Testing the ego-depletion effect in optimized conditions.

The observation that exerting self-control in an initial task impairs subsequent self-control performance in a following task has been used to explain a wide range of phenomena. If evidence for this "ego-depletion" effect was initially believed to be strong, it is now questioned. Recent meta-analyses indicated that this effect was sensitive to publication bias and that it was greatly reduced after control for this bias. In a pre-registered replication attempt where an ego-depletion protocol was conducted in multiple independent laboratories, the effect was not distinguishable from zero. Here, a different approach is adopted to examine the validity of this effect by improving the experimental protocol with the addition of important methodological precautions: 1) a pre-test measurement, 2) a learning period, 3) a prolonged depleting task, 4) a similar control condition, and 5) valid indexes of self-control. Accordingly, a well-learned Simon task was done before and after 1h of continuous practice of a Stroop task in a high inhibition demands condition (75% of incongruent trials) or in a control condition (0% of incongruent trials). Datasets from between-subjects (Study 1, N = 82) and within-subjects (Study 2, N = 52) experiments were analyzed using generalized linear mixed models. A significant ego-depletion effect was found in Study 1 (greater interference effect and accuracy decline in high inhibition demands than in control condition) but not in Study 2. Because it is difficult to explain this difference in results, the findings suggest that, even in a context chosen to optimize the observation of an ego-depletion effect, it seems difficult to be conclusive about the existence of this effect.