Flexibility and stability defined in this way have traditionally been viewed as opposing ends of a single spectrum, representing an unavoidable tradeoff (Cools, 2016; Dreisbach, 2012; Dreisbach & Fr\u00f6ber, 2019; Goschke, 2003, 2013; Hommel, 2015; Paul et al., 2021). We refer to this conceptualization as the unidimensional framework of metacontrol. According to the unidimensional framework, flexibility and stability are considered antagonistic, always varying inversely. However, when relating the unidimensional framework to real-world scenarios, the framework would suggest that a student switching smoothly between the two goal-relevant, beneficial tasks (listening to the lecture and taking notes) cannot effectively shield against the goal-irrelevant task of checking their phone. Treating flexibility and stability as a zero-sum tradeoff does not reflect how daily life often requires both simultaneously.<\/p>\n\n\n\n
To allow flexibility and stability to vary independently, we propose an alternative conceptualization dubbed the dual-dimension framework (DDF; see also Geddert and Egner, 2022; Egner, 2023 for similar proposals). However, this is not a fully elaborated model, but rather a conceptual framework for organizing findings already covered by the unidimensional framework and those beyond its scope. Therefore, the rest for this paper adheres to the following structure: (1) a review of the rationale and evidence for the unidimensional framework; (2) a review of the evidence that is inconsistent with the unidimensional framework; (3) an in-depth description of the DDF; and (4) a discussion of the implications and benefits should the DDF be supported by future evidence.<\/p>\n\n\n\n
The Unidimensional Framework of Flexibility And Stability<\/h2>\n\n\n\n
In this section, we elaborate on the unidimensional perspective. We first summarize the key ideas and rationale. We then describe common experimental paradigms for evaluating flexibility and stability, followed by the body of evidence supporting the unidimensional framework. Throughout this section, we also occasionally highlight the limitations of the evidence supporting the unidimensional approach, paving the way for the next section, titled Challenges to the Unidimensional Framework.<\/em><\/p>\n\n\n\n The rationale for the unidimensional framework is best understood through the lens of task-sets, defined as a collection of mental representations of stimuli, rules, and responses needed to produce goal-appropriate behavior. A task-set therefore focuses attention on task-relevant stimuli features and away from irrelevant features. (Dreisbach & Haider 2008, 2009; Dreisbach & Wenke 2011). A task set can be instantiated strongly, making it \u201cshielded\u201d against other task-irrelevant stimuli or distractions, or less strongly, making it less shielded. Strong shielding is synonymous with cognitive stability. However, if one\u2019s goal suddenly changes, strong shielding becomes detrimental, synonymous with impaired flexibility. The difference between stability and impaired flexibility is merely a semantic value judgment: The terms denote identical information processing strategies for the task-set. The key is that high shielding produces stability and, consequently, reduced flexibility. This description is a central feature of the unidimensional account: a single parameter\u2014the degree of task-set shielding\u2014determines both flexibility and stability such that they are locked in a tradeoff.[1]<\/sup><\/a><\/p>\n\n\n\n Flexibility\/stability as broad and general descriptors has been the subject of much psychological research (e.g., for reviews, see Braem & Egner, 2018; Eppinger et al., 2021; Goschke, 2013; Hommel & Colzato, 2017). Due to this diversity of research, flexibility and stability are conceptualized and operationalized in many different ways, creating ambiguity and confusion when interpreting research findings (e.g., Ionescu, 2012). To avoid this problem, we narrow our focus to examine flexibility and stability at the level of the \u201ctask-set.\u201d Accordingly, we primarily focus on paradigms that are designed to examine control over task-sets, especially in \u201ccued task switching paradigms\u201d where participants are cued on which task to perform for each trial. Task switches typically incur longer response times and more errors compared to task repeats, indicating the involvement of cognitive control processes. These processes include activating new task-sets in working memory<\/a> and overcoming lingering activations from previous task-sets (Allport et al., 1994; Meiran, 1996b; Rogers & Monsell, 1995). When these processes are upregulated to promote better switching among task-sets, switch costs are reduced. Within our circumscribed definition of flexibility, decreases in switch costs serve as the operational definition of heightened flexibility.