Attentional modulation of sensorimotor processes
in the absence of perceptual awareness
Petroc Sumner*†, Pei-Chun Tsai‡, Kenny Yu‡, and Parashkev Nachev‡
*School of Psychology, Cardiff University, Tower Building, Park Place, Cardiff CF10 3AT, United Kingdom; and ‡Division of Neuroscience,
Faculty of Medicine, Imperial College London, St. Dunstan’s Road, London W6 8RP, United Kingdom
Edited by Dale Purves, Duke University Medical Center, Durham, NC, and approved May 19, 2006 (received for review March 10, 2006)
Attention modulates visual perception and is generally considered
inextricably linked with conscious awareness: we become aware of
stimuli as we attend to them, and we attend to stimuli as we
become aware of them. Recent evidence suggests that attention
can also modulate the effects of stimuli that remain invisible, and
a natural explanation is that attention enhances weak perceptual
representations, bringing them closer to conscious threshold even
if they do not reach that threshold. However, there is also the
possibility that attention may modulate neural processes that are
entirely separate from those supporting conscious perception:
sensorimotor mechanisms that do not create awareness however
much they are enhanced. Here we provide evidence in support of
this second hypothesis by showing that attentional cueing can
modulate the behavioral response to invisible stimuli in a way that
is distinct from enhancing their visibility. We used a masked-prime
paradigm that produces a negative or positive compatibility effect
depending on the perceptual strength (duration or brightness) of
the prime. We found that attention enhanced the effect of both
visible and invisible primes and also increased the likelihood of
detecting the prime (i.e., boosted perceptual strength). Crucially,
the pattern of attentional influence on priming could not be
explained by attentional modulation of the prime’s perceptual
strength but was predicted by a direct attentional influence on the
nonconscious priming process itself. Therefore, in addition to
regulating what we perceive, attention seems to influence our
behavior through sensorimotor processes that are not involved in
conscious awareness.
priming 兩 unconscious 兩 motor inhibition 兩 action
T
hroughout the study of cognition it has been widely assumed
that attention and conscious awareness are closely linked
(1–7). A wealth of behavioral, neurophysiological, and imaging
evidence has shown that attention facilitates the processing of
consciously perceived stimuli (8–10) and can also make us aware
of stimuli that might otherwise escape awareness (11, 12).
Conversely, stimuli that are not attended are usually not perceived, and this is reflected in the names given to neurological
disorders or experimental paradigms that diminish conscious
awareness without directly interfering with sensory processing:
e.g., ‘‘attentional blink’’ (13), ‘‘inattentional blindness’’ (14), and
‘‘inattention’’ in visual extinction or neglect (15, 16). Moreover,
neural processes that appear to proceed automatically and
unconsciously are sometimes referred to as ‘‘preattentive,’’
meaning outside the influence of attention (17). This seemingly
close union between attention and consciousness, both physiologically and conceptually, makes it natural to assume that they
are interdependent. Indeed, it is hard to resist the teleological
notion that the principal function of attention is to be ‘‘the sentry
at the gateway of consciousness’’ (18).
However, recent evidence suggests that attention may enhance
neural processes irrespective of whether they reach consciousness. Initial suggestive evidence came from a ‘‘blindsight’’ patient, GY, who, despite cortical damage that renders him partially blind, can successfully guess the orientation of stimuli he
denies seeing. When the location of the unseen stimulus was
10520 –10525 兩 PNAS 兩 July 5, 2006 兩 vol. 103 兩 no. 27
indicated by a preceding attentional cue, GY performed such
‘‘blind’’ orientation discrimination faster (19, 20). In people
without brain damage, nonconscious processing can be assessed
by using brief stimuli that are masked and thus rendered
impossible to identify. It is known that subthreshold processing
of such ‘‘masked primes’’ can nevertheless affect subsequent
responses to perceived stimuli, and some studies have found that
focusing attention in time or space can modulate these priming
effects (21–23). A natural explanation for such attentional
modulation of invisible stimuli might be that attention boosts the
perceptual representation of weak stimuli toward conscious
threshold, enhancing their ‘‘perceptual strength,’’ and thus their
priming abilities, even if they still do not quite reach awareness.
In this explanation, attention still acts on the same neural
processes that lead to perceptual awareness, in line with existing
models and concepts of attention and consciousness (24–26).
