Covariation bias, classical conditioning, and phobic fear.

The present study investigated whether phobics show an illusory correlation (IC) between phobia-relevant stimuli and aversive events. Nineteen treated and 19 untreated spider phobics were exposed to a series of 72 slides. Three different categories were used: Phobia-relevant slides (spiders), alternative fear-relevant slides (weapons), and neutral slides (flowers). Slides were randomly paired with either a shock, a siren, or nothing at all. All slide/outcome combinations occurred equally frequently. A posteriori recorded contingency estimates indicated that untreated phobics dramatically overestimate the covariation of spiders and shock. On-line recorded outcome expectancies revealed that the bias to overestimate the spider-shock contingency is highly resistant to extinction. The covariation bias was accompanied by differentially heightened electrodermal first interval responses (FIR) and unconditioned electrodermal responses (third interval responses: TIR) on phobia-relevant trials. Treated phobics did not show a covariation bias, indicating that such bias can be modulated by behavioral treatment. The present findings sustain the hypothesis that phobic subjects process information in a fear-confirming way.