Visual discrimination, and interpretation in terms of detectability theory.

Geometric squares, and rectangles (squares vertically elongated by 2%), were presented singly in a random sequence to three observers. By using the detectability-model parameter d′ as a measure, average discriminability was found to be 1.32. Two-, three-, and four-category response scales were shown to give equivalent discriminability data. Discriminability was not significantly affected by the inclusion of an extraneous stimulus (a square vertically elongated either 1% or 3%) interspersed randomly in the square-rectangle sequence without the observers’ knowledge. The data were interpreted as a favorable empirical test of detectability-theory assumptions applied to a visual discrimination task.