Biomedical Engineering Reference
In-Depth Information
location (for example, a scene in which a jet airplane is present and one of its reac-
tors has been erased from one of the two photographs to be compared). Although the
alteration is obvious when one directly attends to it, it takes naive observers several
tens of seconds to locate it. Not unexpectedly, instances of this experiment which are
the most difficult for observers involve a change at a location that is of little interest
in terms of understanding and interpreting the scene (for example, the aforemen-
tioned scene with an airplane also contains many people, who tend to be inspected
in priority).
These experiments demonstrate the crucial role of attention in conscious vision:
unless we attend to an object, we are unlikely to consciously perceive it in any detail
and detect when it is altered. However, as we will see below, this does not necessarily
mean that there is no vision other than through the attention bottleneck.
19.4.2
Attentional modulation of early vision
A number of psychophysical end electrophysiological studies indicate that we are
not entirely blind outside the focus of attention. At the early stages of processing,
responses are still observed even if the animal is attending away from the receptive
field at the site of recording [54], or is anesthetized [21]. Behaviorally, we can also
perform fairly specific spatial judgments on objects not being attended to [4, 14],
though those judgments are less accurate than in the presence of attention [31, 61].
This is in particular demonstrated by dual-task psychophysical experiments in which
observers are able to simultaneously discriminate two visual stimuli presented at two
distant locations in the visual field [31].
While attention thus appears not to be mandatory for early vision, it has recently
become clear that it can vigorously modulate, top-down, early visual processing,
both in a spatially-defined and in a non-spatial but feature-specific manner [10, 34,
55]. This modulatory effect of attention has been described as enhanced gain [54],
biased [33, 36] or intensified [31] competition, enhanced spatial resolution [61], or
as modulated background activity [12], effective stimulus strength [43] or noise [17].
Of particular interest in a computational perspective, a recent study by Lee et
al. [31] measured psychophysical thresholds for five simple pattern discrimination
tasks (contrast, orientation and spatial frequency discriminations, and two spatial
masking tasks; 32 thresholds in total). They employed a dual-task paradigm to mea-
sure thresholds either when attention was fully available to the task of interest, or
when it was poorly available because engaged elsewhere by a concurrent attention-
demanding task. The mixed pattern of attentional modulation observed in the thresh-
olds (up to 3-fold improvement in orientation discrimination with attention, but only
20% improvement in contrast discrimination) was quantitatively accounted for by a
computational model. It predicted that attention strengthens a winner-take-all com-
petition among neurons tuned to different orientations and spatial frequencies within
one cortical hypercolumn [31], a proposition which has recently received additional
experimental support.
These results indicate that attention does not implement a feed-forward, bottom-
up information processing bottleneck. Rather, attention also enhances, through feed-
Search WWH ::
Custom Search