A simple two-stage model predicts response time distributions.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Psychology

Background:

• Reaction time (RT) has been modeled using two distinct mechanisms: random-walk for difficult stimuli and linear rise-to-threshold for easier stimuli influenced by higher-level factors.
• Previous models did not fully reconcile these two approaches, leading to an incomplete understanding of neural decision processes.

Purpose of the Study:

• To develop and validate a unified model of reaction time that integrates both noisy sensory integration and decision-making processes.
• To reconcile previously disparate models of reaction time based on stimulus detectability and cognitive factors.

Main Methods:

• A novel computational model was developed, incorporating a sensory detector (random-walk) followed by a decision-maker (linear rise-to-threshold).
• The model's predictions for mean response times, RT distributions, and decision error rates were rigorously tested against empirical data across a range of stimulus detectability levels.

Main Results:

• The proposed sequential model successfully predicted not only mean response times but also the detailed distribution of these times.
• The model accurately accounted for the rate of decision errors across a wide spectrum of stimulus detectability.
• This unified model reconciles findings from previous, distinct reaction time theories.

Conclusions:

• A unified model, combining noisy signal integration and linear rise-to-threshold decision, provides a more complete explanation of reaction time.
• This research advances our understanding of the neural mechanisms underlying human decision-making and reaction time.
• The findings bridge the gap between models for easy and difficult stimulus detection scenarios.