Does Initial Performance Variability Predict Dual-Task Optimization with Practice in Younger and Older Adults?

Abstract

Background/Study Context: The variability associated with reaction time (RT) is sometimes considered as a proxy for inefficient neural processing, particularly in old age and complex situations relying upon executive control functions. Here, it is examined whether the amount of variability exhibited early in practice can predict the amount of improvement with later practice in dual-task performance, and whether the predictive power of variability varies between younger and older adults.

Methods: To investigate the relationship between variability and practice-related improvement, RT mean and variability data are used, obtained from an experiment in which younger and older adults performed two tasks in single-task and dual-task conditions across seven practice sessions. These RT and variability data were related to the single-task and dual-task practice benefits. These benefits were computed as follows: dual-task/single-task RTs at the beginning of practice minus dual-task/single-task RTs at the end of practice.

Results: In both age groups, dual-task processing was speeded up with practice and variability associated with the means was reduced. Most important, independent of mean RTs, variability allowed predicting dual-task practice benefit in both age groups under specific conditions.

Conclusion: These findings suggest that the relationship between performance variability and executive control functions under some specific conditions. Implications of these results for models of practiced dual tasks are discussed.

Notes

¹ The data of 10 of these participants were used in Strobach et al. (2012a).

² The data of 10 of these younger participants were used in Strobach et al. (2012b).

³ In the present analyses, we did not exclude posterror trials because the mix of single-task and dual-task trials in mixed blocks makes the definition of posterror trials very critical. Potential definitions of posterror trials could be the exclusion of (1) only single-task trials, (2) only dual-task trials, (3) single-task and dual-task trials, or (4) only the corresponding component tasks (in dual tasks) after erroneous single and/or dual tasks. Since the definition of posterror trials is thus critical, we refrained from their exclusion.

⁴ RT distributions have been visually inspected and showed unimodal and normal distributions in each component task as well as task condition.

Figure 1. Group-mean reaction times (RTs, represented by open symbols) and individual RTs (represented by filled symbols) in milliseconds (ms) in single-task blocks and dual-task trials for (left panels) the visual task and (right panels) the auditory task across Sessions 1–7 for older and younger adults. Both group-mean and individual RTs decrease with practice, with a stronger trend in dual-task data.