Context-specific adjustment of cognitive control: Transfer of adaptive control sets

ABSTRACT

Cognitive control protects processing of relevant information from interference by irrelevant information. The level of this processing selectivity can be flexibly adjusted to different control demands (e.g., frequency of conflict) associated with a certain context, leading to the formation of specific context–control associations. In the present study we investigated the robustness and transferability of the acquired context–control demands to new situations. In three experiments, we used a version of the context-specific proportion congruence (CSPC) paradigm, in which each context (e.g., location) is associated with a specific conflict frequency, determining high and low control demands. In a learning phase, associations between context and control demands were established. In a subsequent transfer block, stimulus–response mappings, whole task sets, or context–control demands changed. Results showed an impressive robustness of context–control associations, as context-specific adjustments of control from the learning phase were virtually unaffected by new stimuli and tasks in the transfer block. Only a change of the context–control demand eliminated the context-specific adjustment of control. These findings suggest that context–control associations that have proven to be adaptive in the past are continuously applied despite major changes in the task structure as long as the context–control associations remain the same.

KEYWORDS:

• Cognitive control
• Context-specific proportion congruence
• Control regulation
• Dual-task
• proportion congruence
• Stimulus-driven control

Acknowledgements

The authors are grateful to Heike Stein and Benedikt Langenbach for assistance in data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. From previous studies it is known that context-specific control adjustments transfer to items that are unbiased by conflict frequency, i.e., items that appear with equal frequency of conflict/non-conflict at both locations (Crump & Milliken, 2009). Such frequency-unbiased items are processed with the same task set as frequency-biased items, and thus the same location-associated control settings are applied.

2. The same result pattern reported so far has been confirmed when increasing statistical power by combining Task 1 and Task 2 performance. For this we repeated all analyses and included the factor Task (Task 1 vs. Task 2) in the ANOVA. For Experiments 1 and 3, the CSPC effect did not reach significance for Part 1 of the learning phase (p = .163 and p = .110) but was reliably present at the end of the learning phase (p < .001 and p = .016). For Experiment 2 the CSPC effect was present for both parts of learning (p = .007 and p = .001). For the transfer block, CSPC effects were found for Experiment 1 and 2 (p < .001 and p = .027) but not for Experiment 3 (p = .592). We thank an anonymous reviewer for suggesting these analyses.

3. Note that findings of Experiment 3 most likely also predict eliminated CSPC effects for transfer blocks with unbiased conflict frequencies. Again, this would constitute a change of the abstract features of the context that are the basis for the activation of context-associated control sets. We thank an anonymous reviewer for raising this issue.

Additional information

Funding

This research was supported by a grant of the Deutsche Forschungsgemeinschaft [Grant number CRC 940, Project A3].