<\/p>\n\n\n\n Based on this definition, a common manipulation used to shift flexibility is adjusting the number of switch trials: Switch costs are reduced when switches are frequent versus rare (e.g., Dreisbach et al., 2002; Dreisbach & Haider, 2006; Kang & Chiu, 2021; Liu & Yeung, 2020; Monsell & Mizon, 2006; Schneider, 2016; Schneider & Logan, 2006). This has been demonstrated, for example, when spatial locations or stimulus identity are used as contextual cues associated with biased switch probabilities (Chiu, 2019; Crump & Logan, 2010; Leboe et al., 2008). The effect has also been demonstrated using lists (blocks of trials) with frequent vs. rare switches (e.g., Dreisbach et al., 2002; Dreisbach & Haider, 2006; Kang & Chiu, 2021; Liu & Yeung, 2020; Monsell & Mizon, 2006; Schneider, 2016; Schneider & Logan, 2006; Yu-Chin, 2022). This effect, known as the list-wide switch probability effect, reflects increased flexibility as participants become more adept at switching between tasks. Lists serve as contextual cues that modulate the efficiency of control over task-sets.<\/p>\n\n\n\n The same task-switching paradigms can also be used to assess stability, so long as the paradigm utilizes bivalent stimuli. Bivalent stimuli afford two different tasks with overlapping responses. For example, the digit \u201c3\u201d can be classified as \u201codd\u201d in an odd\/even task and \u201csmall\u201d in a larger\/smaller-than-5 task. When one task is relevant and the other is irrelevant, bivalent stimuli can be either congruent (producing the same response in both tasks) or incongruent (producing different responses). These congruency differences lead to differences in performance, known as the congruency effect (Sudevan & Taylor, 1987). This effect arises either from interference caused by past experiences of different responses to the same stimulus (Logan, 1988, 2002; Yamaguchi & Proctor, 2011) or from categorizing every stimulus based on both task rules (Schneider, 2015, 2018). Regardless of the source, the performance cost observed on incongruent trials reflects the interference of irrelevant and detrimental information in goal-relevant processing. Cognitive control is needed to suppress irrelevant information and shield the relevant task-set. When these processes are upregulated, congruency effects are reduced. Decreases in congruency effects are commonly used to operationalize heightened stability.<\/p>\n\n\n\n Stability is also commonly modulated by manipulating the proportion of incongruent trials in a paradigm (e.g., Bejjani et al., 2022; Botvinick et al., 2001; Dreisbach & Fr\u00f6ber, 2019; Geddert & Egner, 2022; van Steenbergen, 2015). Mirroring list-wide switch probability effects for switch costs, differences in congruency effects between frequent vs. rare incongruent lists are referred to as list-wide proportion congruent effects. These effects are often researched in Stroop paradigms (Bugg et al., 2011; Bugg & Hutchison, 2013; Chiu et al., 2017; Jacoby et al., 2003; for reviews see Bugg, 2017; Bugg & Crump, 2012), but have also been demonstrated in the congruency effects of task-switching paradigms (e.g., Bejjani et al., 2022; Botvinick et al., 2001; Braem, 2017; Dreisbach & Fr\u00f6ber, 2019; Geddert & Egner, 2022; van Steenbergen, 2015). Together, list-wide switch probability and list-wide proportion congruent manipulations are commonly used to alter metacontrol and provide working illustrations of the context-dependent changes that we use to operationalize metacontrol.<\/p>\n\n\n\n However, researchers differ on how they interpret congruency effects as metacontrol indices. In accordance with the unidimensional framework, some researchers consider congruency effects to reflect both flexibility and stability. This idea assumes that the unidimensional framework is correct: Any increases in congruency effects must indicate a decrease in flexibility. To illustrate, this tradeoff is analogous to how movement on one end of a see-saw must unavoidably be compensated on the other end. While this is an efficient approach from within the unidimensional framework, it is not suitable for testing the veracity of the unidimensional framework itself; the result could only be circular. To avoid this problem, others have suggested that flexibility and stability must be indexed separately. In this approach, context-dependent changes in congruency effects are used to measure stability only, while changes in switch costs measure flexibility only (Geddert & Egner, 2022). In this paper, we adopt the latter approach and advocate for the separate assessment of flexibility and stability as distinct constructs. This separation allows for an unbiased examination of the relationship between flexibility and stability. In the next subsection, we review relevant findings that satisfy this criterion, enabling an exploration of the relationship between the two metacontrol states.