Dear All,
The CEU Department of Cognitive Science and the Center for Cognitive Computation invites you to the following talk:
Title: Inverse optimal control for modeling continuous perception and action
Abstract: Normative models of behavior strive to explain why behavior unfolds the way it does. These models have been highly successful in explaining many phenomena in neuroscience,
cognitive science, and related fields. The power of these approaches derives from the combination of controlled experimental designs with their associated normative models, e.g. forced-choice psychophysics with Bayesian observer models. Unfortunately, these
tasks do not have much in common with naturalistic behavior and require many trials with binary responses, often by highly trained participants. Continuous psychophysics is a recently developed experimental approach that abandons the rigid trial structure
and replaces it with a more naturalistic and intuitive continuous tracking task. It produces more temporally fine-grained measurements and allow efficient data collection even with untrained participants. However, while highly controlled classical psychophysics
tasks allow using normative models to estimate perceptual uncertainty, continuous tasks introduce additional cognitive and motor factors such as action variability, internal behavioral costs, and subjective internal models. In this talk, I will introduce a
normative analysis framework based on Bayesian inverse optimal control that accounts for these factors. This approach enables estimating perceptual uncertainty, action variability, behavioral costs, and subjective beliefs about the task dynamics from behavior
in a tracking task. I will then discuss the limitations of the method and show recent methodological extensions that address these limitations and allow applying inverse optimal control to a wider range of tasks including information-seeking behavior. In summary,
these methods open up the possibility of fitting normative models to more naturalistic continuous behavior.
Chair: József Fiser
Time and date: 5 PM, Tuesday, 8 October 2024
Venue: CEU Budapest site (1051 Budapest, Nádor u. 15.) N15. room 101. Quantum
Zoom Meeting: https://ceu-edu.zoom.us/j/96878075997?pwd=hQTHk0baU8S19Peavk94XLILP9Pmr0.1
Meeting ID: 968 7807 5997 Passcode: 393441
Please, be informed that video/photo recording might take place at the event and the edited version of the video material might be published to communicate or promote CEU PU's activities. Please, find our Privacy Notice
here.
Best regards,