Non-monotonic Temporal-Weighting Indicates a Dynamically Modulated Evidence-Integration Mechanism

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Modeling

Background:

• Perceptual decisions rely on integrating noisy evidence over time.
• The temporal weighting of evidence significantly impacts decision quality.

Purpose of the Study:

• To quantify temporal weighting profiles in perceptual decision-making.
• To investigate how stimulus duration influences temporal weighting.
• To compare empirical findings with existing computational models.

Main Methods:

• Participants judged the overall brighter of two fluctuating disks presented for 1, 2, or 3 seconds.
• A signal-perturbation method with controlled temporal dispersions was used to quantify weighting.
• Temporal weighting profiles were derived from participant responses.

Main Results:

• A primacy bias was observed for 1- and 2-second stimuli.
• A non-monotonic temporal weighting profile, showing both primacy and recency, emerged for 3-second stimuli.
• These results challenge predictions from drift-diffusion and Ornstein-Uhlenbeck models.

Conclusions:

• Existing computational models may not fully capture temporal weighting in perceptual decisions.
• A dynamic leaky-competing accumulator model is proposed as a potential explanation.
• The findings suggest complex neural mechanisms underlying temporal integration in decision-making.