I have eyes only for you: Inattentional blindness reviewed for meaningful stimuli
Sameer Joshi
University of Central Florida

Abstract
It is now a well-studied fact that we are blind to many stimuli in our daily experience, which may be in plain sight. This has been studied under the rubric of Inattentional Blindness (Mack and Rock, 1998). Though this phenomenon is generally attributed to attention, we explore the role of meaningfulness in causing it. We correlate inattentional blindness with the importance of the stimuli in light of schemas. The main questions we propose are, if an unattended stimulus is meaningful to the observer, Will the stimulus pop out, or will the observer be blind to it just like any meaningless stimulus? And if so, how much can it be degraded before it is indistinguishable from a meaningless stimulus? We found that the meaningfulness of a stimulus is indeed an important factor in noticing it.

Introduction
There is a disparity between the feature rich world around us and our representation of it. Sometimes, we completely ignore a stimulus that is easily visible to us. For example, ignoring a friend waving at us in plain sight in a movie theatre. This is quite perplexing as it suggests that we do not observe the world around us flatly. Rather dramatic studies of this phenomenon have been made of this phenomenon over the past 26 years, which include observers failing to perceive a gorilla in the middle of a ball game (Simons and Chabris 1999), and not seeing a woman with an umbrella in the middle of a basketball game (Neisser 1979). Simons (2000) observes that attentional capture may not be caused by the abrupt onset of some unexpected stimuli. This phenomenon has been studies under inattentional blindness, change blindness, selective looking, and inattentional amnesia.
At least four factors can be conjectured to be involved in this phenomenon. The first is memory. It can be reasoned that the unexpected event was viewed bur promptly forgotten. There is evidence disfavoring this view of “inattentional amnesia” (Wolfe 1999), as Mack and Rock (1998) have shown. Also, the forgetting would have to be inseparable from perception in order for this view to hold. In the experiments conducted by Simons and Chabris (1999), the subjects noticing the unexpected events displayed external signs of recognition, while others did not.
The second is attention. Mack and Rock (1998) assert that there is no perception without attention. The failure to observe a stimulus can be attributed to the attention being tied up in another task. This observation seems intuitive and is supported by multiple other findings. However, it is unlikely that attention is the sole cause, for that would not explain why sudden visual onsets grab attention (as observed by Braun, 2001).
Expectation may also play a role in this affair. Mack and Rock (1998) assert that under the right conditions, a lack expectation is a sufficient condition to yield inattentional blindness. Braun (2001) also suggests that a ‘pure’ manipulation of expectation would produce blindness whereas a ‘pure’ manipulation of attention will not. Observers do not see unanticipated objects and events in both computer and video based studies (Simon and Chabris, 1999).
The fourth factor is meaningfulness. The experiments by Mack and Rock (1998) have shown that if the unexpected stimulus is one’s own name, it is perceived even under conditions of inattention. The work on selective looking (Neisser and Becklen 1975) also points to the semantic importance of the unexpected stimuli being a factor in their perception. Their findings suggest that we attend to things that fit our schema. The cocktail party effect (Cherry 1953) also accentuates the importance of meaningfulness.
The third and fourth causes may be closely related to schemas, as suggested by Neisser and Becklen (1975). The findings of Simons and Chabris (1999) can also be interpreted to support the schema theory. They found that sustained inattentional blindness occurs more often if the display is transparent. Since transparent images do not easily fit into out schemas this would be consistent with the schema theory. The schema explanation would also work for change blindness.
Also, Most, Simons, Scholl, Jimenez, Clifford and Chabris (2001) have found that objects that are similar to the ongoing dynamic events are more likely to be noticed. This explains the earlier findings of Simons and Chabris (1999), where a woman with an umbrella was more visible than a gorilla.
Of all these factors, meaningfulness is amongst the least studied. Some work has been done by Mack and Rock (1998), in which they show that an observers’ own name is less prone to inattentional blindness. According to the “early selection” theory, the name is a special case, and does not require much information for its perception. According to the “late selection” theory the attentional gate occurs late, after the meaning has been processed. Testing subjects with images important or meaningful to them, other than their names, may provide further evidence in this matter. Mack and Rock (1998) have done some work in this direction, testing subjects with images and faces. However, no work has been done testing this in a dynamic environment.
Mack and Rock (1998) also found that distorting the observers’ name even by one letter led to a marked decline in the rate of noticing. This suggests that the meaningfulness of a stimulus shows strong negative correlation with its distortion. This again, has not been repeated in a dynamic environment. Also, these findings have not been extended to important meaningful stimuli other than names.
The main questions I propose to address can be framed thus. What if the unexpected stimulus in the background is meaningful or important to the observer? Will the stimulus “pop-out” then? And if so, how much can it be degraded before it is indistinguishable from an unrelated stimulus? It is expected that the findings for the above will show that semantically important stimuli indeed stand out.

Methodology

Participants
A group of 160 observers was selected from the married graduate-student population. Also, unknown to the participant, his/her spouse too was recruited in the experiment. Care was taken that they had been married for at least two years. The observers were in the age group 24-30. Both the participants and their spouses were financially compensated.

