Change Blindness and Eyewitness Testimony: The Effects of

QuaesitUM QuaesitUM

Volume 3

Issue 1

Article 7

3-1-2016

Change Blindness and Eyewitness Testimony: The Effects of Change Blindness and Eyewitness Testimony: The Effects of

Instruction, Time Intermission, and Gender-matched Observers Instruction, Time Intermission, and Gender-matched Observers

Blake Mitchell

Follow this and additional works at: https://digitalcommons.memphis.edu/quaesitum

Recommended Citation Recommended Citation

Mitchell, Blake (2016) "Change Blindness and Eyewitness Testimony: The Effects of Instruction, Time

Intermission, and Gender-matched Observers,"

QuaesitUM

: Vol. 3: Iss. 1, Article 7.

Available at: https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

This Article is brought to you for free and open access by University of Memphis Digital Commons. It has been

accepted for inclusion in QuaesitUM by an authorized editor of University of Memphis Digital Commons. For more

information, please contact khgger[email protected].

174

Blake Mitchell graduated summa cum laude from the University of Mem-

phis in May 2015, obtaining a Bachelor of Arts degree in Psychology. As

a proud student at the UofM Lambuth campus, he was a member of the

Psychology Club and provided tutoring in a variety of subjects. Through-

out his academic discourse, he was absolutely fascinated with the science

concerning the brain—neuroscience. His interests sit on the frontier of

modern neuroscience: the mysteries surrounding the nature of conscious-

ness and the mechanics that store and retrieve our memories. He also takes

interest in quantum mechanics and how our brain functions at its smallest

scale. Currently, he plans to attend Vanderbilt University, where he hopes

to study cognitive neuroscience and earn his Ph.D. After this endeavor, he

dreams of procuring a career of emphatic devotion to scientic investiga-

tion and becoming a renowned interrogator of nature and an artful commu-

nicator of its reluctant testimonies. He is also the recipient of a Quaesitum

outstanding paper award.

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

175

Blake Mitchell

Change Blindness and Eyewitness Testimony: The Effects of

Instruction, Time Intermission, and Gender-matched Observers

Faculty Sponsor

Dr. Cheryl Bowers

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

176

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

177

Abstract

The purpose of this research was to further examine the parameters of

change blindness and its relevance to eyewitness identication by inves-

tigating the effects of instruction, time intermission, and gender-matched

observers. Participants were shown a short video depicting an opportu-

nity theft in which the identity of the burglar, and that of his accomplice,

changed throughout the lm. Half of the participants were given instruc-

tions to pay close attention, while the other half were not. All were tested

for awareness of change and content recall. As hypothesized, 82% of

participants did not notice substitutions among the perpetrators. Awareness

of change had some unexpected but interesting relationships with our oth-

er independent variables, the results of which are discussed in detail along

with corrective measures for future research. Collectively, the ndings of

this study delve deeper into the mechanics of change blindness, thereby

further illuminating the sophisticated interrelation between change blind-

ness and eyewitness testimony.

Note from author: This study was a collaborative effort to fulll requirements for a

PSYCH Stats & Research class at the University of Memphis Lambuth Campus. While

this paper is solely authored by me, the conception, preparation, and execution of its

contents are credited to each contributor equally: Naomi Adams, Mary Church, Kathryn

Cochran, Brittney Goode, Penny Hensley, Madison Kuykendall, Blake Mitchell, Brianna

Morton, Dianna Moss, Andrew Rush, Christy Williams, with faculty sponsor Dr. Cheryl

Bowers.

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

178

Introduction

Change blindness is a psychological phenomenon that occurs when

a change in a visual stimulus goes unnoticed by its observer. This phenom-

enon can be likened to continuity errors in movies, where an object might

erroneously change location or position from one cut to another (e.g. a

vase is knocked over in one frame but then is standing upright in the next).

In most cases, this change in the background, or even foreground, is not

often noticed by the audience until after it is pointed out to them; in psy-

chology, this is known as change blindness. In recent years, research has

been conducted on this phenomenon in efforts to better understand some

of its qualities. In one of the earliest studies involving change blindness

(Levin & Simons, 1997), participants watched a short video of a person

getting up to answer a telephone call. Just before the actor reaches the

phone, a change in camera positioning coupled with a lm cut allows the

actor to swap identities with a completely new actor. Even though these

two actors differed dramatically in appearance, most of the viewers did not

report noticing this rather substantial, but intentional, continuity error. The

ndings of this study raised awareness of change blindness and ignited a

great interest within the scientic community, especially from groups who

were particularly interested in selective attention and short term memory.

