Processing two tasks with varying task order: Central stage duration influences central processing order

Abstract

In a recent study, Sigman and Dehaene (2006, PLoS Biology, 4, 1227–1238) reported that perceptual processing duration influences processing order of two tasks in the psychological refractory period paradigm (PRP). The present study examines whether the duration of central processes also influences processing order. For this purpose, we employed letter tasks with different central processing durations and varied task order in the PRP. In one part of the experiment, a tone discrimination task was combined with a similar time-consuming letter discrimination task. In the other part, the tone task was combined with a more time-consuming letter task, which required a mental rotation of the letter thereby prolonging central processing. If the duration of central processes influences processing order, participants should perform the tone task first more often when it is presented with the more time-consuming mental rotation task, than when it is presented with the similar time-consuming letter task. The results clearly confirm this prediction and thus show that not only perceptual processing duration but also central processing duration affects processing order in a dual-task situation. We suggest that the reported effect mirrors a tendency of participants to minimize total reaction time to both tasks by reducing slack time.

Introduction

When people execute two tasks at the same time, task performance is often worse than when they execute each of the two tasks in isolation. To examine the temporal micro structure of these limitations in dual-task processing, researchers often employ the psychological refractory period (PRP) paradigm (Pashler, 1994). In a typical PRP experiment, two stimuli, say S1 and S2, are presented in close temporal proximity, and each stimulus requires a separate response, R1 and R2, respectively. The typical finding is that the reaction time of Task 2 (RT2) increases when the stimulus onset asynchrony (SOA) between S1 and S2 is shortened — the PRP effect. In contrast, the RT of Task 1 (RT1), is relatively uninfluenced by a variation of SOA.

This pattern of results is often explained within the framework of bottleneck models. According to these models, two tasks cannot be processed entirely at the same time because certain cognitive operations require a bottleneck process that can handle only one task at a time (Welford, 1952). More specifically, the central bottleneck model proceeds from the assumption that each of the two RT tasks requires three successive processing stages: (a) the perception of the stimulus, (b) the selection of the response, and (c) the execution of motor response (Pashler, 1984, Pashler and Johnston, 1989). Moreover, this model assumes that the response selection stage constitutes a central bottleneck process. So, if Task 2 requires access to this bottleneck stage while the bottleneck is still occupied by Task 1, processing of Task 2 has to be postponed. Accordingly, RT2 should decrease with increasing SOA, whereas RT1 should remain constant (for a review, see Pashler, 1994). Although alternative accounts to the central bottleneck model have been put forward (e.g., Hazeltine et al., 2002, Huestegge and Koch, 2009, Meyer et al., 1995, Meyer and Kieras, 1997a, Meyer and Kieras, 1997b, Navon and Miller, 2002, Tombu and Jolicoeur, 2003), it is fair to say that this model provides a parsimonious account of several RT phenomena associated with the PRP paradigm.

The present study explores how people determine the processing order of the two tasks, when they can no longer rely on a constant S1–S2 presentation order as in the standard PRP paradigm. More precisely, it examines whether the processing order of dual-tasks is influenced by the duration of central processes. To do so, we employed the PRP extension used in the study of Sigman and Dehaene (2006, see also De Jong, 1995). In contrast to the standard paradigm, these authors randomly varied the presentation order of two PRP tasks (i.e., pitch discrimination task and number comparison task), employing SOAs ranging from − 1000 to 1000 ms. In trials with negative (positive) SOAs, the number task occurred SOA ms before (after) the tone task. Therefore, the presentation order of the two task stimuli is no longer predictable for the participants and so they have to decide on-the-fly, which task to perform first. Thus, the question arises, how people determine the processing order of the two tasks, in this paradigm with variable presentation order.

