Task-Dependent Mouse Behavioral Dynamics in Navigational Decision-Making

Abstract

Effective navigation in their environment requires animals to process and integrate diverse inputs across varying timescales while adaptively responding to their surroundings. This ability to integrate information over a diversity of timescales is fundamental to perception, cognition, and motor planning. However, behavioral responses are influenced not only by external inputs but also by experience, expectation, and internal cognitive states. Understanding neuronal computations in the brain, especially regarding complex higher-order cognitive processes, requires a comprehensive characterization of animal behavior that allows us to meaningfully interrogate the neural circuits underlying them. In this work, we have developed a complementary set of behavioral paradigms designed to isolate task-dependent cognitive processes from external inputs by imposing distinct temporal demands while ensuring that sensory stimuli and required behavioral outputs are as similar as possible between tasks. We trained mice to perform three navigational decision-making tasks based in visual virtual reality and established tools to quantify their behavior. We showed that animal running speed was modulated by task context and that training history impacted decision-making strategies. Moreover, we identified behavioral variabilities that were suggestive of latent cognitive states, such as confidence and indecision. Finally, by analyzing running trajectories and employing behavioral modeling approaches, we inferred timescales of choice formation aligned with the presentation of task-relevant stimuli. Our novel behavioral tasks, which can be combined with the extensive genetic and optical techniques that have been developed for rodents, are thus valuable tools for the study of the temporal dynamics of information integration and perceptual decision-making, facilitating deeper insight into the underlying neural circuits.