The Use of Generalization Gradients for the Study of Mediational Processes

Area of Science:

• Cognitive Psychology
• Perception Research

Background:

• Generalization is a fundamental concept in learning theory, describing the tendency for a learned response to occur with stimuli similar to the original training stimulus.
• Verbal labeling has been explored for its potential to influence generalization gradients, but findings are sometimes inconsistent regarding the role of imagery.

Purpose of the Study:

• To investigate the effect of verbal labels with varying imagery (high vs. low) on generalization gradients.
• To determine if verbal pretraining, irrespective of label imagery, inhibits or modifies generalization patterns.

Main Methods:

• Experiment I: College students (N=60) trained on a 60° stimulus, labeled with high-imagery words, low-imagery words, or no label (control). Tested for generalization on a line-tilt dimension.
• Experiment II: College students (N=180) used high- and low-imagery labels during training, with a control group. Generalization was assessed similarly.

Main Results:

• Experiment I: High-imagery labels completely inhibited the central tendency shift observed in controls. Low-imagery labels showed a partial inhibition effect.
• Experiment II: Both high- and low-imagery labels resulted in complete inhibition of the central tendency shift, suggesting verbal pretraining's inhibitory effect is independent of label imagery.

Conclusions:

• Verbal pretraining, particularly with high-imagery labels, significantly inhibits generalization gradients.
• The acquired-distinctiveness-of-cues hypothesis provides a framework for understanding how verbal labels influence perceptual learning and generalization.