Beyond feature binding: interference from episodic context binding creates the bivalency effect in task-switching

Predictive alternating runs and random task-switching sequences produce dissociative switch costs in the Consonant-Vowel/Odd-Even task

Task-switching is commonly used to investigate working memory and attentional control processes. The current study compares predictive versus non-predictive task-sequencing effects on task-switching performance. Participants completed four blocks of the Consonant-Vowel/Odd-Even (CVOE) task: Two single task pure blocks, a predictable switch block where task switching occurred every two trials, and a random switch block where switching was unpredictable. In addition to mean error rates and response times (RTs), we assessed sequence effects on local switch costs (i.e., switch vs. nonswitch trials) and global costs (i.e., nonswitch vs. pure trials) for both error rates and RTs along with their underlying distributions. Overall, we show that while predictive and random switching produced similar patterns for mean error rates and RTs, a dissociation occurred in RT switch costs. When switching was random, local costs were inflated. In contrast, predictive switching increased global costs. Increased local costs for random versus predictive switching reflect an increase in task-set reconfiguration processes as participants struggle to reconfigure to an unpredictable task type in working memory on a subsequent trial. Separately, increased global costs for predictive switching reflect declines in task-set maintenance processes, as participants must maintain both task types in working memory while simultaneously monitoring their progress through the trial sequencing.

Different impact of task switching and response-category conflict on subsequent memory

The impact of cognitive control demands on long-term memory is mixed, with some conflicts leading to better, others leading to worse subsequent memory. The current study was designed to investigate how different types of cognitive control demands modulate the effects on memory. At study, participants had to switch between two classification tasks and later, free recall performance was assessed. The stimuli consisted of two interleaved words, one word had to be categorized and the other word had to be ignored. In four experiments, the congruency between target and ignored words was manipulated by changing the distractor category. This allowed us to investigate the impact of different types of conflict (i.e., task switching, perceptual load, response-category conflict, stimulus-category conflict). The results revealed that task switching impaired memory in all experiments. In Experiment 1, higher perceptual load also impaired memory. Experiments 2–4 showed that the co-activation of two words which required different responses (i.e., response-category conflict) enhanced memory performance but only when the conflict stimuli were presented in pure blocks. Overall, memory performance seems to depend on attentional policies. Withdrawing attention from target encoding results in lower memory performance. In contrast, focusing attention on the target results in enhanced memory performance.

Monitoring and control in multitasking

The idea that conflict detection triggers control adjustments has been considered a basic principle of cognitive control. So far, this "conflict-control loop" has mainly been investigated in the context of response conflicts in single tasks. In this theoretical position paper, we explore whether, and how, this principle might be involved in multitasking performance, as well. We argue that several kinds of conflict-control loops can be identified in multitasking at multiple levels (e.g., the response level and the task level), and we provide a selective review of empirical observations. We present examples of conflict monitoring and control adjustments in dual-task and task-switching paradigms, followed by a section on error monitoring and posterror adjustments in multitasking. We conclude by outlining future research questions regarding monitoring and control in multitasking, including the potential roles of affect and associative learning for conflict-control loops in multitasking.

How long-lasting is the post-conflict slowing after incongruent trials? Evidence from the Stroop, Simon, and flanker tasks

The purpose of the present study was to determine how long-lasting the post-conflict slowing following incongruent stimuli is. In previous research, incongruent stimuli have been used to induce a conflict because they have relevant features for two different response alternatives. So far, the post-conflict slowing following incongruent stimuli has mainly been assessed up to one trial. In the first two experiments, we assessed the persistence of the post-conflict slowing across several trials. To this end, we presented a few incongruent stimuli among non-conflict stimuli. The results showed a consistent slowing for the first few trials immediately following the incongruent trials. In addition, a sporadic slowing was still found on later trials. In two subsequent experiments, we investigated to what extent the infrequency of incongruent trials - rather than their conflict - induced this slowing. To determine this, we used the same design as in the first two experiments, but we presented non-conflict stimuli as infrequent stimuli. The results showed a slowing on one subsequent trial, ruling out the possibility that the post-conflict slowing following incongruent trials was only caused by infrequency. Together, the findings of the present study indicate that the conflict induced by incongruent trials can have a longer lasting impact on subsequent trials than previously thought.

Post-conflict slowing effects in monolingual and bilingual children

Previous research has shown that bilingual children outperform their monolingual peers on a wide variety of tasks measuring executive functions (EF). However, recent failures to replicate this finding have cast doubt on the idea that the bilingual experience leads to domain-general cognitive benefits. The present study explored the role of disengagement of attention as an explanation for why some studies fail to produce this result. Eighty children (40 monolingual, 40 bilingual) who were 7 years old performed a task-switching experiment. In the pure blocks, three simple non-conflict tasks were performed in which children responded by pressing one of two response keys. In the conflict block, occasional bivalent stimuli appeared and created conflict because the irrelevant dimension was mapped to the incorrect response key. The results showed that these bivalent stimuli affected subsequent performance in the conflict block. For monolinguals, the effect of conflict was found for up to 12 trials after the appearance of the bivalent stimulus, but for bilinguals the effect disappeared after only two trials. The results are interpreted as evidence for faster disengagement of attention by bilingual children. Most studies examining EF in monolingual and bilingual children do not examine trial-by-trial adjustments following conflict, but these are essential considerations because relevant processing differences are masked when analyses are applied to data averaged across entire blocks.

