The impact of secondary tasks on multitasking in a virtual environment

https://doi.org/10.1016/j.actpsy.2005.09.002Get rights and content

Abstract

One experiment is described that examined the possible involvement of working memory in the Virtual Errands Test (McGeorge et al. (2001). Using virtual environments in the assessment of executive dysfunction. Presence, 10, 375–383), which requires participants to complete errands within a virtual environment, presented on a computer screen. Time was limited, therefore participants had to swap between tasks (multitask) efficiently to complete the errands. Forty-two undergraduates participated, all attempting the test twice. On one of these occasions they were asked to perform a concurrent task throughout (order of single and dual-task conditions was counterbalanced). The type of secondary task was manipulated between groups. Twenty-one participants were asked to randomly generate months of the year aloud in the dual-task condition, while another 21 were asked to suppress articulation by repeating the word “December”. An overall dual-task effect on the Virtual Errands Test was observed, although this was qualified by an interaction with the order of single and dual-task conditions. Analysis of the secondary task data showed a drop in performance (relative to baseline) under dual-task conditions, and that drop was greater for the random generation group than the articulatory suppression group. These data are interpreted as suggesting that the central executive and phonological loop components of working memory are implicated in this test of multitasking.

Introduction

The term “multitasking” can be used to refer to a situation where a person has to complete multiple tasks, but cannot execute them sequentially (due to time limitations) or simultaneously (due to physical or cognitive limitations). The tasks must therefore be interleaved with one another, each being suspended and then resumed after appropriate intervals (Burgess, 2000a, Burgess, 2000b). Everyday domestic examples are cooking and shopping, but multitasking is also necessary for many jobs, for example, emergency medicine (Chisholm, Collison, Nelson, & Cordell, 2000) or management (Seshadri & Shapira, 2001). A number of studies have shown that patients with brain damage, particularly in the frontal lobes, can have great difficulty in applying efficient strategies in multitasking situations (Alderman et al., 2003, Burgess et al., 2000, Crépeau et al., 1996, Fortin et al., 2003, Goldstein et al., 1993, Knight et al., 2002, Levine et al., 2000, Levine et al., 1998, Shallice and Burgess, 1991). However, little research has yet been conducted into the factors that constrain multitasking performance among healthy adults, and how the cognitive system deals with these complex situations (Law et al., 2004). Therefore, our aim was to investigate the cognitive demand of multitasking using a well-established theoretical framework—the multiple-component model of working memory (e.g., Baddeley & Logie, 1999). Specifically, we used dual-task methodology to investigate the involvement of the phonological loop and central executive components of working memory in a test of multitasking. The test used was a “Virtual Errands Test” created by McGeorge and colleagues (McGeorge et al., 2001) and based on the “Multiple Errands Test” of Shallice and Burgess (1991).

The Multiple Errands Test (MET) was created by Shallice and Burgess (1991) in response to the observation that some patients with frontal lobe lesions who had disruptions to everyday life skills nevertheless performed normally on traditional “executive” tests, which were supposed to be sensitive to frontal lobe damage. The idea was to tap cognitive processes analogous to those involved in real life open-ended planning situations, but to have a quantifiable measure of performance. The original version of the Multiple Errands Test involved taking participants to a local shopping centre and giving them a list of tasks. Most of these were easy, for example, “buy a brown loaf”, but others were more difficult, for example, “find out the name of the coldest place in Britain yesterday”. An important part of a time-limited shopping trip is finding an efficient route through the shopping precinct. Excessive backtracking will result in the time elapsing before all errands are completed. Therefore the errands have to be interleaved in an efficient manner, rather than simply tackled in the sequential fashion in which they are presented. Shallice and Burgess found that their three patients were impaired in their ability to attempt this task effectively compared to a group of control participants—the patients tended to break more rules and to be more inefficient.

