|
[back]
Part
1/ Rapid Visual Categorization in the Absence of Awareness
Rufin VanRullen
Humans
can categorize natural scenes on the basis of the presence of a target
object (i.e. animal) so rapidly (150 ms) that such processing has been
proposed to rely on the feed-forward propagation of information collected
during the first milliseconds of visual stimulation. According to this
view, early motor responses should be mostly unaffected by masking the
visual stimulus after a few tens of milliseconds. We asked our subjects
to respond to masked (SOA 26.6 ms) and unmasked natural scenes when
they contained an animal. In addition, subjects rated their confidence
in perceiving the contents of each masked image. For a majority of the
scenes, masking effectively prevented awareness of the stimuli, as indicated
by the fact that confidence ratings did not predict categorization accuracy
(Kunimoto et al, 2001). For the same scenes however subjects responded
significantly above chance level to the presence of animals. In addition,
motor responses started to reflect correct categorizations at the same
time for masked and unmasked stimuli, indicating that early responses
in "normal" (unmasked) visual categorization probably also rely on the
first milliseconds of stimulation. Similar results were obtained with
simpler displays for which we could control stimulus and mask contrast.
In that case the earliest motor responses to "perceived" and "unperceived"
targets showed virtually identical distributions, strongly supporting
the feed-forward model: information about the first milliseconds of
visual stimulation can propagate throughout the visual system, unaffected
by later changes, and determine behavior even when it is not (or not
yet) available to consciousness.
Figure
1. Natural scene categorization (animal vs. non-animal) with and
without a mask (3 subjects). A. Proportion of trials associated with
each possible confidence rating. Zero confidence (more than 50% of the
trials) corresponds to a situation where masking the scenes prevents
visual awareness, as indicated by the fact that further confidence judgments
for these images will fail to predict categorization accuracy. For all
images including those associated with zero confidence ratings, subjects
can perform the categorization task above chance (p<.0005). B. Distribution
of reaction times to masked and unmasked scenes (20 ms time bins). Reaction
times to masked targets differ from reaction times to masked distractors
at the same time (Ôdiscrimination onsetÕ) as for unmasked stimuli. This
suggests that in the general case of unmasked natural scenes, the visual
system must also make use of the information extracted during the very
first milliseconds of visual stimulation.
Figure
2. By using letters as stimuli we can make sure that the contrast
of the mask is comparable to the contrast of the stimulus. Under these
conditions a purely feed-forward model predicts that fine temporal differences
between stimulus and mask onset should be preserved up to the highest
levels of the processing hierarchy, and observed in motor responses
as well. A. Subjects are required to respond as fast as possible when
the letter P is presented, and whithold responding when the letters
R or B are displayed. In congruent trials, letters are flashed for 52ms,
while in incongruent and control trials 2 distinct letters are flashed
successively for 26 ms (the target followed by a distractor for incongruent
trials, a distractor followed by the target for control trials). Under
these conditions only the distractor letter is consciously perceived,
due to backward- and forward-masking effects occuring respectively in
incongruent and control trials. B. As predicted by the feed-forward
model, responses to incongruent (masked) target trials follow the distribution
of responses to congruent (unmasked) targets for a certain period of
time after the discrmination onset. During this period, behavioral responses
are only determined by the first 26 ms of stimulation. After this period
the masking letter begins to affect responses, but it is only after
more than 400 ms that reaction times will reflect the subjectÕs perception
of the stimulation.
Part 2/ Processing capacity for natural scenes and objects in the
human visual system.
When a visual scene containing many discrete objects is presented to
our retinae, only a subset of these objects will be explicitly represented
in visual awareness. The number of objects accessing short-term visual
memory might be even smaller. Finally, it is not known to what extent
"ignored" objects (those that do not enter visual awareness) will be
processed Ðor recognized. By combining free recall, forced-choice recognition
and visual priming paradigms for the same natural visual scenes and
subjects, we were able to estimate these numbers, and provide insights
as to the fate of objects that are not explicitly recognized in a single
fixation. When presented for 250 ms with a scene containing 10 distinct
objects, human observers can remember up to 4 objects with full confidence.
If forced to guess a further number of objects, they can reliably report
between 2 and 3 more objects above chance level. These numbers depend
on various factors such as target objects size, eccentricity or familiarity.
Finally, even the objects that the subjects consistently failed to report
elicited a significant negative priming effect when presented in a subsequent
task, suggesting that their identity was represented in high-level cortical
areas of the visual system, before the corresponding neural activity
was suppressed during attentional selection.
top
|
