Automaticity and conscious control in single and choice response time versions of the Perruchet effect

The Perruchet effect refers to a dissociation between conscious expectation of an event and the strength or speed of the response elicited by that event, which is revealed through sequential analyses of consecutive trials. For instance in a cued go/no-go task, over a run of consecutive go responses, participants perform the go task with increasing speed but display decreasing expectancy of the go response when asked to rate the likelihood of its occurrence, in line with a gambler's fallacy. The effect has conventionally been shown in procedures that measure expectancy and responding in separate blocks of trials (separate measurement) or measure both on the same trial (concurrent measurement). Here we directly compared the trends produced in separate and concurrent measurement procedures, using single response and two-choice response procedures. While the single response showed the same trends regardless of measurement style, the Perruchet dissociation disappeared when concurrent measurement was used with the choice response task. Furthermore, in the concurrent measurement versions of both tasks, those participants who produced a consistent gambler's fallacy did not show the standard effects of positive recency that underpin the Perruchet effect in response time. The results suggest that explicit expectation of events and facilitation based on recent trial history have dissociable but competing effects on response performance.