<\/p>\n\n\n\n The unidimensional framework has intuitive appeal and, without a doubt, is supported by several lines of behavioral evidence. A particularly recent example comes from a line of research demonstrating the susceptibility of metacontrol to operant conditioning. Braem (2017) frequently rewarded one group of participants immediately following successful task switches during cued task switching, while the other group was instead rewarded frequently after task repeats. In a later part of the experiment, both groups were asked to freely choose either to switch or to repeat tasks. The results revealed that participants frequently rewarded for task switches earlier had a higher voluntary switch rate compared to those frequently rewarded for task repeats. Moreover, Braem (2017) also found an increased congruency effect in the group that had been conditioned for flexibility, indicating a flexibility-stability tradeoff. Similar conditioning effects have been shown in the opposite direction as well: Held et al. (2023) rewarded participants frequently on incongruent trials and rarely on congruent trials. Participants who had been conditioned for stability then showed reduced congruency effects as well as an increase in switch costs, indicating a flexibility-stability tradeoff once more. Merging these two conditioning procedures, Bartossek et al. (2023) replicated both Braem\u2019s (2017) and Held et al.\u2019s (2023) findings and furthermore revealed that the loss in stability (or flexibility) was proportional to the volitional increase in stability (or flexibility).<\/p>\n\n\n\n As previously mentioned, a very common metacontrol manipulation is to manipulate the frequency of task switches or incongruent trials. This method occasionally offers support for the unidimensional account. Particularly strong support can be found in a recent study by Qiao et al. (2023), who manipulated the frequency of incongruent trials by including task-irrelevant distractors to evaluate its impact on the flexibility-stability tradeoff.\u00a0 They reported, and subsequently replicated, that an increase in the frequency of distractors resulted in both decreased distractor effects and increased switch costs. Moreover, the authors modeled participants\u2019 control learning and found that the best-fitting model assigned both flexibility and stability to a single parameter\u2014guaranteeing a tradeoff. Together, these recent studies provide support for the unidimensional account.<\/p>\n\n\n\n Above, we have described instances where changes in external context led to changes in cognitive control settings. Changes in internal context, such as affect, can also modulate cognitive control. Among the most often cited when discussing a flexibility-stability tradeoff, Dreisbach and Goschke (2004) directed participants to perform a single categorization task on a target stimulus indicated by a pre-defined ink color while ignoring a differently colored distractor. The mappings between ink color and target\/distractor changed mid-way through the experiment: In one condition, the novel color designated the target, but the previous target color became the distractor color after the switch. This setup was intended to measure perseveration, or the degree to which participants continue to pay attention to the previous, but now irrelevant, target color. In the other condition, the new color designated the distractor, and the old distractor color became the new target color. This setup was intended to measure distractibility, or the degree to which one is inappropriately attracted to the novel distractor color. The results confirmed that each condition produced perseveration or distractibility, respectively. Armed with these diverging conditions, positive affect was induced among participants in Experiment 2. Affect had been selected as an established means of shifting metacontrol towards flexibility. The affect manipulation had opposite impacts in the two conditions: reducing perseveration in the first but increasing distractibility in the second condition (Dreisbach & Goschke, 2004). This is among the earlier studies finding that facilitated switching comes at the cost of intrusions from unwanted task-sets \u2013 i.e., a flexibility-stability tradeoff.<\/p>\n\n\n\n Besides these findings, many additional studies documented a flexibility-stability tradeoff (Brown et al., 2007; Chiew & Braver, 2014; Cools et al., 2010; Dreisbach et al., 2005; Dreisbach, 2006; Fischer & Hommel, 2012; Goschke, 2000; Goschke & Bolte, 2014; Hefer & Dreisbach, 2016, 2017; Locke & Braver, 2008; Stoet & Snyder, 2003, 2007b; Tharp & Pickering, 2011; Watzek et al., 2019; Yahya & \u00d6zkan Ceylan, 2022). In each of these studies, a manipulation or individual difference which impacts flexibility had an inverse impact on stability, suggesting that the two constructs traded off. All of the evidence cited above in support of the unidimensional framework does<\/em> satisfy the requirement for separate measurement of flexibility and stability. However, as we elaborate next, a considerable body of research that also satisfies the separation requirement did not align well with the unidimensional framework.