When Did It Sound, and Which Sound Was It? Timing an Action's Outcome

A finger presses a key, and a tone follows after a short delay. Knowing whether that tone will be high or low is different from knowing when it will arrive. Choosing when to start the sequence adds another distinction: you can predict an event without controlling its onset. Two experiments published in 2012 by Andrea Desantis, Gethin Hughes and Florian Waszak used these differences to examine the perceived timing of an action's auditory outcome. Their question concerned which information contributes to a shift in reported time, not whether a computer actually plays a sound earlier. [1]

The research addressed intentional binding, a phenomenon in which an action and its consequence can seem closer together in time. This particular study focused on the perceived onset of the consequence. Participants reported where a moving clock hand had been when a target tone began. Comparing that report with the recorded physical onset supplied a timing error. It was not a direct measurement of free will, a neural recording or a complete test of every process involved in predicting movement. [1]

Separating a predictable time from a predictable identity

Twenty-four people took part in the first experiment. They watched a clock hand complete a revolution in 2,560 milliseconds. In two conditions, a left or right key press initiated the sequence leading to a target tone. In a third, externally generated sounds replaced the participant's action. The clock was a reference for locating the target's onset, rather than an invitation to rate how quickly an entire session had passed. [1]

One action condition provided a consistent mapping: a particular key produced a particular high or low tone. Selecting the action therefore supplied information about the target's identity. In the other action condition, either key could produce either tone with equal probability. Participants still determined when to act, but the selected key did not reliably specify which target they would hear. In both conditions, the target began 450 milliseconds after the action's onset. [1]

In the externally initiated condition, a preceding sound signaled when the target would follow. Participants could use that cue to anticipate the target's timing, but did not decide when the cue itself would appear. The timing of these precursor events was linked to action-production records to organize the comparison. The contrast was therefore not simply between an entirely predictable sequence and random noise. It separated predicting the target's time from initiating the sequence through an action. [1]

Participants reported the target's onset after the clock hand had stopped and disappeared. Occasional catch trials instead asked whether the target had been high or low. Those trials checked attention to the distinction that the identity manipulation depended on. They did not constitute an independent measure of every action–effect association or a report about the participant's feeling of authorship. [1]

What the first comparison showed

Target tones were perceived earlier in both action conditions than in the externally cued condition. However, a stable key-to-tone mapping did not produce a statistically significant additional shift compared with the action condition in which target identity was random. Under these conditions, knowing precisely which tone the chosen action would produce did not appear necessary for the observed timing shift. A nonsignificant contrast is not proof that every possible prediction effect must be exactly zero. [1]

Catch-trial accuracy was approximately ninety-nine percent across the three conditions. That makes simple failure to listen to the targets an implausible explanation of the missing identity difference. It does not make the experiment immune to every alternative interpretation. Strong performance on a specific attention check supports a bounded inference, rather than demonstrating that all relevant mental processes were measured. [1]

The design also left a more interesting possibility. When both tones were possible, participants might anticipate both outcomes rather than anticipate neither. Failing to identify one particular forthcoming sound is not equivalent to having no expectation at all. The researchers consequently designed a second experiment in which some outcomes respected a previously learned association and others violated it. That comparison tested the accuracy of identity prediction through a different arrangement. [1]

Hearing an outcome that breaks the learned pairing

Thirty-two people entered the second experiment. One was excluded from further analysis because of very poor catch-trial performance. In acquisition phases, participants learned associations between left or right actions and target tones. In a separate condition, they learned which initial sound predicted which subsequent tone. The test phase then included both congruent outcomes, matching the learned association, and incongruent outcomes that broke it. The target followed the preceding action or sound by 400 milliseconds, not the 450-millisecond interval used in the first experiment. [1]

Short blocks kept the required action or initial sound constant. For example, an action block required only one of the two keys. This arrangement was intended to simplify identifying the precursor and anticipating the associated target. Across the test, approximately one fifth of trials were incongruent. Most outcomes therefore preserved the learned relationship, while the remaining trials allowed a comparison with violated expectations. Catch trials again checked identification of the target tone. [1]

The action-to-tone condition again produced earlier reported target times than the tone-to-tone condition. Yet congruent and incongruent targets did not differ significantly in reported onset within either condition, and the interaction between condition and congruency was also nonsignificant. The results supported the first experiment's conclusion through a manipulation of whether the heard identity matched the learned prediction. They concerned timing judgments, not the loudness, pleasantness or emotional meaning of the sounds. [1]

A proposed explanation is not another measurement

The authors discussed temporal control as a possible basis for the observed differences. Someone who chooses when to act may prepare attention differently from someone awaiting an external cue. Deciding when to begin could reduce distraction before the target appears, whereas an unpredictable precursor might itself require an orienting response. These are interpretations of the pattern, not separately demonstrated neural mechanisms in these experiments. No brain recording or independent measurement of attentional allocation established that explanation directly. [1]

The paper also considered whether associations had been learned sufficiently well, whether unexpected outcomes might acquire new associations during testing, and whether short blocks changed the predictive setting. The authors argued that their procedure supported learning and made violations noticeable. Those arguments explain their interpretation, but should not be silently converted into a direct learning measure that the timing task did not provide. [1]

The relationship between binding and a sense of agency likewise remained open. Reporting a tone earlier does not by itself specify how strongly someone feels responsible for producing it. The authors explicitly regarded a definitive answer about that relationship as premature. Their results constrain a particular account of binding based on predicting a specific motor outcome; they do not abolish prediction throughout motor control or resolve philosophical questions about agency. [1]

The study's lasting methodological point is concrete: initiating a sequence, anticipating its timing and knowing its precise sensory outcome need not be treated as one variable. In these two experiments, more specific outcome information and agreement with a learned identity prediction did not significantly strengthen the timing shift. Understanding what happened between a key and a tone required separating the information available to the participant, not simply observing that an action had a consequence. [1]

Sources

[1] Desantis, A., Hughes, G., and Waszak, F. (2012). Intentional Binding Is Driven by the Mere Presence of an Action and Not by Motor Prediction. PLOS ONE, 7(1), e29557. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0029557