Classification errors and response times over multiple distributed sessions as a function of category structure.

Area of Science:

• Cognitive Psychology
• Concept Learning
• Categorization Research

Background:

• Concept learning research traditionally uses classification accuracy to order learning difficulty.
• Stability of these orderings over extended learning and the role of response times (RTs) are less explored.

Purpose of the Study:

• To investigate the stability of learning difficulty orderings for categorical stimuli across multiple sessions.
• To examine the influence of classification accuracy (errors) and response times (RTs) on perceived learning difficulty.

Main Methods:

• Two experiments used a well-established family of category structures with three binary dimensions and four positive examples.
• Participants learned categories of 'flasks' varying in size, shape, and color.
• Experiment 1: Random presentation of all category types across three sessions.
• Experiment 2: Repeated exposure to single category types across sessions.

Main Results:

• Experiment 1 showed stable difficulty orderings for both errors and RTs across sessions with mixed stimuli.
• Experiment 2 revealed that error-based orderings disappeared with repeated exposure, while RT-based orderings emerged.
• This suggests processing difficulty persists even with reduced errors.

Conclusions:

• Learning difficulty orderings can be stable or dynamic depending on stimulus presentation and the measure used (errors vs. RTs).
• Response times may offer a more sensitive measure of persistent learning difficulty than accuracy alone.
• Findings inform concept learning and perceptual expertise theories.