<\/p>\n\n\n\n In this section, we highlight literature that is inconsistent with the unidimensional framework. We divide our review of these findings into two broad categories: within individuals versus between individuals. It is important to consider both within-individual and between-individual tradeoffs (and the lack thereof) because separate patterns could conceivably emerge (e.g., Mekern et al., 2019). These within-individual studies include, in order: (1) flexibility manipulations failing to impact congruency effects; (2) flexibility manipulations failing to impact attentional capture; and (3) flexibility and stability often loading onto separate latent factors. The between-individual studies include: (1) individual differences in flexibility failing to predict stability measures; and (2) neuropsychological double dissociations between flexibility and stability.<\/p>\n\n\n\n Based on the unidimensional framework, changes in flexibility should always be offset by opposite changes in stability. However, this assumption was challenged by Geddert and Egner (2022)\u2019s recent study. The authors independently manipulated both the probability of task switches and the proportion of task-rule incongruent trials within-subjects<\/em>\u2014two manipulations that have produced robust behavioral effects, i.e., the list-wide switch probability effect indexing flexibility, and the list-wide proportion congruent effect indexing stability, respectively. Thus, the study allowed measurement of both (1) changes in task-rule congruency effects as switches became more frequent and (2) changes in switch costs as incongruent trials became more prevalent. According to the unidimensional framework, there should be an interaction between the two manipulations and specifically a tradeoff relationship: A larger congruency effect in the frequent switch condition than in the rare one, and a larger switch cost in the frequent incongruent condition than in the rare one. Unexpectedly, neither effect was observed\u2014 (1) adaptation to the higher switch probability did not increase congruency effects, and (2) adaptation to the higher proportion incongruence did not increase switch costs. When compared to a model including an interaction term, Bayesian statistics indicated evidence 7 times greater supporting a model with no interaction. This finding of no interaction was replicated using different kinds of task-rule congruency. These findings revealed instances where flexibility and stability varied independently.<\/p>\n\n\n\n Switch probability manipulations\u2019 lack of impact on congruency effects have also been documented in earlier studies (Chiu & Egner, 2017; Kang & Chiu, 2021; Siqi-Liu & Egner, 2020). While shifting flexibility (not stability) was the primary focus, stability could still be indexed by congruency effects because bivalent stimuli were used. From our own lab, we consistently found smaller switch costs in the frequent switch condition than in the rare one, but have never found a significant difference in the congruency effect between the two conditions (Chiu & Egner, 2017; Kang & Chiu, 2021). Taken together, there is consistent evidence that individuals adapt to situations requiring cognitive flexibility without relaxing stability. Regarding the lack of modulation of congruency effects in these studies, a straightforward explanation is that there was no demand for changing stability across different switch probability conditions. This explanation fits with the idea of demand avoidance (e.g., Kool et al., 2010; Schouppe et al., 2014; Shenhav et al., 2013) whereby cognitive effort is a key determinant for engaging cognitive control and metacontrol.<\/p>\n\n\n\n Because these acute experiences with frequent switching can induce shifts towards flexible metacontrol states, more chronic exposure may be expected to similarly reduce switch costs. Indeed, Zhao et al. (2020) revealed significantly diminished switch costs after 21 days of task-switching training. This benefit was observed both on the trained tasks and a transfer task with different stimuli and response rules. This transfer means that participants became better at switching in general, perhaps by associating the laboratory with the demand for a flexible metacontrol state and retrieving it in the transfer task. Critically, this flexible state did not come at the cost of decreased stability: Training had no impact on stability indices derived from the Stroop or Flanker paradigms. A similar pattern was demonstrated among children (Zhao et al. 2018). Intensive training can also cause a shift towards greater stability but no complimentary shift away from flexibility. Talanow and Ettinger (2018) reported improved Stroop performance over 8 training sessions, but there was no change in (untrained) switch costs. In the absence of training or a biased switch probability manipulation, flexibility can also be increased by inducing positive affect (Tae et al., 2021, Experiment 2) without modulating stability.