However, there is a more intriguing possibility. In addition to
any effect on perceptual threshold, attention may also act on
sensorimotor processes that are independent of representations
that support conscious awareness. Thus, attention may act at two
conceptually distinct places in the neural processing that follows
presentation of a weak stimulus: it may boost perceptual representations, making the stimulus more likely to be consciously
perceived, and it may in parallel boost any sensorimotor processes activated by the stimulus. Such sensorimotor mechanisms
may be wholly unconscious, and boosting them may have no
effect on how visible the stimulus is.
Unfortunately, measuring attentional modulation in conventional priming tasks cannot distinguish between these two ways
in which attention might affect subliminal processing because
any attentional boost to priming will appear to act in the same
way as increasing the prime’s brightness or duration: both will
enhance the priming effect. Thus, it might be natural to assume
that attention modulates priming precisely because it enhances
the prime’s perceptual strength in a similar way to modulating
brightness or duration. To tell whether attention might actually
modulate subliminal processing in another way, we have to give
attention a chance to modulate the priming effect in a way that
is dissociable from the effect of boosting the prime’s perceptual
strength toward conscious threshold.
To address this difficulty we exploited an interesting priming
task in which the priming effect changes from negative to
positive as the prime’s perceptual strength is increased (27, 28).
In this task, participants simply make left or right responses to
arrows pointing leftward or rightward. Each target arrow is
preceded by a brief prime arrow, which is backward-masked by
an array of randomly arranged lines. Under these conditions,
primes with a duration sufficient to make them consciously
visible produce a positive ‘‘compatibility effect,’’ such that
Conflict of interest statement: No conflicts declared.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: RT, response time; cdm⫺2, candelas per square meter.
†To
whom correspondence should be addressed. E-mail: [email protected].
© 2006 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601974103
responses are faster if the prime and target point in the same
direction (compatible) than if they point in opposite directions
(incompatible). However, briefer primes that are rendered invisible (nondiscriminable) by the mask produce a negative
compatibility effect, such that compatible primes cause slower
responses than incompatible primes.
To study unconscious processing, the choice of mask is important because it must successfully render the prime invisible,
but some masks have been criticized for interacting with the
prime to produce confounding priming effects (29, 30). These
problematic masks were composed of overlapping arrows similar
to the primes and targets themselves. Instead, we used masks
composed of overlapping lines, which have recently been shown
to produce robust negative compatibility effects without the
problematic confound produced by arrow masks (31).
The negative compatibility effect produced by the invisible
primes has been modeled as an automatic inhibitory mechanism
that suppresses motor initiation elicited by subthreshold stimuli
(27, 28). It parallels other automatic negative effects known in
cognitive science, such as ‘‘inhibition of return,’’ in which
attention is biased against returning to recently attended locations. In the masked-prime case, however, the inhibition seems
to occur within the motor domain rather than on perceptual
representations, because it is ‘‘effector-specific’’; primes associated with left or right foot movements do not cause inhibition of
manual responses (32). The priming effect turns positive for
stronger, visible primes either because stronger positive priming
simply overcomes the inhibition or because the inhibitory mechanism is overruled by the conscious representation of the prime’s
occurrence (27, 28).
For our purposes it does not matter that invisible primes
happen to produce a negative effect and visible primes produce
a positive effect, rather than vice versa, or, indeed, that the
crossover from negative to positive seems to coincide with the
threshold for consciously perceiving the prime (27). What is
crucial is that invisible stimuli can produce a robust measurable
effect that is distinct from, and not just a weaker version of, the
effect of stronger stimuli. In other words, because enhancing the
prime’s perceptual strength (by presenting it for longer) does not
simply enhance the priming effect, an attentional enhancement
of the priming effect can be distinguished from an attentional
enhancement of perceptual strength. If attention makes a weaker
Sumner et al.
prime more like a stronger one, the priming effect should
become more like that produced by a stronger prime: less
negative and more positive. If attention independently enhances
the sensorimotor priming process elicited by the weak prime, the
negative effect of the subliminal prime should be enhanced
(becoming more negative, not more positive). Thus, the slightly
complex pattern of priming (negative then positive) known to
occur as the invisible prime’s perceptual strength is increased
allows us to make opposite predictions for attentional modulation of the prime’s perceptual strength and for attentional
modulation of the subliminal priming effect itself.
