ABSTRACT
The n − 2 repetition cost seen in task switching is the effect of slower response times performing a recently completed task (e.g. an ABA sequence) compared to performing a task that was not recently completed (e.g. a CBA sequence). This cost is thought to reflect cognitive inhibition of task representations and as such, the n − 2 repetition cost has begun to be used as an assessment of individual differences in inhibitory control; however, the reliability of this measure has not been investigated in a systematic manner. The current study addressed this important issue. Seventy-two participants performed three task switching paradigms; participants were also assessed on rumination traits and processing speed—measures of individual differences potentially modulating the n − 2 repetition cost. We found significant n − 2 repetition costs for each paradigm. However, split-half reliability tests revealed that this cost was not reliable at the individual-difference level. Neither rumination tendencies nor processing speed predicted this cost. We conclude that the n − 2 repetition cost is not reliable as a measure of individual differences in inhibitory control.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Additional analysis limited to the first 360 trials per paradigm was also conducted, to examine to what extent unequal number of trials in task switching paradigms affected the results. Differences from additional analyses are reported in footnotes.
2. One participant had missing data for the Processing Speed measure; to maintain power, we kept this participant and imputed their value using the mean score for the Processing Speed test. Removing this participant changes none of the conclusions.
3. Transforming RRS and the processing speed scores into z-scores and then inputting them into the regression analysis also yielded non-significant results.
4. Equal trials correlation analysis: the n − 2 repetition cost correlated only between the Visual and Numeric Judgments paradigms (r = .28, p = .01). The reported correlations remained unchanged when we controlled for individual differences in processing speed via partial correlations: Target Detection paradigm did not correlate with the Visual Judgement paradigm (r = .07, p = .56), but it did correlate with the Numeric Judgement paradigm (r = .25, p = .03); the Visual Judgement paradigm correlated with the Numeric Judgement paradigm (r = .30, p = .01). Note these latter correlations do not remain significant when using Bonferroni corrections for multiple correlations.
5. We are grateful to Cai Longman for suggesting this possibility.
6. We are grateful to Cai Longman for suggesting this possibility.
7. We are grateful to an anonymous review for suggesting we discuss this.