Response Selection Can Feed Back on Task Selection Through Episodic Retrieval

In task switching, response repetitions (RR) usually yield performance benefits compared to response switches, but only when the task repeats. When the task switches, RR benefits vanish or even turn into costs, yielding an interaction between repeating versus switching the task and the response. Several theoretical accounts were proposed to explain this interaction. In this study, we tested a prediction derived from binding and retrieval accounts which assumes that task and response are bound in each trial N. Moreover, when the task or the response repeats, it retrieves the task-response binding formed in the N−1 trial. A previous study, we showed that repeating the task indeed retrieves the response. Namely, we found higher probability of erroneously repeating the same response as in N-1 when the repeats than when it switches. This can explain why RR benefits are found in task repetitions. We now wish to examine whether repeating the response retrieves the previous task. To this aim, we will examine whether the probability of erroneously performing the n-1 task is higher in response repetitions than switches. This could explain why RR costs are found in task switches. Such an erroneous selection of the N-1 task in response repetitions might reflect retrieval of the previous episode and thus bring further support to the episodic retrieval account for the RR effect in task switching.

In task switching, response repetitions (RR) usually yield performance benefits compared to response switches, but only when the task repeats. When the task switches, RR benefits vanish or even turn into costs, yielding an interaction between repeating versus switching the task and the response. Several theoretical accounts were proposed to explain this interaction. In this study, we test a prediction derived from binding and retrieval accounts which assumes that task and response are bound in each trial N. Moreover, when the task or the response repeats, it retrieves the task-response binding formed in the N−1 trial. In a previous study, we showed that repeating the task indeed retrieves the response. Namely, we found a higher probability of erroneously repeating the same response as in N-1 when the repeats than when it switches. This can explain why RR benefits are found in task repetitions. We now conduct the second of a series of experiments to examine whether repeating the response retrieves the previous task. To this aim, we will examine whether the probability of erroneously performing the n-1 task is higher in response repetitions than switches. This could explain why RR costs are found in task switches. Such an erroneous selection of the N-1 task in response repetitions might reflect retrieval of the previous episode and thus bring further support to the episodic retrieval account for the RR effect in task switching.

This is the project folder of the manuscript Response Selection Can Feed Back on Task Selection Through Episodic Retrieval. Abstract: Goal-directed behaviour is thought to subsume integration, or binding, of perceptual and action features. In task-switching settings, this entails forming a task–response binding in each trial that can then be retrieved in the following trial. Accordingly, repeating the same response in a trial supposedly retrieves the previously relevant task (the N−1 task). In task switches, the retrieved task mismatches with the current task, which causes costs for response repetitions in task-switch trials (RR costs). In the present study (two re-analyses of published data: N = 255, N = 39, and two new experiments: Ns = 96 each), we tested such a binding and retrieval account of the RR costs by isolating specific task confusion errors, namely the erroneous re-application of the N−1 task. Coupled with the use of Multinomial Processing Tree (MPT) models, we could test the prediction, unique to the binding account of RR costs, that selecting a repeating response triggers retrieval of the N−1 task. Coherent with this prediction, the MPT model results showed a larger probability of selecting the N−1 task when the response should be repeated compared to switched. These results challenge strict feedforward processing flowing from task selection to response selection. In fact, selecting a repeating response may divert task selection from the N task towards the N−1 task, via retrieval of the bound N−1 task. Taken together, this study provides novel evidence for episodic retrieval in task switching while specifying the interplay of task and response selection.