Content uploaded by Ronald A Rensink
Author content
All content in this area was uploaded by Ronald A Rensink on Oct 17, 2017
Content may be subject to copyright.
a0005 Attention : Change Blin dness and Inatt entional Blindnes s
R A Rensink, University of British Columbia, Vancouver, BC, Canada
ã2009 Elsevier Inc. All rights reserved.
Glossary
g0005 Change blindness – The failure to visually
experience changes that are easily seen once
noticed. This failure therefore cannot be due
to physical factors such as poor visibility;
perceptual factors must be responsible.
Focused attention is believed to be necessary
to see change, with change blindness
resulting if such attention is not allocated to
the object at the moment it changes.
g0010 Diffuse attention – A type of attention that is
spread out over large areas of space. It is
believed to be space-based rather than
object-based.
g0015 Focused attention – A type of attention
restricted to small spatial extents. It is
believed to act on small areas of space or on
relatively small objects.
g0020 Implicit perception – Perception that takes
place in the absence of conscious awareness
of the stimulus. It is generally believed to take
place in the absence of any type of attention.
g0025 Inattentional blindness – The failure to
visually experience the appearance of an
object or event that is easily seen once
noticed. Attention (likely, diffuse attention) is
thought to be necessary for such an
experience. Inattentional blindness typically
occurs when attention is diverted, such as
when the observer engages in an
attentionally demanding task elsewhere, and
does not expect the appearance of the object
or event.
s00 05 Introductio n
p0005 As obser ver s, we generally have a strong impres-
sion of seeing everything in front of us at any
moment. But compelling as it is, this impression
is false – there are severe limits to what we can
consciously experience in everyday life. Much of
the evidence for this claim has come from two
phenomena: change blindness (CB) and inatten-
tional blindness (IB).
CB refers to the failure of an observer to visu-
ally experience changes that are easily seen once
noticed. This can happen even if the changes are
large, constantly repeat, and the observer has been
informed that they will occur. A related phenome-
non is IB – the failure to visually experience an
object or event when attention is directed else-
where. For example, observers may fail to notice
an unexpected object that enters their visual field,
even if this object is large, appears for several
seconds, and has important consequences for the
selection of action.
Both phenomena involve a striking failure to
report an object or event that is easily seen once
noticed. As such, both are highly counterintuitive,
not only in the subjective sense that observers have
difficulty believing they could fail so badly at
seeing but also in the objective sense that these
findings challenge many existing ideas about how
we see. But as counterintuitive as these phenom-
ena are, progress has been made in understanding
them. Indeed, doing so has allowed us to better
understand the limitations of human perception in
everyday life and to gain new insights into how our
visual systems create the picture of the world that
we experience each moment our eyes are open.
Change Blindness
Background
The ability to see change is extremely useful in
coping with everyday life: we can monitor the
movement of nearby automobiles (as drivers or
pedestrians), notice sudden changes in the posture
or location of people in front of us, and notice that
the sky is quickly darkening. Given the importance
1
of perceiving change and the fact that most
humans can survive reasonably well in the world,
it follows that our ability to perceive change must
be such that few events in the world escape our
notice. This agrees nicely with our impression that
we perceive at each moment most, if not all,
objects and events in front of us.
p0025 Failures to see change have long been noticed,
but they were usually taken to be temporary aber-
rations, with nothing useful to say about vision.
This attitude began to change in the early twenti-
eth century, when film editors discovered an inter-
esting effect: when the audience moved their eyes
across the entire screen (e.g., when the hero exited
on the left side and the femme fatale entered on
the right) almost any change made during this time
(e.g., a blatant change of costume) would often go
unnoticed. A similar blindness to change could be
induced by making it during a loud, sudden noise
(e.g., a gunshot), during which the audience would
momentarily close their eyes.
p0030 The scientific study of this ef f ect began in the
mid-1950s, with work on position changes in dot
arrays and other simple stimuli. Here, a change in
one of the items was typically made contingent on a
temporal gap that lasted several seconds. A separate
line of studies was also begun that investigated
the perception of displacements made contingent
on an eye movement (or saccade). In all cases,
observers were found to be surprisingly poor at
detecting changes made under such conditions.
The next wave of studies, begun in the 1970s, was
based on a more systematic examination of these
effects. This work uncovered a general limit to the
ability to detect gap-contingent changes under a
wide variety of conditions; this eventually formed
the basisfor the proposalof a limited-capacity visual
short-term memory (vSTM). Likewise, a general
lack of ability was found for detection of saccade-
contingent changes, which was traced to a limited
transsaccadic memory. Both lines of research
were eventually linked by the proposal that trans-
saccadic memory and vSTM were in fact the same
system.
A third wave began in the mid-1990s, extending
the methodology and results of earlier work in
several ways. To begin with, stimuli were often
more complex and realistic: images of real-world
scenes or dynamic events were used in place of
simple figures. Next, changes were often repeated
rather than occurring just once, allowing the use of
time as well as accuracy to measure performance
(Figure 1). Third, blindness was induced via sev-
eral new kinds of manipulation, such as making
changes during a film cut or during the appearance
of sudden splats elsewhere in the image. Finally, all
these effects – as well as the earlier ones – were
accounted for by the proposal that attention is
One-shot paradi
g
m Flicker paradi
g
m(a) (b)
f0005 Figure 1 Example of a technique to induce CB. Here the change is made during a brief gap between the original and
the modified stimulus. (a) One-shot paradigm. A single alternation of the stimuli is used, and the observer must
respond to it. Performance is measured by accuracy of detection (or identification or localization). (b) Flicker
paradigm. Stimuli
Au2 are continually cycled under the observer responds. Performance is measured by the time taken
until the change is detected (or identified or localized). Both types of measurement paradigm can also be applied to
other techniques, such as changes made during saccades or splats.