Results
Exp. 1. We manipulated attention to the primes with a ‘‘precue-
ing’’ procedure (33), which required the primes to be presented
randomly in one of two locations (see Fig. 1). The targets also
occurred randomly in one of these locations, and a control
experiment, without attentional cues, showed that the negative
compatibility effect was the same whether the primes and targets
were in the same location or different locations: For primes and
targets in the same location, mean response time (RT) in
compatible and incompatible trials was 374 ms and 355 ms,
respectively, giving a negative compatibility effect of 19 ms,
whereas for primes and targets in different locations, mean RT
in compatible and incompatible trials was 377 ms and 357 ms,
respectively, giving a negative compatibility effect of 20 ms (for
more details see Supporting Methods, which is published as
supporting information on the PNAS web site). This finding was
expected because the priming effect is thought to be elicited by
subthreshold response initiation after the prime, not from a
perceptual interaction between the prime and target stimuli
(28, 32).
We then introduced attentional cues by brightening a rectangle around one of the possible prime and target locations (see
Fig. 1). These ‘‘exogenous’’ or ‘‘stimulus-driven’’ attentional cues
were very similar to those used in many previous studies (33, 34),
but they were much nearer fixation than is normally the case,
because we wished to keep the masked primes and targets near
fixation. A second control experiment confirmed that attention
was successfully biased to the cued location: in the absence of
masks and targets, responses to primes in the cued location were
speeded relative to the uncued location (341 vs. 355 ms; t ⫽ 6.7,
PNAS 兩 July 5, 2006 兩 vol. 103 兩 no. 27 兩 10521
PSYCHOLOGY
Fig. 1. Schematic illustration of a single trial. In Exp. 1 the duration of the prime was 20 ms (invisible) or 70 ms (visible). In Exp. 2 the duration of the prime was
varied from 10 ms to 70 ms across the threshold of conscious perception. In Exp. 3 prime’s duration was 40 ms, and its brightness was varied instead.
Fig. 2. Attentional effect on visible and invisible primes in Exp. 1. The
compatibility effect (mean RT in incompatible trials ⫺ mean RT in compatible
trials) was modulated by cueing in opposite directions for visible and invisible
primes. Thus, cueing the invisible stimuli did not make their effect more similar
to that of visible primes. Error bars depict the standard error of the cueing
effect. Mean RT for targets after invisible compatible and incompatible
primes, respectively, were 427 ms and 402 ms for cued primes, and 421 ms and
407 ms for uncued primes. For visible compatible and incompatible primes,
mean RT was 415 ms and 437 ms for cued primes and 427 ms and 439 ms for
uncued primes. The overall error rate was low (4.6%), and, like RT, the errors
displayed negative compatibility effects for invisible primes (⫺2.1% and
⫺1.9% for cued and uncued primes) and positive effects for visible primes
(2.2% and 3.5% for cued and uncued primes). This main effect of prime
visibility was significant (F(1,11) ⫽ 12.7, P ⬍ 0.01), but there was no effect of
cueing (F(1,11) ⫽ 1.0), nor was there an interaction between visibility and
cueing (F(1,11) ⫽ 0.8).
P ⬍ 0.001; see Supporting Methods for more details). It is worth
noting that this attentional effect, although statistically significant, is smaller than that normally seen in cueing experiments,
presumably because cued and uncued locations were only 4°
apart (cueing paradigms normally use between 10° and 20°).
We manipulated the perceptual strength (visibility) of the
primes by altering their duration, choosing 20 ms for ‘‘invisible’’
primes predicted to cause a negative compatibility effect, and 70
ms for ‘‘visible’’ primes predicted to cause a positive priming
effect. In objective tests of conscious perception, participants
showed chance performance for discriminating the direction of
the 20-ms primes (mean accuracy 50% and 51% for cued and
uncued primes, respectively), whereas performance was comfortably above chance for the 70-ms primes (mean accuracy 92%
and 88% for cued and uncued primes, respectively).