The sensitivity of the Multiple Errands Test as a measure of executive impairment has also been demonstrated in other studies with brain-lesioned patients (Alderman et al., 2003, Goldstein et al., 1993, Knight et al., 2002). These studies ask patients to attempt a task with high ecological validity in a real world setting, but this of course makes it a difficult and time-consuming test to administer. Some patients have become distressed or behaved in socially inappropriate ways while attempting the Multiple Errands Test (Alderman et al., 2003). In addition, unforeseen events in a real shopping centre can result in little experimental control. With these difficulties in mind, McGeorge and colleagues (2001) assessed the ability of five dysexecutive patients and five matched controls to undertake a version of the Multiple Errands Test in a virtual environment presented on a computer screen. They compared performance in this condition with performance in a real building on which the virtual environment was based. Patients were recruited on the basis that care staff reported that they had problems with planning. In the “real” condition, which was attempted first, participants had to move around inside the physical building completing a series of simple office-type errands (e.g., meet a colleague in one room and take him to another, send a fax from the main office). Participants then attempted a similar set of errands within the virtual replica of the building. Patients completed significantly fewer errands than controls, and the type of environment (real or virtual) had no effect on the results.

The present study used secondary task methods (random generation and articulatory suppression) to investigate the working memory demands of the Virtual Errands Test (VET) for healthy adults. Given that McGeorge et al.’s “dysexecutive” patients performed poorly on the VET, a direct prediction is that performance will be sensitive to an executively-demanding secondary task. However, it is important to demonstrate this with behavioural data from a sample of healthy participants. Random generation (RG) was chosen because it is becoming widely used in the working memory literature (Towse & Neil, 1998) and is thought to engage executive resources within working memory (Baddeley, 1996). Letters or numbers are often chosen as the response set from which participants have to generate random sequences, while inhibiting the well learned sequences such as alphabetical or numerical order. Neither letters nor numbers were suitable for the present experiment however, because the Virtual Errands Test involved remembering room identifiers that involved a letter and a number (e.g., F5, T15). Therefore, the response set chosen for the RG task was months of the year; these were not involved in the Virtual Errands Test but fulfilled the requirements for a RG task because there is a limited number of alternatives in the set with an over-learned sequence (i.e., calendar order).

For the present experiment, secondary task responses were generated orally because manual responding might have interfered with the use of the mouse-based interface in the Virtual Errands Test. This meant that the random generation task would also prevent sub-vocal rehearsal during the Virtual Errands Test, and any disruption to multitasking could be attributed to participants being unable to use inner speech. This possibility was investigated by asking half of the participants to suppress sub-vocal rehearsal by repeating the word “December”. Articulatory suppression is thought to have less executive involvement than random generation, but still places an extra demand on participants because they are unable to use inner speech to rehearse current task goals. The word “December” was chosen for the articulatory suppression task because this matched one of the longest alternatives in the random generation set. Some studies (e.g., Baddeley et al., 2001, Phillips et al., 1999) have used the over-learned sequence as the articulatory suppression control to a random generation task, for instance they have asked participants to count from 1 to 9 if the task is random number generation, or to recite the months of the year in calendar order. However, using a single word created a bigger difference in executive demand between the two secondary tasks, while keeping them comparable in terms of overt vocalization.

In the task switching paradigm, where participants are required to switch frequently between two simple cognitive tasks, articulatory suppression has been shown to increase the time-cost of these switches (Baddeley et al., 2001, Emerson and Miyake, 2003, Goschke, 2000, Miyake et al., 2004). Emerson and Miyake (2003) argue that people make use of inner speech to help retrieve and keep active a phonological representation of the next goal they have to accomplish. If this is the case in task switching, then it is also likely that people will recruit inner speech during multitasking, where participants have to determine their own schedule of switching between multiple sub-tasks, rather than switching back and forth between two tasks in response to a cue. It was therefore expected that there would be interference between the Virtual Errands Test and the secondary task of articulatory suppression, but that the interference would be greater with the secondary task of random generation. In terms of the multiple-component model of working memory (e.g., Baddeley & Logie, 1999), the articulatory suppression task would be expected to load the phonological loop, while the random generation task would be expected to load both the phonological loop and central executive. All participants attempted two versions of the Virtual Errands Test, one version without a concurrent task (single-task condition) and the other version with a concurrent task (dual-task condition). Half the participants were given articulatory suppression as their concurrent task, while the other half were given random generation. Secondary task performance was also recorded.