2Attention: Change Blindness and Inattentional Blindness
necessary to see change. This work therefore
showed that this CB is a general and robust phe-
nomenon that can last for several seconds, and can
be connected to known perceptual mechanisms. As
such, it supported the idea that change perception
is an important part of perception, and that study-
ing its failure can cast light on mechanisms central
to our experience of the world.
s00 20 Conceptual Distinctions
p0045 A c lear under standing of CB – and its inver se,
change perception – has been slow to emerge.
This is partly due to the nature of change itself.
Although this concept appears simple, attempts to
formalize it over the years – from the earliest
Greek thinkers to modern philosophers – have
generally encountered difficulties. However, a
number of important distinctions have become
clear, which have helped our understanding of CB.
s00 25 Change versus motion
p0050 The human perce ption of temporal variation is
handled by two separate systems: a motion-
detection system for variation in regards to a partic-
ular location and a change-detection system for
variation in regards to structure. Temporal varia-
tions in the world activate both systems to some
extent. For example, a moving automobile will acti-
vate motion detectors over the relevant part of the
visual field, although these will not be able to deter-
mine whether there is an enduring spatiotemporal
structure that remains constant. Conversely, a whir-
ling dust devil will be seen as a coherent moving
structure, even though there is nothing beyond
the flow of dusty air in space. The key challenge in
perceiving motion and change is to separate out the
contributions of these two systems, so that temporal
variation is assigned to the correct substrate.
s00 30 Change versus difference
p0055 Another impor tant distinction is that between
change and difference. In both cases, reference is
to a particular structure – an object or event of
some kind. However, while change refers to trans-
formation of a single structure, difference refers to
the comparison of two or more separate structures.
p0060 More precisely, change is based on the proper-
ties of the same structure at separate points in
time, and can be perceived as a single dynamic
visual event. For example, when a person is seen
walking, their legs are seen at various positions
over time. This kind of perception suggests –
although does not prove – that the underlying
representation has a spatiotemporal continuity of
some kind. In contrast, difference is based on an
atemporal comparison of structures that may or
may not exist simultaneously, for example, com-
paring the height of two people. This kind of
perception allows for the possible involvement of
long-term memory elements that are only inter-
mittently engaged.
This distinction is important. Change detection
and difference detection can be distinguished, at
least conceptually; it is important to consider the
type of experiment used when discussing either one.
Spotting the difference between two side-by-side
stimuli may not engage the same perceptual
mechanisms as detecting a change in successively
presented stimuli. Failure to make this distinction
can lead to erroneous conclusions being drawn
about the mechanisms involved.
Experimental Approaches and Results
All studies to date have been based on much the
same design: an initial stimulus is first presented
(e.g., a picture), an altered version is then shown
(e.g., some object in it is removed), and the ability
of the observer to perceive the change is then
measured (see Figure 1). Various aspects of per-
ception have been explored in this manner based
on particular selections among a relatively small
set of design parameters.
Type of task
CB is the failure to perceive a change. Since there
are several different aspects to perception, there
can be – at least in principle – several different
kinds (or levels) of blindness. These can be inves-
tigated by giving the observer the appropriate kind
of task:
.Detection. This is the most basic and the most
widely studied aspect, concerned with the sim-
ple noticing of a change. No properties of the
change itself are necessarily perceived: the
observer is simply asked to report whether
it occurs.
Attention: Change Blindness and Inattentional Blindness 3
.Identification. This concerns reporting the
properties of the change, that is, seeing what
type it is (e.g., a change in color, a change in
orientation). This can in principle be separated
from detection: the observer could be asked to
guess the type of change even if the change
itself was not detected. Identification of change
appears to be more difficult than detection,
indicating that somewhat different mechanisms
are involved.
.Localization. This is concerned with reporting
the location of the change. This can be
decoupled from detection and from identifica-
tion, at least in principle. Relatively little work
to date has been done on this aspect of percep-
tion. Some results suggest that a separate mem-
ory for location may exist, although this has not
yet been fully established.
s00 45 Type of response
p0080 Another way to engage dif f erent mechanisms is to
use different aspects of the response of an observer
to a change in the external world. These effectively
test different perceptual subsystems:
.Explicit percept. This is the approach used in
most perceptual experiments, involving the
conscious visual experience of the observer.
A high degree of blindness can usually be
induced. The proposal that attention is needed
to see change is concerned with this type of
experience.
.Semiexplicit percept. Some observers can have
a ‘gut feeling’ for several seconds that change is
occurring, even though they do not yet see it
(i.e., do not have a picture of it). The basis of
this is controversial, but it may involve nonat-
tentional or subattentional systems.
.Implicit percept. Change that is not experi-
enced consciously may still be perceived
implicitly. If so, this must be measured by its
effect on other processes. For example, some
studies indicate that an unseen change can
influence forced-choice guessing about its pos-
sible location. The existence of this form of
perception is controversial, although evidence
for it appears to be increasing.
