Adaptation to implied tilt: extensive spatial extrapolation of orientation gradients

Area of Science:

• Visual perception
• Neuroscience
• Computational vision

Background:

• The visual system integrates local orientation estimates to perceive global image structure.
• Understanding how structure influences local orientation coding is crucial but underexplored.
• Previous work showed adaptation to structured patterns induces spatially remote tilt-aftereffects (TAEs).

Purpose of the Study:

• To investigate the spatial extent and properties of visually induced adaptation.
• To determine if adaptation to structure influences local orientation coding in unseen areas.
• To elucidate the role of orientation gradients in visual adaptation.

Main Methods:

• Adaptation to noisy circular textures with manipulated element distribution.
• Measurement of spatially remote tilt-aftereffects (TAEs).
• Analysis of orientation tuning properties of remote TAEs.
• Manipulation of orientation gradients in adapting stimuli.

Main Results:

• Remote TAEs were driven by orientation structure extrapolation over large visual regions (>20°).
• These effects were not limited to polar structures but required an orientation gradient.
• Adaptation mechanisms exploit orientation gradients to predict pattern content in occluded regions.

Conclusions:

• Visual adaptation actively extrapolates existing orientation gradients to infer structure in unseen areas.
• The visual system predicts local orientation in unfilled regions based on surrounding structure.
• This predictive coding mechanism is fundamental to visual perception and scene understanding.