The first main result was that the negative compatibility effect
elicited by invisible primes was significantly modulated by the
attentional cues: it was enhanced when the location of the
invisible prime was cued compared with when the opposite
location was cued (t ⫽ 4.0, P ⬍ 0.01; see Fig. 2). More
importantly, the cueing effect for invisible stimuli was opposite
to that found with visible stimuli: The compatibility effect for
invisible primes was made more negative by cueing, whereas the
effect for visible primes became more positive (ANOVA interaction: F(1,11) ⫽ 20.4, P ⫽ 0.001; see the Fig. 2 legend for more
details on RT and error rate). If attention merely boosted the
sensory representation of the invisible stimulus, thus making it
more likely to be consciously perceived, we would expect attentional cueing to reduce the size of the negative compatibility
effect, making it more similar to the positive effect of the
stronger visible prime. Thus, the extra negative effect found is
difficult to explain as a consequence of an attentional enhancement of perceptual strength toward conscious threshold. Rather,
10522 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601974103
Fig. 3. Performance in prime discrimination blocks in Exps. 2 and 3. As the
perceptual strength of the primes was increased [by manipulating duration in
Exp. 2 (A) and brightness in Exp. 3 (B)], discrimination performance rose from
chance (50%) to ⬇90%. Note that cued primes reach threshold earlier than
uncued primes; the attentional effect at nominal threshold (75%) is equivalent to ⬇10 ms duration in Exp. 2 and 20 – 40 cdm⫺2 in Exp. 3.
it seems that the attentional effect may come from enhancing the
automatic, unconscious sensorimotor process that the invisible
stimuli elicit, rather than from making the stimulus representations more likely to penetrate through to consciousness.
Exp. 2. In a second experiment we aimed to more fully characterize the relationship between attentional modulation of invisible stimuli and perceptual strength by measuring the attentional
effect on primes as we manipulated their perceptual strength
stepwise across the conscious threshold. We did this by changing
the duration of the prime from 10 ms to 70 ms in 10-ms steps.
When participants were asked to guess the direction of the prime
(after the main experiment), discrimination performance rose
steadily from chance for 20-ms primes to ⬎90% for 60-ms
primes, and the probability of success was increased by cues at
the prime’s location compared with cues at the other location
(Fig. 3A). A cued prime of 30 ms, for example, produced
performance approximately equivalent to an uncued prime of 40
ms, and a cued prime of 40 ms was like an uncued prime of 50
ms. Thus, the magnitude of the attentional modulation was
equivalent to ⬇10 ms (or ⬇25%) extra prime duration. These
results confirm previous evidence that attention can modulate
perceptual threshold (12, 35) and are consistent with models in
which attention enhances perceptual contrast or saliency (24,
25). Such enhancement is reported to occur in many areas of
visual cortex (10) and may operate on neuronal activity that is
below the threshold for awareness as well as on suprathreshold
activity.
The critical question is this: Does this attentional modulation
of perceptual threshold explain attentional modulation of subSumner et al.
Fig. 4. Hypothetical and observed attentional modulation of the relationship between perceptual strength and the compatibility effect. (A) Predicted
pattern of results if attention modulates subliminal priming by means of
modulating the prime’s perceptual strength toward or away from conscious
threshold. An uncued prime needs greater duration to have the same effect
as a cued prime, shifting the results along the axis of prime duration (illustrated with black arrows). We have depicted an absolute shift of 10 ms
(alternatively, a percentage shift of, e.g., 25% would mean that the horizontal
gap between cued and uncued curves would be smaller than depicted for
short prime durations and larger than depicted for long prime durations). The
dashed gray lines mark the prime durations probed in Exp. 1, illustrating how
Sumner et al.
the opposing cueing effects for invisible and visible primes (Fig. 2) might have
arisen with this hypothesis. (B) Predicted pattern of results if attention modulates subliminal priming independent of perceptual strength and thus independent of the representations supporting conscious awareness. Cueing modulates the amplitude of the compatibility effect at each prime duration
(illustrated with black arrows). For comparison with A we have depicted the
same curve for the cued condition (we use the cued condition as baseline
because in most masked-prime experiments the location of the prime is
attended). As above, the dashed gray lines mark the prime durations probed
in Exp. 1. (C) Results of Exp. 2, which conform to prediction B: Although there
is perhaps a small shift along the x axis between the cued and uncued
conditions, the main attentional effect is a modulation of priming amplitude.
Thus, the attentional modulation cannot be explained as a modulation of
perceptual strength alone (prediction A), and this result supports the idea that
attention can act on representations that are independent of those supporting conscious awareness. Error bars are standard errors of the cueing effect at
each prime duration. See Table 1 for details of RT and error rates in each
condition. (D) Results of Exp. 3, which also conform to prediction B and
replicate all of the essential elements of the previous results with a different
manipulation of perceptual strength (brightness instead of duration). Error
bars are standard errors of the cueing effect at each prime duration. See Table
2, which is published as supporting information on the PNAS web site, for
details of RT and error rates in each condition.
