What is change blindness?
Change blindness is a phenomenon in visual perception in which very large changes occurring in full view in a visual scene are not noticed.
Recently a number of studies have shown that under certain circumstances, very large changes can be made in a picture without observers noticing them. What characterizes the experiments showing such "Change Blindness" in visual scenes is the fact that the changes are arranged to occur simultaneously with some kind of extraneous, brief disruption in visual continuity, such as the large retinal disturbance produced by an eye saccade, a shift of the picture, a brief flicker, a "mudsplash", an eye blink, or a film cut in a motion picture sequence. These phenomena are attracting an increasing amount of attention from experimental psychologists and from philosophers, because they suggest that humans' internal representation of the visual world is much sparser than usually thought.
Experimental work on change blindness:
In the first experiments that triggered interest in Change Blindness (McConkie & Currie [1996]), observers viewed high-resolution, full-colour everyday visual scenes presented on a computer monitor, while their eye movements were being measured. The computer could make changes in the scene as a function of where the observer looked. For example, when the observer looked from the door of a house to the window, say, the window (or some other element of the scene such as the sky, or the car parked in front of the house) changed in some way: it could disappear, be replaced by a different element, change colour, change position, etc.
It was found that when the change occurred during an eye movement, surprisingly large changes could be made without observers noticing them. Elements of the picture that occupied as much as a fifth of the picture area would not be seen. At first, the explanation of the phenomenon was assumed to have something to do with the mechanisms the brain uses to combine information from successive eye fixations to form a unified view of the visual world. In particular, every time the eye moves, the retinal image shifts. Some mechanism in the brain may correct for such shifts in order to create a stable view of the world. However the mechanism could be imperfect and not take into account certain differences in the visual content across the shift, thereby explaining why changes made during saccades might sometimes go unnoticed.
But a subsequent set of experiments showed that in fact, the change blindness phenomenon was not specifically related to eye movements. Rensink, O'Regan, & Clark [1997] used what they called the "flicker" technique, in which, instead of an eye movement, a brief flicker was introduced between successive images. A first picture (picture A) would be shown for, say, 250 ms, followed by a modified picture (picture B). Inbetween A and B, a brief blank screen (bl) would be shown. This would cause a flicker, lasting about 80 ms, that is, a duration similar to that of an eye movement. The cycle A-bl-B-bl-... was then repeated. Observers were told that something was changing in the picture every time the flicker occurred, and they were asked to search for it.
Under conditions where no flicker was inserted inbetween the pictures (A-B-A-B-…) the change was immediately visible and totally obvious
Experimental work on change blindness:
It was found that when the change occurred during an eye movement, surprisingly large changes could be made without observers noticing them. Elements of the picture that occupied as much as a fifth of the picture area would not be seen. At first, the explanation of the phenomenon was assumed to have something to do with the mechanisms the brain uses to combine information from successive eye fixations to form a unified view of the visual world. In particular, every time the eye moves, the retinal image shifts. Some mechanism in the brain may correct for such shifts in order to create a stable view of the world. However the mechanism could be imperfect and not take into account certain differences in the visual content across the shift, thereby explaining why changes made during saccades might sometimes go unnoticed.
But a subsequent set of experiments showed that in fact, the change blindness phenomenon was not specifically related to eye movements. Rensink, O'Regan, & Clark [1997] used what they called the "flicker" technique, in which, instead of an eye movement, a brief flicker was introduced between successive images. A first picture (picture A) would be shown for, say, 250 ms, followed by a modified picture (picture B). Inbetween A and B, a brief blank screen (bl) would be shown. This would cause a flicker, lasting about 80 ms, that is, a duration similar to that of an eye movement. The cycle A-bl-B-bl-... was then repeated. Observers were told that something was changing in the picture every time the flicker occurred, and they were asked to search for it.
Under conditions where no flicker was inserted inbetween the pictures (A-B-A-B-…) the change was immediately visible and totally obvious
However with the flicker, it was often extremely difficult to locate the change. This was particularly true for changes which concerned aspects of the scene which were not of "central interest". For example, the reflection of houses in a lake scene, though occupying a very large part of the picture, would not be considered to be what the picture was about. Observers sometimes were unable to see such changes at all, even after searching actively for as long as one minute. On the other hand, the changes were perfectly visible once they were pointed out to observers.
