Intentional binding

Intentional binding is an interesting effect. The perceived interval between an voluntary action and its effect is shorter if there is a sense of agency involved. Berberian, Sarrazin, Blaye and Haggard (see citation) have investigated this effect.


They used an aircraft mockup with stages: conflict detection, command decision (how much to change heading), command implementation (set the heading), command engagement (go button – action), command result (confirmation of success/failure – effect), followed by interval estimation (between action and effect)and rating of feeling of agency. The decision, implementation and engagement can be automated or not, giving four situations: from all three under control to all three automated. The interval between action and effect was fixed at three different times. What was measured was how the interval estimation was affected by the level of automation, the length of the interval and the feeling of agency.

First, we observed a strong relation between measures of IB (intentional binding) and different levels of system automation. Second, our data revealed a gradual increase in temporal estimation with the increasing level of automation. The more the system was automated, the longer the action-effect interval was perceived. Third, the effect of automation level on intentional binding was dependent on the actual action-effect delays…Our findings confirm the interest of the intentional binding as implicit measure of agency. First, we provide evidence that quantitative changes in binding are strongly associated with progressive changes in actual level of control, and also with quantitative changes in explicit ratings of agency. Second, we replicate the basic binding effect in a situation with high face-validity, in which action-event sequences paralleled those that participants might meet in their everyday lives….Our results reveal a gradual increase of interval estimates with the increasing level of automation. In contrast, many previous studies of agency have relied on binary agency-attribution judgments…. our result suggests action selection, planning and intention realisation may all contribute to sense of agency. In this way, our results are consistent with recent arguments that sense of agency is based on cue integration….We propose that temporal contiguity is task dependent and that intentional binding occurs in a specific ‘‘window of opportunity’’ which may vary across tasks, and may also depend on the range of action-effect delays experienced in a given setting. Operant learning is similarly sensitive to the natural time delays of the system linking actions to effects, even for systems as familiar as one’s own body.


Measures of intentional binding are a useful addition to reports of a feeling of agency for future experimentation.


But what does this study say about the consciousness of agency? Agency is not an all or nothing feeling – it can be graduated. One way to look at this would be that knowledge of agency is important to learning. We have to be able to accurately assess cause and effect in order to learn to get the outcomes we want. Therefore our understanding and memory must connect those things that are causally connected and avoid connecting those that are not. One way to do this is to store the memories of the cause and the effect (temporarily at least) in the same ‘time slot’. This in itself may shorten the interval between action and effect in memory when we recall it to estimate the interval. If there is a need to bias memory with probability of agency, that is a very good reason for the need for a conscious feeling of agency when this is appropriate. Some unconscious process must calculate how appropriate it is, because this calculation is not part of the conscious feeling of agency.

Berberian B, Sarrazin J-C, Le Blaye P, & Haggard P (2012). Automation Technology and Sense of Control: A window on Human Agency PLoS ONE DOI: 10.1371/journal.pone.0034075

I'm on ScienceSeeker-Microscope

Leave a Reply

Your email address will not be published. Required fields are marked *