Temporal dynamics of short-term neural adaptation across human visual cortex

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Neural adaptation is a fundamental property of sensory systems, crucial for processing continuous streams of information.
• While adaptation is widespread in the visual cortex, its hierarchical organization and underlying computational mechanisms remain incompletely understood.

Approach:

• Analyzed intracranial electroencephalography (iEEG) data from human participants viewing naturalistic images.
• Characterized short-term neural adaptation signatures, including adaptation rate and recovery time, across visual areas (V1-V3, ventral-, and lateral-occipitotemporal cortex).
• Developed and validated an augmented delayed divisive normalization (DN) model to capture category-specific adaptation dynamics.

Key Points:

• Higher visual areas (ventral- and lateral-occipitotemporal cortex) exhibit slower adaptation and prolonged recovery compared to early visual areas (V1-V3).
• Recovery from adaptation is slower for preferred stimuli in category-selective regions.
• The augmented DN model, incorporating category-dependent input strength and area-specific normalization dynamics, accurately predicts observed neural responses.

Conclusions:

• Demonstrated systematic differences in temporal adaptation across the human visual hierarchy.
• Showcased that a unified computational model of history-dependent normalization dynamics can explain these hierarchical differences with area-specific parameters.