Implicit task sequence learning with auditory stimuli

Abstract

We investigated implicit task sequence learning with auditory stimuli. In previous studies only visual stimuli have been used and thus learning may have been due to visuoperceptual learning. Further, we explored the generality of the correlated streams account which holds that correlated streams of information are necessary for implicit sequence learning to occur. We used three classification tasks with auditory stimuli. The presence or absence of a task sequence was orthogonally manipulated with that of a response sequence. Sequence-specific learning was found, but only in the condition with both a task and a response sequence. No learning was found in the conditions with a single task sequence and with a single response sequence. These results show that task–response sequence learning occurs with auditory stimuli and that visuoperceptual learning is not necessary. Moreover, they underscore the importance of correlated streams of information for implicit sequence learning.

Keywords:

• Correlated sequences
• Incidental learning
• Task switching

Acknowledgements

This work was supported by a grant from the Swiss National Science Foundation (Grant No. 130104) to BM. We thank Josias Bucher, Anina Bundi, Kora Dietsche, Mathias Doerig, and Sven Schmutz for data collection, and Alodie Rey-Mermet for helpful comments on an earlier version.

Notes

¹The disruption score of the seq-ran condition was not significantly different from zero, the effect size (d _z ) was 0.26. The required sample size to reach significance (with alpha level set to .05, one-tailed) would be 165 participants (G*Power Version 3.1.2; Faul, Erdfelder, Lang, & Buchner, 2007). Thus, the lack of finding a learning effect is not due to a power problem.

²We ran additional analyses to test whether sequence learning in the seq-seq condition affected n–2 task repetition costs. In task switching studies, it has been shown that n–2 task repetitions (e.g., ABA) are associated with a performance cost relative to nonrepetitions (e.g., CBA; see Koch, Gade, Schuch, & Philipp, 2010, for a recent review; Mayr & Keele, 2000). We compared n–2 task repetition costs of Block 7 (transfer sequence) to mean n–2 task repetition costs of Blocks 6 and 8 (training sequence). A two-factorial ANOVA with block (7 vs. 6 and 8 averaged) and n–2 task repetition (n–2 task repetition vs. nonrepetition) as within-subject factors revealed a significant effect of block (mean RTs were slower in Block 7 than in Blocks 6 and 8 averaged; 1008 ms vs. 956 ms), F(1, 19)=7.77, p=.012, η²=.29, but no effect of n–2 repetition, with p=.072 (mean RTs of n–2 task repetition trials did not significantly differ from n–2 nonrepetition trials; 995 ms vs. 969 ms). Importantly, the Block×n–2 repetition interaction was not significant, with p=.304. Thus, task inhibition did not decrease as a function of sequence learning. This result is in line with findings from Koch et al. (2006), who found reduced task inhibition in a TSL paradigm only in intentional-learning participants with explicit knowledge but not in incidental-learning participants. We thank Iring Koch for suggesting this analysis.

³For the between-experiments comparison, we used a different data filter than in the original study by Meier and Cock (2010). Therefore, the disruption scores of the “visual” experiment differ slightly from the previously published data.