Prolonged exertion of self-control causes increased sleep-like frontal brain activity and changes in aggressivity and punishment

The study investigates whether prolonged exertion of self-control (executive functions) induces cognitive fatigue leading to shifts from deliberation toward more impulsive behaviors in social interactions. Prior work shows mixed evidence on the role of intuition versus deliberation in fostering cooperation and on the ego depletion effect, with replications yielding inconsistent results. The authors propose that effects attributed to ego depletion may be underpinned by local, use-dependent sleep-like slow-wave activity in frontal brain areas during wakefulness, which impairs executive control. They test whether extended practice with tasks requiring self-control (≈45 minutes) increases frontal delta/theta activity and alters choices in one-shot, incentivized economic games compared to similar tasks not requiring self-control.

Research on social decision-making highlights a tension between intuitive and deliberative processes, with mixed findings on whether intuition promotes cooperation or whether deliberation is central to prosocial choices, and with antisocial acts often characterized as impulsive. Ego depletion paradigms commonly use tasks like Stroop, emotion suppression, and high-conflict decision-making to tax self-control, reporting subsequent impairments and greater reliance on automatic responses; however, large-scale replications have questioned the robustness of ego depletion effects, while reanalyses suggest certain manipulations may be more effective. Parallel literature on local sleep demonstrates that extended task practice and sleep restriction increase delta/theta slow waves during wakefulness, particularly in frontal areas, and that such waves are linked to performance impairments. Frontal regions involved in executive functions appear especially vulnerable to fatigue. This body of work motivates a reinterpretation of ego depletion as a spectrum whose behavioral manifestations may reflect local, task-dependent neural fatigue rather than depletion of a global willpower resource.

Design: Two preregistered studies with matched experimental conditions: Frontal Fatigue (FF; tasks requiring self-control) versus No Fatigue (NF; similar tasks not requiring self-control). Study 1 included behavioral and high-density EEG; Study 2 replicated behavioral findings in a larger sample. Participants: Total N=447 healthy adults (Study 1: N=44, 22 females, mean age=30.3, SD=7.8; Study 2: N=403, 251 females, mean age=23.3, SD=3.4). Prescreening excluded medical/neurological/psychiatric conditions. Ethics: Study 1 protocol No. 1485/2017 (Area Vasta Nord Ovest); Study 2 protocol No. 03/2022 (Joint Ethical Committee for Research of Scuola Normale Superiore and Scuola Superiore Sant'Anna). Procedure (Study 1): hd-EEG cap (64 electrodes); baseline block (~20 min) including questionnaires, resting EEG (T0), Go/NoGo; fatigue manipulation: three ~15-min tasks requiring (FF) or not requiring (NF) self-control—emotion suppression, false response, and Stroop; post-manipulation block (T1) similar to baseline; one-shot, anonymous, incentivized economic games (~15 min); final resting EEG (T2); next-day online questionnaires (personality, empathy, aggressivity, impulsiveness). Economic games and measures: Dictator Game split; Ultimatum Game proposer offer; Ultimatum Game minimum acceptance threshold (MAT); Public Goods Game (PGG) contributions without punishment; PGG contributions with punishment; punishment decision in PGG with punishment; Hawk and Dove game (peaceful vs aggressive choice). EEG analyses: Low-frequency (delta 1–4 Hz; theta 4–8 Hz) power changes T1/T0 compared between FF and NF; topography, cluster-based statistics; source localization highlighting peak differences (e.g., left inferior frontal gyrus [IFG] and right middle/superior temporal gyri); assessment of T2 changes post-EG. Behavioral analyses: Nonparametric tests (Pearson’s chi-squared), probit regressions, multinomial logit for punishment modalities, diff-in-diff regressions for Go/NoGo accuracy and response times; multiple testing considerations; clustered standard errors in probit linking EEG changes to behavior.

EEG (Study 1): - Delta (1–4 Hz) power increased significantly more in FF than NF over a frontal electrode cluster (cluster-based correction; Fig. 1B). Source localization: delta power differences peaked in left IFG (MNI −36, 16, 28) and right middle/superior temporal gyri (P<0.01, cluster-based correction; Fig. 2). - Theta (4–8 Hz): No significant T0→T1 differences across groups. At T2 (after EG), FF showed stronger theta increases over temporoparietal and occipital electrodes (corrected P<0.05; peak in right temporal lobe, P<0.01). Behavior (Study 1, N=44): - Hawk and Dove game: Proportion choosing the peaceful “dove” option dropped from 86% (NF) to 41% (FF); P=0.0017 (Pearson’s chi-squared). Robust in probit (P=0.0020). - Association between frontal delta-power increase (T1–T0) and aggressive choice (hawk): probit P=0.0408 (SE clustered at individual level). - No significant differences in other games; total of 10 tests, results robust to multiple-testing corrections. - No differences in beliefs between groups; Go/NoGo diff-in-diff showed no differences in average accuracy or response times. Behavior (Study 2, N=403): - Hawk and Dove game: Significant difference between NF and FF (P<0.001, Pearson’s chi-squared), robust in probit (P<0.001). - Punishment modalities in PGG with punishment: Distribution differs by group (P<0.001). Prosocial punishment reduced (P<0.001), spiteful punishment increased (P=0.0019), total punishment unchanged (P=0.2961). Robust to probit (P<0.001; P=0.0022) and multinomial logit. - No significant differences in other games; Go/NoGo diff-in-diff showed no differences in accuracy or response times.

Findings indicate that extended exertion of self-control increases aggressive choices and shifts punishment behavior away from prosocial toward antisocial/spiteful modalities without changing overall punishment propensity. Concurrently, the FF condition elicits increased sleep-like delta activity in frontal regions central to executive control, consistent with local, use-dependent neuronal fatigue. This supports reconceptualizing ego depletion as a spectrum whose observable effects depend on manipulation duration/intensity and the nature of measured behaviors. The local sleep framework offers a physiological mechanism linking cognitive fatigue to altered social behavior. However, the study does not establish a direct causal link between sleep-like activity and decision-making, which remains for future work.

Extended self-control practice (~45 min) can induce frontal sleep-like EEG activity (delta) and increase aggressive choices in social interactions, with shifts from prosocial to spiteful punishment. These results help reconcile inconsistent ego depletion findings by emphasizing manipulation duration and suggest a physiological basis—local sleep-like episodes in executive networks—for observed behavioral changes. The work advances a perspective of sleep as locally regulated during wakefulness with societally relevant implications. Future research should test causal links between local sleep-like activity and decision-making, examine repeated interactions and compensatory mechanisms, and explore individual/population differences in vulnerability to task-dependent cognitive fatigue and local sleep.

- Control condition: While NF tasks did not require self-control, the study lacked a control condition with comparable cognitive demands but distinct functions, leaving a residual possibility of nonspecific mental fatigue; however, self-reported sleepiness, fatigue, mood, and affects showed no condition or interaction effects (only an uncorrected trend for motivation). - Behavioral scope: Clear significant effects were observed primarily in the Hawk and Dove game; lack of effects in Dictator and Ultimatum games contrasts with some prior studies, potentially reflecting task-specific sensitivities and complexity. - One-shot design: The protocol used one-shot games; interactions with repeated-play dynamics and potential compensatory strategies remain unexplored. - Go/NoGo performance: No between-condition differences were detected; absence of a practice/learning phase may have allowed learning and fatigue to offset each other. - No direct causality: The study cannot demonstrate a causal link between sleep-like activity and decision-making.