The flicker technique is very easy to implement using widely available computer software (for example software to make video presentations), and so lends itself to easy experimentation. Pictures as well as symbolic or text material can be used. The timing of the flicker between original and modified images is not critical.
What is important about the flicker technique is that it shows that change blindness can be obtained without the change being being synchronized with eye movements. This shows that in the earlier experiments where changes were synchronized with eye movements, the inability to detect the change was probably not not specifically related to the eye movement and to the mechanisms that the brain uses to combine successive images of the world during eye explorations.
Following the discovery that CB was not specifically related to eye movements, but to the brief disruption that is inserted between the two versions of the picture, considerable interest in the phenomenon developed, and a large number of further experiments have been performed. These can be classified according to the nature of the disruption that is used between successive images: global disruptions, local disruptions, and progressive changes. A review of different change blindness experiments can be found in Simons & Levin [1997].
Global disruptions are ones in which the picture change is accompanied by a disruption which covers the whole area of the picture. The experiments where the changes occurred during eye movements were global disruption experiments, since the whole retinal image is completely smeared during the time of approximately 20 to 80 ms that it takes for the eye to move from one fixation point to the next. The flicker experiments are also global disruption experiments, since the blank displayed briefly between original and modified images covers the whole picture.
Other examples of experiments with global disruptions are experiments involving eye blinks, picture shifts, and film cuts. In the blink experiments, observers' eye blinks, registered by online computer monitoring, are used to trigger the picture change. The blink produces a global disruption similar, though somewhat longer in duration, to the disruption caused by an eye movement. In the picture shift experiments, a picture is suddenly shifted in position, and a change made at the same time. Here a global disruption is caused by the retinal smearing that accompanies the eye movement that observers make to refixate the shifted picture.
The flicker technique is very easy to implement using widely available computer software (for example software to make video presentations), and so lends itself to easy experimentation. Pictures as well as symbolic or text material can be used. The timing of the flicker between original and modified images is not critical.
What is important about the flicker technique is that it shows that change blindness can be obtained without the change being being synchronized with eye movements. This shows that in the earlier experiments where changes were synchronized with eye movements, the inability to detect the change was probably not not specifically related to the eye movement and to the mechanisms that the brain uses to combine successive images of the world during eye explorations.
Following the discovery that CB was not specifically related to eye movements, but to the brief disruption that is inserted between the two versions of the picture, considerable interest in the phenomenon developed, and a large number of further experiments have been performed. These can be classified according to the nature of the disruption that is used between successive images: global disruptions, local disruptions, and progressive changes. A review of different change blindness experiments can be found in Simons & Levin [1997].
Global disruptions are ones in which the picture change is accompanied by a disruption which covers the whole area of the picture. The experiments where the changes occurred during eye movements were global disruption experiments, since the whole retinal image is completely smeared during the time of approximately 20 to 80 ms that it takes for the eye to move from one fixation point to the next. The flicker experiments are also global disruption experiments, since the blank displayed briefly between original and modified images covers the whole picture.
Other examples of experiments with global disruptions are experiments involving eye blinks, picture shifts, and film cuts. In the blink experiments, observers' eye blinks, registered by online computer monitoring, are used to trigger the picture change. The blink produces a global disruption similar, though somewhat longer in duration, to the disruption caused by an eye movement. In the picture shift experiments, a picture is suddenly shifted in position, and a change made at the same time. Here a global disruption is caused by the retinal smearing that accompanies the eye movement that observers make to refixate the shifted picture.
In film cut experiments, observers view motion picture extracts, and at the moment when the camera 'cuts' from one view to another, a large change is made -- for example an actor is replaced by a different actor. The camera cut produces a global disruption similar to the blank in the flicker experiments.