<\/p>\n\n\n\n In addition to task-rule congruency effects, stability can be operationalized as the extent of shielding against task-irrelevant distractors. Inconsistent with the unidimensional framework, studies have shown that increasing flexibility does not invariably increase bottom-up attentional capture by these distractors. For instance, in a recent study by Sali and Key (2021), participants switched between two categorization tasks, and the probability of switches was varied across blocks in order to modulate switch costs, similar to previous studies. Unlike others, the target stimulus for the categorization task was inside a pre-defined target shape in an array with other shapes (e.g., a circle among diamonds). Thus, before performing the categorization task, participants had to locate the target stimulus first. Critically, in half of the trials, both target and distractors were in the same color (i.e., a distractor absent condition), while in the other half, one of the distractors was in a unique color (i.e., a distractor present condition). This distractor present condition has been shown to capture attention and slow responses to the target (e.g., Theeuwes, 1992). Thus, comparing distractor absent versus present conditions provides an index of stability. In two experiments, the switch probability manipulation modulated switch costs with frequent switches incurring smaller switch costs, as expected. However, the performance cost due to distractors was not significantly different when switches were frequent versus when they were rare. These results showed that increased flexibility did not come with decreased stability as indexed by attentional capture.<\/p>\n\n\n\n Besides considering the direct effects of flexibility manipulations like switch probability on stability indices, we can also examine the factor structure of these constructs. Factors can be derived from behavioral paradigms such as those using list-wide switch probability\/list-wide proportion congruent manipulations, or from self-report data. After each are discussed in turn, both types of data will point to a similar factor structure: flexibility and stability represent distinct\u2014albeit positively correlated\u2014latent variables.<\/p>\n\n\n\n Bejjani et al. (2022) recently investigated whether flexible and stable metacontrol states can be explained by one versus two factors. Participants performed two separate paradigms: One paradigm featured a list-wide switch probability manipulation and the other featured a list-wide proportion congruent manipulation. The authors examined correlations across participants between the list-wide switch probability effect and the list-wide proportion congruent effect. A structural equation modeling analysis was performed to find out whether individual performance differences in both tasks could be explained by (1) a single-factor structure, (2) a two, negatively correlated factor structure, or (3) a two, positively correlated factor structure. The first and second structures would be consistent with the unidimensional framework, while the third would not. Challenging the unidimensional framework, Bejjani et al. (2022) found that the list-wide switch probability and list-wide proportion congruent effects are best explained by a two, positively correlated factor structure. Namely, flexibility and stability in metacontrol are separate constructs, and it is possible for the same individuals to be proficient or deficient at both.<\/p>\n\n\n\n Self-report studies offer a separate line of converging evidence challenging the unidimensional framework by measuring individuals\u2019 perceptions of their own cognitive flexibility and stability. In particular, Derryberry and Reed (2002) measured self-report flexibility and stability in terms of shifting versus focusing of attention. Earlier, Derryberry and Rothbart (1988) noticed that people tend to self-identify themselves as skilled in both shifting and focusing, or unskilled in both; but rarely as prioritizing one in such a way that requires tradeoff. To formally test this, Derryberry and Reed (2002) later developed an Attentional Control Scale, asking participants to report their agreement with a series of statements related to shifting, such as, \u201cIt is easy for me to alternate between two different tasks\u201d as well as statements related to focusing, such as, \u201cWhen concentrating, I can focus my attention so that I become unaware of what\u2019s going on in the room around me.