NACS Seminar: Dr. Ralf Haefner

Behavioral and neural signatures of approximate hierarchical inference

Since motion can only be defined relative to a reference frame, which reference frame guides perception?
A century of psychophysical studies has produced conflicting evidence: retinotopic, egocentric, world-centric, or even object-centric. I will present our recent work on a hierarchical Bayesian model mapping retinal velocities to perceived velocities as well as the psychophysical data that support critical predictions of our model, and that allow us to infer the subjective set of references frames used by individual observers. Our model provides a quantitative normative justification for key Gestalt principles providing inspiration for building better models of visual processing in general.
If time permits, I will also present work on the neural basis of our model, both by linking it to divisive normalization, as well as by describing a general method for testing Bayesian models using neural data.

Dr. Ralf Haefner is an Associate Professor at the University of Rochester.


NACS Seminars are free and open to the public.