Attentional release in the saccadic gap effect

Adam Reeves,

Northeastern University

Can a "release of attention" from fixation help explain the saccadic 'gap effect', the
shortening of saccadic latency (SL) when the fixation target is extinguished just before
saccade target onset, as postulated by Fischer & Weber (1993)?  A role for attention is
suggested by Pratt, Lajonchere & Abrams (2006)'s finding of even shorter SLs when an
attended, rather than unattended, line segment in a fixation cross was removed in the gap
condition. Our well-practiced observers generated saccades and button presses to one of
four 10^o eccentric targets. They ran in "overlap" (fixation spot stays on), "gap0" (fixation
offsets at target onset), and "gap200" conditions; in gap200, the fixation spot was removed,
dimmed, expanded, or brightened, always 200ms before target onset.  Our SL data excluded
speed-accuracy tradeoffs, express saccades, stimulus salience, and oculomotor readiness,
as explanations of the gap effect. Fixation offset and general warning had minor influences
on SLs, leaving attention release as a default.  Supporting this idea, we found that in saccade
trials, finger-press reactions to foveal probe dots presented after the fixation spot was
brightened (to hold attention) were faster than those made after the spot was removed (to
release attention); in these two conditions, warning effects were equated.  Varying the time
from gap onset to the probe dot mapped out the time course of the putative attentional
release, which takes ~140 ms.  All very good; but we are left with the puzzle of why attention
should be held by an incredibly boring stimulus - an unmoving,  unchanging fixation spot on
a black background- when the task (maintaining fixation in preparation for making a sacade)
is so simple and has already been repeated innumerable times.