Mostly Biomathematics Lunchtime Seminar

Visual Texture: a bridge between algorithmically-constructed and naturalistic stimuli

Speaker: Jonathan Victor, Weill Cornell Medical College

Location: Warren Weaver Hall 1314

Date: Tuesday, November 24, 2026, 12:45 p.m.

Synopsis:

A central goal of visual neuroscience is to understand the computations that enable the extraction of visual form from sensory inputs.  Simple mathematically-motivated stimulus sets, such as sine waves and noise ensembles, have proven effective for analyzing the earliest stages of processing. However, their  effectiveness is limited for probing more central stages, likely because they lack the features of natural scenes to which central neurons and circuits are tuned. The use of natural scenes as stimuli directly addresses this issue, but they are challenging to characterize mathematically and their features are intertwined, leading to other analytic challenges. Visual texture provides a bridge between these strategies: they have explicit mathematical properties and can be specified by a flexible range of visual features, including not only features in direct correspondence to those of natural scenes, but also, analytically-useful out-of-sample features. Their use enables a dissection of intermediate stages of visual processing via neurophysiological and psychophysical studies.