Post-conflict slowing after incongruent stimuli: from general to conflict-specific

Encountering a cognitive conflict not only slows current performance, but it can also affect subsequent performance, in particular when the conflict is induced with bivalent stimuli (i.e., stimuli with relevant features for two different tasks) or with incongruent trials (i.e., stimuli with relevant features for two response alternatives). The post-conflict slowing following bivalent stimuli, called "bivalency effect", affects all subsequent stimuli, irrespective of whether the subsequent stimuli share relevant features with the conflict stimuli. To date, it is unknown whether the conflict induced by incongruent stimuli results in a similar post-conflict slowing. To investigate this, we performed six experiments in which participants switched between two tasks. In one task, incongruent stimuli appeared occasionally; in the other task, stimuli shared no feature with the incongruent trials. The results showed an initial performance slowing that affected all tasks after incongruent trials. On further trials, however, the slowing only affected the task sharing features with the conflict stimuli. Therefore, the post-conflict slowing following incongruent stimuli is first general and then becomes conflict-specific across trials. These findings are discussed within current task switching and cognitive control accounts.

Turning univalent stimuli bivalent: Synesthesia can cause cognitive conflict in task switching

In this study we investigated whether synesthetic color experiences have similar effects as real colors in cognitive conflict adaptation. We tested 24 synesthetes and two yoke-matched control groups in a task-switching experiment that involved regular switches between three simple decision tasks (a color decision, a form decision, and a size decision). In most of the trials the stimuli were univalent, that is, specific for each task. However, occasionally, black graphemes were presented for the size decisions and we tested whether they would trigger synesthetic color experiences and thus, turn them into bivalent stimuli. The results confirmed this expectation. We were also interested in their effect for subsequent performance (i.e., the bivalency effect). The results showed that for synesthetic colors the bivalency effect was not as pronounced as for real colors. The latter result may be related to differences between synesthetes and controls in coping with color conflict.

Support for a history-dependent predictive model of dACC activity in producing the bivalency effect: an event-related potential study

In the present study, we examine electrophysiological correlates of factors influencing an adjustment in cognitive control known as the bivalency effect. During task-switching, the occasional presence of bivalent stimuli in a block of univalent trials is enough to elicit a response slowing on all subsequent univalent trials. Bivalent stimuli can be congruent or incongruent with respect to the response afforded by the irrelevant stimulus feature. Here we show that the incongruent bivalency effect, the congruent bivalency effect, and an effect of a simple violation of expectancy are captured at a frontal ERP component (between 300 and 550ms) associated with ACC activity, and that the unexpectedness effect is distinguished from both congruent and incongruent bivalency effects at an earlier component (100-120ms) associated with the temporal parietal junction. We suggest that the frontal component reflects the dACC's role in predicting future cognitive load based on recent history. In contrast, the posterior component may index early visual feature extraction in response to bivalent stimuli that cue currently ongoing tasks; dACC activity may trigger the temporal parietal activity only when specific task cueing is involved and not for simple violations of expectancy.

More conflict does not trigger more adjustment of cognitive control for subsequent events: a study of the bivalency effect

Encountering a conflict triggers an adjustment of cognitive control. This adjustment of cognitive control can even affect subsequent performance. The purpose of the present study was to determine whether more conflict triggers more adjustment of cognitive control for subsequent performance. To this end, we focussed on the bivalency effect, that is, the adjustment of cognitive control following the conflict induced by bivalent stimuli (i.e., stimuli with relevant features for two tasks). In two experiments, we tested whether the amount of conflict triggered by bivalent stimuli affected the bivalency effect. Bivalent stimuli were either compatible (i.e., affording one response) or incompatible (i.e., affording two different responses). Thus, compatible bivalent stimuli involved a task conflict, whereas incompatible bivalent stimuli involved a task and a response conflict. The results showed that the bivalency effect was not affected by this manipulation. This indicates that more conflict does not trigger more adjustment of cognitive control for subsequent performance. Therefore, only the occurrence of conflict--not its amount--is determinant for cognitive control.

An orienting response is not enough: Bivalency not infrequency causes the bivalency effect

When switching tasks, occasionally responding to bivalent stimuli (i.e., stimuli with relevant features for two different tasks) slows performance on subsequent univalent stimuli, even when they do not share relevant features with bivalent stimuli. This performance slowing is labelled the bivalency effect. Here, we investigated whether the bivalency effect results from an orienting response to the infrequent stimuli (i.e., the bivalent stimuli). To this end, we compared the impact of responding to infrequent univalent stimuli to the impact of responding to infrequent bivalent stimuli. For the latter, the results showed a performance slowing for all trials following bivalent stimuli. This indicates a long-lasting bivalency effect, replicating previous findings. For infrequent univalent stimuli, however, the results showed a smaller and shorter-lived performance slowing. These results demonstrate that the bivalency effect does not simply reflect an orienting response to infrequent stimuli. Rather it results from the conflict induced by bivalent stimuli, probably by episodic binding with the more demanding context created by them.

A role for recency of response conflict in producing the bivalency effect

The bivalency effect is a block-wise response slowing that is observed during task-switching when rare stimuli that cue two tasks (bivalent stimuli) are encountered. This adjustment in response style affects all trials that follow bivalent stimuli, including those trials that do not share any features with bivalent stimuli. However, the specific stimulus and response properties that trigger the bivalency effect are not well understood. In typical bivalency effect experiments, bivalent stimuli can be congruent or incongruent with respect to the response afforded by the irrelevant stimulus feature, and this distinction has never been examined. In the present study, we show that cognitive load defined by the response incongruence on bivalent trials plays a critical role in producing the subsequent response slowing observed in the bivalency effect, as well as maintaining the magnitude of the bivalency effect across practice. We propose that the bivalency effect reflects a process involved in predicting future cognitive load based on recent cognitive load experience. This is in line with a recent proposal for a role of the ACC in monitoring ongoing changes in the environment to optimize future performance (Sheth et al., in Nature 488:218-221, 2012).