.Visuomotor response. This involves the res-
ponse of a visually guided motor system to a
change that is not consciously experienced. Sys-
tems used are almost always manual pointing or
eye fixation. Both kinds of visuomotor response
are faster than consciously mediated ones. They
also appear to be more accurate, suggesting the
existence of representations with higher-capac-
ity memory.
Attentional manipulation
A central tenet in change perception is that atten-
tion is needed to consciously experience change.
In normal viewing, the local motion signals accom-
panying a change attract attention to its location,
allowing the change to be seen immediately; this is
why it helps to wave at a friend from across the
room. If these local signals can be neutralized so
that the automatic drawing of attention cannot
help, a time-consuming attentional scan of the
display will be needed. The observer will conse-
quently be blind to the change until attention is
directed to the appropriate item. A variety of tech-
niques have explored this:
.Gap-contingent techniques. The change is
made during a blank field or mask briefly dis-
played between the original and altered stimu-
lus, which swamps the local motion signal (see
Figure 1). Observers are very poor at detecting
change if more than a few items are present;
results suggest only 3–4 items can be seen to
change at a time.
.Saccade-contingent techniques. The change is
made during a saccade of the eyes. Observers
are generally poor at detecting change if more
than a few items are present; again, a limit of
3–4 items is found. CB can also be induced for
position change with even one item present,
provided no global frame of reference exists.
.Blink-contingent technique. The change is
made during an eyeblink. Again, observers are
generally poor at detecting such changes. Inter-
estingly, blindness can be induced even if the
observer is fixating the changing item.
.Splat-contingent techniques. These make the
change at the same moment as the appearance
of brief distractors (or splats) elsewhere in the
image. The blindness induced in this manner is
relatively weak, but still exists, showing that it
4Attention: Change Blindness and Inattentional Blindness
can be induced even when the change is
completely visible.
.Gradual-change techniques. Here, the transi-
tion between original and altered display is
made slowly (e.g., over the course of several
seconds). Observers generally have difficulty
detecting such changes, even though no disrup-
tions are used.
The results from all these approaches are consis-
tent with the proposal that attention is needed to
see change. The finding that at most 3–4 items can
be seen to change at a time is consistent with the
capacity of attention obtained via other techniques
such as attentional tracking.
s00 55 Perceptual set
p0095 An impor tant par t of perce ption is the perce ptual
set of the observer, which strongly affects the
mechanisms engaged for a given task. The issue
of set is important for the question of which
mechanisms are involved in everyday vision, and
how these are related to performance as measured
in the laboratory:
.Intentional set. Observers are instructed to
expect a change of some kind. They are assumed
to devote all their resources to detecting the
change, which provides a way to determine per-
ceptual capacities. A further refinement is
controlling expectation for particular types of
change. Changes for presence and location are
not affected by expectation of type, whereas
changes for color are.
.Incidental set. Observers are given some other
task as their primary responsibility (e.g., count
the number of sheep in an image); there is no
mention of a possible change until after the task
is over. The engagement of perceptual mechan-
isms is believed to be more representative of
their use in everyday life. The degree of blind-
ness found under these conditions is higher
than under intentional conditions, indicating
that relatively little is attended – or at least
remembered – in many real-life tasks.
s00 60 Implications for Perceptual Mechanisms
p0100 At one level, CB is an impor tant phenomenon in
its own right: among other things, it illustrates the
extent to which we can potentially miss important
changes in everyday life. Indeed, most people are
unable to believe the extent to which they are
unable to see change – in essence, they suffer
from ‘CB blindness.’
However, CB itself can also be harnessed as a
powerful tool to investigate the mechanisms by
which we see. The exact conclusions obtained
from such studies are still the subject of debate,
but their general outlines are becoming clear.
Visual attention and short-term memory
All experiments on CB are consistent with the
proposal that attention is needed to see change.
Experiments on carefully controlled stimuli sug-
gest that 3–4 items can be seen to change at a time,
consistent with the limit on attention obtained via
other techniques, such as attentional tracking.
However, while a limit of some sort is involved,
the nature of this limit is still somewhat unclear.
Contrary to subjective impression, change percep-
tion is not an elementary process. Instead, it
involves – at least in principle – a sequence of
several steps: enter the information into a memory
store, consolidate it into a form usable by
subsequent processes, hold onto this for at least a
few hundred milliseconds, compare it to the cur-
rent stimulus at the appropriate location, clear the
memory store, and then shift to the next item.
A limit on any of these steps would limit the entire
process, making it difficult to determine the rele-
vant step in a given situation.
Most proposals for mechanisms have been
couched in terms of either visual attention or
vSTM. Both are similar in the results they cite
and the mechanisms they propose. Part of this is
caused by the extensive overlap that appears to
exist between the mechanisms associated with
attention and vSTM. Indeed, the difference may
be largely one of terminology, caused by the
vagueness in the definition of attention. As used
in CB studies, attention is defined by the formation
of representations coherent over space and time;
such representations are not that different from
those posited as the basis of vSTM.
In any event, interesting new issues are
emerging. One is whether the limiting factor
applies to the construction of the coherent repre-
sentation, or to its maintenance once it is formed,
Attention: Change Blindness and Inattentional Blindness 5
or perhaps both. Another issue is the nature of the
elements that are attended and held in vSTM.