PNAS 兩 July 5, 2006 兩 vol. 103 兩 no. 27 兩 10523
PSYCHOLOGY
liminal priming? We can answer this question because increasing
the prime’s perceptual strength is predicted to change the
priming effect gradually from negative to positive (27, 28). If
attention acts on the invisible primes by modulating their
perceptual strength, then modulating attention and modulating
prime duration should have equivalent effects (as occurred for
discrimination performance; Fig. 3A). This means that the
results for uncued primes should simply be shifted along the axis
of perceptual strength by ⬇10 ms relative to cued primes,
because the effect of having attention drawn away from the
prime is the same as the effect of decreasing the prime’s
duration: both affect the perceptual representation of the prime.
As illustrated in Fig. 4A, an uncued prime of 40 ms should behave
like a cued prime of 30 ms, and this shifts the results for uncued
primes to the right, such that the duration required to create a
given compatibility effect is slightly more for uncued primes than
for cued primes.
Alternatively, if attention acts on sensorimotor processes that
are independent from the representations that support conscious
awareness, the subliminal priming effect produced by any given
prime should simply be enhanced by attention. Thus, as illustrated in Fig. 4B, the pattern of compatibility effects for cued
primes would simply be an exaggeration of the pattern for
uncued primes: a change in amplitude rather than a shift along
the axis of perceptual strength.
The results of the second experiment show that attracting
attention toward or away from the primes modulated the amplitude of the compatibility effect, and there is only a small hint,
if anything, of a shift along the axis of perceptual strength (Fig.
4C; for more details see Table 1, which is published as supporting
information on the PNAS web site). The most crucial results are
those for primes of 30 ms, 40 ms, and 50 ms, which in the
perceptual strength account (Fig. 4A) are predicted to produce
a peak of negative compatibility for uncued primes similar to the
peak produced by cued primes with shorter duration. Instead,
consistent with the prediction in Fig. 4B, the results show that the
effect of uncued primes is simply reduced in amplitude relative
to cued primes, such that the relative effect of cueing reverses
near the point at which the priming itself crosses from negative
to positive. Thus, the attentional effect is not explained by
boosting the perceptual strength of the subliminal stimuli toward
conscious threshold. Instead, it indicates that attention can act
on sensorimotor processes that are independent of the representations that support consciousness awareness.
Exp. 3. In Exps. 1 and 2 we manipulated perceptual strength by
manipulating duration. In Exp. 3 we tested whether manipulating
perceptual strength in an alternative way, changing the brightness
(or contrast) of the prime, produced an equivalent pattern of
results. We fixed the duration of the prime at 40 ms and made all
of the stimuli gray instead of white, allowing us to choose four levels
of brightness for the prime that spanned discrimination threshold
(Fig. 3B). Again, primes in the cued location were discriminated
slightly better than uncued primes (tested after the main experiment), consistent with attention lowering threshold by enhancing
perceptual contrast (12, 24, 25, 35).
The main results of Exp. 3 mirrored those of Exps. 1 and 2 in
all of the essential elements (Fig. 4D). Low-contrast invisible
primes produced a negative compatibility effect, which changed
to a positive effect with brighter primes. Cueing modulated these
compatibility effects in opposite directions, making the effect of
weak primes more negative and the effect of strong primes more
positive (for more details see Table 2). Importantly, the pattern
of attentional modulation took the form of a change in amplitude
(as depicted in Fig. 4B) rather than a shift along the axis of
perceptual strength (as depicted in Fig. 4A). Thus, it is not
explained by a boost in perceptual strength (contrast or brightness in this case); instead, the pattern of results supports the idea
that attention can act independently on nonperceptual sensorimotor representations.
Discussion
It has long been known that attracting attention to a certain
location makes us more likely to consciously perceive weak
stimuli in that location. In other words, attentional orienting
modulates perceptual threshold, boosting the perceptual
strength of near-threshold stimuli (12, 35). This attentional
inf luence was demonstrated for our paradigm by improved
discrimination performance when participants were asked to
guess the direction of masked arrows (Fig. 3). This type of
finding has been extended recently by measuring attentional
modulation of priming effects elicited by invisible stimuli (22,
23) and also by measuring attentional modulation of adaptation to nondiscriminable illusory lines (36). It seems that the
inf luence of attention extends beyond near-threshold stimuli
to subthreshold stimuli that still remain invisible even after
attentional enhancement.
