Navigation, Learning, and Decision-making under Uncertainty

I am a computational neuroscientist in the Computational and Biological Learning Lab, Department of Engineering at University of Cambridge (Lengyel group). In 2019, I was elected a Junior Research Fellow at Wolfson College, University of Cambridge. Before Cambridge, I obtained my PhD supervised by Michael Shadlen at Columbia University. Prior to PhD, I was trained in biology (MD at Seoul National University, South Korea) and informatics (Summer/Winter Schools for International Olympiad in Informatics, South Korea).

I study how the brain guides behavior under uncertainty. I make computational models of the neural activity and behavioral patterns and test them with data from the literature or from experimental collaborators. I also perform behavioral experiments to test the models.

I build theories first from a normative perspective, to find the optimal way to perform a task (e.g., navigation, learning, or decision-making). This provides an upper-bound for the performance, and exposes biological constraints that limit the actual performance (e.g., decisions/neural responses made less accurate and slower than optimal). I also develop computational methods to efficiently test hypotheses using modern machine learning methods.

I found that:

• Conscious awareness: people become aware that they reached a decision when evidence for the decision is accumulated up to a threshold, and that we can predict the accuracy of their decision from the timing of the awareness (Kang et al. Current Biology 2017; the Independent)
• Dual task: two decisions about one object cannot be made simultaneously; they are made one by one, and evidence for each accumulate in an interleaved fashion (Kang et al. bioRxiv 2020; under revision at eLife)
• Episodic memory: memory of a unique episode is retained with a graded sense of uncertainty, which is later used to weight the reliability of the memory on retrieval (Kang et al. CogSci 2019)
• Spatial navigation: grid fields (so-called “ruler in the brain”), which were thought to “deform” depending on the geometry of the environment, in fact represent uncertainty about the animal’s location, which can be predicted by a unifying theory that considers how the geometry of the environment is perceived from a first-person view. The same theory jointly accounts for the bias and variability in homing behavior (Cosyne 2020 & 2021).

When inspiration calls, I like making artworks, in and outside science.

Collaborators

• Máté Lengyel (University of Cambridge)
• Daniel Wolpert (Columbia University)
• Michael Shadlen (Columbia University)
• Gergely Csibra (Central European University, Birbeck University of London)
• Dora Angelaki (NYU)