\u201d A confirmatory factor analysis revealed two factors (Chiorri & Vannucci, 2019), supporting the proposal that flexibility and stability in attentional control are dissociable states of an individual. Such findings would be unexpected based on the unidimensional framework. Notably, in subsequent studies using the Attentional Control Scale, the two subscales were positively correlated across individuals (Carriere et al., 2013; Chiorri & Vannucci, 2019; Jessup et al., 2021; Ralph et al., 2014, 2017; Sansevere & Ward, 2021; Qiao & Liu, 2020). Like much of the behavioral data, psychometric data suggest that flexibility and stability are not locked in a tradeoff.<\/p>\n\n\n\n Inter-individual relationships between flexibility and stability provide another way to evaluate metacontrol. That is, the unidimensional framework predicts a negative correlation between individual differences in flexibility and stability. Recall that flexibility and stability are metacontrol states, indexed via \u201cchanges\u201d in cognitive control. Therefore, evidence presented in this section will relate to individual difference factors which produce \u201cchanges\u201d in cognitive control indices, rather than the direct correlation between the indices themselves. However, if any, we did find several documented instances of nonsignificant or positive correlations between switch costs and Stroop interference (e.g., S\u00e1nchez-Cubillo et al., 2009; Ward et al., 2001; Zhao et al., 2021). While not speaking directly to metacontrol as we have defined it here, these findings do suggest that there are at least separable cognitive control mechanisms engaged in overcoming Stroop interference and in switch costs. Such distinct mechanisms at the control level are likely necessary for independent modulation at the metacontrol level, serving as an assumption check lending additional credence to the individual difference studies considered next.<\/p>\n\n\n\n At the level of metacontrol, several quasi-experimental studies provide additional evidence challenging the unidimensional framework. Such studies can reveal whether individual differences in flexibility also have an impact on stability in the same group of participants. In some studies, the individual difference variable of interest might be considered flexibility per se. For example, Wiradhany et al. (2020) found no relationship between self-reported media multitasking (flexibility) and distractor suppression (stability). However, it may be argued that self-report media multitasking taps different constructs than differences in switch costs, and including a separate index of flexibility (other than the individual difference variable of interest) would allow for stronger conclusions. Accordingly, all following papers in this section report the impacts of individual difference variables on common indices of both flexibility and stability. Although the impacts of media multitasking on cognitive measures are strongly contested and directionally inconsistent (e.g., Luo et al., 2021, 2022; Murphy & Shin, 2022; Parry & Le Roux, 2021; Schneider & Chun, 2021; Wiradhany & Nieuwenstein, 2017), when reported, such effects tend to selectively impact switch costs and not laboratory stability indexes (e.g., Alzahabi & Becker, 2013; Ophir et al., 2009; Wiradhany & Nieuwenstein, 2017). A similar pattern appears when using self-report flexibility and stability as opposed to task-based measures: Reported flexibility, but not stability, differs based on media multitasking (Luo et al., 2022).<\/p>\n\n\n\n Another individual difference factor with bearing on flexibility and stability is reappraisal ability, defined as the capacity to adaptively change one\u2019s interpretation of negative information. Higher reappraisal ability predicted increased switch costs, but no change in Stroop effects (McRae et al., 2012). In a similar vein, school classes taught in two languages predicted decreased switch costs, but no change on congruency effects (Christoffels et al., 2015). Although cognitive advantages of bilingualism are inconsistent and contested (Dick et al., 2019; Mas-Herrero et al., 2021; Paap et al., 2017; Sanchez-Azanza et al., 2017), in this particular study, students enrolled in bilingual classes produced smaller switch costs compared to those instructed in only one language. While switch costs differed as a function of bilingual education, the congruency effect did not. Similarly, Dong and Liu (2016) investigated the impacts of a semester-long class on language interpreting. Upon post-test of cognitive abilities, students in this class showed reductions in switch costs, but no change in Stroop effects. In other words, reappraisal ability and language training selectively influenced flexibility without impacting stability. There are also examples of independence in the opposite