neuroscience. I would be interested in deliberate, analytical thinking, for example how one approaches solving difficult math problems, plays chess, or works on a new research problem. This type of thinking happens over long timescales, at least minutes to hours, and seems to me to be uniquely human. This problem is known as “System 2 Thinking” in Artificial Intelligence, and researchers are hopeful that AI models will soon be able to engage in this type of deep thought.
9. Do you think artificial intelligence and neuroscience will merge in a way that allows us to fully simulate a brain in the future?
Yes, I do, but I think the path forward for this merging may not be straightforward. Although AI models are rapidly improving and bypassing human performance on many tasks, it is likely that such models achieve intelligence in a very different way than human brains. However, there are likely fundamental concepts that are shared, and I think the fields of neuroscience and AI will continue to benefit from each other. For example, powerful AI models may help in designing and implementing realistic brain simulations which may be too complex to do with “classical” methods, and breakthroughs in circuit neuroscience may reveal clever implementations (designed by evolution) that could increase the efficiency and performance of AI agents.
Personal Journey & Perspective
10. What advice would you give to young scientists interested in circuit neuroscience and neuroimaging techniques?
Two important points.
About the Author
Paula Albeda ('26) is a senior at The University of California, San Diego concentrating in psychology.
As a postdoctoral fellow at Harvard Medical School under the mentorship of Dr. Wei-Chung Lee and Dr. Christopher Harvey, he developed novel electron microscopy and X-ray imaging techniques which have allowed for high-resolution neural circuits mapping and more insight into the neural basis of decision-making. He earned his PhD in Applied Physics at Harvard University.
Through his research and interdisciplinary perspective, Dr. Kuan continues to employ state-of-the-art imaging techniques to decode the complex wiring of the brain and its implications for cognition and disease.
Research & Scientific Contributions
1. Your background is in applied physics, yet your research is deeply rooted in neuroscience. How has this interdisciplinary approach shaped the way you investigate the brain?
My training in the physical sciences encourages me to think about the limits of current technologies as opportunities, rather than barriers. I like to ask, what are the fundamental physical limits to the resolution and speed with which we can map the brain? Can we design new technologies that vastly expand our capabilities, opening new paths for understanding brain function?
2. What inspired you to transition from physics to neuroscience?
While I was a graduate student in physics, I became fascinated with electron microscopy. The ability to see the nanoscale world felt amazing and surreal. I was introduced to neuroscience research by my wife Sue, who was also a graduate student at the time. I was fascinated to learn that brain function relies fundamentally on nanoscale neural circuitry, and that emerging 3D electron microscopy techniques would be crucial in the quest to map and understand the brain.
3. Your lab focuses on applying circuit-mapping techniques at brain-wide scales. What are some key questions you hope to answer in the coming years?
Brain wiring has been studied at different scales for many decades, but we have had limited means to bridge these scales. For example, we can measure whole-brain connectivity non-invasively in humans with MRI, but with very coarse resolution. This is like looking at “Google Maps” but zoomed all the way out so that you only see the highways. You know how to get from Chicago to New York but you have no idea how to get to the grocery store. On the other hand, with high resolution microscopy techniques like electron microscopy, we can resolve each and every neuronal connection, but only for local circuits. This is like a local town map where you can find the ice cream shop but have no idea how to get to another city. With current technologies, we don’t have ways of relating these disparate scales, so it’s hard to understand holistically how the brain functions. In my lab, we’re developing imaging technologies that can bridge the top-down and bottom-up views of the brain. I think this is a key unanswered question in neuroscience: how does long-range wiring interact with local connectivity to give rise to brain function?
4. You’ve developed new electron microscopy and X-ray imaging techniques. Could you walk us through how these innovations improve circuit mapping?
Electron microscopy has been the “gold standard” for high resolution imaging for decades. Many of the pictures you see in textbooks showing cellular structures or synapses in detail are electron micrographs (a picture taken by the electron microscope). Generally, each such picture only captures an extremely tiny field of view, so we need to slice brain samples into ultra-thin sections (~40 nm, about 1000 times thinner than a human hair), then stitch together millions of individual images from these sections to form a contiguous 3D volume. Our automated electron microscopy systems streamline this process, so that we can take pictures rapidly enough to build up a volume large enough to contain neural circuits. In some sense, these systems are like the little car that drives around and takes pictures for the “Google Maps” 3D view. It can capture the important fine details, but it takes an incredible amount of work to build up the whole map.