Much evidence suggests that these are proto-
objects that already have a considerable local bind-
ing of features; if so, the function of attention and
vSTM would be to create a representation with
extended spatial and temporal coherence. Results
also suggest that the 3–4 items are not indepen-
dent, but, rather, may have a higher level of inter-
action such that their contents are pooled into a
single collection point (or nexus) that supports the
perception of an individual object.
s00 70 Scene perception
p0130 Given that only a f ew items may have a coherent
representation at any time, our subjective impres-
sion of seeing everything that happens in front of
us cannot be correct. But care must be taken in the
particular inferences drawn. Since it deals only
with dynamic quantities, CB cannot say anything
about the static information that may or may not
be accumulated. On the other hand, it does show
the existence of severe limits on the extent to
which changes are represented in those subsystems
accessible to conscious perception.
p0135 To account for the impression that we see all the
changes that occur in front of us – not to mention
that we can actually react to many of these – it has
been proposed that scene perception is handled by
a virtual representation. Here, coherent represen-
tations of objects – needed for change perception –
are created on a ‘just in time’ basis, that is, formed
whenever they are needed for a task and then dis-
solved afterward. Coordination of this process can
be achieved via a sparse schematic representation
of the scene – perhaps a dozen or so items, each
with some properties – formed independently of
the coherent representations. Guidance could be
done on the basis of both high-level factors (e.g.,
schemas that enable testing of expected objects)
and low-level factors (e.g., motion signals that
draw attention to unexpected events). All results
on CB are consistent with this proposal, and results
from other areas (e.g., research on eye movements)
appear consistent with this as well.
p0140 The status of static scene information is still
unclear. In principle, it could be accumulated to
create a dense description that would match our
subjective impression. However, such accumulation
is unnecessary: a virtual representation could handle
most if not all aspects of scene perception. Further-
more, no results to date have clearly shown storage
of information beyond the relatively sparse informa-
tion used for guidance and the contents of attention
and vSTM. Some information about the prior state
of a changed item appears to be stored in a longer-
term memory not used for the perception of change;
the amount of this is information not known. More
generally, many scenes might be stored in long-term
memories of various kinds, but the information den-
sity of each could still be quite low – perhaps a
relatively limited amount of information from each
of a dozen or so locations.
Whether the dynamic element is total or partial,
it nevertheless plays an important role in each indi-
vidual’s perception of a given scene. CB has there-
fore been useful as a tool to investigate individual
differences in perception: the faster a change is
detected, the more important it is deemed to be.
Studies have shown an effect of training – including
culture – on the encoding of objects and the impor-
tance attached to them, along with an influence of
the particular task undertaken. There is also an
emerging connection here with the design of inter-
active visual interfaces: these owe much of their
success to the engagement of these dynamic mech-
anisms, and are therefore subject to many of the
same considerations regarding operator and task.
Inattentional Blindness
Background
It has been known for thousands of years that
people engaged in deep thought can fail to see
something directly in front of their eyes. Such
blindness can be easily induced when an observer
intensely attends to some event, for example, wait-
ing to see if an oncoming automobile will stop in
time. Under such conditions, much of the visual
field can effectively disappear from consciousness,
even if it contains objects that are highly visible.
It was long believed that such blindness might be
due to the image of an unseen object falling onto
the periphery of the eye, which is relatively poor at
perceiving form. But decades of research have
shown that location is unimportant – the key factor
is attention.
6Attention: Change Blindness and Inattentional Blindness
p0155 One of the ear liest lines of research that
encountered this IB was the study of head-up dis-
plays for aircraft pilots, which superimpose
instructions and information over a view of the
external world. Simulator studies in the early
1960s discovered that when pilots carried out dif-
ficult maneuvers requiring attention to outside
cues, they did not see the unexpected appearance
of instructions to alter course. Later studies
showed the converse effect: if pilots focused on
the displayed instructions, they often failed to see
highly visible aircraft unexpectedly rolled out onto
the runway in front of them.
p0160 The fir st perce ptual studies to investigate this
effect took place in the 1970s. These used selective
looking tasks, where observers were presented
with two superimposed visual events (e.g., a sports
game and the face of someone talking) and asked
to attend to just one of these. Results showed that
observers could easily report what happened in the
attended event, but did not do well in the other:
they often missed large unexpected events, such as
the appearance of a woman carrying an umbrella.
This happened even when the unseen stimulus
was in the center of the visual field, showing that
the cause of this blindness effect was not optical,
but attentional.
p0165 Although this effect was surprising, the use of
superimposed stimuli caused it to be considered
somewhat artificial, and interest in it never really
developed. But in the 1990s a new wave of studies
did manage to kindle interest. These differed from
the earlier studies in several ways. First, they used
opaque rather than superimposed (or transparent)
stimuli, greatly reducing concerns about artificial-
ity. Indeed, it was found that even large unex-
pected events – such as the appearance of a
human in a gorilla suit – were still not noticed by
most observers under these conditions. Second, in
addition to videos of events, techniques were also
developed based on brief-presented static images
(Figure 2), which showed that IB could apply to
both static and dynamic stimuli. Third, simple
stimuli were often used in both the static and
dynamic case, allowing more experimental control.
And finally, there was greater examination of the
effects caused by the stimuli that were not seen
consciously. Together, these developments showed
that IB could indeed occur in everyday life, and
that the mechanisms involved play a major role in
everyday perception.
Conceptual Distinctions
Work on the theoretical and experimental aspects
of IB have emphasized several conceptual distinc-
tions. Some apply not only to IB but are also
relevant to more general issues of awareness.