direction: Stroop interference increased with age, but switch costs showed no changes (Hirsch et al., 2016, Experiment 1<\/em>; Hutchison et al., 2010; Reimers & Maylor, 2005; Wasylyshyn et al., 2011, but see Georgiou-Karistianis et al., 2006, who used a different paradigm). While the individual differences studies in this section did not involve direct experimental manipulation of flexibility or stability, they follow the same pattern as the within-subject studies: Certain individual difference variables appear to increase or decrease one metacontrol state across individuals without causing opposite changes in the other metacontrol state in those same individuals. The unidimensional framework is not well suited to explain this selectivity of modulation.<\/p>\n\n\n\n Double dissociations are a direct means of demonstrating that two different cognitive capabilities are subserved by separate neural systems. Indeed, Stuss et al. (2000) showed that lesions in distinct parts of the human frontal cortex are linked to distinct flexibility\/stability impairments as measured in the Wisconsin Card Sorting Test (WCST; Grant & Berg, 1948; Milner, 1963). Like cued task-switching, participants occasionally switched between different categorization rules to apply to multivalent stimuli; but unlike cued task-switching, the appropriate task was not explicitly instructed and instead needed to be inferred by performance feedback. Participants had to continuously adapt in a context-sensitive manner. We reason that this is similar to experimental manipulations of switch probability (or proportion congruency): The WCST should also prompt flexibility-stability metacontrol. With this premise in mind, flexibility and stability (or the lack thereof) in the WCST are operationalized by two different kinds of errors. Perseverative errors occur when participants do not adapt to changes and continue using a sorting rule after performance feedback indicates the rule is no longer relevant\u2014an intuitive index of inflexibility. Set-loss errors, on the other hand, occur when participants apply new sorting rules in the absence of feedback indicating a change\u2014a failure of task-set maintenance indicating inadequate stability. Lesions limited to the inferior medial frontal cortex were associated with no changes in perseverative errors but increased set-loss errors (Stuss et al., 2000, replicated also in monkeys by Dias et al., 1997), while lesions in the superior medial frontal cortex were instead associated with greatly increased perseverative errors but relatively small increases in set-loss errors (Stuss et al., 2000).<\/p>\n\n\n\n Although less localized than the lesions studies by Stuss et al. (2000), other neurological conditions also reveal dissociations between flexibility and stability. We summarize some of them in Table 1. For example, impaired flexibility but preserved stability are seen in conditions such as amyotrophic lateral sclerosis (Abrahams et al., 1997; Lange, et al., 2016d; Lange et al., 2016a; Seer et al., 2015) and schizophrenia (Manoach et al., 2002; Sullivan et al., 1993). Whereas, conditions associated with impaired stability but preserved flexibility include dyslexia (Kapoula et al., 2010; Protopapas et al., 2007; Stoet et al., 2007b), alcoholism (Sullivan et al., 1993) and sleep-related hypermotor epilepsy (Licchetta et al., 2018). Double dissociations are of neuropsychological value for localizing brain areas and systems responsible for behaviors. Unfortunately, the studies presented here provide neither consistent nor converging neuroanatomical loci of flexibility and stability, but instead, a diverse set of brain changes. This is unsurprising, given that flexibility and stability are higher-level metacontrol states emerging from recruitment of other basic processing abilities, not freestanding functions (c.f., Ionescu, 2012; Schneider & Logan, 2009). Metacontrol likely involves distributed and overlapping brain networks (e.g., Barbey et al., 2013; Collette & Van der Linden, 2002). While the localization of flexibility and stability is<\/p>\n\n\n\nOverview and Rationale<\/h3>\n\n\n\n
Operationalizing Flexibility and Stability<\/h3>\n\n\n\n
Empirical Evidence for the Unidimensional Account<\/h3>\n\n\n\n
Challenges to the Unidimensional Framework<\/h2>\n\n\n\n
Within Individuals<\/h3>\n\n\n\n
Flexibility Manipulations do not Influence Task-Rule Congruency Effects<\/h4>\n\n\n\n
Flexibility Manipulations do not Increase Attentional Capture.<\/h4>\n\n\n\n
Flexibility and Stability Load Onto two Separate Factors<\/h4>\n\n\n\n
Between Individuals<\/h3>\n\n\n\n
Individual Difference Variables Modulating Flexibility do not Trade off With Stability<\/h4>\n\n\n\n
Flexibility and Stability can be Doubly Dissociated<\/h4>\n\n\n\n
![\"The](\"https:\/\/advances.in\/psychology\/wp-content\/uploads\/Figure1_aip00007.webp\")
<\/a><\/figure>\n\n\n\n