Neurally constrained modeling of perceptual decision making

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Perceptual decision-making relies on accumulating evidence over time.
• Stochastic accumulator models are a leading framework for understanding response times.
• The neural basis of evidence accumulation and its integration with perceptual input remains incompletely understood.

Purpose of the Study:

• To test alternative computational models of evidence accumulation in perceptual decision-making.
• To investigate the role of frontal eye field (FEF) neural activity in processing perceptual evidence and accumulation.
• To determine how perceptual evidence is integrated into motor preparation.

Main Methods:

• Monkeys performed a visual search task while their FEF neural activity was recorded.
• Firing rates of FEF visual and movement neurons were mapped to perceptual evidence and accumulation, respectively.
• Computational models, including those with evidence gating, were evaluated against behavioral (response time) and neural data.

Main Results:

• Models incorporating the gating of perceptual evidence to accumulating units provided the best fit to both behavioral and neural data.
• Specific neural populations in the FEF were identified as potentially representing distinct processing stages.
• Alternative model architectures were effectively ruled out by the neurophysiological data.

Conclusions:

• Evidence gating is a crucial mechanism in the neural processing of perceptual decisions.
• Distinct neural populations in the FEF support discrete stages of evidence accumulation and motor preparation.
• Neurophysiological data serve as a powerful tool for selecting between computational models, bridging neural and computational explanations.