Expression versus suppression
Given that attention is needed to see something (i.e.,
consciously experience it), two general possibilities
Non-critical trial Critical trial
(a) (b)
f0010 Figure 2 Example of a technique to induce IB. A static test stimulus is presented while visual attention is focused on a
primary task; here, the task is to report which of the two lines is longer. (a) Noncritical trial. The primary lines alone are
presented for 200 ms, followed by a mask to prevent further processing. The observer must report which is the longer
line. Two or three such trials are usually given to allow the observer to get into the appropriate mental state. (b) Critical
trial. Along with the primary lines, an unexpected test stimulus is presented nearby. Blindness is measured by
detection (or identification or localization) of the test stimulus. An additional round of trials is often presented, in which
the test stimulus is now expected; this is believed to correspond to a condition of divided attention.
Attention: Change Blindness and Inattentional Blindness 7
exist as to how it acts. First, attention might cause
selected stimuli to be expressed consciously, for
example, they enter awareness by being activated
so as to exceed some threshold. In this view, atten-
tion does not just facilitate the conscious perception
of a stimulus – it also enables it. On the other hand,
attention might act to selectively suppress some of
the stimuli, so that they vanish from the conscious
mind. (In both cases, the status of unattended items
is open: they might be weakly expressed in some
way, or available implicitly.) It is worth noting
that the two possibilities are not exclusive: it could
be that both are used, with attention acting to
express – or at least emphasize – some items while
suppressing others.
s00 95 Restricted versus unrestricted effects
p0180 A related issue is the extent of any expression or
suppression. It was once believed that everything
registered in the visual system was experienced
consciously. However, given that implicit percep-
tion of various kinds has been found, an important
issue is whether the selective effects underlying IB
are restricted to conscious experience, or whether
they spill over to other aspects of perception.
p0185 For example, results show that attentional inhi-
bition of a given location will increase the amount
of blindness there. But does such inhibition atten-
uate the input strongly enough to also affect
implicit processes at that location? Virtually all
studies to date have assumed that selective effects
are restricted to conscious experience, but this
assumption has not been tested.
s01 00 Experimental Approaches and Results
p0190 At the most general level, all studies of IB use the
same basic approach. An observer is given a pri-
mary task that engages their visual attention, an
unexpected test object (or event) then appears in
the display, and the response of the observer to that
appearance is then tested. Because expectation is
an important factor, several trials without a test
stimulus are generally presented first. Results are
taken from the trial in which the test stimulus first
appears (the critical trial). After this, the object is
no longer entirely unexpected, and the blindness
levels in subsequent trials are usually much lower,
consistent with the idea that attention is now
divided among the expected stimuli.
Type of task
As in the case of CB, several different aspects of
perception can be distinguished, each of which can
be tested by appropriate selection of task:
.Detection. This is the most basic aspect of per-
ception, concerned with the simple noticing of
something present; no properties of the stimu-
lus itself need be involved. For isolated static
items that are small, that is, have an extent of
less than about 1of visual angle, observers
often fail to see anything at all in conditions
that induce IB. For larger items, however,
detection remains generally good. For dynamic
events, informal observations suggest that the
unattended event does not necessarily disap-
pear entirely – observers often notice ‘some-
thing else’ going on, but cannot say what it is.
.Identification. This is concerned with deter-
mining various properties of the stimulus. For
example, observers can be asked to report the
color or location of an item; in principle –
although not often in practice – this can be
done whether or not it was detected. Observers
can usually report the color of a small item that
is detected; identification of shape appears to be
more difficult. Detection without identification
has been reported, but it has been suggested
that these are simply false positives.
.Localization. Here, the task is to locate the test
stimulus that appears in the critical trial. In
principle, this can be tested independently of
detection and identification, although this is
rarely done. Observers are able to report the
location of a small item that is detected,
showing that there may be some interaction
between these two aspects of perception.
Type of response
Performance can also be studied in terms of the
kind of percept experienced. Operationally, this
can be done by testing different aspects of the
observer’s response to the appearance of a stimu-
lus. These different aspects effectively involve dif-
ferent perceptual subsystems:
.Explicit percept. This is the kind of percept
used in most IB experiments. The observer is
asked to respond to their conscious experience,
8Attention: Change Blindness and Inattentional Blindness
that is, the picture they have of something. The
usual definition of IB is in terms of the failure to
have such an experience; the proposal that
attention is needed to see is likewise concerned
with this aspect.
.Implicit percept. Items that are not experienced
consciously may still be perceived implicitly.
Given that the observer is blind to the stimulus,
responses cannot be measured via direct subjec-
tive experience; they must instead be based on
indirect effects. For example, relatively little
blindness is found for emotionally laden words
or pictures. This suggests that such words have
been perceived implicitly, with attention drawn to
them on the basis of their meaning. Similarly, a
length illusion can occur in a set of lines perceived
consciously, even if the background lines that
induce the illusion are themselves not reported.
.Motor response to primary task. The response
to the appearance of a test stimulus can be
measured by its effect on the speed or accuracy
of the primary task. This is essentially the
approach developed to study attention capture,
which measures the effect of a new item on
motor response times for the main task, regard-
less of whether the new item is seen. (An inter-
esting variation would be to determine if such
effects are conditional on the experiential state
of the test stimulus.)
