The Perceptual Maze Test Revisited: Evaluating the Difficulty of Automatically Generated Mazes

Area of Science:

• Psychology
• Cognitive Science
• Psychometrics

Background:

• The Elithorn perceptual maze test is a common clinical tool.
• Limited psychometric data and difficulty in creating new mazes hinder its use.

Purpose of the Study:

• To rigorously evaluate 18 automatically generated mazes using item response theory.
• To assess the psychometric properties of a novel Automatic Perceptual Maze Test.

Main Methods:

• Utilized an R software package for automatic maze generation.
• Applied various item response theory models (Rasch, LLTM, LLTM+Error) to analyze difficulty parameters.
• Correlated maze test scores with a nonverbal intelligence test.

Main Results:

• The Rasch model best fit the data.
• The linear logistic test model (LLTM) identified significant sources of maze difficulty.
• The LLTM plus error model was the most parsimonious.
• The Automatic Perceptual Maze Test showed moderate correlation with nonverbal intelligence.

Conclusions:

• The Automatic Perceptual Maze Test, with its expanded maze set, offers improved assessment of cognitive performance.
• This tool has potential for future clinical and research applications in cognitive studies.