Note that this seminar will be held in Wu & Chen Auditorium (Levine 101).

Humans and animals exhibit a range of interesting behaviors in complex environments, and it is unclear how the brain reformats dense sensory information to enable these behaviors. To gain traction on this problem, new recording paradigms now facilitate the ability to record and manipulate hundreds to thousands of neurons in awake, behaving animals. Consequently, a pressing need arises to distill these data into interpretable insights about how neural circuits give rise to intelligent behaviors.

To engage with these issues, I take a computational approach, known as “task-optimized modeling”, that leverages recent advancements in artificial intelligence (AI) to express hypotheses for the evolutionary constraints of neural circuits. These constraints guide the iterative optimization of artificial neural networks to achieve a specific behavior (“task”). By carefully analyzing the factors that contribute to model fidelity in predicting large-scale neural response patterns, we can gain insight into why certain brain areas respond as they do, and what selective pressures over evolutionary and developmental timescales give rise to the diversity of observed neural responses.

In this talk, I apply this approach to examine the functional constraints of brain areas involved in the perception-action loop across multiple timescales: 1. the role of recurrent processing in rapid object recognition (within 250 ms), and 2. visually-grounded mental simulation of future environmental states (within several seconds). Finally, I conclude with future directions towards closing the perception-action loop by extending task-optimized modeling to build integrative, embodied agents to gain a systems-level understanding of an organism’s brain. These agents would serve as normative accounts of how brain areas collaborate to enable meaningful actions in the physical world. Their design will elucidate the algorithmic principles of natural intelligence conserved across species, and yield safer, more grounded embodied AI algorithms.