.Visuomotor response. Here, the appearance of an
item causes the eye to move toward it. This is
another approach developed in the area of atten-
tion capture; again, this response is measured
regardless of whether the item is seen. This
might be useful for studying IB if measurement
were made conditional on whether the test stim-
ulus is consciously experienced.
s01 15 Attentional manipulation
p0205 An impor tant f actor in inducing IB is to present
the test stimulus while the observer has their visual
attention engaged on an unrelated task. This has
been done in several ways:
.Superimposed stimuli. Two independent dyn-
amic events (e.g., a basketball game and two pairs
of hands playing a game) are presented simulta-
neously. In earlier studies, this was often done via
half-silvered mirrors; more recent studies do this
electronically. The result is perceived as a set of
transparent or ghostly images. Observers can eas-
ily attend to one of these, but can often miss events
in the other.
.Interspersed stimuli. Here, two different sets of
stimuli are presented; all are opaque and appear
on the same display (see Figure 2). Tests on
static stimuli generally use this method. Obser-
vers are asked to carry out a primary task on
one of the sets (e.g., make a length judgment on
a pair of lines); a test stimulus is presented near
these lines a few trials later. Dynamic events
have also been tested this way. A high degree of
blindness can be found, even when the items in
the two sets of stimuli are intermingled.
.Dichoptic presentation. Here, two independent
events are presented, each to a different eye; the
observer is asked to pay attention to one of
them. Events in the unattended eye often fail
to be reported. In contrast with superimposed
stimuli, where observers often have an impres-
sion of ‘something else’ going on, observers here
fail to experience anything of the unattended
set – it is simply not there. This may be related
to the blindness experienced in binocular
rivalry suppression.
Perceptual set
An important aspect of IB is the perceptual set of
the observer. There are several ways this could
influence performance:
.Control of selectivity. This is the extent to which
selection – either expression or suppression – is
invoked in the primary task. For example, obser-
vers can be required to attend only to the white-
shirted players in a game, while ignoring the
black-shirted players. In such tasks, blindness
appears to be due at least in part to observers
suppressing the features of the ignored stimuli:
the greater the similarity of the test item to the
ignored items, the greater the blindness. In non-
selective tasks, there is no need to screen out any
stimuli, at least up to the critical trial. Blindness
is still induced here, but it remains unknown
whether this is due to a failure of expression or
an invocation of suppression.
.Control of capture. It has been suggested that
high-level control may be exerted over the kind
Attention: Change Blindness and Inattentional Blindness 9
of stimuli that can capture attention, and thus
be seen. Attention is usually drawn by the
appearance of a new item or by the presence
of a unique property. However, work on atten-
tion capture has shown that this can be over-
ridden by a high-level attentional set. Results
on IB are consistent with the proposal that an
attentional set governs what is consciously seen,
with distinctive stimuli experienced con-
sciously only if they fit into the observer’s
expectations.
s01 25 Implications for Perceptual Mechanisms
p0215 The finding that obser ver s can often f ail to see
highly noticeable objects and events touches on
several issues concerning the way we cope with
our world. For example, it suggests that we might
not be aware of the extent to which we fail to see
various aspects of our immediate environment,
even if these are important and highly visible.
(This might be termed IB blindness.) This has
obvious implications for tasks such as driving,
where the ability to accurately perceive objects
and events is literally a matter of life and death,
and where knowing about our limitations could
well affect how careful we are.
p0220 Meanwhile, work on IB can also tell us a bout
the mechanisms involved in visual perception. In
particular, it can provide a unique perspective on
the mechanisms that underlie our conscious expe-
rience of the world.
s01 30 Visual attention
p0225 All results to date are consistent with the proposal
that attention is needed to see an object or event.
For example, the degree of blindness has been
found to increase with distance from the center of
the attended location, in accord with space-based
models of attention. In addition, the greater the
attentional load of the primary task, the greater
the blindness to the test stimulus, which is consis-
tent with the proposal of a limited attentional
capacity. Indeed, the inability of an observer to
follow more than one coherent event supports the
proposal that only one complex object or event can
be attended at a time.
p0230 Results also provide tentative suppor t for the
proposal that the high-level control of attention is
achieved via an attentional set which determines
the kinds of information that can capture attention,
and perhaps also the kinds of information that can
enter conscious awareness. This development
potentially connects work on IB to work on atten-
tion capture.
Another possible connection is with preatten-
tive vision, usually studied by the rapid detection
(or pop-out) of unique items in a display. Work
here has pointed toward a considerable amount of
processing achieved in the absence of attention.
But this assumes that little or no attention is
given to most items in a display at any moment.
Results on IB suggest that this supposedly inatten-
tional condition may be better viewed as a case of
diffuse attention, with the inattentional condition
characterized as one where no conscious percep-
tion exists.
However, there are several findings that make
these connections less than certain. For example,
an observer in an IB experiment can usually detect
the individual items in a group, although the
grouped pattern itself cannot be identified. In
addition, the pop-out of a unique item in a group
occurs only if the group appears in the critical
trial; if this group is shown earlier in noncritical
trials (with the unique item the same as the
others), pop-out no longer occurs. So what is the
effect of the unique item here? If it is to draw
attention to the group, why are the individual
elements already seen in the earlier, noncritical
trials? And why should pop-out occur in one con-
dition, but not the other?
Scene perception
Relatively little work has studied the aspects of
natural scenes perceived under conditions of inat-
tention. Studies using briefly presented scenes as
test stimuli found that observers could usually
report the scene gist (i.e., its overall meaning,
such as being an office or a forest), along with
several objects of indeterminate description.
Some confabulation is also found. This is broadly
similar to results on the ability of observers to
rapidly perceive the gist of scenes from brief expo-
sures of 100 ms or less, which is also likely done
with little or no focused attention.
