On the distinction of different types of between-task interference in dual-tasks

Abstract

The widely-accepted response selection bottleneck model of dual-tasking assumes that two tasks gain access to the stage of central processing in a strictly serial manner. However, a frequent observation that contradicts this assumption is that performance in Task 1 is already influenced by certain aspects of Task 2. Such observations were thus termed backward crosstalk effects (BCEs). For instance, response times (RTs) in Task 1 are shorter when the required Response 1 and Response 2 overlap spatially (the R1-R2 BCE) and when Stimulus 1 and Response 2 overlap conceptually (the S1-R2 BCE) than when they do not – the compatibility based BCEs. Similarly, RTs in Task 1 are shorter when a Response 2 is given (go-trial) than when it is withheld (no-go trial) – the no-go BCE. The main question of the present dissertation was: Can we distinguish different types of backward crosstalk? To answer this question the nogo BCE was compared to the compatibility-based BCEs based on its underlying processing stages (Study 1), the mechanism it is caused by (Study 2), and the way in which task processing is adjusted following a no-go trial (Study 3). The results of Studies 1-3 indicate that the no-go BCE results from temporal overlap of Task 1 motor execution with Task 2 central stage, that it is caused by the inhibition of a prepared Response 2, and that the preparation state for Task 2 iis adjusted following a no-go trial. As the no-go BCE differs fundamentally from the compatibility-based BCEs in the three aspects investigated here, both should be seen as two different types of BCEs. Furthermore, the R1-R2 BCE and the S1-R2 BCE were compared based on their underlying processes (Study 4). Results of Study 4 suggest that even though both phenomena arise in Task 1 central stage, they are based on different processes, and hence should be seen as two different types of compatibility-based BCEs. To answer the present main question: Different types of backward crosstalk can indeed be distinguished. As each type of backward crosstalk contradicts the notion of an encapsulated central stage in a different way, the present results support capacity-sharing over strictly serial processing.