Materials
The films were divided into three categories. The first category, the distracter tasks, consisted of two films of 100 seconds each. One of them consisted of six people playing a game of catch with an ordinary football. The second film consisted of people walking across the screen in either direction. On an average, there were 2-3 people on the screen at a time. There were never more than five people on the screen and neither was the screen empty at any time. The second category was the unexpected stimuli. This consisted of one film of a woman with an umbrella walking across the screen. The woman appeared on the screen holding an open umbrella over her head, walked around for some time before walking off. The entire event lasted for about 5 seconds. The third category was the meaningful unexpected stimuli. This consisted of the observers’ spouse walking across the screen. A separate film was made for each observer. The spouse walked around the screen for approximately 5 seconds before walking off.
To prepare the films for final display, the films from category one were merged with one from either category two or category three. The two distracter tasks were superimposed. After about 50 seconds of play, the third tape was superimposed over the first two. Next, the films from category three were blurred to various degrees and were used as usual. Two forms of blurring were used. In the first case, the bottom half of the face was blurred and the rest was left as is. In the second case, the entire face was blurred, but not to a degree so as to make it unrecognizable.
Like Simons and Chabris (1999), two styles of video were used, transparent and opaque. In the first style, all the films were transparent. In the opaque style, a more natural means was adopted. The figures partially occluded each other as if all of them were present on the screen at the same time.

Apparatus
Video cameras were used to record the films and video-editing software was used to superimpose them onto each other. Standard video editing software was also used to blur the images in a controlled manner. The films were displayed to the observers on a standard 18” screen.

Procedure
All observers were tested individually. The observers were divided into four groups. Group-A was shown a film with the woman with an umbrella. Group-B was shown the films with their spouses. The observers in group-C were also shown films with their spouses in them, except that the bottom halves of their spouses’ faces had been blurred. Group-D was shown the film with the spouses’ face blurred entirely.
All groups were asked to keep count of the passes in the game of catch. At every catch, the observers also had to press a button. This was done to ensure honesty.
After viewing the videotape the observers were asked answer a questionnaire. The first question asked them to write down the number of passes counted. The second question asked them if they had observed anything unusual about the game. If yes, then they were asked to write down what it was. At this point the observers were asked if they had participated in any similar experiment before. If they had, the data collected from them was discarded from the final result.
Ten experimenters tested the observers. A common protocol for debriefing was used to ensure uniformity of testing procedures.

Results
Data from fifteen observers was discarded because either the observer had heard of the phenomenon before, or his/her responses were not clear.
Of all the observers 82% noticed the unexpected stimulus, and 18% failed to do so.
The unexpected event was observed the least in Group-A. Only 55% of the observers noticed the woman with the umbrella in the transparent condition, and 91% in the opaque condition.
Observers in Group-B noticed the unexpected event better than all other groups. 89% of the observers noticed the unexpected condition in the transparent condition. All the observers noticed their spouses in opaque condition without fail. In fact, the unexpected event was accompanied by a sudden capture of attention by the background task. In many cases the observers showed physical signs of shifting attention. This is why, unlike the earlier studies by Simons and Chabris (1999), button pressing was an essential ingredient of this study. In many cases the attention shifted completely and the observer lost count of the number of passes. Most observers reported the event without being asked.
Group-C showed strong decline in noticing the unexpected event with even a little amount of distortion. Though all the observers reported the unexpected stimuli in the opaque case, only 64% noticed something out of the ordinary in the transparent case. However, the performance of Group-C was still better than group-A and Group-D.
Group-D’s performance was significantly lower than Group-C, but still higher than Group-A. 97% of the observers noticed their spouses in the opaque case and 60% in the transparent case.
The unexpected event was noticed more in the opaque case across all three groups. 67% of all observes noticed the unexpected event in the transparent case as opposed to 97% in the opaque case.

Table 1.
Percentage of observers noticing the unexpected event.
Group-A Group-B Group-C Group-D
Transparent 55 89 64 60
Opaque 91 100 100 97