Section snippets

Participants

All 42 participants were first year psychology undergraduates who received course credit for their participation. There were 26 women and 16 men, with equal proportions allocated randomly to each group (articulatory suppression (AS) or random generation (RG)). They ranged in age from 18 to 26, with a mean age of 19.52 years (SD = 2.05).

Design

The experiment employed a 2 × 2 mixed design where the within-participants factor was the condition under which the Virtual Errands Test was attempted (single vs.

Virtual errands score

The overall score for each participant was the number of correct errands (or part-errands where the errand involved two steps) completed, minus the number of errors. Errors were either going into an incorrect room, or picking up an incorrect object. A high score therefore indicates better Virtual Errands performance. As an initial exploration of the data suggested possible interactions with the order of single and dual-task conditions, a 2 × 2 × 2 mixed analysis of variance was carried out with the

Discussion

Analysis of the Virtual Errands Test data offers some support for the hypothesis that the concurrent tasks would have a negative impact on performance, in that there is an overall impact of a secondary task. This conclusion is qualified by the interaction with order of conditions, in that the dual task impairment was most evident for those participants who had attempted the dual-task condition first. One possible account is that participants relied heavily on working memory resources when they

Acknowledgements

We are grateful to Dr. Peter McGeorge and Dr. Louise Phillips at the University of Aberdeen for making the Virtual Errands Test available to us, and for their help in setting up the equipment and materials.

References (40)

  • A. Baddeley et al.

    Working memory and the control of action: Evidence from task switching

    Journal of Experimental Psychology: General

    (2001)
  • A. Baddeley et al.

    Working memory: The multiple component model

  • P. Barrouillet et al.

    Time constraints and resource sharing in adults’ working memory spans

    Journal of Experimental Psychology: General

    (2004)
  • P.W. Burgess

    Real-world multitasking from a cognitive neuroscience perspective

    Attention and Performance

    (2000)
  • P.W. Burgess

    Strategy application disorder: The role of the frontal lobes in human multitasking

    Psychological Research-Psychologische Forschung

    (2000)
  • C.D. Chisholm et al.

    Emergency department workplace interruptions: Are emergency physicians “interrupt-driven” and “multitasking”?

    Academic Emergency Medicine

    (2000)
  • F. Crépeau et al.

    Disorganisation of behavior in a multiple subgoals scheduling task following traumatic brain injury

    Brain and Cognition

    (1996)
  • K.J. Gilhooly et al.

    Planning processes and age in the five-disc Tower of London task

    Thinking and Reasoning

    (1999)
  • L.H. Goldstein et al.

    Unilateral frontal lobectomy can produce strategy application disorder

    Journal of Neurology, Neurosurgery, and Psychiatry

    (1993)
  • T. Goschke

    Involuntary persistence and intentional reconfiguration in task-set switching

  • Cited by (39)

    • Evaluating the dual-task decrement within a simulated environment: Word recall and visual search

      2023, Applied Ergonomics
      Citation Excerpt :

      This task was based on the multiple errands task (MET, Shallice and Burgess, 1991) which required participants plan and perform a variety of errands such as buying items within a certain time limit as a test of executive abilities. Law et al. (2006) examined multitasking performance utilizing the VET paired with a long-term memory task (recalling months of the year) and were able to observe a dual-task decrement where participants chose to sacrifice performance on the secondary task rather than on the VET. In a more recent study, Law et al. (2013) used the EVET to further explore multitasking in simulated environments, where they found that both a sound localization task and a random generation task disrupted dual-task performance when paired with the EVET.

    • Utility of recorded guided imagery and relaxing music in reducing patient pain and anxiety, and surgeon anxiety, during cutaneous surgical procedures: A single-blinded randomized controlled trial

      2016, Journal of the American Academy of Dermatology
      Citation Excerpt :

      With subjects presumably preoccupied listening to guided imagery or relaxing music, surgeons may be more at ease focusing solely on the technical aspects of the procedure. Recent studies suggest that in many activities, even those who suspect themselves of being able to multitask objectively perform better when they engage in only 1 type of work.15 Surgeons, their serene confidence notwithstanding, may be susceptible to this same phenomenon, and when the need to multitask (eg, entertain the patient while excising or repairing) is removed, they may be more at ease, more efficient, and less tense.

    View all citing articles on Scopus
    View full text