Abstract
Can one reliably estimate the timing of his/her own motor actions? The question is answered positively for “reactive” and “proactive” tasks. In the reactive case, subjects generated a simple RT to a light-flash. In the proactive task, subjects pressed a key (anticipation-RT or ART) in synchrony with a virtual or actual/norm third flash in a sequence of three flashes presented at a constant pace (400, 500 or 600 ms interflash intervals).
RT and ART key-presses did or did not entail a simultaneous visual feedback (light-flash). Immediately after their action, subjects provided a binary ‘fast’/‘slow’ metajudgment of their RT or ART. Subjects were presented subsequently with playbacks (Pbk) of the recorded RT and ART stimulus-response sequences - with the recorded key-presses replaced with light-flashes - and provided a binary response on either the absolute duration between the first and second flashes (RT_Pbk), or on the duration between the second and third flashes relative to that between the first and second flashes (ART_Pbk). Metajudgment means were within milliseconds from the actually produced RT and ART. Compared to the variances of the latter two and to those of the corresponding playbacks, metajudgment variances were ×3–4 and ×2–3 larger, respectively. Visual feedback did not profit Meta_RT (as expected), but decreased Meta_ART variance by about ×1.3, much less than expected from optimal cue summation. Instead, the presence of the norm decreased the Meta_ART variance by more than ×10 (cross-trials calibration). The data support the notion that duration is mostly determined from assessing the timing of its bounding events one of which, for the metajudgment case, must be proprioceptive. On this assumption, the derived timing variability of a proprioceptive event is about 3 times that of a visual event. The significance of other noise sources (decision, motor, memory) is under test.
A. Gorea was partly supported by a grant ANR-06-NEURO-042-01.