Researchers were eager to explore the boundaries of change blind-

ness. Some posited that the awareness of change existed in the observer’s

subconscious but simply wasn’t available for retrieval (Simons, et al.,

2002). With regard to subconscious retention, “the observer must form a

consciously accessible visual representation of the original stimulus that

can be tracked over time and form a basis for later recognition” (Most, et

al., 2005). Subsequent work began to address on the seeming interrelation

between change blindness and eyewitness testimony. The idea was that

this aw in our attenuation system had some relevance to the clockwork

behind faulty eyewitness testimony and the misidentication of perpetra-

tors. Recent research has illustrated this interrelation between the eyewit-

ness and change blindness literatures (Davies & Hine, 2007), identifying

multiple risk factors of change blindness. In their study, Davies and Hine

created a 2-min video clip depicting an opportunity theft from a student

house. Midway through the lm, the burglar’s identity changed (aided by

lming techniques). The participants in this study were split randomly into

two groups: the intentional group and the incidental group. Participants in

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

179

the intentional group were told to pay close attention to the contents of the

lm because they would be tested later on. Participants in the incidental

group were given an ominous description of the video along with a false

pretense for its purpose. The paradigm was set up this way in order to test

whether or not priming participants to focus and pay attention would have

a signicant effect on awareness of change rates and content recall scores.

In total, only 39% of all the participants noticed the burglar’s identity

change during the lm. Participants who noticed the change were, as de-

scribed by Davies and Hine, “drawn disproportionally from the intentional

condition, a highly signicant effect” showing that the priming received

by the intentional group was effective. Participants who were in the inten-

tional group also received signicantly higher content questionnaire scores

than did those in the incidental group. Lastly, in an effort to explicitly

demonstrate the interrelation between change blindness and eyewitness

testimony, Davies and Hine also created a photo lineup from which par-

ticipants were asked to identify the perpetrator. The results conrmed that

participants in the intentional condition were signicantly more likely to

correctly select both actors than were participants in the incidental group.

Davies & Hine report that “all the participants who recognized both actors

detected the change, whereas none of participants who did not notice se-

lected both burglars from the lineup.” From this research, one can reason-

ably conclude that priming participants to pay attention does in fact have a

positive effect on notice of change and content recall. Furthermore, Da-

vies and Hine adequately demonstrated a signicant relationship between

change blindness and eyewitness testimony by incorporating a photo

lineup in their paradigm. The current research wishes to build on these

ideas—further testing their validity—and also incorporate new indepen-

dent variables such as time intermission and gender-matched observers.

Method

Hypotheses and Design

The research detailed by this paper was modeled after the research

done by Davies & Hine (2007), including the creation of a video that

depicts an opportunity theft and a follow-up content questionnaire. In our

study, there were two main dependent variables: (a) awareness of change

and (b) memory score on the content questionnaire. In addition to Davies

and Hine’s burglar, we added another factor to the paradigm: a lookout

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

180

for the burglar. As Davies and Hine did, we hypothesized that those in

the intentional group would be more likely to notice the changes and less

prone to change blindness. Going beyond Davies and Hine’s research, we

also hypothesized that the interval between change, or the time intermis-

sion that occurred between the identity swaps, would have an impact on

awareness of change. Specically, participants were expected to identify

the change in the burglar more often than the lookout because the burglar

change happened within a few seconds. Alternatively, participants were

expected to fail to notice the change in the lookouts since approximately

two minutes elapsed between the switch. Also, beyond Davies and Hine’s

analysis, we intentionally selected male actors to play the burglar and fe-

male actresses to play the lookout. We hypothesized that male and female

participants would perform differently on their notice of change based on

the gender of the target. While expecting a difference, we did not predict

whether males would do better at noticing the male burglar or female

lookout identity change. Likewise, while expecting a difference, we did

not predict whether females would do better at noticing the change in the

male burglar or female lookout.

Participants

A community college sample of 55 participants were pooled to-

gether from four separate General Psychology classes. We designated two

of the classes to the intentional group, while the remaining two classes be-

came the incidental group. The groups were not perfectly equal in volume,

although they were relatively close, and they did not contain an equal ratio

of male to female participants (40 males and 15 females).

