Accounting for speed-accuracy tradeoff in perceptual learning

Area of Science:

• Cognitive Psychology
• Neuroscience
• Perception

Background:

• Perceptual learning (PL) studies often focus on accuracy (d') or response times (RT) separately.
• Cognitive skill practice research emphasizes RT improvements.
• The speed-accuracy tradeoff (SAT) phenomenon, where faster responses yield more errors, necessitates considering both measures.

Purpose of the Study:

• To advocate for the integrated analysis of accuracy and RT in perceptual learning experiments.
• To investigate the utility of the diffusion model in explaining SAT within PL.
• To identify a more robust index for quantifying perceptual learning.

Main Methods:

• Applied the diffusion model to analyze data from a yes-no coherent motion detection task.
• Examined the relationship between stimulus discriminability, response accuracy, and response time.
• Assessed changes in diffusion model parameters, specifically drift rate and evidence accumulation threshold.

Main Results:

• The diffusion model successfully explained the observed speed-accuracy tradeoff.
• Results indicated that observers did not maintain a fixed threshold for evidence accumulation.
• Changes in accuracy alone may not accurately reflect learning due to variable thresholds.

Conclusions:

• Accounting for both accuracy and response time is crucial in perceptual learning research.
• The diffusion model provides a framework for understanding SAT and quantifying learning.
• Drift rate emerges as a promising and sensitive index for measuring perceptual learning gains.