“Priming” refers to a pervasive phenomenon in which the repetition of a particular stimulus, response, or thought process facilitates its subsequent use. Might this phenomenon extend to more “executive” capacities as well?
In a recent article from theJournal of Experimental Psychology: Human Perception and Performance, Verbruggen, Logan, Liefooghe & Vandierendonck explore the phenomenon of repetition priming in the stop-signal paradigm. The stop signal paradigm requires subjects to make one or more “Go” responses to a particular stimulus class (typically, to press one button in response to one shape, and another button in response to a different shape) unless these Go stimuli are followed by some kind of “stop signal” (typically, an auditory “beep”). If they hear the stop signal, subjects must try to prevent themselves from responding, which is increasingly difficult the later in time the stop signal is presented (since subjects are then closer to making a response).
Previous work in this paradigm showed that Go responses on trials without a stop signal are slower following trials with a stop signal than following trials without a stop signal. One conclusion is that subjects are engaging inhibition, and that this lingers in a way that makes subsequent responding slower. But there are other reasons this might occur.
For example: subjects might slow down because they are more likely to make an error on stop signal trials – known as post-error slowing. Or subjects might simply have their attention captured by this infrequent stimulus, and thus be slower on trials that occur after it. Or the occurrence of the signal might remind subjects that they need to monitor for its occurrence, thus correlating with a general slowing of behavior. These possibilities have made previous work difficult to interpret.
In clarifying this state of affairs, Verbruggen et al hypothesized that those Go trials which follow correctly inhibited responses should be slower than those which do not follow a stop trial – regardless of whether the exact Go stimulus was repeated between trials. This hypothesis implies that the aftereffects (“priming”) of response inhibition must reflect a rapidly formed stimulus-response association – or, more accurately in this case, between a stimulus and the absence of a response.
A first experiment showed not only that responses were slower on Go trials which followed correctly-inhibited stop trials (relative to those following another Go trial) and that this was true only when the same go stimulus was repeated between trials. Conversely, responses were also slower on those Go trials which followed unsuccessfully-inhibited stop trials, regardless of whether the stimulus was repeated – implying a conflict monitoring or post-error slowing effect. A second experiment replicated the slowing on Go trials following a signal trial, but this did not differ depending on whether the same Go stimulus was repeated.
In a third experiment, Verbruggen et al demonstrated that these effects were due to stimulus repetitions (and not merely the repetition of the response those stimuli required) by adopting a “data reduction” paradigm (in which 4 possible stimuli were mapped to 2 possible responses, allowing for the presence of response repetition in the absence of stimulus repetition). They also replicated all previous effects (including post-error slowing), but response-repetitions yielded no slowing on Go trials when responses repeated and stimuli had switched from the preceding stop trial. In contrast, unsuccessful inhibition was always followed by slowing, again indicating post-error slowing or conflict monitoring. In yet another experiment, Verbruggen et al demonstrated that this stimulus-repetition cost was similar regardless of whether the exact stimulus was repeated or merely its abstract category.
In a final experiment, Verbruggen et al examined whether attentional capture might explain the results by adopting a “selective stop” paradigm, in which only one of two infrequent signals requires subjects to stop. If the aftereffects are due solely to the presentation of a infrequent signal, there should be no differential slowing depending on the type of signal presented on the previous trial. In contrast, if aftereffects have to do with response inhibition per se, Go trials following validly-signalled trials should be slower than those following invalid and “no signal” trials, but no differences between invalid and no signals. Verbruggen et al actually observed a slowing in invalid and valid cases, though the latter was larger, suggesting roles for both attentional capture and response inhibition. Alternatively, attentional capture may be greater when rare signals have greater task-relevance.
Subsequent median-split analyses of invalid trial RTs showed that subjects were slower on a trial following an invalidly-signalled trial with a fast RT (relative to the median). The authors suggest this pattern is the opposite predicted by the “always inhibit, and then respond if signal is invalid” theory: according to that account, subjects would have used response inhibition on those slow trials, and thus should be more slowed following those trials. On the other hand, this result is consistent with an account where the signal reminds subjects of the need to monitor for the occurrence of a signal. And of course, the presence of unpredicted effects raises the possibility that perhaps no theoretical account is correct.
To summarize, Verbruggen et al showed that presentation of rare signals leads to a slowing of responses, particularly if those signals require the need to inhibit and/or are task-relevant. In addition, this subsequent slowing is limited to stimulus-repetitions, suggesting that the aftereffects of inhibition (beyond attentional capture) are due to stimulus-based associations.
Frederick Verbruggen, Gordon D. Logan, Baptist Liefooghe, Andre Vandierendonck (2008). Short-term aftereffects of response inhibition: Repetition priming or between-trial control adjustments? Journal of Experimental Psychology: Human Perception and Performance, 34 (2), 413-426 DOI: 10.1037/0096-15188.8.131.523