Drift-diffusion models for multiple-alternative forced-choice decision making.

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Decision Neuroscience

Background:

• The drift-diffusion model (DDM) is a standard computational tool for understanding two-alternative forced-choice decision-making.
• The DDM models decision-making as accumulating sensory evidence over time.

Purpose of the Study:

• To generalize the drift-diffusion model (DDM) for scenarios involving more than two choices.
• To explore the mathematical underpinnings and potential applications of a multi-alternative DDM.

Main Methods:

• Derivation of the multi-alternative DDM from coupled linear firing rate equations.
• Equivalence established between the multi-alternative DDM and a Bayesian sequential probability ratio test with time-varying thresholds.
• Analysis of nonlinear neuronal systems leading to lower-dimensional diffusion processes.

Main Results:

• A novel multi-alternative drift-diffusion model (DDM) was derived, accounting for correlated noise across decision variables.
• The model demonstrates equivalence to a generalized Bayesian sequential test.
• For nonlinear systems, the DDM dynamics were shown to correspond to particle motion on a potential landscape.

Conclusions:

• The generalized DDM provides a flexible framework for modeling multi-alternative decision-making.
• This work bridges computational models of decision-making with neurobiological implementations.
• The findings offer analytical insights into the dynamics of complex choice behavior.