Characterizing Goal-driven Dynamics Underlying Naturalistic Behavior

Understanding how the brain drives behavior is a key goal of neuroscience. Traditionally, the field has focused on
simple behaviors, but recent research is shifting towards more naturalistic paradigms, such as navigation and
foraging. This opens up exciting possibilities for studying natural behaviors and uncovering the neural
mechanisms underlying them.

Animal behavior is inherently complex and dynamic, making it difficult to interpret. However, recent literature has
shown that behavior is driven by an animal’s internal state—such as fear, hunger, etc. These internal states
provide a framework for quantifying behavior over extended periods of time, and have been linked to changes in
neural activity. However, inferring these states in natural settings is extremely challenging. For example, when
animals forage, their behavior is unconstrained, and they can shift between different internal goals. Current
studies on spontaneous behavior have identified distinct motifs, or "syllables," that compose natural behavior.
However, these syllables last less than a second and lack a normative framework for explaining how behavior
evolves over longer timescales.

My postdoctoral work aims to fill this gap and uncover the organizational principles that govern natural behavior.
As a graduate student, I developed a method to model behavior as driven by an animal’s time-varying internal
goal states. By integrating this with the concept of syllables—where their evolution is influenced by internal goal
states and environmental factors—I aim to create a comprehensive model of long timescale animal behavior.
Using statistical modeling and reinforcement learning, I will develop a goal-driven model to predict behavior in
dynamic, naturalistic settings, and apply it to foraging in rhesus macaques and rodents. This will result in a
unified framework for behavior quantification across species, enabling deeper study of the neural mechanisms
driving natural behavior, and