ABSTRACT
The phonological loop plays an important role in task switching. Until now, however, it has been unclear whether it is the maintenance of the task sequence or the retrieval of the task goal that requires phonological processing. In the present study, 60 participants (20 children, 20 adolescents, 20 young adults) performed three phonological working-memory tests before they switched between two tasks in a cue-present condition (random task sequence) and in a cue-absent condition (predictable task sequence). Results revealed that better phonological and central-executive working-memory performance predicted lower switch costs. In line with age-differentiation of cognitive abilities switching was less coupled with working memory in adults compared to children and adolescents. Interestingly, however, relations were only demonstrated in the cue-present but not in the cue-absent condition, indicating that it is more the retrieval of the task goal than the maintenance of the task sequence that requires the involvement of phonological processing.
Acknowledgment
The authors would like to thank Beate Biskup-Ackermann, Alena Engel, and Patricia Erb for their help with conducting the experiment.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data of the experiment are available via the following link: https://osf.io/6kquw/?view_only=774710ecac814a6eadf8f23e25e38ee2.
Notes
1 At this point it is important to note that the performance costs in the study by Kray et al. (Citation2004) were not switch costs but so-called mixing costs. In contrast to switch costs, mixing costs are defined as the performance differences between single-task and mixed-task conditions. That is, mixing costs represent the cognitive demands for maintaining two task sets in working memory and selecting the appropriate one, whereas switch costs represent the costs arising from actually switching between two task sets.
2 The sample size was based on power considerations. Using G*Power 3.1.9 (Faul, Erdfelder, Lang, & Buchner, Citation2007), we estimated that a linear regression model with multiple correlation coefficient ρ2 = 0.2 (medium-sized effect), one predictor (per model), an α error probability = .05, and a power = .95, requires N = 59 participants.
3 The Mottier Test is a well-established instrument in clinical diagnostics, which is used to examine children’s phonological working-memory. That is, standard values of the Mottier-Test are restricted to 5–17 years-old children and adolescents. However, we decided to use the test also for the third age group (i.e. for young adults) because another comparable test for German with standard values for older participants is lacking.
4 Computing proportional costs based on error rates was not possible because of the very low error rates in repetition trials (for 11 participants the error rate in repetition trials was zero in at least one of the cells of the ANOVA).
5 The reason behind using raw scores instead of standard values is that standard values of the Mottier-Test are restricted to 5–17 years-old children and adolescents (see also Footnote 3). Using raw scores ensured an equal procedure for all participants and all three working-memory tasks. However, in order to make sure that participants in the current study were representative for their respective age groups, we compared their performances with the standard values of the two digit span tasks. In these tasks, participants were mostly within normal ranges of their respective comparison groups: 85.8% of the standardised values of our sample were within ±1 SD above/below the standardised mean.
6 We defined pressing the wrong key and responding too slowly (i.e. more than 3500 ms after stimulus presentation) as an error.
7 All analyses were additionally conducted with switch costs instead of proportional switch costs and the pattern of results did not change.