Generalization of extinction of a generalization stimulus in fear learning.

Two experiments examined whether extinction of a generalization stimulus (GS) after single cue fear conditioning would in turn generalize to other stimuli, relative to a control group that received regular extinction of CS + itself. We found only a weak effect of such "generalization of GS extinction" either back to CS + or to a different GS, on either US expectancy or skin conductance measures. In other words, despite extinction trials with a stimulus highly similar to CS+, participants showed a return of fear when tested with CS + or a novel GS. However this responding declined rapidly over non-reinforced test trials. Trait anxious participants showed higher overall US expectancy ratings in the extinction and test phases, and slower extinction of expectancy, relative to low anxious participants. These results may help explain why exposure therapy, which is unlikely to reproduce the exact stimuli present at acquisition, sometimes fails to transfer to other fear-eliciting stimuli subsequently encountered by anxious clients. The generalization of GS extinction paradigm might provide a useful testbed for evaluation of interventions designed to enhance transfer, such as exposure to multiple diverse exemplars.