BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Inverse normative modeling of continuous perception and action - D
 ominik Straub\, TU Darmstadt
DTSTART:20250211T150000Z
DTEND:20250211T154500Z
UID:TALK228037@talks.cam.ac.uk
CONTACT:Daniel Kornai
DESCRIPTION:Normative models of behavior strive to explain why behavior un
 folds the way it does and have been highly successful in explaining many p
 henomena in neuroscience\, cognitive science\, and related fields. The pow
 er of these approaches derives from the combination of controlled experime
 ntal designs with their associated normative models\, e.g. forced-choice p
 sychophysics experiments with Bayesian observer models. Unfortunately\, th
 ese tasks do not have much in common with real-world behavior\, as they di
 vide behavior into independent trials with discrete responses\, often by h
 ighly trained participants. In naturalistic tasks\, however\, behavior is 
 typically continuous and sequential. While highly controlled classical psy
 chophysics tasks allow using normative models to estimate perceptual uncer
 tainty and biases\, naturalistic tasks introduce additional cognitive and 
 motor factors such as action variability\, intrinsic behavioral costs\, an
 d subjective internal models. To account for these factors\, I propose to 
 apply inverse normative modeling\, i.e. to infer the components of normati
 ve models from behavior. In this talk\, I will first present recent work t
 hat extends Bayesian models of perception to more general cost functions i
 ncluding intrinsic behavioral costs. I will then apply inverse normative m
 odeling to continuous psychophysics. This recently developed experimental 
 approach abandons the rigid trial structure of classical psychophysics and
  replaces it with a more naturalistic and intuitive continuous tracking ta
 sk. It produces more temporally fine-grained measurements and allows effic
 ient data collection even with untrained participants. Using Bayesian inve
 rse optimal control\, perceptual uncertainty\, action variability\, behavi
 oral costs\, and subjective beliefs about the task dynamics can be estimat
 ed from behavior in a tracking task. Finally\, I will discuss some limitat
 ions of the method and show recent methodological extensions that address 
 these limitations and allow applying inverse optimal control to a wider ra
 nge of tasks. In summary\, these methods open up the possibility of fittin
 g normative models to more naturalistic continuous behavior.
LOCATION:CBL Seminar Room\, Engineering Department\, 4th floor Baker build
 ing
END:VEVENT
END:VCALENDAR