Our main finding is that attentional modulation of subliminal
priming cannot be fully accounted for by the established idea
that attention modulates the perceptual representations of the
stimulus, boosting them over, or nearer to, conscious threshold.
Attention enhanced the priming effect elicited by invisible
stimuli (Figs. 2 and 4 C and D) in a pattern that was not predicted
by an enhancement of the prime’s perceptual strength (as
illustrated in Fig. 4A). Instead, the pattern of results was
predicted by a direct attentional enhancement of the unconscious sensorimotor priming effect (Fig. 4B). There was one
previous hint of this dissociation between the effects of attention
and perceptual strength in masked priming: Schlaghecken and
Eimer (21) (Exp. 3) used cueing while investigating differences
between primes presented at fixation and in the periphery, and
they found a slightly greater positive compatibility effect for cued
primes. In a different study, on the other hand, increasing the
perceptual strength of peripheral primes by delaying mask onset
was found to reverse the compatibility effect (37).
Thus, we have measured two separate attentional effects on
invisible stimuli: one that enhanced their visibility (Fig. 3) and
one that independently enhanced unconscious sensorimotor
processes initiated by the invisible stimuli (Fig. 4 C and D). This
second type of attentional modulation potentially represents a
wider role for attention outside the perceptual realms traditionally studied, and the scope and limits of such attentional
modulation of unconscious processes will need to be thoroughly
10524 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601974103
investigated. From a practical point of view, it may be necessary
to control for attentional effects in any study of unconscious
sensorimotor processing.
A clue to how the second type of attentional modulation may
occur can be found in the apparent independence of some
sensorimotor processes from perceptual processes. For example,
whereas visual illusions may affect our perception of an object’s
size or position, it has been reported that visually guided
reaching can be relatively immune from such perceived distortions (38, 39). Similarly, patients with brain damage may be able
to act appropriately on objects despite dramatic perceptual
disruption that would seem to preclude such accurate actions
(40). Furthermore, neurons in some sensorimotor areas are
activated by visual stimuli more quickly than neurons in many
areas of visual cortex (41), indicating that sensorimotor mechanisms can be initiated without relying on a hierarchy of visual
processing to feed them perceptual information. Thus, there
appears to be some dissociation between ‘‘vision for perception’’
and ‘‘vision for action’’ (42), giving scope for attention to act on
these systems independently.
In fact, many components of our behavior seem to be underpinned by fast automatic sensorimotor processes that operate
outside the conscious domain (43). These automatic mechanisms
allow behavior to be efficient but are generally thought to
proceed inflexibly (44), so that goal-directed behavior must be
supervised by a ‘‘top-down’’ conscious attentional system. Our
data, combined with other recent evidence, show that unconscious processing is more flexible than previously suspected.
Moreover, there may be two kinds of unconscious process:
subthreshold activation that would reach awareness if boosted in
strength, and wholly unconscious sensorimotor mechanisms that
never reach awareness however strongly activated. Our results
suggest that the latter, as well as the former, is susceptible to
attentional modulation. This would explain how the priming
effect we studied, an automatic sensorimotor phenomenon, can
be enhanced independently from enhancing the perceptual
representations that support conscious awareness. Because so
much of sensation and action remains unconscious, we might
speculate that the unstudied role of attention in this domain may
prove to be as functionally important as its role in guiding
conscious behavior.
Nonconscious processes have also proved to be more flexible
than previously supposed in their response to context. Greenwald et al. (45) found that subliminal priming in a number
classification task was sensitive to contextual memory resulting
from previous practice. Kunde et al. (46) demonstrated that the
impact of subliminal priming depends on the participants’ cognitive intentions to respond to certain stimuli. Similarly, Eimer
and Schlaghecken (47) found context specificity effects in their
original experiments with the masked-prime paradigm we have
used: arrow primes have an effect only if the participant currently
wishes to respond to arrow targets. If instead the participants
must respond right and left to the letters R and L, arrow primes
have no effect, despite previous training that associated the
arrows with right and left responses. A recent functional MRI
study suggests that the source of such cognitive contextual
influences may be an area of frontal cortex known as the
presupplementary motor area (pre-SMA) (48). It is possible that
this contextual modulation, like the attentional modulation we
report, may affect sensorimotor processes that are independent
of the perceptual representations supporting awareness. Further
study will be needed to investigate this interesting possibility.
