Prospective decision making for randomly moving visual stimuli.

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Decision Science

Background:

• Humans frequently attempt to predict future events despite inherent uncertainty.
• Understanding the cognitive mechanisms behind prospective decision-making biases is crucial.

Purpose of the Study:

• To investigate the computational strategies underlying human biases in predicting future positions of randomly moving objects.
• To explore how observers utilize visual motion information for future event prediction.

Main Methods:

• A psychophysical task involving observers predicting the future position of a Gabor target after stimulus offset.
• Reverse correlation analysis to determine which parts of the target's motion history influenced decisions.
• Modeling behavioral data using a leaky-integrator model with adaptive gain control.

Main Results:

• Observer predictions were heavily influenced by the target's most recent velocity, particularly after direction reversals.
• When a linear trend was present in the motion, observers incorporated this information into their predictions.
• A leaky-integrator model successfully explained the observed behavioral patterns.

Conclusions:

• Prospective decision-making relies on accumulating recent sensory evidence, similar to retrospective judgments.
• The findings suggest shared computational principles between predicting future and recalling past events.
• Adaptive gain control in evidence accumulation may play a role in future event prediction.