An additional, amusing, variant of the experiments with global disruptions are experiments in which the change occurs in real life. In a typical scenario described by Simons & Levin [1998], the experimenter stops a person in the street and asks for directions. While the person is speaking to the experimenter, workers carrying a door pass between the experimenter and the person, and an accomplice takes the place of the experimenter. The person usually goes on giving directions after the interruption, and very often does not notice that the experimenter has been replaced by the accomplice.
Change blindness experiments with local disruptions are experiments in which, at the moment of the change, five or six small, localized disturbances are superimposed on the picture, like mudsplashes on a car windscreen (O'Regan, Rensink, & Clark [1996]. The disturbances can be small in comparison to the size of the change and they need not coincide with the location of the change: the change takes place in full view. As for change blindness with global interruptions, changes are very often not noticed.
Experiments with slow changes are experiments in which there is no local or global disruption at all. Instead, the change is made so slowly that the attention-grabbing processes that would normally cause attention to be attracted to the change location can no longer operate. Again, it is found that in many cases, changes are hard to detect [Quicktime video: http://nivea.psycho.univ-paris5.fr/ECS/sol_Mil_cinepack.avi]
Related paradigms:
The change blindness phenomenon is strongly related to a well-established line of research in experimental psychology that started in the 1970's and concerns visual visual short term memory (for review cf. Haber [1983]).
In this literature, experiments analogous to the change blindness experiments had been performed using briefly displayed arrays of simple elements such as letters. These experiments showed that although observers have the impression of seeing all the letters in, say, a 12-element array, in fact they notice changes or report the identity of only about four or five letters. It appears that there is a kind of attentional "bottleneck" which limits information transfer into memory: only a fraction of the information available in a scene is transferred into visual storage for later report or comparison. Further work additionally showed that the code in which the information is stored in visual visual short term memory is not a visual code, but a code in which only the category or identity of the elements is available. This work was also coherent with another line of research showing that information from successive eye fixations is combined only in categorical form, and not as a picture-like composite image (for review cf. Irwin & Andrews [1996]).
We shall see in the next section that the conclusion from these experiments, showing that visual storage is sparse and categorical, is also applicable to the change blindness results. Because in the case of change blindness, natural, highly detailed visual scenes are used as stimuli, the conclusion is all the more striking than it was in the older literature using simple stimuli.
Change blindness, in addition to links with research on visual short term memory, also has relations with several more recent lines of research showing that attentional capacity in short term visual processing is severely limited, both in spatial extent, and in the way it extends over time. Thus in "inattentional blindness" (Mack & Rock [1998]), observers do an attention-demanding visual task. At a given moment, a large, unexpected visual event takes place. Even though such an event would be totally obvious under normal circumstances, and even though the event takes place in full view, it is often not noticed. For example Simons & Chabris [1999] used a task in which observers look at a film of two groups of players, a black-clad and a white-clad group, each playing with their own ball in the same small room. The observer’s task is to try to track the number of times one group exchanges the ball. While the observer is doing this task a woman with an umbrella walks through the room, in full view. Observers often fail to notice this totally obvious event.
An example of a temporal restriction on the deployment of attention is the "attentional blink" (review cf. Shapiro & Terry [1998]): in this, an observer is required to identify a target letter in a stream of rapidly presented letters. It is found that the observer often fails to report the occurrence of a second target letter if the second target follows the first by less than about 450 ms: it is as though attention had to recover for a brief period after having been sollicited.
In "repetition blindness" (Kanwisher [1987]), a visual stimulus like a letter, symbol, picture, or word tends not to be noticed if it is the second of two identical occurrences of the item in a rapidly presented series.
Another field of research that has connections to change blindness is the extensive literature on memory and cognitive descriptions (review cf Pani [2000]). Part of the explanation for change blindness may reside in memory limitations rather than in perceptual limitations. If this is so, then we expect that change blindness will be affected by factors similar to factors that affect memory. This is compatible with the finding that changes made to elements in a scene which are of "central interest" will in general be easier to dectect than "marginal interest" changes. Other work has shown that variables like semantic coherence, observer familiarity, and task to be achieved, affect change blindness in a way similar to how they affect normal memory.
