4 Summary of studies 1 – 3


Based on the outlined state of the art in the research on the functional neuroanatomy of dual-task processing, three basic questions concerning the functional role of the lPFC in interference processing in dual tasks remain open. These three questions concern:

  1. the generality of the lPFC involvement in dual-task processing across stimulus-response modality pairings and different types of dual-task coordination
  2. the dissociability of different control processes involved in dual-task processing in the lPFC
  3. the interaction of the lPFC with posterior task-relevant brain regions to deal with dual-task interference


These three issues form the starting point for the three fMRI studies that are presented in detail as original articles below. Each study focuses on one of these issues but also has implications for the other questions outlined above. Next, a short overview will be given on the specific research questions of the three articles and the obtained results will be shortly summarised.

4.1  Study 1: “The neural effect of stimulus-response modality compatibility on dual-task performance: an fMRI study (Stelzel et al., 2006)”

Research Question:

Study 1 aimed at investigating the generality of the lPFC involvement in dual-task processing on two levels. First, generality with respect to lPFC involvement across different types of interference and second, generality with respect to the type of dual-task coordination associated with dual-task-related activity in the lPFC.

As outlined above, some dual-task fMRI studies with the PRP paradigm tried to specify the type of dual-task interference related to lPFC activity. Whereas Jiang (2004) and Herath et al. (2001) found evidence for perceptual and motor interference, at the point of Study 1, no unambiguous evidence was present for central bottleneck interference in the lPFC. This is because all studies that compared dual-task-blocks with single-task-blocks used either overlapping perceptual or motor modalities in the component tasks. Therefore, the obtained dual-task-related activity changes in the lPFC might also be related to the resolution of perceptual or motor interference instead. Aim 1 of Study 1 was to exclude perceptual and motor interference and to investigate whether dual-task-related lPFC activity can also be found in situations with non-overlapping stimulus and response modalities. For that purpose tasks with visual and auditory stimuli and manual and vocal responses were combined to compare the BOLD-signal changes between dual-task and single-task blocks. The finding of dual-task-related activity in such a task situation would then be attributable to the processing associated with the response selection bottleneck.


Other dual-task studies (Szameitat et al., 2002; 2006) related the lPFC to the coordination of the temporal order of two tasks. Coordination, however, might also have other aspects. One important aspect is to coordinate the concurrent mapping of sensory information onto corresponding motor responses on an abstract level. When central task representations, like abstract verbal or spatial codes of two tasks, overlap, increased processing demands might be related to the coordination of the concurrent processing of the two tasks. Additional coordinative demands might keep overlapping contents separated (Hazeltine et al., 2006). Aim 2 of Study 1 therefore was to investigate whether differences in lPFC activity can be found for two tasks with or without such central content overlap. For that purpose the stimulus-response modality compatibility between the two component tasks was manipulated. Importantly, modality incompatible tasks both required spatial coding and therefore are assumed involve the postulated processes of task coordination. In sum, modality compatibility offers the opportunity to manipulate the degree of central task overlap while keeping the applied stimuli and responses constant. Increased lPFC was expected for modality incompatible compared to modality compatible tasks.

Results & Discussion:

The comparison of dual-task and single-task blocks revealed dual-task-specific activity predominantly in lateral frontal and parietal regions. That is, even in dual-task situations without perceptual and motor overlap, lateral prefrontal regions surrounding the left inferior frontal sulcus showed increased dual-task-related activity. Thus, Study 1 supports the idea that the involvement of the lPFC in interference processing is rather general and that the lPFC is also related to the processing of central bottleneck interference. This complements previous findings by Herath et al. (2001) and Jiang (2004) and is also in accordance with later findings by Dux et al. (2006).

In addition, modality incompatible dual tasks that overlap with respect to central task representation revealed increased activity in the IFS compared with modality compatible task.


