Neuronal Adaptation Reveals a Suboptimal Decoding of Orientation Tuned Populations in the Mouse Visual Cortex

Area of Science:

• Systems Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Cortical neurons encode sensory information, but how this is used for perceptual choices remains unclear.
• Visual adaptation in the primary visual cortex (V1) offers a model to study perceptual decision-making computations.

Purpose of the Study:

• To determine the neural computation underlying orientation discrimination.
• To investigate how V1 adaptation affects task performance and neural representations.
• To identify the computations consistent with behavioral and neural data.

Main Methods:

• Designed a stimulus paradigm to vary V1 neuronal adaptation during an orientation discrimination task.
• Measured V1 population responses and mouse behavior (both sexes).
• Used decoding methods to link neural activity to choices and tested computational models.

Main Results:

• Adaptation increased behavioral thresholds despite enhancing neural orientation representation reliability.
• Decoding revealed an overreliance on target-preferring neurons and failure to discount distractor-preferring neurons.
• Behavior was susceptible to distractor orientation and V1 optogenetic suppression, consistent with an all-positive computation.

Conclusions:

• The perceptual choice circuit uses a suboptimal, task-specific computation that discards relevant information.
• This "all-positive computation" may be favored for its learning simplicity and processing speed.
• This strategy highlights how sensory circuits can deviate from optimal information integration in decision-making.