Temporal Binding in the Libet Clock Paradigm: A Psychometric and Mechanistic Point of View

The sense of agency allows us to feel and recognize ourselves as the ones planning and executing our intentional actions. The phenomenon of intentional or temporal binding (the subjective shortening of the temporal interval between an action and its outcome) has been proposed to be an objective measure of the sense of agency. The two most used paradigms to investigate temporal binding are the Libet clock method and interval estimation. There is an ongoing debate about the underlying mechanisms of temporal binding and there is little information on the psychometric properties of these paradigms. The four studies presented here aim to contribute to filling this research gap. In Study 1 we employed a correlational within-participant design to compare temporal binding scores from the Libet clock method and interval estimation. Both paradigms were completed by 46 healthy sports science students. Sample size was determined adaptively using Bayes factors. Correlational analyses indicated no positive relationship between action or outcome binding in the Libet clock method and binding in the interval estimation paradigm. These results are backed by good to excellent reliability of binding measures. Study 2 repeats the design of Study 1 but aims to equalize experimental parameters between the two temporal binding paradigms. These include variance in the action-outcome delay and the presence of passive conditions. We preregistered Study 2. The adjusted paradigms were completed by 36 healthy sports science students with the sample size being determined via Bayes factors again. Despite these amendments, Study 2 confirms the results of Study 1. Study 1 and Study 2 suggest that temporal binding in the Libet clock method and in interval estimation do not share the same underlying processes. Concentrating on the Libet clock method, Study 3 and Study 4 aim to test a novel theory that clock hand positions during target events are stored in spatial working memory to be reported later. In the contingent condition where both events occur and highlight neighboring clock hand positions, these positions might interfere pro- and retroactively with each other during the retention interval causing the typical convergent shifts in reported clock hand position. To test this theory, Study 3 uses a preregistered correlational within-participant design. 104 healthy sport science students completed the Libet clock method and one visual and one spatial 2-back task. Visual working memory performance was tested to control for more general attentional compared to specific spatial working memory processes. Additionally, they completed a novel task similar to the Libet clock trying to separate the confound between the temporal and spatial dimensions of the Libet clock’s continuously moving clock hand. We found action and outcome binding in the Libet clock task and outcome binding in the novel binding task. These were, however, not correlated. Outcome binding in the Libet clock showed a negative correlation to spatial working memory performance but evidence against a correlation with visual working memory performance was inconclusive according to Bayes factors. For outcome binding in the novel binding task the opposite was found. It correlated negatively with visual working memory performance but evidence regarding spatial working memory performance was inconclusive. Study 4 complimented Study 3 with an experimental approach. We embedded each trial of the Libet clock task in the retention interval of a spatial (Corsi block task) or visual (sequences of Chinese characters) memory task to put load on the specific memory modality. We expected to find increased binding with higher spatial memory load but not with higher visual memory load. 94 healthy sport science students completed the Libet clock task twice, embedded in the respective memory task. We used Bayesian mixed models to analyze our data. We preregistered Study 4. The results suggest that action binding does not increase with spatial memory load, but the Bayes factor indicates inconclusive evidence. Bayes factors show substantial evidence against an increase in action binding with higher visual memory load. Likewise, outcome binding does not seem to increase with spatial memory load. Again, Bayes factors remain inconclusive for this result but show conclusive evidence against an increase of outcome binding with higher visual memory load. Study 1 and 2 show that temporal binding in the Libet clock method seems to stem from different processes than temporal binding in the interval estimation paradigm. The results of Study 3 and 4 do not confirm the theory that views the binding effect in the Libet clock method as a result of spatial working memory processes. We discuss implications of the results of the four studies presented here for theoretical considerations, practical application and future research.