The effects in the IFS were investigated in individual ROIs based on the dual-task minus single-task contrast. That way inter-individual variability in the exact location of dual-task specific regions was taken into account. In addition, this effect was accompanied by strong behavioral effects of modality compatibility in the dual task conditions but not in the single tasks. Importantly, modality compatible and incompatible dual tasks did not differ with respect to the timing of the two tasks as it was the case for the manipulation of task order in the studies by Szameitat et al. (2002; 2006). Hence, effects of dual-task coordination in the lPFC are not limited to situations of changing task order but can also be shown for coordination related to overlapping central contents.

4.2  Study 2: “Dissociable neural effects of task order control and task set maintenance during dual-task processing (Stelzel et al., in press)”

Research Question:

The aim of Study 2 was to dissociate the neural effects of different control functions associated with dual-task performance: task order control and task set maintenance. As summarized above, there is evidence for a consistent involvement of the lPFC in dual-task performance. Usually, this was measured by comparing BOLD-signal changes in dual-task blocks with single-task blocks (Schubert & Szameitat, 2003; Szameitat et al., 2002). The results of this subtraction method, however, may reflect any difference between dual tasks and single tasks. Although Szameitat et al. (2002; 2006) showed that one such difference between dual tasks and single tasks – the demand to control the task order – is related to activity changes in the lPFC, other differences between dual tasks and single tasks are conceivable. Jiang et al. (2004) argued that simply the requirement to maintain two task sets simultaneously may be the crucial factor underlying dual-task-related activity in the lPFC. In Study 2, the contribution of task order control and task set maintenance to activity changes in the lPFC was investigated while participants performed dual tasks. For that purpose, a parametrical manipulation of task order control and task set maintenance was realized in an integrated experimental design performed by one group of participants. Specifically, task order control was measured in the comparison of dual-task blocks with random and fixed temporal order of the component tasks. Task set maintenance was manipulated via the number of relevant stimulus-response mappings per component task. Note that such an integrated design has the advantage that the corresponding activity changes can be compared directly within the same group of participants. Inferences about the overlap in functional localization of different control functions in the lPFC therefore exclude differences between participants, between the applied paradigms or scanning and analysis procedures as they may be present in the comparison of activity peaks between studies.

Results & Discussion:


The fMRI data revealed a functional-neuroanatomical dissociation of both factors in the lPFC (see Figure 3). Regions surrounding the inferior frontal sulcus and the middle frontal gyrus were exclusively associated with task order control but not with increased demands on task set maintenance during dual-task processing. The only lPFC region associated with task set maintenance was located in the left anterior insula. Outside the lPFC, there were dissociable regions for task order control and task set maintenance bilaterally in the premotor cortices

Figure 3: Resuts of Study 2. Whol-brain analysis fort he comparison of dual-task blocks with random order and fixed order (red) and blocks with set sizes of 8 vs. 4 S-R mappings (green) as well as the conjunction of both factors (blue).

with more rostral premotor activity for task order control and more caudal premotor activity for task set maintenance. These data clearly contradict the assumption that lPFC activity during dual-task processing is simply related to the requirement to maintain the task sets of two tasks simultaneously (Jiang et al., 2004). Instead, the results of Study 2 suggest
that task order control is a separable cognitive mechanism in dual-task situations that is related to activity changes in the lPFC and that can be dissociated from task set maintenance.

4.3 Study 3: “Neural mechanisms of attentional task setting in dual tasks
(Stelzel et al., submitted for publication)”

Research Question:


