Quantifying the effect of intertrial dependence on perceptual decisions.

Area of Science:

• Perceptual sciences
• Cognitive psychology
• Psychophysics

Background:

• Perceptual decisions are influenced by stimuli and internal factors.
• Internal factors can cause serial dependencies, complicating causal inference.
• Previous studies often neglect serial dependencies due to a lack of reliable quantification tools.

Purpose of the Study:

• To develop a statistical method for detecting, estimating, and correcting serial dependencies in behavioral data.
• To investigate the extent to which serial dependencies influence perceptual decisions, even in trained observers.

Main Methods:

• Development of a novel statistical approach to quantify serial dependencies.
• Application of the method to behavioral data from psychophysical experiments.
• Analysis of decision variance to differentiate stimulus-dependent and history-dependent components.

Main Results:

• A new statistical method effectively detects and corrects for serial dependencies.
• Trained psychophysical observers exhibit significant history dependence.
• A considerable portion of decision variance, especially for difficult stimuli, is attributable to experimental history rather than the current stimulus.

Conclusions:

• Serial dependencies significantly impact perceptual decisions, even in controlled experimental settings.
• The developed method provides a reliable tool for accounting for these internal factors.
• Correcting for serial dependencies is crucial for accurate causal inference in perceptual research.