A neurocomputational model for the processing of conflicting information in context-dependent decision tasks

Area of Science:

• Computational Neuroscience
• Cognitive Science
• Decision Neuroscience

Background:

• Neural systems exhibit context-dependent computation, crucial for adaptive behavior and information discrimination.
• Flexible information routing and context-dependent associations are key challenges in computational neuroscience.
• Conflicting tasks highlight the need for adaptive behavioral responses and stimulus discrimination.

Purpose of the Study:

• To extend a context-dependent associative memory model for decision-making with conflicting, noisy stimuli.
• To integrate noisy dynamics into the context-dependent associative memory framework.
• To validate the model's efficacy using behavioral data from a monkey decision-making experiment.

Main Methods:

• Utilized Kronecker tensor product for multiplying input and context vectors.
• Embedded the context-dependent associative memory within a leaky competing accumulator model for noisy dynamics.
• Validated the model against a behavioral experiment involving monkeys performing a context-dependent conflicting decision-making task.

Main Results:

• The extended model successfully reproduced monkey behavior in a context-dependent conflicting decision-making task.
• Demonstrated the model's capability to handle noisy, multi-attribute stimuli.
• Showcased the model's power in simulating context-dependent decision processes.

Conclusions:

• The developed tensor context model aligns with recent experimental findings on functional flexibility in neural organization.
• The model provides a feasible neural explanation for context-dependent computation and decision-making.
• Suggests tensor product operations as a potential mechanism for flexible information routing in the brain.