A neural circuit framework for economic choice: from building blocks of valuation to compositionality in multitasking
Preprint
Abstract
SUMMARY: Value-guided decisions are a cornerstone of cognition, yet the underlying circuit-level mechanisms remain elusive. We used reinforcement learning to train recurrent neural network models endowed with Dale’s law on a battery of economic choice tasks, which revealed a two-stage computational framework. First, value estimation occurs at input level where learned weights store subjective preferences and approximate the nonlinear multiplication of reward magnitude and probability to yield expected values. This feedforward mechanism enables generalization to novel choice options. Second, option values are compared within the recurrent network, where specific connectivity patterns mediate robust winner-take-all decisions, with both excitatory and inhibitory neurons exhibiting value and choice selectivity. By training a single network on multiple tasks, we show compositional representations combining a shared computational schema with specialized neural modules. Reproducing key neurophysiological findings from the primate orbitofrontal cortex, our model unifies value computation, comparison, and generalization into a coherent framework with testable predictions. IN BRIEF: Battista et al. use biologically plausible RNNs to uncover circuit mechanisms of economic choice. They propose a two-stage framework where feedforward inputs compute offer values and recurrent inhibition drives comparison. This architecture explains how the brain generalizes preferences and multitasks using compositional neural codes, offering a unified theory of decisionmaking. HIGHLIGHTS: Biologically plausible RNN reveals a two-stage economic choice framework Input weights store preferences and multiply features to compute offer value Competitive recurrent inhibition mediates winner-take-all comparison Multitasking relies on compositional shared and specialized neural modulesKeywords
All Related Versions
- Published version: Neuron, 114 (5), 952.
Funding Information
- Swartz Foundation
- ONR (N00014-23-1-2040)
- NIH (R01MH104494, R01DA032758, U19NS123714, R01DA055709)
This publication has 50 references indexed in Scilit:
- Logistic analysis of choice data: A primerNeuron, 2022
- Artificial Neural Networks for Neuroscientists: A PrimerNeuron, 2020
- Orbitofrontal Cortex: A Neural Circuit for Economic DecisionsNeuron, 2017
- Specialized Representations of Value in the Orbital and Ventrolateral Prefrontal Cortex: Desirability versus Availability of OutcomesNeuron, 2017
- Neuronal Origins of Choice Variability in Economic DecisionsNeuron, 2013
- Goal-directed decision making as probabilistic inference: A computational framework and potential neural correlates.Psychological Review, 2012
- Recurrent neuronal circuits in the neocortexCurrent Biology, 2007
- Neurons in the orbitofrontal cortex encode economic valueNature, 2006
- Probabilistic Decision Making by Slow Reverberation in Cortical CircuitsNeuron, 2002
- Neural correlates of decision variables in parietal cortexNature, 1999