Individual differences in the coding of scenes and
events can be measured in terms of how blindness
10 Attention: Change Blindness and Inattentional Blindness
varies with the primary task. For example, experts in
basketball could better detect the appearance of an
unexpected object while they were attending to a
basketball game. This suggests that they had
encoded the scenes and events in a way that allowed
them to divert some of their attention to occasion-
ally monitor other items, without seriously affecting
performance on the primary task.
p0255 The model of scene perce ption often used to
account for results on IB is the perceptual cycle.
Here, sustained attention is believed to activate the
conscious percept of a stimulus; once this has been
done, the stimulus has entered the cycle, and helps
select the appropriate schema to determine what
information to admit next. In contrast, stimuli
that do not become part of the predictive cycle
may never be seen at all. Such stimuli – especially
low-level signals – can guide the process, but
do not enter awareness on their own. Similarly,
information assembled at a preattentive level can
also guide this process, although it too does not
enter awareness automatically. This model has
some similarities to the dynamic scene representa-
tion often used to account for CB, as well as the
reentrant models used to account for conscious
experience.
s01 40 General Issues
s01 45 Change Blindness versus Inattentional
Blindness
p0260 CB and IB both involve a f ailure to perceive things
that are easily seen once noticed, and both are
believed to be due to a lack of attention. It is
therefore likely they are related. But exactly how?
p0265 Fir st of all, the dif f erence between CB and IB
does not depend on the kind of input. Both can be
found using dynamic images, and both can be found
using static images. And the particular contents of
the input do not matter for either. Instead, the criti-
cal difference between the two is the status of the
information under consideration. IB is entirely
concerned with first-order information – the simple
presence of quantities. In contrast, CB involves
second-order information – the transitions between
these quantities. These can be separated: an alter-
nating sequence of two images, say, could be experi-
enced as a 50% presence of each image over time
(same first-order distributions), but with different
amounts of change (different second-order distri-
butions) if the alternation rates are not the same.
And just as CB can say little about first-order (static)
information, IB can say little about second-order
(changing) information. The two therefore refer to
largely complementary aspects of the visual world.
This distinction has consequences for the per-
ceptual mechanisms involved. For example, the
kind of attention required for each aspect of per-
ception may be different – or at least, have different
effects. The kind of attention involved in IB is space
based: the degree of blindness increases with
increasing distance from the center of attention.
It is also easily diverted – hence the common use
of a test stimulus that is completely unexpected.
In contrast, the kind of attention needed to per-
ceive change is object-based and is much less eas-
ily diverted (or at least slower), since telling the
observer that a change will occur – and even giv-
ing them practice at perceiving it – does not
affect performance to any great extent. Loosely
speaking, IB might be identified with the absence
of diffuse attention, and CB with the absence
of focused attention. But a final determination of
this must await a better understanding of atten-
tion itself.
Blindness versus Amnesia
An important issue in regards to the status of these
induced failures is whether they are failures of
perception or of memory. It might be, for instance,
that an observer did experience something under
conditions that induced CB or IB, but then forgot
it before they could make their report. If so, these
effects would not be forms of blindness, but forms
of amnesia.
In the case of CB, the resolution of this issue is
reasonably straightforward. Perception of change
is often measured by asking the observer to
respond to the change as soon as they see it; all
that is needed to trigger this is a minimal conscious
experience. When observers are asked to respond
to a single change, only a few hundred millise-
conds exist between its presentation and the initi-
ation of the report (e.g., the pressing of a button).
Thus, if the experience of change is forgotten in
the absence of attention, it would be exceedingly
Attention: Change Blindness and Inattentional Blindness 11
brief and incapable of causing a response to be
initiated. To all intents and purposes it would be
as if it never existed, at least at the conscious level.
(Note that if taken seriously, the possibly of such a
fleeting perception would not be restricted to CB –
it would apply to any failure of perception.)
p0285 The situation for IB is more complex. In one
sense, this issue was resolved by the finding that
unattended – and thus unseen – items are indeed
perceived, in that they can indirectly affect aspects
of conscious experience. Entry into conscious
experience itself, however, is not addressed by
this. The observer is usually asked about a possible
item only after the primary task has been com-
pleted, which allows several seconds to possibly
forget it. And, the observer cannot be asked to
prepare to respond to the test stimulus, since this
sets up an expectation of the item, which severely
diminishes the attentional diversion, and thus the
degree of blindness.
p0290 This issue has been grappled with in several
ways. One is to present a highly surprising or
meaningful item, and hope that the observer will
spontaneously report it. However, this has not
generally been successful: even if a human walks
around in a gorilla suit or if an airplane is wheeled
out onto the runway where the pilot is about to
land, most observers still do not respond. It has
been proposed that the observers do see the visual
elements, but do not assign meaning to them. In
essence, this phenomenon becomes one of inatten-
tional agnosia. (This may explain the reports
in some selective looking experiments, where
observers see ‘something else’ going on, but do
not know what it is.)
p0295 Another per spective derives from studies of the
neural systems involved. Patients with lesions to
particular parts of the cortex can suffer conditions
such as neglect and extinction, in which attention
cannot be easily allocated to objects. Such patients,
however, do not appear to experience forgetting –
they simply do not report perceiving such stimuli,
even when asked with the object in full view. In
addition, functional imaging of the brains of nor-
mal observers shows that words are not con-
sciously identified in the absence of attention,
even if the observer is looking directly at them.
