Shih-Wei Wu

As a graduate student at NYU, I worked with Larry Maloney on modeling how humans plan movements under risk and uncertainty. Intrigued by many results that reported near-optimal performance in motor tasks and classical results detailing systematic deviations in decision-making from standard models for choice, we became interested in comparing how people make motor decisions with how they make economic decisions. This eventually led to one of the first studies that formally compare decision-making between these two domains with mathematically equivalent tasks.

I then became interested in how humans integrate information presented separately in time or from different sources, and use the integrated knowledge to guide choices. As a postdoc working with Antonio Rangel, John O’Doherty, and Shin Shimojo at CALTECH, we studied the neural mechanisms for sequential integration of value and the interaction between systems involved in updating and choice using fMRI.

I am currently an assistant professor at the Institute of Neuroscience, National Yang-Ming University, Taiwan. My lab is working on several things:

1. The tradeoff between information and time: In a lot of situations, being more patient is a good thing. This is especially true when valuable information just keeps coming at you. If you wait and let information accumulates, your chance of making better decisions increases. However, time is often a limited resource. In this case, a competition between the gain of information accumulation and the loss of time emerges into the deicsion-making process. The question is timing: when is the best time to make a decision?
2. Integration of prior knowledge and sensory evidence for reward statistics: There is a growing literature on identifying the neural mechanisms for updating computations and using the Bayesian framwork to characterize the computations. There is not a lot of work on the integration of prior knowledge and current evidence, which is conceptually related but distinct from sequential updating in the presence of new information. We are particularly interested in such integrative computations concerning reward statistics and how they guide value-based decisions.
3. Using knowledge of perceptual performance to make value-based choices: From the work of Newsome, Shadlen, and many others, we now know a great deal about perceptual decision-making. Relatively little is known about how we use knowledge about our performance in perceptual tasks to make value-guided decisions and how neural systems involved in estimating perceptual performanc interact with valuation network.
4. Value coding in decision-making networks: We currently attempt to combine theories from behavioral economics on reference dependence and models that respect and take into account neurobiological constraints to investigate issues on value coding in the ventromedial prefrontal cortex and related circuits.