REGISTER HERE

How big is that bagel? Differences in size constancy for perception and action.

Speaker: Mel Goodale
The University of Western Ontario, Canada

February 26, 2025
11:15 AM - 1:00 PM U.S. Eastern Time 

Abstract 
The images of people and objects on our retina are constantly shrinking, expanding, and changing shape as we move through the world.  Yet remarkably we see a world that is stable, and things are most often perceived to be (roughly) the size they really are.  This is a good thing because otherwise our perception of the world would be chaotic and impossible to interpret. Our ability to see the real-world size of objects despite dramatic changes in the images captured by our eyes is called “size constancy.”  It’s thought that our brain creates size constancy by taking into account how far away an object is and combining that information with the size of the object’s retinal image.  As a consequence, even though the image of a car driving away from us becomes smaller and smaller on our retina, we don’t see it as shrinking in size but instead as a car that is the same size, but further away.

It is seldom appreciated, however, that object constancies must also operate for the visual guidance of goal-directed action. For example, when we reach out to pick up an object, our hand’s in-flight aperture scales to the size of the goal-object and is refractory to the decrease in retinal image size with increased viewing distance, a phenomenon we have named ‘grip constancy.’ It turns out that the visual cues and the underlying neural circuits used for perceptual size constancy and grip constancy are not always the same. Although primary visual cortex (V1) and the ventral visual stream are necessary for perceptual size constancy, they are unnecessary for grip constancy, which is mediated instead by separate visual inputs to dorsal-stream visuomotor areas that bypass V1. Understanding how the brain maintains size constancy in perception and action can help engineers who are trying to devise machine vision systems for everything from robots to self-driving cars. 

Return to the full 2024-2025 program.