Cognitive flexibility in and out of the laboratory: task switching, sustained attention, and mind wandering

The paper examines how cognitive flexibility supports everyday behavior by enabling people to switch tasks, disengage from irrelevant activities, and re-engage after lapses of attention. It frames cognitive flexibility as a core function of cognitive control, largely supported by prefrontal cortex and large-scale brain networks including the frontoparietal (FPN), dorsal attention (DAN), ventral attention (VAN), and default mode network (DMN). The authors highlight that task performance suffers during switches (switch costs) and that flexible control relies on activating and reconfiguring task-set representations that encompass attentional, perceptual, memory, motoric, goal, motivational, and contextual information. The review integrates research on laboratory task switching with work on sustained attention and mind wandering to understand the neural dynamics that mediate flexible control both between tasks and within a task.

The review synthesizes extensive behavioral and neuroimaging literatures on task switching and sustained attention/mind wandering. Behavioral work shows robust switch costs across procedures: predictable (alternating runs), task-cueing (with cue–stimulus interval effects reducing switch costs with longer CSIs), and voluntary task selection; switch costs index the demands of activating/reconfiguring task sets and interference or competition among sets. Neuroimaging meta-analyses implicate FPN and attentional networks as flexible hubs (e.g., posterior parietal cortex in DAN; ACC in VAN; DLPFC in FPN), with specific regions such as left inferior frontal junction (IFJ), intraparietal sulcus, pre-supplementary motor area, and precuneus linked to task switching. Trial-by-trial coding work (e.g., RSA) shows frontoparietal recoding of task sets; individuals with greater IFJ variability show smaller switch costs. The DMN, typically active during internally oriented states and deactivated during externally focused tasks, exhibits dynamic relationships with control networks relevant to flexibility. Variability in FPN–DMN functional connectivity relates to cognitive control differently at rest vs task. Evidence suggests DMN can be recruited during switching, especially for cross-domain transitions, possibly reflecting release from sustained attention and reconfiguration to broader task context. Ecologically valid paradigms (e.g., multistep everyday tasks, naturalistic film viewing) dissociate DMN representations of broad task context from FPN representations of step/item-level details. During film viewing, findings diverge by content: suspense tends to increase VAN and decrease DMN activity, whereas emotionally engaging content can increase DMN activity and DMN–FPN coupling; disengagement can appear as coactivation of FPN and DMN. Research on sustained attention links spontaneous network fluctuations (QPPs ~20 s) and anticorrelations between DAN/TPN and DMN to vigilance and engagement; behavioral variability paradigms (in-the-zone vs out-of-the-zone) map onto antagonistic DMN–DAN relationships, modulated by FPN and VAN. These converging results indicate that flexible control depends on dynamic, state-dependent interplay among DMN, DAN, FPN, and VAN across simple tasks and complex, naturalistic contexts.

This is a narrative review synthesizing behavioral and neuroimaging studies on task switching, sustained attention, and mind wandering. The authors summarize: (1) laboratory task-switching procedures (predictable/alternating runs, task-cueing with manipulations of cue–stimulus intervals, and voluntary task selection) to quantify switch costs; (2) neuroimaging meta-analyses and targeted fMRI studies identifying domain-general and specific regions (e.g., IFJ, intraparietal sulcus, pre-SMA, precuneus) and applying multivariate methods (e.g., representational similarity analysis) and measures of brain signal variability; (3) resting-state and task-based functional connectivity analyses examining anticorrelations and time-varying dynamics among DMN, DAN, FPN, and VAN; (4) quasi-periodic pattern (QPP) analysis to characterize ~20 s fluctuations in large-scale network activity and their relation to attentional states; (5) ecologically valid paradigms including multistep task sequences and naturalistic film viewing to assess engagement, emotion, suspense, and narrative processing while relating them to network dynamics. The review integrates across these methodological approaches to infer mechanisms of cognitive flexibility within and between tasks.

