The goal of the Naturalistic Neuroscience Team is to understand cognition and behaviors that occur under naturalistic settings. We develop naturalistic paradigms, in which rodents, nonhuman primates, and humans engage in active exploration of the environment, strategic planning, and continuous decision making. By adopting a cross-species approach and utilizing ecologically valid tasks and advanced computational methods, we aim to investigate the neural mechanisms that underlie ongoing interactions of complex cognitive functions and adaptive behaviors in a multidimensional real world. This team consists of “Reinforcement learning using virtual reality in rodents”, “Naturalistic behaviors in nonhuman primates”, and “Naturalistic perception, action, and cognition in humans” Unit.

(1) Reinforcement learning using virtual reality in rodents (PI: HyungGoo Kim)

Throughout our lives, we make decisions that we believe are the best at the given moment. Using rodents, we aim to elucidate the biological mechanisms underlying adaptive behaviors in uncertain environments. Reinforcement learning theories provides quantitative explanations of how the brain can achieve adaptive behaviors. We plan to investigate neural basis of learning in a more naturalistic situations using virtual reality. Navigation tasks in virtual reality are well-suited to study the neural basis of reinforcement learning. For example, we have developed a virtual world where reward and aversive stimuli often co-exists, and rodents learn to avoid bad things and approach positive goals. Leveraging the computational theories, we will focus on elucidating neural implementation of adaptive behaviors. The results of the study will provide critical insights in treating mental disorders caused by malfunctions of the reward circuitry, such as addiction, depression, and obsessive-compulsive disorder.

(2) Naturalistic behavios in nonhuman primates (PI: Seng Bum Yoo)

There are two methods for scaling up neuroscience: (1) using simplified open-loop tasks where the input and actions are non-sequential, and (2) using math to understand single-neuron tuning profiles. However, since the brain operates as an open system and natural behavior are sequential, these methods may not be suitable for scaling up. Furthermore, understanding the tuning profiles of single neurons may not be very useful in finding organizational and computational principles of the brain. Instead, we aim to investigate the population dynamics and computational principles of naturalistic behaviors in nonhuman primates. We will achieve this by designing innovative closed-loop tasks, developing interpretable computational models that capture complex behavior, and analyzing nonhuman primate electrophysiology data under the principle of computational through dynamics (CtD). Ultimately, our research could provide diagnostic insights for psychiatric disorders, which cannot be achieved using simplified and unnatural tasks.

(3) Naturalistic perception, action, and cognition in humans (PI: Won Mok Shim)

Cognitive neuroscience research has greatly benefited from highly controlled laboratory stimuli and tasks that are designed to examine individual cognitive processes in isolation. However, our brain has evolved to actively explore the environment interacting with other objects and agents and makes hierarchical plans and decisions to achieve complex behavioral goals. We aim to investigate the neural mechanisms and computational principles of perception, action, and cognition as an integrative process that continuously unfolds in naturalistic settings. By combining interactive naturalistic paradigms with fMRI and computational methods, we will examine how the human brain builds internal models at multiple levels, from low-level visuospatial coding to high-level contextual beliefs, to support continuous decisions and adaptive actions in our multidimensional world.

 

SELECTED PUBLICATIONS

1. Cognitive and neural state dynamics of narrative comprehension, Song H. et al. (2021), J Neurosci, 41, 8972-8990

2. Neural representations of perceptual color experience in the human ventral visual pathway, Kim I. et al. (2020), Proc Natl Acad Sci USA, 117, 13145-13150

3. A Unified Framework for Dopamine Signals across Timescales. Kim H.G. et al. (2020), Cell, 183, 1600-1616

4. The neural basis of predictive pursuit, Yoo S.B.M. et al. (2020), Nat Neurosci, 23, 252-259

5. The transition from evaluation to selection involves neural subspace reorganization in core reward regions. Yoo S.B.M. & Hayden B.Y. (2020), Neuron, 105, 712-724