The importance of mixed selectivity in complex cognitive tasks.

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Single-neuron activity in the prefrontal cortex (PFC) exhibits mixed selectivity, responding to multiple task aspects.
• This mixed selectivity is heterogeneous and challenging to interpret, hindering understanding of PFC functions.

Purpose of the Study:

• To investigate the role of mixed selectivity in cognitive functions associated with the PFC.
• To analyze neural activity in monkeys during an object sequence memory task.

Main Methods:

• Analysis of neural activity recorded from monkeys performing an object sequence memory task.
• Decoding task-relevant aspects from neuronal population activity.
• Assessing the computational advantages of mixed selectivity.

Main Results:

• Mixed selectivity neurons encode distributed information across all task-relevant aspects.
• Task aspects are decodable from neuronal populations even after eliminating single-cell selectivity.
• Mixed selectivity provides computational benefits due to diverse nonlinear selectivity and high-dimensional representations.
• Neural representation dimensionality predicts animal behavior, collapsing during error trials.

Conclusions:

• Mixed selectivity neurons are crucial for PFC cognitive functions, encoding distributed information and offering computational advantages.
• The high dimensionality of these neural representations is linked to behavioral performance.
• Future research should focus on analyzing mixed selectivity neurons rather than solely on interpretable single-cell tuning.