- Task switching reliably induces performance costs (slower, less accurate on switch vs repeat trials). Longer cue–stimulus intervals in task-cueing reduce switch costs, indicating preparatory control. - Meta-analytic and empirical neuroimaging link task switching to FPN and attentional networks, with specific regions implicated: left IFJ (FPN), intraparietal sulcus (DAN), pre-SMA (DAN), precuneus (DMN), ACC (VAN), and DLPFC (FPN). FPN acts as a flexible hub for task engagement; posterior parietal cortex supports attentional shifting. - Multivariate analyses (RSA) show frontoparietal recoding of task sets: more similar activation patterns for task repeats than switches; representational similarity predicts performance. Greater IFJ signal variability associates with smaller switch costs, reflecting enhanced flexibility. - DMN’s role is context-dependent: it generally increases during internally focused states and decreases during externally focused task engagement; however, it can be recruited during demanding executive control and cross-domain task switches, potentially reflecting release from sustained attention and representation of broad task context. - FPN vs DMN dissociation in multistep tasks: DMN encodes task-level context (e.g., which recipe), whereas FPN encodes step/item-level details (e.g., wash vegetables), supporting hierarchical control. - Naturalistic viewing findings diverge by content: suspense increases VAN activity and decreases DMN; emotionally engaging content increases DMN activity and DMN–FPN coupling; disengagement can appear as coactivation of FPN and DMN. - Sustained attention relates to spontaneous network dynamics: DMN and TPN (DAN/FPN/VAN) show quasi-periodic ~20 s anticorrelated fluctuations (QPPs) at rest; targets detected faster when TPN is more active than DMN. Stronger DMN–DAN anticorrelation associates with better sustained attention. - Behavioral variability indexes engagement: in-the-zone (low RT variability) vs out-of-the-zone (high variability) states map onto antagonistic DMN–DAN relationships. During engaged performance, DMN and FPN are anticorrelated and DMN–VAN correlated; during less engaged states, DMN–FPN become positively correlated and DMN–VAN negatively correlated. - Across simple lab tasks and complex, ecologically valid paradigms, common patterns of network antagonism and modulation (DMN vs DAN/FPN/VAN) underpin cognitive flexibility and attentional state changes.

The findings converge on a model in which cognitive flexibility emerges from dynamic interactions among large-scale brain networks rather than isolated regional activations. External, between-task flexibility (task switching) relies on preparation and reconfiguration within FPN and attention networks, with switch costs reflecting interference and competition among task sets. Internal, within-task flexibility (sustaining or shifting attention, mind wandering) reflects moment-to-moment fluctuations in the balance of DMN vs task-positive systems (DAN/FPN/VAN). The antagonistic relationship between DMN and DAN supports engaged, externally oriented performance, while FPN flexibly modulates these interactions to enable shifting between broad context and step-level control. Ecologically valid paradigms demonstrate that these mechanisms generalize beyond simple stimuli-response tasks, though content (e.g., suspense vs emotional arousal) can differentially recruit DMN and its coupling with FPN. Together, this framework addresses how neural dynamics mediate maintaining a task set versus switching between sets, uniting task-switching and sustained-attention literatures under common network principles.

Flexible cognitive control depends on the coordinated, time-varying interactions among DMN, DAN, FPN, and VAN. Task switching and sustained attention/mind wandering share core network dynamics, particularly DMN–DAN antagonism modulated by FPN and VAN. Evidence from meta-analyses, multivariate fMRI, variability-based measures, QPP analyses, and naturalistic tasks indicates that both broad task context (DMN) and step-specific control (FPN) are necessary for adaptive behavior. The recurrence of similar network patterns across laboratory and real-world tasks supports the ecological validity of simple paradigms and encourages leveraging complex, naturalistic contexts. Future work should clarify when and why DMN increases during engagement (e.g., emotional content), delineate causal mechanisms linking network dynamics to behavior, and develop paradigms that bridge controlled manipulations with ecological validity.

As a narrative review, the paper integrates heterogeneous methods and paradigms, limiting causal inference. Neural mechanisms of task switching remain incompletely understood, and findings regarding DMN’s role during engagement vary by task content (e.g., suspense vs emotional arousal). Many results are correlational (functional connectivity, QPPs), and generalizability from specific tasks (e.g., serial tapping, film viewing) to other real-world behaviors requires further validation. Publication details indicate ongoing work and under-review findings, underscoring the need for replication and convergence across methods.