While Studies 1 and 2 showed the involvement of the lPFC in the control of dual-task processing, they left open how exactly the lPFC controls the dual-task processing stream. Study 3 aimed at specifying the interaction of the lPFC with task-relevant sensory regions in order to better understand the neural dynamics involved in dual-task processing. The computational dual-task models described above suggest that control mechanisms like attentional task setting (Sigman & Dehaene, 2006) serve the coordination of two interfering tasks (see also Logan & Gordon, 2001; Meyer & Kieras, 1997). These models suggest that S1 processing is not affected by a secondary task because – due to its temporal precedence or because of the task instructions – its processing is attentionally more focused. At the same time, the PRP effect in S2 processing emerges due to delayed focussing on S2. Importantly, the assumption of attentional task setting implies different temporal dynamics for S1 and S2 processing with respect to the devoted attentional resources at different temporal overlaps. Due to the S1-related task setting mechanism, attentional focussing is assumed to be increased for S1 processing when there is temporal overlap with S2 compared to situations without such temporal overlap. Attentional focussing on S2, in contrast, should not depend on the temporal overlap.

Until now, the neural mechanism associated with these task setting mechanisms are not well understood. In Study 3, two plausible neural mechanisms that might reflect such attentional task setting with respect to S1 and S2 processing in a PRP situation were tested. First, task-relevant sensory regions were tested for a differential up-regulation in S1-relevant sensory regions in dual-task situations with high temporal overlap (see Egner & Hirsch, 2005; but also Desimone & Duncan, 1995). Second, changes in functional connectivity of S1-relevant sensory regions with dual-task-related regions in the lPFC might represent a second neural mechanism reflecting attentional task setting in dual tasks. Functional coupling was assessed between lPFC regions related to dual-task control and task-relevant sensory regions in the Fusiform Face Area (FFA) for S1-processing and Visual Word Form Area (VWFA) for S2-processing. In particular, the functional coupling was measured for condition-specific contrasts between different SOAs in the PRP paradigm. To control for perceptual differences in the three SOA conditions (100, 300, 1000 ms), a control condition was included where the secondary stimulus was completely irrelevant for task performance (dual-task vs single-task blocks). High interference at short SOA was expected to be associated with increased functional coupling of dual-task-related lPFC regions and S1-relevant regions in the FFA compared to low or non-overlapping tasks. At the same time, deficient coupling of lPFC with S2-relevant regions in the VWFA was expected to be related to the performance costs in this secondary task.

Results & Discussion:

As a first result of Study 3, we replicated the behavioral PRP effect. The activity in S1-relevant regions in the FFA depended on the degree of temporal overlap of the two stimuli with signal increases at high and no temporal overlap. However, identical effects were found for DUAL TASKS and SINGLE TASKS, indicating that the SOA-effects did not solely reflect the PRP-effect or associated task setting mechanisms present in the DUAL-TASK condition but rather perceptual effects of simultaneous stimulus presentation. S2-relevant activity in the VWFA was modulated by the task relevance of S2 with increased activity in situations where participants also responded to S2. This fits nicely to known top-down effects of sustained attention on activity in sensory regions (Desimone & Duncan, 1995).


For the dual-task-related regions in the plPFC, we found effects of SOA and task relevance. However, there was no interaction of both factors. Agiain, this suggests that the activity differences between the SOAs in the plPFC are not related to the PRP effect and attentional task setting in dual tasks only. Instead, effects related to the processing of the more or less simultaneously presented second stimulus seem to be a more plausible account for the obtained SOA effects in the plPFC (see also Jiang et al., 2004).

Most importantly, in Study 3 significant differences in the functional coupling across SOAs were present between S1-relevant regions in the FFA and dual-task-related regions in the right lPFC. Functional coupling was strongest at SOA100 as expected from the theory of attentional task setting. In contrast, no SOA-related up-regulation of functional coupling with the lPFC was present for the VWFA.

In addition, the degree of functional coupling of the right plPFC and the FFA was negatively correlated with the error rates in dual tasks but not in single tasks. That is, participants with increased functional coupling of plPFC and FFA at short SOAs made generally less dual-task errors and were thus more efficient in dual-task processing.


Taken together, the results of Study 3 support the idea that the presence of transient changes in functional coupling of control regions in the plPFC and S1-relevant regions together with the absence of such coupling for S2-relevant regions in situation of temporal overlap is associated with the PRP performance pattern.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated: