Optimal policy for multi-alternative decisions.

Area of Science:

• Cognitive Neuroscience
• Decision Science
• Computational Neuroscience

Background:

• Everyday decisions often involve choosing from numerous alternatives.
• The optimal strategy for multi-alternative decision-making remains largely unknown.
• Existing models struggle to explain complex decision behaviors.

Purpose of the Study:

• To derive the normative policy for general multi-alternative decisions.
• To propose a neural circuit implementation for optimal decision-making.
• To explain key features of decision-making neurons and reaction time patterns.

Main Methods:

• Derived a normative policy based on evidence accumulation to nonlinear, time-dependent bounds.
• Proposed a simple neural circuit model with normalization, fixed bounds, and an urgency signal.
• Tested the model against empirical data on neural responses and reaction times.

Main Results:

• The optimal strategy involves evidence accumulation to time-dependent bounds.
• A geometric symmetry in bounds allows implementation via a simple neural circuit.
• The model accurately predicts Hick's law and captures 'irrational' behaviors like the similarity effect.

Conclusions:

• The derived policy provides a normative framework for multi-alternative decisions.
• The proposed neural circuit offers a plausible mechanism for optimal choice selection.
• The model successfully accounts for both rational and seemingly irrational decision-making phenomena.