Decoding the future from past experience: learning shapes predictions in early visual cortex.

Area of Science:

• Neuroscience
• Cognitive Science
• Visual Perception

Background:

• Understanding environmental structure is key for scene interpretation and event prediction.
• Brain mechanisms for translating environmental statistics into sensory predictions are largely unknown.

Purpose of the Study:

• To investigate how learning temporal regularities influences neural representations in the visual cortex.
• To determine if these representations support sensory prediction abilities.

Main Methods:

• Functional magnetic resonance imaging (fMRI) decoding was used to analyze brain activity.
• Participants predicted stimulus orientation after viewing sequences of oriented gratings.
• Decoding accuracy for predicted versus cued orientations was compared between structured and random sequences.

Main Results:

• fMRI decoding identified brain patterns associated with visual predictions, distinct from stimulus-driven activity.
• Decoding of predicted orientations improved after training with structured sequences.
• Decoding of cued orientations after random sequences showed no significant change.

Conclusions:

• Learning temporal structures shapes neural representations in early visual cortex, enhancing sensory prediction.
• These predictive representations are linked to the same neural populations encoding actual stimuli.
• Findings suggest that primary visual cortex plays a role in predicting future events based on learned temporal sequences.