Abstract
It has been repeatedly asserted that motor responses can be triggered without ‘conscious’ perception. Scrutinizing such studies always reveals methodological and/or logical drawbacks. We re-examine the issue with a standard 2AFC detection task combined with a speeded simple or choice Response Time task (sRT & cRT). Observers were presented with two high contrast (‘mask’) annuli (S2) displayed at ±6° and ±2° along the horizontal and vertical meridians and with one Gaussian target blob (S1) set at its detection threshold (81% correct) and preceding the annuli by 50 ms. S1 was displayed either within (masked condition, M) or 4° below or above one of the two annuli (non-masked condition, nM). Observers had first to react to the S1+S2 complex by pressing one (sRT), or one of two, left/right button(s) contingent on S1 location (cRT) and then to specify/confirm this location (perceptual 2AFC). cRT were 26 ms longer than sRT and yielded close to chance performance. Both sRT and cRT were shorter (1) for correct than for incorrect perceptual responses (22 and 14 ms for nM & M conditions; with sRT[incorrect responses] = sRT in the absence of S1) and (2) for M than for nM conditions (3 and 12 ms for correct & incorrect perceptual responses). Observation (1) supports the notion that motor responses do not ‘profit’ from non-detected targets, disproving the perceptual-motor dissociation. Together with observation (2), it sustains a sensorimotor race model positing that the two processes operate on the same internal response but are based on distinct decision variables, one fixed (motor threshold) and the other context-dependent (perceptual criterion) and well above the former in the M relatively to the nM case. The chance performance obtained with cRT requires that perceptual decisions be taken once the internal response reaches its asymptotic level, well after it exceeds the motor threshold.