Discussion
This study brings out the startling extent to which our schemas modify our perception. There is a heavy bias towards noticing meaningful stimuli as opposed to semantically neutral ones. The findings replicate the IB results by Simons and Chabris (1999) for meaningless stimuli. Additionally, this study generalized and extended the findings of Mack and Rock (1998) in showing that meaningfulness is an important factor in perception, and the distortion of stimuli leads to a sharp decline in the capture of attention due to meaningfulness. The subjects displayed significantly lower thresholds for meaningful stimuli than for visually salient but meaningless stimuli. In fact, even if the meaningful stimuli were distorted, they were still more readily noticed than meaningless stimuli. This effect declined sharply as the stimuli were distorted further.
The results were more pronounced for opaque displays than for transparent displays. This finding can be attributed to visual recognition or the schema theory. Schema theory would suggest that this effect is due to the unnaturalness of the transparent display, which does not fit any natural schemas. Alternatively, it can be said that this effect is a by-product of the transparent display condition and is more due to the manner in which we perceive visually. In other words, it is more of a perceptual phenomenon than a cognitive one. However, as observed by Simons and Chabris (1999), this explanation does not explain the fact that even in the opaque display a large number of observers were blind to plainly visible but unexpected stimuli. We hold that the schema theory, on the other hand, is a more complete explanation as it explains both ends of the phenomenon.
This study also brought out the dramatic effects of distortion of meaningful stimuli. The capture of attention by the meaningfulness of stimuli dropped sharply as the stimuli were distorted. This effect is similar to but not as pronounced as in the recognition of one's own name, as studied by Mack and Rock (1998). This fact may be attributed to the word superiority effect. However, it is still surprising that a little amount of distortion can drop the rate of noticing by a large amount. The figures were not highly distorted. All the subjects in Group-C and Group-D were shown the film of their spouse alone after the experiment had been conducted. None of them had any trouble recognizing their spouse despite the distortion. The non-linear relationship between the amount of distortion and the decline in noticing the stimuli suggests that meaningfulness depends highly on the clarity of the image.
The similarity of the unexpected stimuli to the attended stimuli is also a factor in attentional capture, as proposed by Most, Simons, Scholl, Jimenez, Clifford and Chabris (2001). They found that if the unexpected stimulus is more like the attended stimuli, it is more likely to capture attention. This factor was not an important one in our study as all the unattended stimuli were similar to the distracter task.
Much of the previous work in similar experiments lays emphasis on attention as the prime factor in this phenomenon. Inattentional blindness may not be affected by attention as much as previously thought. This study shows that even with a lack of attention, unattended stimuli can be captured and processed consciously if they are meaningful enough. But as Braun (2001) observes, expectation plays an important part. Furthermore, he asserts that a pure manipulation of expectation would produce blindness, whereas a pure manipulation of attention would not. Though attention is undoubtedly highly important, our findings support Braun’s conjecture. Mack and Rock (1998) identify two factors in inattentional blindness, attention and expectation. Since the observers in all the groups were tested over the condition of undivided attention alone, it is safe to conclude that expectation played a major role in the disparity observed between the various Groups.
The results of this study favor the late selection models of attention. The high rate of noticing unattended stimuli, and the apparentness of semantically meaningful stimuli suggests that even unattended input is processed to a cognitive level.
Our findings go against the notion of inattentional amnesia. Many of the observers in group-B remembered the unexpected event without being asked. Overall, across all groups, if the observers did not notice the stimulus they did not give any outward signs of observing it either.
These findings have important practical implications for design of interfaces. The role of IB in display of information has been explored earlier. For example, Haines (1989) performed a study in which pilots failed to notice another plane on the runway when the information was superimposed on the windshield. Our findings can help further refine such human-interface design.

Directions for future work
The role of familiarity versus meaningfulness remains to be completely explored in this area. It is not very clear if the lowering of thresholds for certain stimuli is due to their familiarity or their meaningfulness. Mack and Rock (1998) have discovered that in that case of static displays of unexpected objects, the subjects displayed inattentional blindness towards familiar but meaningless objects.
Other important questions are raised by the findings of this paper. What makes a stimulus important, is it just meaningfulness or does it have to be personally important to the observer? The fine distinction of what causes something to be semantically important has yet to be made.
More work is needed to understand how all the factors attributed to causing Inattentional Blindness integrate with the schema theory. This may be particularly rich field as, according to Neisser and Becklen (1975), not only do we exhibit selective looking, but we also invent things to fit in with our schema. This can be further explored by performing similar studies with children, who have very different schemas from adults.

References
Braun, J. (2001). It’s great but not necessarily about attention. Psyche, 7(06).
Mack, A., & Rock, I. (1998). Inattentional Blindness. Cambridge, MA: MIT press.
Most, S.B., Simons, D.J., Scholl, B.J., Jimenez, R., Clifford, E., & Chabris, C.F. (2001). How not to be seen: The contribution of similarity and selective ignoring to sustained inattentional blindness. Psychological science, Vol 12, no. 1, 9-17.
Simons, D.J. & Chabris, C.F. (1999). Gorillas in our midst: Sustained inattentional blindness for dynamic events. Perception, 28, 1059-1074.
Simons, D.J. (2000). Attentional capture and inattentional blindness. Trends in Cognitive Sciences, 4(4), 147-155.
Wolfe, J. M. (1999). Inattentional amnesia, Fleeting Memories (pp.71-94). Cambridge, MA: MIT Press
Neisser, U., & Becklen, R. (1975). Selective looking: Attending to visually specified events. Cognitive psychology, 7, 480-494.
Cherry, E.C. (1953). Some experiments upon the recognition of speech, with one and two ears. Journal of the Acoustical Society of America, 25, 975-979.
Haines, R.F. (1989). A breakdown in simultaneous information processing. In Presobyopia Research: From molecular biology to visual adaptation. Eds G. Obrecht, L.W. Stark (New York: Plenum) pp 171-175