Abstract
Removing the fixation spot some 200 ms before the onset of a target speeds saccades to the target (the ‘gap effect’ of Fisher, ). Last year (Reeves & Jin, VSS 2006) we tested and rejected six (oculo-motor and other) accounts of the gap effect, and were left with a hypothetical default explanation based on disengaging attention from fixation. This year we tested the role of attention directly, by presenting a dim probe dot near fixation on a random 20% of trials. The fixation spot was manipulated 200 ms before target onset either by removing it altogether (‘blank gap’ condition) or by leaving it on but changing its color (‘no-gap’ condition). (Our no-gap condition is superior to the often-used ‘overlap’ condition, in which the fixation spot remains unchanged, in that it provides the same temporal cue as the gap condition.) The probe occurred 80 ms before, or 20, 80, 140, or 240 ms after the fixation spot was manipulated. Probes were always presented before the saccade occurred. Compared to the blank-gap condition, in which attention is hypothetically released from fixation, holding attention at fixation in the no-gap condition both slowed saccadic latencies to the target (the gap effect) and sped up reaction times to detect the probe. There were no speed-accuracy trade-offs. Thus, the latency data support the hypothesis that attention remains attached to the fixation area in no-gap, thus speeding processing of the probe, but is disengaged in the gap condition, thus delaying processing of the probe. Moreover, differences between the saccadic latencies in two gap conditions decreased with probe-dot onset up to 140 ms, while key-press reaction time differences increased, as if attentional disengagement requires 0.1 to 0.2s to complete.