Differential contributions of set-shifting and monitoring to dual-task interference

Abstract

It is commonly argued that complex behaviour is regulated by a number of “executive functions”, which work to coordinate the operation of disparate cognitive systems in the service of an overall goal. However, the identity, roles, and interactions of specific putative executive functions remain contentious, even within widely accepted tests of executive function. The authors present two experiments that use dual-task interference to provide further support for multiple distinct executive functions and to establish the differential contributions of those functions in two relatively complex executive tasks—random generation and the Wisconsin Card Sorting Test. Results are interpreted in terms of process models of the complex executive tasks.

Keywords:

• Executive functions
• Set-shifting
• Monitoring
• Dual-task interference
• Random number generation
• Wisconsin Card Sorting Test

Acknowledgments

^†Karolina Wutke is now at the Department of Clinical Psychology, Coventry University, UK.

This research was supported by a faculty research grant from Birkbeck, University of London, to Richard P. Cooper. We are grateful to Erik Altmann, Ines Jentzsch, André Vandierendonck, and two anonymous reviewers for constructive comments on early drafts of this manuscript.

Notes

¹ IFI is Bollen's (1989) Incremental Fit Index. Miyake et al. (2000) suggest that a value of more than 0.95 indicates a good fit.

² See Towse and Neil (1998) for formulae for the less obvious measures (R, RNG, and TPI). Note that in comparison to Towse and Neil (1998), TPI values reported here are divided by 100 so that, like most other measures of randomness used here, they lie within the 0 to 1 range.

³ Some RT data for the secondary tasks were lost due to lack of sensitivity of the voice-key. The analyses reported here are thus based on a subset of participants for which RT data were available. This accounts for the variation in reported degrees of freedom across comparisons.

⁴ RR showed the same pattern of effects, though for this dependent measure the effects were not statistically significant: 2-back versus digit-switching, t(35) = 1.65, p = .053; 2-back versus go/no-go, t(35) = 1.60, p = .059; digit-switching versus go/no-go, t(35) = 0.54, p = .295. The failure of these tests to reach statistical significance is, we suggest, the result of a floor effect in the RR measure.

⁵ An alternative possibility is that set-shifting is itself dependent on memory processes (Altmann & Gray, 2008). This may account for similar effects of the digit-switching and 2-back tasks on the R score. It would suggest, however, that all dependent measures should pattern similarly with respect to these secondary tasks. This is not what was observed.

⁶ It is possible that lower than chance RR scores contribute to higher than chance AA scores. Adjacent associate responses may be due in part to inhibition of a repeat response and the resultant selection of a response that is spatially close but distinct from the repeat response. Such a response would be an AA response. We are grateful to Erik Altmann for raising this possibility. However, while such a mechanism coupled with increased inhibition of repeat responses may contribute to the increase in AA scores in the dual-task conditions, it cannot be the only factor, as the increase in AA scores is far greater than the decrease in RR scores.

⁷ The use of WCST in a within-subjects design requires some justification. The task's novelty, and in particular the unannounced sorting criterion changes, are commonly considered to be part of the difficulty of the task. However, the task has been used before in a within-subjects context (Stuss et al., 2000; see also Reverberi et al., 2005). Its use here is justified because (a) all participants have a chance to learn the task rules during the control condition, (b) order of tasks in the experimental conditions is counterbalanced, and (c) we are not interested in the processes concerned with learning that the sorting criterion changes—merely in the behaviour following a change. Note also that even in the standard administration participants are only naïve to the first criterion change—participants then negotiate up to five more criterion changes during the task.

Additional information

Notes on contributors

Karolina Wutke

Karolina Wutke is now at the Department of Clinical Psychology, Coventry University, UK.