X-ray imaging is perhaps more elegant in the sense that it can penetrate thick samples and reveal the interior structure of samples without cutting them into sections. This is why they are commonly used for medical and dental imaging. However, X-ray imaging currently typically can’t get to quite as high of a resolution as electron microscopy. Recently, we’ve demonstrated that new X-ray imaging techniques performed at synchrotrons (a type of particle accelerator that produces intense and coherent X-ray beams) can indeed reach levels of detail useful for mapping brain wiring. In our “Google Maps” analogy, X-ray techniques are a bit like satellite imaging – they can very efficiently map moderately sized features, but not the smallest details (yet).
5. What are some of the biggest technical challenges in mapping neural circuits at such a large scale?
Brain circuits are really unforgiving because the wiring can be extremely thin (~100 nm) but traverse extremely long distances (~10 cm in the human brain). This means that the accuracy of tracing needs to be extremely high. Any mistakes can cut off large portions of these axons, which could significantly affect analysis of connectivity. Thus, for mapping neural circuits, we have to aspire to extremely high standards at every step of the pipeline, including sample preparation, image quality, image processing, data analysis, and proofreading.
6. Decision-making is a complex process that involves multiple brain regions. What have your studies revealed about how neural circuits encode and execute decisions?
In our recent work, we focused on micro-circuits within the Posterior Parietal Cortex (PPC), a cortical brain region known to be involved in decision-making. Because decision-making indeed involves many brain regions, it is still unclear exactly what role the PPC plays. We found an interesting pattern of connectivity (circuit motif) where inhibitory neurons in PPC mediate competition between neurons that are active on right or left trials. Using circuit modeling, we showed that this inhibitory motif pushes the network towards decisively-right or decisively-left activity patterns, which improves the accuracy of decisions. This discovery is interesting because it showed that inhibitory neurons play a much more specific role than previously appreciated, and may be involved in the specialization of cortical regions to particular tasks such as visual processing, decision-making, language, etc.
7. How do you see your work contributing to our understanding of cognition and neurological diseases?
Our work aims to identify basic and fundamental principles of how neural circuits give rise to neuron activity and cognition. We use these types of knowledge when we look at an electronic circuit diagram. We can understand how the overall function is built up from simple circuit elements and connections, and accurately simulate its function. When something goes wrong, we can test different elements and isolate where the malfunction originates. The goal is to be able to do similar things with brain circuits – understand cognition as an emergent property of basic circuit motifs, accurately simulate brain circuits based on detailed circuit maps, and understand the cause of neurological diseases in order to support therapeutics.
Other Questions
8. If you could map the neural circuits of any cognitive function with unlimited resources, what would it be and why
Given unlimited resources, I would try to study something that isn’t commonly studied in systems neuroscience. I would be interested in deliberate, analytical thinking, for example how one approaches solving difficult math problems, plays chess, or works on a new research problem. This type of thinking happens over long timescales, at least minutes to hours, and seems to me to be uniquely human. This problem is known as “System 2 Thinking” in Artificial Intelligence, and researchers are hopeful that AI models will soon be able to engage in this type of deep thought.
9. Do you think artificial intelligence and neuroscience will merge in a way that allows us to fully simulate a brain in the future?
Yes, I do, but I think the path forward for this merging may not be straightforward. Although AI models are rapidly improving and bypassing human performance on many tasks, it is likely that such models achieve intelligence in a very different way than human brains. However, there are likely fundamental concepts that are shared, and I think the fields of neuroscience and AI will continue to benefit from each other. For example, powerful AI models may help in designing and implementing realistic brain simulations which may be too complex to do with “classical” methods, and breakthroughs in circuit neuroscience may reveal clever implementations (designed by evolution) that could increase the efficiency and performance of AI agents.
Personal Journey & Perspective
10. What advice would you give to young scientists interested in circuit neuroscience and neuroimaging techniques?
Two important points.
- Find the type of work that gets you up in the morning. You really need to love what you are doing to do your best work. Pursuing a scientific career can be stressful and competitive, so you need to make sure that you really enjoy what you are doing most days. Don’t be afraid to switch fields or topics if you are not inspired! Moreover, make sure you seek out environments where you will be supported and valued. Your mental well-being is critical to your productivity, so don’t downplay it when you make choices of where or what to study. The choice of environment is equally important to scientific projects.
- Learn how to get things done. Being talented and creative is important, but in my opinion the trait that most successful scientists share is being effective at finishing tasks, from answering emails to performing experiments to writing manuscripts and grants. These tasks can be ill-defined or nebulous, lack clear timelines or have factors outside our control. Nevertheless, you need to find your own internal motor to continuously make progress, and know when to give up and try something different. This is a high-level life skill that takes years to cultivate, but is extremely important. Scientific careers are a marathon, not a sprint!
About the Author
Paula Albeda ('26) is a senior at The University of California, San Diego concentrating in psychology.