One possibility would be that central processing order is determined by a first-come, first-served principle at the bottleneck (Sigman & Dehaene, 2006, p. 1229). In brief, the stimulus that arrived first at the bottleneck also enters the bottleneck first. The results of Sigman and Dehaene are consistent with this principle. In particular, they assessed the response order of the two tasks as a function of SOA. In other words, they computed for each level of SOA the fraction of trials where the response to the tone task occurred before the response to the number task. This function increased in a sigmoidal fashion from 0 (at large negative SOAs) to 1 (at large positive values). From this function, Sigman and Dehaene (2006) computed the level of SOA at which the fraction of tone responses was equal to 50%, that is, the SOA level at which the two response orders (tone response first and number response first) were equally likely.

Theoretically important for the present study are the effects on 50% SOA of two manipulations affecting the duration of the number task. First, Sigman and Dehaene (2006) presented the number either as an Arabic digit or as a spelled-out word. They expected that this notation manipulation should affect the perceptual latency of the number task and therefore shift the response order function. Consistent with this notion, they observed a 38 ms shift in the expected direction. Consequently, when perceptual processing was fast (i.e., number presented in Arabic digits), the tone had to be presented earlier to reach the 50% SOA. Second, they manipulated the numeric distance of the presented numbers from the target number (i.e., 45). Although, this distance manipulation had even a larger effect on mean RT than the notation manipulation, this distance manipulation did not shift the response function. The reason for this might be that distance manipulation affects central rather than perceptual processes. Thereby, this manipulation would not influence the arrival order of the two tasks at the central bottleneck. To sum up, in the study of Sigman and Dehaene, central processing order seemed to be influenced by the duration of perceptual processes, but not by the duration of central processes. Altogether, the analysis of the response function provides a promising tool for analyzing how people determine task processing order in dual-task situations.

Although a first-come, first-served principle at the bottleneck is an appealing account for determining task processing order, other research has indicated that the duration of central processes also influences central processing order. Specifically, a study by Leonhard, Ruiz Fernández, Ulrich, and Miller (2011) provided evidence for this notion. These authors combined a more time-consuming first task with a less time-consuming second task in a PRP paradigm with constant task orders (i.e., SOA > 0). Task 1 was a mental rotation task (Cooper and Shepard, 1973, Shepard and Metzler, 1971) and participants judged whether a rotated letter was presented in its normal or in its mirror-imaged form. This task leads to particular long RTs (e.g., Kail & Park, 1990) and mental rotation is assumed to require the central bottleneck stage (e.g., Heil and Rolke, 2002, Ruthruff et al., 1995). In the less time-consuming Task 2, participants judged whether a tone was presented to their left or to their right ear. Leonhard et al. expected that the different task demands influence central processing order because processing Task 2 before Task 1 would reduce Task 2 waiting time at the bottleneck (see Miller, Ulrich, & Rolke, 2009). Their results showed an influence of the different task demands on central processing order providing evidence for the notion that the duration of central processes influences central processing order.

The results of the study by Leonhard et al. (2011) are in contrast to the results of the study by Sigman and Dehaene (2006). While Leonhard et al. found that duration of the central stage influenced central processing order, the latter authors did not find evidence for this notion. The aim of the present study is to examine these apparently discrepant results. For this purpose, the present study combines the extension of the PRP paradigm used in the study of Sigman and Dehaene with a mental rotation task similar to the one used by Leonhard et al. As in the study of Leonhard et al., we used a mental rotation task as a particular time-consuming task. This task consisted of a single capital letter that was rotated 90° clockwise or anti-clockwise from the upright position and was presented in its normal or in its mirror-imaged form. Participants were asked to perform a mirror-normal judgment of the rotated letter. In one part of the experiment, this task was combined with the less-time consuming tone discrimination task, in which participants judged if a tone was presented to their left or to their right ear (dissimilar condition). In the other part, participants were presented two tasks with a comparable central processing duration (similar condition). In this baseline condition, a letter discrimination task, in which participants decided whether the letter X or Y had been presented, was combined with the tone discrimination task. If the duration of central processes influences central processing order, as in the study by Leonhard et al., one should observe that participants tend to process the less time-consuming task first more often in the dissimilar than in the similar condition. That is, the 50% SOA level should be shifted between the two conditions.