Sunreeta Bhattacharya - 2026 Three Minute Thesis (3MT) Championship Presentation at CMU

Duration: 3:08 438 transcript words EN summary

Video summary

Sunreeta Bhattacharya's presentation explores how the human brain functions as an information processing machine designed to resolve uncertainty by identifying hidden patterns and gathering evidence, a capability that is often compromised in individuals with anxiety disorders. To understand these mechanisms, her research focuses on deciphering the specific algorithms the brain uses to integrate prior knowledge with new sensory information when making decisions under uncertain conditions. By investigating whether organisms rely on simple memorization or flexible combination of existing knowledge with current observations, she aims to reveal the fundamental steps involved in learning and decision-making processes. To study these complex cognitive functions, Bhattacharya designed an interactive maze game for mice that required them to find hidden rewards through repeated trial and error, effectively creating a simplified version of exploration similar to video games like Minecraft. The research revealed that mice do not merely memorize locations but actively explore their environment to build sophisticated mental maps that allow them to predict outcomes and reduce uncertainty when faced with doubt. A particularly surprising finding was that even after weeks of experience, the mice continued to actively explore and fine-tune these mental models, demonstrating that active exploration is a crucial strategy for maintaining flexibility in a constantly changing world. These insights into mouse behavior provide valuable parallels to human cognition and suggest potential pathways for addressing anxiety disorders by understanding how uncertainty processing can become paralyzing rather than helpful. Furthermore, deciphering these natural algorithms offers opportunities to develop adaptive artificial intelligence systems that learn and adapt in ways that mimic human flexibility rather than relying on static data. The core principle emerging from this work is that the ability to actively explore and update one's internal model of the world is essential for navigating life's difficult choices effectively. Ultimately, Bhattacharya concludes that the brain is inherently equipped to help individuals navigate uncertainty and make informed decisions, even when information is limited, as illustrated by everyday scenarios like choosing between different pizza options late at night. By remembering that our brains are designed to process ambiguity and find solutions through active engagement with our surroundings, we can better appreciate the resilience of our cognitive systems. This understanding not only sheds light on the biological basis of learning but also encourages a perspective where uncertainty is viewed as an opportunity for growth rather than a source of fear or paralysis.

Read the full video transcript

Our 10th and final presenter is Sunritha Bhattacharaya of the Neuroscience Institute whose title is deciphering the brain's uncertainty processing algorithms for learning and decision-making. It's Friday night. After a long week in a new city, you're craving pizza. You're torn between your friend's recommendation and the pizzeria next door where there's a long line and the delicious smell of fresh pizza. Which one do you choose? How do you make this very important decision based on the very limited information that you have? As it turns out, our brains are information processing machines that are usually great at resolving uncertainty by finding hidden patterns and gathering evidence to reach a decision. However, uncertainty can be difficult or even paralyzing for some individuals such as those struggling with anxiety disorders. To better understand why that happens, we first need to understand the fundamental steps our algorithms our brains use to learn and decide under under uncertainty. My PhD thesis aims to decipher these algorithms in the brain. To be more specific, I'm interested in how our brains integrate prior knowledge and new information to make decisions. To answer this question, I designed an interactive game for mice where they have to find hidden rewards in a maze through repeated trial and error. Think Minecraft but for mice. So, the central question of my project is do mice use simple strategies like memorization or do they actually flexibly combine what they know with what they see to reach an informed decision? So, here's what we find. Mice don't simply memorize but instead they actively explore their surroundings to make a sophisticated mental model of their surroundings. When they're unsure about what to do, they use this mental map to make predictions about the odds of different outcomes and reduce their uncertainty. This is very similar to how you predict the odds of rain when you have to go out on a cloudy day. But what really surprised us was that even after weeks, mice continued to actively explore their surroundings and fine-tune the um mental map they had before. This reveals a fundamental principle, that active exploration could be a key strategy to staying flexible in a changing world. These insights allow us to draw parallels between mouse brains and human brains. Knowledge of these algorithms could help us tackle anxiety disorders and also build adaptive AI that learns like we do. So, the next time you're stuck with a difficult choice, like where to get pizza, remember that your brain is actually designed to help you navigate uncertainty and it will help you make a decision. Thank you.