Well under control: Control demand changes are sufficient for metacontrol

Abstract

Metacontrol arises from the efficient retrieval of cognitive control by environmental cues that are predictive of the upcoming control demands. Previous studies have demonstrated that proactive and reactive metacontrol can be indexed by a list-wide switch probability (LWSP) and an item-specific switch probability (ISSP) effect, respectively. However, what triggers metacontrol in the first place has not been clearly articulated. While a “mere-experience” hypothesis attributes metacontrol to changes in control demands, an “affective-signaling” hypothesis suggests that high control demands are aversive and aversiveness drives metacontrol. In two experiments, we adjudicated between these hypotheses by considering the modes of metacontrol (proactive vs. reactive) and temporal dynamics of background valence (sustained vs. transient and positive vs. negative). We induced metacontrol (proactive or reactive) in a task-switching paradigm and created background valence by using positive and negative images as stimuli. With valence being an irrelevant aspect of the task, the design allows us to test whether (task-irrelevant) background valence would modulate metacontrol. While we were able to replicate the LWSP effect in Experiment 1 and the ISSP effect in Experiment 2, we did not find valence modulating either effect, regardless of the background valence being a sustained (Experiment 1) or a transient one (Experiment 2). These findings together suggest that negative valence (i.e., aversiveness) does not necessarily benefit metacontrol, and control demand variations are sufficient to induce metacontrol.

Learned switch readiness via concurrent activation of task sets: Evidence from task specificity and memory load

“…Here, our LWSP design aimed to induce proactive metacontrol instead and test for the similar within-task generalizability. The inclusion of diagnostic items with unbiased switch frequencies was used to demonstrate that the control state is applied proactively, even before a stimulus is encountered (e.g., Kang & Chiu, 2021;Kang & Yu-Chin, 2022). Unlike Fröber et al (2021), our switch probability manipulation was tied to the stimuli, which remained consistent throughout the learning and transfer phases, increasing the likelihood of detecting transfer.…”

“…Subsequent studies further qualified that the increased switch readiness in a frequent switch list is partly a result of proactive processing, which is applied to all trials in the same list and might be engaged even before a trial begins. The evidence supporting this interpretation regarding proactive processing came from studies demonstrating the LWSP effect in so-called "diagnostic" items, which are part of switchprobability-biased lists yet are themselves unbiased, i.e., associated with a 50% chance of switching (e.g., Kang & Chiu 2021;Kang & Yu-Chin, 2022). Such findings underscore the usefulness of switch cost measurement in cued task switching paradigms for investigating metacontrol shifts in flexibility.…”