Theories of change blindness:
The currently accepted explanation of change blindness owes much to the work done in the 1960's and 70's showing how visual information is transferred via an attentional "bottleneck" to a very low capacity short term visual storage (e.g. Gegenfurtner & Sperling [1993]).
Within this context, the explanation of change blindness involves two components: a component related to what is called "visual transients", and a component related to the way a scene is encoded in memory.
Visual transients are fast changes in luminance or colour in the retinal image, such as would be produced by a sudden appearance or disappearance, or through motion of an element of the scene. It is known that such transients are detected in the first levels of the visual system, and that attention is automatically attracted to the location where they occur.
Under normal viewing conditions therefore, when a change occurs, it produces a visual transient which attracts attention to the change location. The transient thus provides information that a change has occurred, and it says where it occurred, but it does not provide information about what the change was.
In order for an observer to be able to determine what the change was, he or she will have had to have encoded into visual memory what was at the change location before the change occurred, and compare it to what is there after the change.
There are thus two things that can go wrong in change detection: either the transient that attracts attention to the change location may be interfered with (thereby causing a deficit in detection that or where the change has occurred), or the encoding and comparison process may be interfered with (causing a deficit in determining what the change was).
Both these mechanisms may be at work in the change blindness experiments. In the paradigms using global disruptions, like the flicker, blink, and film cut experiments, the global disruption presumably creates a large number of transients all over the picture, which mask or compete with the local transient corresponding to the sought-for change, and which prevent attention being automatically drawn to it. The change will only be immediately noticed if an observer happens to have been attending to the changing element at the moment it changes. Failing this, in order to find a change, the observer must search through the scene looking for an element which is different from what was previously encoded about the scene. However, because of the limitations in short term visual memory, very little of the scene is likely to have been previously encoded, and the chances of success are very limited.
In change blindness paradigms using local disruptions, the situation is very similar, with the difference that the local transient corresponding to the change location is missed by observers, not because it is swamped by a global transient, but because the mudsplashes act as "decoys", attracting attention to locations other than the true change location.
In change blindness paradigms with slow changes, the change occurs so slowly that no local transient is generated. Attention is thus not attracted to the change location, and again, the observer must rely on the very sparse information that he or she has encoded about the scene in order to locate the change.
Whereas researchers working in change blindness will broadly agree on the explanation just outlined of the phenomenon, further work is necessary to ascertain the relative roles of the different component mechanisms involved. To what extent does the flicker in the flicker paradigm act to mask or "wipe out" the internal representation? Or does it act essentially like the mudsplashes in the mudsplash paradigm to create local transients that act as decoys? Exactly how much information is encoded concerning the initial and final views of the scene? Is the overall "gist" of a scene coded in some way? Does what is encoded depend on the observer's attentional state, on the task, on viewing strategies, on the semantic relation between the gist of the scene and the element that is changing? Are certain aspects of elements (their layout? their colour?) automatically encoded and easier to detect when they change? Even if little information is available to make conscious judgments about display changes, could it be that some information is retained unconsciously? A number of recent lines of research are investigating these issues (cf. Bibliography).
Relevance of change blindness for consciousness and cognitive science:
Change blindness raises an important question: If the information that is encoded about a visual scene is so sparse, how is it that we have the subjective impression of visual richness, that is, of seeing everything there is to see in our field of view, so to speak in "glorious technicolor and cinemascope"?
Perhaps the most natural view to take is to suppose that what we have the subjective impression of seeing is not the very sparse, more semantically coded, content of visual memory, but the content of a shorter-lived but higher quality, image-like replica or "icon" of the visual scene. The impression of richness that we have from the world would derive from this high-quality icon. On the other hand, only a small portion of the icon's contents, namely the parts that have been attended to, would at any moment be transferred into memory and be available for doing tasks like change detection -- the rest would be forgotten. This view of visual processing has been called "inattentional amnesia" (Wolfe [1999]): the idea is that we see everything, but forget most of it immediately.
