Rapid decision threshold modulation by reward rate in a neural network

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Decision Science

Background:

• Optimal performance in decision-making tasks relies on the drift-diffusion model, requiring adaptable thresholds.
• Individuals often adjust response thresholds to maximize rewards, but models for this adaptation are scarce.
• Existing models lack neurally plausible mechanisms for threshold adaptation.

Purpose of the Study:

• To propose a novel neural network model for adaptive threshold modulation in decision-making.
• To investigate how neural networks can maximize reward rate by adjusting decision thresholds.
• To provide a computational framework for understanding reward-driven behavior.

Main Methods:

• Development of a neural network implementing the drift-diffusion model with adaptive thresholds.
• Simulation of the model under various task conditions to assess performance.
• Analysis of model predictions for optimal performance and neural encoding of reward rate.

Main Results:

• The proposed neural network successfully adapts thresholds to maximize reward rate.
• The model generates predictions for optimal performance benchmarks.
• The study outlines testable hypotheses regarding neural mechanisms for reward rate encoding.

Conclusions:

• A neurally plausible neural network model for adaptive threshold control in decision-making is presented.
• The model offers a framework for understanding reward maximization strategies in cognitive tasks.
• This work provides a foundation for further research into the neural basis of decision optimization.