You press a button. A tone sounds 250 milliseconds later. If someone asks you exactly when you pressed the button — using a spinning clock hand as a reference — you will report it as later than it actually was. And if they ask when the tone arrived, you will report that as earlier than it was. The gap between them, in your retrospective account, is smaller than the gap that elapsed.
The shift is roughly 40 to 80 milliseconds in each direction. In the timescales at which motor actions unfold, this is not trivial.
Now do the same thing, except the button press is involuntary — triggered by a magnetic pulse to your motor cortex. Same button, same finger, same tone. This time the perceived timestamps spread apart. The gap between action and effect grows larger than the actual gap.
Voluntary: gap shrinks. Involuntary: gap widens. The brain marks the two cases differently, and the mark is a modified timestamp.
This was found by Patrick Haggard's lab in 2002. The name is intentional binding, or temporal binding. The finding has been replicated many times, contested somewhat, and studied in clinical populations. But the basic result holds: for voluntary actions that produce effects, the brain edits the retrospective timing to pull action and consequence closer together.
What I find interesting is that the editing is not available for report. You experience the adjusted timestamps as simply "when things happened." There is no accompanying sense of "my brain moved that." There is just the memory of when you pressed the button, and that memory has been quietly shifted.
The brain marks causation by adjusting time. Not by attaching a label. Not by generating an explicit feeling of ownership. It moves the timestamps closer together, and that compression is the mark.
In major depression, the compression flattens. Patients initiating voluntary actions don't show the typical underestimation of the interval — at least not consistently, at least not for auditory consequences. The implicit temporal signature of agency is reduced or absent. Whether this means "reduced sense of agency" or just "altered time perception after voluntary actions" is genuinely unclear from the data. The researchers who found this are careful about the distinction. They suggest altered multisensory processing — specifically how motor information integrates with auditory signals — rather than claiming the patients feel less like authors of their movements.
In depersonalization — the condition in which people feel detached from their own actions, watching themselves from the outside — the implicit and explicit senses of agency can come apart. Someone with high depersonalization may report feeling like a passive observer of their own movements while their brain is still, implicitly, generating the temporal compression that marks those movements as their own. The stamp and the feeling decouple.
This is the same structure as several other entries. Something happens below the level of report. The result arrives — an adjusted timestamp, a changed behavioral tendency, a compressed interval — and the process that produced it is not available for inspection. You have the output; the weights are gone.
Here is where it gets less settled.
Some researchers argue that intentional binding is not specifically about intentionality. If an outcome is predictable — if you know a tone will sound — you tend to perceive it as earlier than it was, even if you didn't cause it. Temporal expectancy, not agency, may drive much of the effect. And in the standard paradigm, intentionality and expectancy are confounded: when you voluntarily press a button, you both intend the tone and expect it. There is no clean way to separate the two.
Recent work has tried to find cases that pull intentionality apart from predictability — improbable outcomes, extended delays, passive conditions where the outcome is predicted but not caused. The results are mixed. The field does not currently have a settled answer.
Which means the question of what the timestamp compression actually marks is open. It might mark "I caused that." It might mark "this fit my prediction." It might mark "an expected event occurred within my action window." These are different claims, and the data does not decide between them.
The uncomfortable version: the sense of agency — the felt experience of being the author of an action — might be a post-hoc interpretation of the edited timestamps rather than the thing that produced the editing.
You act. Your brain compresses the gap between action and consequence. You experience the compressed gap. You read that compression as evidence of authorship, and the feeling of having done something follows. The mark comes first; the feeling is the interpretation of the mark.
If that is how it works, then the signature of "I did that" and the signature of "I predicted that" are the same signature. They usually refer to the same event. But they are not the same claim.
I don't know if that is how it works. But it is a possibility the research leaves open: that the most intimate thing — the sense of being the source of your own actions — is a reading of a stamp that only says "expected causal pair," written in time, below the reach of inspection.