The notion that what undelies the richness of vision is a high-quality internal replica of the outside world underlies some of the current work in neurophysiology and neuroanatomy, where cortical sites are being sought which provide the "neural correlate of consciousness". Indeed area V1 of the visual cortex contains a distorted map of the visual field which might be a plausible locus for visual consciousness, possibly in relation to other brain structures.
A more radical answer to the question of why we have the impression of continuously seeing everything in our visual field has also been suggested (O'Regan & Noë [2001]). The idea is that in fact the experience of seeing does not derive from the activation, inside the brain, of an "icon" of the outside world. Rather, the experience of seeing is somewhat like the temporally extended, multifacetted experience of driving a car, involving a kind of "give and take" between the observer and the environment, a kind of attunement to the laws that link the observer's actions to the changes in sensory input.
Under this view, the outside world serves as a form of "external memory". Only those aspects of the environment that are currently being "visually manipulated", are actually available for conscious processing at a given moment. We have the impression of seeing everything because we know we have access to everything, even though without actually accessing something, no detailed information is available about it. This explains the apparent paradox between the feeling of richness we have of our visual environments, and our striking inability, in change blindness experiments, of knowing what has changed.
References:
Gegenfurtner, K. R., & Sperling, G. (1993). Information transfer in iconic memory experiments. Journal of Experimental Psychology: Human Perception & Performance, 19(4), 845-866.
Haber, R. N. (1983). The impending demise of the icon: A critique of the concept of iconic storage in visual information processing. Behavioral and Brain Sciences, 6, 1-54.
Irwin, D. E., & Andrews, R. V. (1996). Integration and accumulation of information across saccadic eye movements. In T. Inui & J. L. McClelland (Eds.), Attention and performance XVI: Information integration in perception and communication (pp. 125-155). Cambridge, MA, USA: Mit Press.
Kanwisher, N. G. (1987). Repetition blindness: Type recognition without token individuation. Cognition, 27(2), 117-143.
Mack, A., & Rock, I. (1998). Inattentional blindness. Cambridge, MA, USA: The Mit Press.
McConkie, G. W., & Currie, C. B. (1996). Visual stability across saccades while viewing complex pictures. Journal of Experimental Psychology: Human Perception & Performance, 22(3), 563-581.
O'Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24(5), in press.
O'Regan, J. K., Rensink, J. A., & Clark, J. J. (1996). "Mud splashes" render picture changes invisible. Invest. Ophthalmol. Vis. Sci., 37, S213.
Pani, J. R. (2000). Cognitive description and change blindness. Visual Cognition, 7(1/2/3), 107-126.
Rensink, R. A., O'Regan, J. K., & Clark, J. (1997). To see or not to see: the need for attention to perceive changes in scenes. Psychological Science, 8(5), 368-373.
Shapiro, K., & Terry, K. (1998). The attentional blink: The eyes have it (but so does the brain). In R. D. Wright (Ed.), Visual attention (pp. 306-329). New York, NY: Oxford University Press.
Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: sustained inattentional blindness for dynamic events. Perception, 28(9), 1059-1074.
Simons, D. J., & Levin, D. T. (1997). Change Blindness. Trends in Cognitive Sciences, 1(7), 261-267.
Simons, D. J., & Levin, D. T. (1998). Failure to detect changes to people in a real-world interaction. Psychonomic Bulletin and Review, 5(4), 644-649.
Wolfe, J. M. (1999). Inattentional amnesia. In V. Coltheart (Ed.), Fleeting memories . Cambridge, Mass.: MIT Press.
Bibliography:
Coltheart V. (Ed.). (1999). Fleeting memories: Cognition of brief visual stimuli. Cambridge, MA: The Mit Press.
O'Regan, J. K. (in press). Thoughts on change blindness. In L. R. Harris & M. Jenkin (Eds.), Vision and Attention. . Berlin: Springer.
Pashler, H. E. (1998). The psychology of attention. Cambridge, MA, USA: The Mit Press.
Special issue of Visual Cognition On Change Blindness (2000), 7
Dan Simons' change detection database: www.wjh.harvard.edu/~viscog/change/Home.html
Blogger Comment
Facebook Comment