Methods
Exp. 1. Participants fixated on a central cross and responded with
a left or right button to the direction of arrows (targets)
presented randomly above or below fixation inside one of two
outline rectangles (see Fig. 1 for illustration and stimulus
Sumner et al.
Exp. 2. We manipulated the duration of the primes from 10 ms
prime duration between 20 ms and 60 ms, producing a psychometric function for each participant from which perceptual
threshold (e.g., the duration needed to produce 75% correct) can
be determined (see Fig. 3A).
Exp. 3. The procedure was similar to Exp. 2 except that we fixed
the duration of the prime at 40 ms and manipulated its contrast
to cross the perceptual threshold. The background was dark gray
[10 candelas per square meter (cdm⫺2)], and other stimuli were
lighter gray: fixation cross, targets, and masks were 40 cdm⫺2;
outline rectangles were 15 cdm⫺2, brightening to 40 cdm⫺2 for
the cue; primes varied from 20 cdm⫺2 to 160 cdm⫺2. As before,
we also measured discrimination performance for the direction
of the prime for each prime contrast (see Fig. 3B).
to 70 ms in steps of 10 ms, crossing the threshold of conscious
awareness. In all other ways the procedure was the same as for
Exp. 1. We also measured discrimination performance for the
direction of the prime, in the absence of the targets, for each
We thank Masud Husain, Geraint Rees, Andy Parton, Elaine Anderson,
and Adrian Burgess for comments; Hilmar Jay for technical support; and
all of the participants for their time. The research was supported by the
Biotechnology and Biological Sciences Research Council and the Wellcome Trust.
1. von Helmholtz, H. (1867) Handbuch der Physiologischen Optik (Voss, Hamburg).
2. James, W. (1890) Principles of Psychology (Holt, New York).
3. Posner, M. I. & Boies, S. J. (1971) Psychol. Rev. 78, 391–408.
4. Desimone, R. & Duncan, J. (1995) Annu. Rev. Neurosci. 18, 193–222.
5. LaBerge, D. (1997) Consciousness Cognit. 6, 149–181.
6. Robertson, L. C. (2003) Nat. Rev. Neurosci. 4, 93–102.
7. Lau, H. C., Rogers, R. D., Haggard, P. & Passingham, R. E. (2004) Science 303,
1208–1210.
8. Pashler, H. E. (1998) The Psychology of Attention (MIT Press, Cambridge, MA).
9. Driver, J. & Frith, C. (2000) Nat. Rev. Neurosci. 1, 147–148.
10. Treue, S. (2003) Curr. Opin. Neurobiol. 13, 428–432.
11. Rensink, R. A., O’Regan, J. K. & Clark, J. J. (1997) Psychol. Sci. 8, 368–373.
12. Hawkins, H. L., Hillyard, S. A., Luck, S. J., Mouloua, M., Downing, C. J. &
Woodward, D. P. (1990) J. Exp. Psychol. Hum. Percept. Perform. 16, 802–811.
13. Raymond, J. E., Shapiro, K. L. & Arnell, K. M. (1992) J. Exp. Psychol. Hum.
Percept. Perform. 18, 849–860.
14. Mack, A. & Rock, I. (1998) Inattentional Blindness (MIT Press, Cambridge,
MA).
15. Driver, J., Mattingley, J. B., Rorden, C. & Davis, G. (1997) in Parietal Lobe
Contributions to Orientation in 3D Space, eds. Thier, P. & Karnath, H.-O.
(Springer, Heidelberg), pp. 401–430.
16. Duncan, J., Bundesen, C., Olson, A., Humphreys, G., Chavda, S. & Shibuya, H.
(1999) J. Exp. Psychol. Gen. 128, 450–478.
17. Ohman, A. & Soares, J. J. (1994) J. Abnorm. Psychol. 103, 231–240.
18. Zeman, A. (2002) Consciousness: A User’s Guide (Yale Univ. Press, New
Haven, CT).
19. Kentridge, R. W., Heywood, C. A. & Weiskrantz, L. (1999) Proc. Biol. Sci. 266,
1805–1811.
20. Kentridge, R. W., Heywood, C. A. & Weiskrantz, L. (2004) Neuropsychologia
42, 831–835.