Materials

We lmed a 4-min video specically for this experiment that de-

picted an opportunity theft from a vacant house. The lm consists of four

actors: two males (burglar) and two females (lookout). The rst female

actress who played the lookout in the rst scene was of average build

with brown hair and a rounded face; she wore a dark shirt with bold white

letters on the front; she sat with casual posture and appeared tranquil. The

second female actress who played the lookout in the nal scene was of

slim build with brown hair and a narrow face; she wore a black shirt with

small letters on the front; she sat with poor posture and appeared worried.

The rst male actor who played the burglar on screen for the rst half

of the lm had a heavier build with dark brown hair and a rounded face;

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

181

he wore a black hoodie with blue jeans, green shoes, and carried a black

backpack; he moved at a slow pace and conducted himself calmly. The

second male actor who played the burglar on screen for the second half of

the lm had an average-to-heavy build with light brown hair and a round-

ed face; he wore a black hoodie with black pants, blue shoes, and carried

the exact same backpack; he moved with haste and was in much more of a

hurry than the rst actor.

The lm opened with a long shot of a white SUV cruising up the

street toward the camera. A cut is made to the driver’s side of the parked

vehicle, in which Lookout 1 is seen glancing down at her phone. The lm

then shows Burglar 1 walking down the street toward a house. The bur-

glar uses a crowbar to enter the house and begins searching for valuables.

He searches the living room and nds a camera, a laptop, and a gaming

controller. Burglar 1 then walks into the kitchen and disappears around a

corner. The camera then changes position to the inside of a bedroom where

Burglar 2 is seen entering. He searches the room and nds an iPad, a stack

of cash, jewelry, and headphones. He glances at his phone for a moment

and then leaves the room and exits the house. The lm cuts to an outside

view of the burglar leaving and pans over to the driver’s side of the white

SUV, where Lookout 2 is seen watching the burglar approach her. The

burglar proceeds to load the vehicle with the stolen items and drives off

with the lookout. The female lookout actresses were on screen for approx-

imately 20 seconds each and the male burglar actors were on screen for

approximately 90 seconds each.

We created a content questionnaire that was prefaced by open-end-

ed questions about what the participant saw. The rst questions, (“Did you

notice anything unusual about the burglar throughout the lm? If so, de-

scribe.”) and (“Did you notice anything change about the burglar through-

out the lm?”) probed for evidence of awareness of changed identity

with regard to the male burglar. This was followed with a set of mirrored

questions related to the female lookout. The rest of questionnaire featured

questions that tested content recall such as (“In the blanks below, please

list 6 items that you remember being stolen in the lm), which we convert-

ed into a memory score.

Procedure

The participants were tested one class at a time by two experiment-

ers: one to handle materials and one for presentation. When the intentional

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

182

condition was being tested, the participants were instructed to pay close

attention to the lm because there would be a content questionnaire to ll

out afterwards. Participants in the incidental condition were simply given

an ominous description of the lm they were about to watch: “A short lm

depicting an opportunity theft that conveys the importance of increased

campus security.” After the lm, all participants were given the same con-

tent questionnaire to complete. The experimenters stressed that everyone

should work on their own and remain silent until everyone had completed

the questionnaire. As a precaution, we made sure to delay debrieng until

all of the participants had been tested to maintain internal validity.

Results

Notice of Change vs. No Notice of Change

A chi-square goodness-of-t test was calculated comparing the fre-

quency of occurrence of participants who noticed the male burglar change

to those who did not notice the change. It was hypothesized that most

participants would not notice the male burglar change. A signicant differ-

ence was found (χ² (1) = 22.273, p < .01) between participants who did not

notice a change (n = 45) and participants who did notice a change (n = 10).

As hypothesized, most participants were affected by change blindness and

did not notice the male burglar change.

A chi-square goodness-of-t test was also calculated comparing

the frequency of occurrence of participants who noticed the female driv-

er change to those who did not notice a change. It was hypothesized that

most participants would not notice the female driver change. A signicant

difference was found (χ² (1) = 36.818, p < .01) between participants who

did not notice a change (n = 50) and participants who did notice a change

(n = 5); as hypothesized, most participants were affected by change blind-

ness and did not notice either change.

