Neural and behavioral mechanisms of foraging in multi- timescale dynamic environments

• Applicant: Dr. Roxana Zeraati
• Subject Area: Cognitive, Systems and Behavioural Neurobiology
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 583405102

Animals live in environments that change over various timescales. Surviving in these dynamic environments requires learning their temporal structure and adapting decisions accordingly. While uncovering the mechanisms of such adaptivity is a central open question in neuroscience, we are still far from a coherent, principled understanding. This problem arises because most experimental and computational approaches rely on simplified cognitive tasks that lack the temporal richness of natural environments, leaving a major gap in our understanding of how the interactions among neural, behavioral, and environmental dynamics across timescales shape adaptive decision processes. This project addresses this gap by establishing a unified computational-experimental framework to study adaptive decision-making during foraging, an evolutionarily conserved behavior that naturally embeds multi-timescale structure. In foraging, animals should decide whether to exploit a depleting resource or explore alternatives. Optimal foraging requires animals to adapt their exploration-exploitation dynamics to environmental dynamics. By jointly modeling and analyzing behavioral and neural dynamics during foraging in multi-timescale environments, this project aims to uncover 1) how animals adapt their decisions to environmental dynamics, 2) how they learn the temporal structure of their environment, and 3) what neural circuit mechanisms shape their multi-timescale decisions. To achieve this, we will first integrate reinforcement learning methods and mechanistic recurrent neural network models to develop a comprehensive theoretical framework that systematically links neural, behavioral, and environmental dynamics during foraging. Our approach will reveal how the temporal dynamics of the environment and learning algorithms jointly shape decision policies and their underlying neural dynamics, while generating testable predictions. Building on this framework, we will design a novel foraging task that incorporates realistic, multi-timescale dynamics. Using this theory-driven task, our collaborators at the Allen Institute will conduct mouse foraging experiments, collecting behavioral and electrophysiological data across multiple brain areas. Model-based analyses of these data will allow us to test our theoretical predictions and uncover how neural circuits integrate information across multiple timescales to guide adaptive foraging decisions. Together, this interdisciplinary project will uncover how neural, behavioral, and environmental dynamics interact across timescales to support adaptive foraging decisions. Combining computational models with theory-driven experiments, our findings will reveal general principles of adaptive decision-making that enable survival in dynamic, volatile environments.

• DFG Programme: Position