21. Schlaghecken, F. & Eimer, M. (2000) Percept. Psychophys. 62, 1367–1382.
22. Naccache, L., Blandin, E. & Dehaene, S. (2002) Psychol. Sci. 13, 416–424.
23. Lachter, J., Forster, K. I. & Ruthruff, E. (2004) Psychol. Rev. 111, 880–913.
24. Carrasco, M., Ling, S. & Read, S. (2004) Nat. Neurosci. 7, 308–313.
25. Treue, S. (2004) Trends Cognit. Sci. 8, 435–437.
26. Koch, C. (2004) The Quest for Consciousness: A Neurobiological Approach
(Roberts, Englewood, CO).
27. Eimer, M. & Schlaghecken, F. (2002) Psychonom. Bull. Rev. 9, 514–520.
28. Eimer, M. & Schlaghecken, F. (2003) Biol. Psychol. 64, 7–26.
29. Lleras, A. & Enns, J. T. (2004) J. Exp. Psychol. Gen. 133, 475–493.
30. Verleger, R., Jaskowski, P., Aydemir, A., van der Lubbe, R. H. & Groen, M.
(2004) J. Exp. Psychol. Gen. 133, 494–515.
31. Schlaghecken, F. & Eimer, M. (2006) J. Exp. Psychol. Gen., in press.
32. Eimer, M., Schubo, A. & Schlaghecken, F. (2002) J. Motor Behav. 34, 3–10.
33. Posner, M. I. & Cohen, Y. (1984) in Attention and Performance, eds. Bouma,
H. & Bouwhuis, D. G. (Erlbaum, Hillsdale, NJ), Vol. 10, pp. 531–556.
34. Sumner, P., Adamjee, T. & Mollon, J. D. (2002) Curr. Biol. 12, 1312–1316.
35. Muller, H. J. & Humphreys, G. W. (1991) J. Exp. Psychol. Hum. Percept.
Perform. 17, 107–124.
36. Montaser-Kouhsari, L. & Rajimehr, R. (2005) Vision Res. 45, 839–844.
37. Schlaghecken, F. & Eimer, M. (2002) Percept. Psychophys. 64, 148–162.
38. Bruno, N. & Bernardis, P. (2003) Exp. Brain Res. 151, 225–237.
39. Dewar, M. T. & Carey, D. P. (2006) Neuropsychologia 44, 1501–1508.
40. Milner, A. D., Perrett, D. I., Johnston, R. S., Benson, P. J., Jordan, T. R.,
Heeley, D. W., Bettucci, D., Mortara, F., Mutani, R., Terazzi, E. & Davidson,
D. L. W. (1991) Brain 114, 405–428.
41. Lamme, V. A. F. & Roelfsema, P. R. (2000) Trends Neurosci. 23, 571–579.
42. Milner, A. D. & Goodale, M. A. (1996) The Visual Brain in Action (Oxford
Univ. Press, Oxford).
43. Koch, C. & Crick, F. (2001) Nature 411, 893.
44. Marcel, A. J. (1983) Cognit. Psychol. 15, 197–237.
45. Greenwald, A. G., Abrams, R. L., Naccache, L. & Dehaene, S. (2003) J. Exp.
Psychol. Learn. Mem. Cognit. 29, 235–247.
46. Kunde, W., Kiesel, A. & Hoffmann, J. (2003) Cognition 88, 223–242.
47. Eimer, M. & Schlaghecken, F. (1998) J. Exp. Psychol. Hum. Percept. Perform.
24, 1737–1747.
48. Wolbers, T., Schoell, E. D., Verleger, R., Kraft, S., McNamara, A., Jaskowski,
P. & Büchel, C. (2006) Cereb. Cortex 16, 857–864.
Sumner et al.
PNAS 兩 July 5, 2006 兩 vol. 103 兩 no. 27 兩 10525
PSYCHOLOGY
timing). Before the target, a prime occurred in one of the two
rectangles for either 70 ms (visible condition) or 20 ms (invisible
condition). This prime was identical to one of the four possible
targets (leftward or rightward arrows in top or bottom rectangle)
and was immediately followed by masks of randomly oriented
lines presented for 100 ms in both rectangles. An attentional cue
was presented 100 ms before the prime by brightening one of the
rectangles. The location of the cue and the location or direction
of the prime had no predictive value for the location or the
direction of the target arrows, and participants were told this.
After the main experiment, we measured perceptual awareness
of the primes by asking participants to guess their direction in
trials where the targets were omitted. For more details see
Supporting Methods.