The Effect of Instruction on Change Blindness

A chi-square test of independence was calculated comparing the

frequency of participants who noticed the male burglar change and fe-

male driver changes in the intentional condition to that of the incidental

condition. It was hypothesized that signicantly more participants in the

intentional condition would notice the male burglar and the female driver

changes than participants in the incidental condition. No signicant rela-

tionship was found between condition and whether or not the participant

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

183

noticed the female driver change (χ² (1) = .066, p > .05). A signicant re-

lationship, however, was found between condition and whether or not the

participant noticed the male burglar change (χ² (1) = 3.194, p < .05). The

results show that participants who were primed to pay attention were more

likely to notice the change in the male burglar, but not the female driver.

The Effect of Instruction on Memory Score

An independent-sample t-test was computed to compare the mem-

ory scores of participants in the intentional condition to memory scores of

participants in the incidental condition. It was hypothesized that memory

scores would be signicantly higher for participants in the intentional con-

dition. No signicant difference was found (t(53) = 1.187, p > .05), how-

ever, between the memory scores of participants in the intentional group

(M = 7, sd = 1.56) and the memory scores of participants in the incidental

group (M = 6.5, sd = 1.42). Being primed to pay attention, therefore, did

not have a signicant impact on content memory.

The Effect of Time Intermission on Change Blindness

A chi-square test of independence was calculated comparing the

frequency of participants who noticed a change in the male burglar (no

time interval) to that of the female driver (two minute interval). A signi-

cant difference was found (χ² (1) = 6.466, p < .05). The results show that

signicantly more participants noticed the male burglar change (n = 10)

than the female driver change (n = 5). This result suggests that time-inter-

mission had a signicant impact on change blindness, but this interpreta-

tion is not fully supported by the statistics used (full analysis found in the

discussion section).

The Effect of Gender-Matched Observers on Change Blindness

A chi-square test of independence was also calculated comparing

the frequency of male participants who noticed the male burglar change to

female participants who noticed the male burglar change. It was hypoth-

esized that there would be a signicant interaction between male partici-

pants and recognition of male burglar change. No signicant interaction

was found (χ² (1) = .490, p > .05), however. Gender matched participants

did not differ signicantly from non-gender matched participants in

whether or not they noticed the male burglar change.

A chi-square test of independence was also calculated comparing

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

184

the frequency of female participants who noticed the female driver change

to male participants who noticed the female driver change. It was hypoth-

esized that there would be a signicant interaction between female partic-

ipants and recognition of female driver change, but no signicant interac-

tion was found (χ² (1) = .636, p > .05). Gender matched participants did

not differ signicantly from non-gender matched participants in whether

or not they noticed the male burglar change.

Discussion

As expected, 82% of participants did not notice the identity change

of the male burglar and 91% of participants did not notice the identity

change of the female lookout. In other words, most participants, regard-

less of which condition they were in, did not show signs of awareness of

the changes made in identities throughout the lm. Although this result

had the highest expectation, we were still surprised by how high the rates

were for change blindness. The changes in clothing and appearance, even

if argued as subtle, were accompanied by radical changes in demeanor

between the actors. We therefore recommend that future research push

change blindness further by creating even more discernable differences

between actors (perhaps incorporate a shift from dull colored attire to

brightly colored attire).

With regard to the effects of instruction, there was a signicant re-

lationship between awareness of change of the male burglar’s identity and

condition, indicating that priming participants to pay close attention had a

positive effect on change blindness. This result was anticipated due to re-

sults obtained by Davies & Hine (2007). However, awareness of change of

female lookout did not share the same relationship with condition. At this

point in the analysis, a bit of ambiguity surrounds this result, for it could

be caused by the time intermission variable that distinguishes the burglar

from the lookout identity swaps, or it might be related to the gender of the

actors. Further speculation also raises a concern of overlay between our

variables, which would justify further study.

In terms of the interaction between instruction and memory score,

we were surprised to nd that there was no signicant difference between

the memory score of participants who were instructed to pay close atten-

tion and that of participants who were not, indicating that ‘priming’ partic-

ipants to remember content did not signicantly affect their memory score.

This result was perhaps the most unexpected because it disagrees with

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

185

Davies’ research. We believe this result may be due to the difculty of our

content questionnaire. Our memory score was heavily based on how many

stolen items the participant could ‘recall’ from their memory after watch-

ing the lm. In any situation, we know that content recall (ll in the blank)

is much more difcult than content recognition (multiple choice). Further-

more, it could be argued that the items stolen in the video were not com-

prised of prototypical objects in their respective category. For instance, the

camera that was snatched up was a digital camera that was small and com-

pact; it could have been mistaken for something else. Some of the objects

were also displayed in poor lighting, which should be addressed in future

paradigms of similar nature. Lastly, we propose a more dynamic memory

score be created for future research, one that includes a mixture of both

recognition and recall questions; this should yield a more effective tool for

accurately conveying both the quality and quantity of content retention.