These results make it highly likely that the failure
is one of perception, and not memory.
Visual Attention versus Visual Experience
CB and IB can be regarded as two forms of the
perceptual failure created by the diversion of
attentional resources. They can be distinguished
at the functional level by the type of information
involved (second- or first-order information,
respectively). They also appear to be distinguished
by the type of attention involved (focused or dif-
fuse) and the kinds of operations (e.g., comparison)
associated with these. This division may corre-
spond to the two modes sometimes proposed for
conscious visual experience: an object mode asso-
ciated with focused attention and a background
mode operating as default. Beyond this, however,
only partial and tentative conclusions can be
drawn regarding the issue of how visual attention
relates to conscious visual experience.
In the case where attention of any kind is
absent, there does not appear to be any conscious
experience of stimuli (second-order quantities for
focused attention; first-order quantities for dif-
fuse). However, results still point to a considerable
amount of processing being carried out. For exam-
ple, work on IB indicates that unattended – and
therefore unseen – items can influence the percep-
tion of attended items. Similarly, some models of
CB posit low-level representations with a degree
of detail and feature binding (proto-objects) that
are formed in the absence of this kind of attention.
It is worth pointing out that observers in IB
experiments often report that they can detect some-
thing about the nonselected stimuli, even though
they cannot always identify it. Importantly, this
kind of experience is found only in those experi-
ments involving superimposed or interspersed
stimuli; for dichoptically presented stimuli, there is
a complete absence of perception of the nonselected
event. This suggests that in the superimposed and
interspersed conditions diffuse attention is given to
nonselected events, with the main event given
focused attention. Ifso, this would suggest that iden-
tification and localization may require more focused
attention (or related resource), and that both diffuse
and focused attention may be allocated simulta-
neously to different stimuli.
This proposal would be consistent with work on
CB. Focused attention is needed only for the per-
ception of complex quantities such as change;
12 Attention: Change Blindness and Inattentional Blindness
background items not given focused attention
might still be seen, but only in regards to detection
and perhaps a limited form of identification based
on relatively fragmented pieces of static items.
s01 60 Conclusions
p0320 Work on CB suggests that focused attention is
needed for the conscious experience of change:
without it, observers will be blind to even large
changes, at least at the conscious level. Some abil-
ity to implicitly perceive change may exist; if so,
this does not appear to require focused attention.
p0325 Similar ly, work on IB suggests that diffuse atten-
tion is needed to detect an unexpected object or
event, that is, to see it as ‘something.’ Some results
indicate that further (attentional) processing may
be needed to more completely identify or locate it.
There also appears to be some ability to pick up
and process information about static items in the
absence of any form of attention, even though such
items are not experienced consciously.
p0330 Beyond this, our current understanding is poor.
More work is needed to expand our empirical
knowledge of the basic phenomena. More work is
also needed on the basic conceptual issues involved,
in particular, on our understanding of terms such as
attention and awareness. However, much exciting
progress is being made on these fronts. And some
of the most powerful sources of these new develop-
ments are the phenomena of CB and IB.
See also: Attention, Awareness, and Neglect (00005);
Attention: Selective Attention and Consciousness
(00007); Perception: Unconscious Influences on
Perceptual Interpretation (00060).
Suggested Readings
Brockmole JR and Henderson JM (2005) Object
appearance, disappearance, and attention prioritization in
real-world scenes. Psychonomic Bulletin & Review 12:
1061–1067.
Irwin DE (1991) Information integration across saccadic eye
movements. Cognitive Psychology 23: 420–456.
Levin DT (2002) Change blindness blindness as visual
metacognition. Journal of Consciousness Studies 9:
111–130.
Luck SJ and Vogel EK (1997) The capacity of visual working
memory for features and conjunctions. Nature 390:
279–280.
Mack A and Rock I (1998) Inattentional Blindness.
Cambridge, MA: MIT Press.
Moore CM and Egeth H (1997) Perception without attention:
Evidence of grouping under conditions of inattention.
Journal of Experimental Psychology: Human Perception
and Performance 23: 339–352.
Most SB, Scholl BJ, Clifford E, and Simons DJ (2005) What
you see is what you set: Sustained inattentional blindness
and the capture of awareness. Psychological Review 112:
217–242.
Neisser U and Becklen R (1975) Selective looking: Attending
to visually significant events. Cognitive Psychology 7:
480–494.
Rensink RA (2002) Change detection. Annual Review of
Psychology 53: 245–277.
Rensink RA, O’Regan JK, and Clark JJ (1997) To see or not
to see: The need for attention to perceive changes in
scenes. Psychological Science 8: 368–373.
Simons DJ (ed.) (2000) Change Blindness and Visual
Memory. Hove, East Sussex, UK: Psychology Press.
Simons DJ and Chabris CF (1999) Gorillas in our midst:
Sustained inattentional blindness for dynamic events.
Perception 28: 1059–1074.
Simons DJ and Rensink RA (2005) Change blindness: Past,
present, and future. Trends in Cognitive Sciences 9:
16–20.
Wilken P (ed.) (1999) Mack & Rock: Inattentional blindness.
Psyche 5–7. http://psyche.cs.monash.edu.au/symposia/
mack_rock/.
Wolfe JM (1999) Inattentional amnesia. In: Coltheart V (ed.)
Fleeting Memories, pp. 71–94. Cambridge, MA: MIT
Press.
Attention: Change Blindness and Inattentional Blindness 13
