Amodal completion and vernier acuity: evidence of 'top-but-not-very-far-down' processes?

Vernier thresholds were measured by means of a forced-choice paradigm for physically separated bars that were perceived either as separate or as amodally completed behind an occluder. The finding that thresholds were not reduced (improved) with amodal completion argues against an early neural code for amodal completion that is equivalent to physical completion of the vernier bars (ie neural propagation/interpolation) and limits the stage at which amodal representations could arise via bottom-up processes to neural sites located centrally with respect to those involved in position coding. An alternative explanation is also considered, that this result reflects the activity of top-down processes that mediate amodal completion by imposing grouping/organizational constraints upon earlier representations of position. In support of this proposition it was found that when tested by means of a non-forced-choice paradigm without feedback, observers demonstrated a reduced tendency to report the presence of a vernier offset when the bars were perceived as amodally completed (ie a shift in response bias occurred). This separability of vernier acuity and response bias suggests the existence of 'top-but-not-very-far-down' grouping processes whose influence does not extend to early representations of relative position.