The Use of Generalization Gradients for the Study of Mediational Processes

In Experiment I three groups (N = 60 college students) were presented with a 60° training stimulus (TS) and subsequently tested for generalization along a line-tilt dimension with the use of an asymmetrical test series. During training, one group labeled the TS with a high-imagery word, another with a low-imagery word, while a third served as a no-label control. While the control group showed an expected central tendency shift, the shift was completely inhibited for the high-imagery group. The low-imagery group's gradient appeared to show a partial inhibition effect, although statistical confirmation for a differential imagery effect was not obtained. In Experiment II (N = 180 college students) three words of high and three of low imagery were employed as labels. While the control condition resulted in a central tendency shift, both labeling conditions produced complete inhibition, indicating that verbal pretraining produces inhibitory effects regardless of the labels employed. The findings were discussed in terms of the acquired-distinctiveness-of-cues hypothesis.