In addressing one of our unique hypotheses, awareness of change

appears to be signicantly related to the time intermission between each

identity change. This interpretation derives from the data analysis show-

ing that signicantly more participants noticed the male burglar identity

change (no intermission) than the female driver change (two-and-a-half

minute intermission), which would indicate that change blindness is, at the

very least, somewhat dependent on time intermission. This result seems

to alleviate our confusion about an earlier result—the absence of a rela-

tionship between awareness of change of female lookout and condition.

However, upon review, this result should still be treated with caution due

to a reasonable possibility of variable confounding. It is possible that other

variables, such as the burglar’s gender or his role in the lm, might have

contributed or had an unwelcomed effect on the recorded result. Therefore,

the analysis does not serve as sufcient evidence to support the claim that

time-intermission signicantly affected awareness of change within our

study; however, it does not evoke a rejection of our hypothesis, either. Fu-

ture research should further investigate the relevancy of time-intermission

on change blindness by taking greater care in controlling for this variable,

for there are still adequate reasons to pursue this facet of change blindness.

Lastly, no signicant interaction was found between gen-

der-matched participants and awareness of change, indicating that aware-

ness of change of either the male or female perpetrators was not inuenced

by the participant’s gender. This result was neither expected nor unexpect-

ed, as we did not have predictions for this variable. It is also worth noting

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7

186

that our participant pool was not made up of an equal number of male and

female participants, which we fear may have skewed the results on this

particular test. Of course, future research would do well to control for an

equal ratio of male to female participants when testing for this variable.

Nevertheless, these results mostly strengthen the ndings of prior

research conducted by Davies & Hines (2007) and adequately demonstrate

the relevance of change blindness on eyewitness identication literature.

At this point in the investigation, it would be logically indefensible to deny

the validity of change blindness or the fact that primed individuals are

less likely to experience it. This study succeeds in further examining the

infrastructure of this phenomenon and makes laudable attempts at explor-

ing the relevancy of two unique variables—the effects of time intermission

and gender-matched observers. Although certainty was not obtained from

every result, the groundwork has certainly been laid for future research to

build on these ideas and learn from the limitation mentioned above.

The study of change blindness is certainly important in understand-

ing more about how attention, perception, and memory are intertwined,

but this line of research has an even greater practical application in our

justice system—where eyewitness misidentication has led to a stagger-

ing number of wrongful convictions. The change blindness found in this

study builds on previous research in demonstrating the fallible nature

of memory and reveals that eyewitness testimony may not be the most

reliable resource in ascertaining the truth. Unfortunately, this is contrary

to what many judges, litigators, and active jurors may currently believe.

Eyewitness testimony is currently held in high regard in terms of accura-

cy and reliability; more disturbingly, it has served as the turning point in

securing thousands of successful, but wrongful convictions. A rst step in

minimizing this problem begins with encouraging researchers in eyewit-

ness testimony to take a closer look at the relevance of change blindness,

and to begin disseminating their ndings to the public and to our judicial

ofcials. The goal is to reform public predispositions about eyewitness

testimony by revealing the fallible nature of memory and instances where

change blindness comes into play.

Mitchell: Change Blindness and Eyewitness Testimony: The Effects of Instruc

Published by University of Memphis Digital Commons, 2016

187

References

Davies, G., & Hine, S. (2007). Change Blindness and Eyewitness

Testimony. Journal of Psychology, 141(4), 423-434.

Levin, D. T., & Simons, D. J. (1997). Failure to detect changes to attended

objects in motion pictures. Psychonomic Bulletin and Review, 4,

501-506.

Most, S. B., Scholl, B. J., Clifford, E. R., & Simons, D. J. (2005). What

you see is what you set: Sustained inattentional blindness and the

capture of awareness. Psychological Review, 112, 217-242.

Simons, D. J., Chabris, C. F., Schnur, T. T., & Levin, D. T. (2002).

Evidence for preserved representations of change blindness.

Consciousness and Cognition, 11, 78-97.

QuaesitUM, Vol. 3, Iss. 1 [2016], Art. 7

https://digitalcommons.memphis.edu/quaesitum/vol3/iss1/7