Inferring latent behavioral strategy from the representational geometry of prefrontal cortex activity.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Behavioral tasks can be solved using multiple strategies, some of which are not directly observable.
• Understanding latent strategies is crucial for deciphering decision-making processes.

Purpose of the Study:

• To infer the latent behavioral strategy employed by monkeys in a working memory updating task.
• To compare neural representations in prefrontal cortex with computational models.

Main Methods:

• Compared representational geometry of lateral prefrontal cortex (LPFC) and prearcuate cortex (PAC) in monkeys.
• Utilized recurrent neural network (RNN) models trained on different task strategies.
• Analyzed neural activity patterns and model strategy alignment.

Main Results:

• Neural activity in both LPFC and PAC aligned with a single, specific latent strategy.
• This suggests monkeys consistently employ this identified strategy for the task.

Conclusions:

• Monkeys utilize a specific latent strategy for working memory updating.
• Findings provide insights into strategy learning and choice mechanisms in the brain.