Change Blindness and its Effect on Eyewitness Testimony



This study examined the relevance of prior research on the perceptual phenomenon of change blindness. The research conducted in this experiment was based from the findings of Davies and Hine (2007). The effects of the phenomenon on correct eyewitness identification and eyewitness testimony under incidental and intentional memory conditions was analysed, this was reminiscent of the conditions present in the original Davies et al. (2007) experiment.  The intentional group would be informed of a facial recognition test after the stimulus was shown, the instructions of the incidental condition claimed a test on objects and events would be follow the stimulus. Participants (N = 88, 32 men and 56 women) viewed a series of 20 photographs in the form of a 2-minute automatic slideshow, at the 1 minute mark the actress playing the burglar changes without prompt therefore potentially priming the participants for change blindness. The author explored the relevance of the research of change blindness on an eyewitness’ account by conducting an experiment to test that. This was achieved by testing participants on their ability to perceive a change in the burglar at the midway point of the slideshow. Similarly to the original work, once the participant had completed viewing the stimulus there were tests to see the recall abilities of specific data, in a content questionnaire.  The tests to see the awareness of change and the successful identification rates of one or both burglar’s were kept consistent with the addition of testing familiarity and the probability of each suspect being the correct perpetrator. 72% of participants however did not detect the change, with the incidental condition shown to have a significant effect when its effects on the detection of change was analysed. Once a participant could detect the change they were then significantly more likely to do better during the identity parades. The results support the initial findings of Davies et al. (2007), and help to demonstrate the convergence of scholarly findings on eyewitness testimonies and change blindness.



Contents Page

  • Introduction (4)
  • Reasons For The Exploration Into The Topic (4)
  • Prior Research (5-17)
  • Research into Memories Formation and its’ Questioned Reliability in the Courtroom (5-7)
  • The Role of Eyewitness Testimonies in the Criminal Justice System (7-9)
  • Perceptual Blindness’ Effect On Eyewitness Testimonies (9-11)
  • Change Blindness and its Subsequent Effects on Visual Stimuli (12-13)
  • Convergence in Literature (14-17)
  • Methods (18-23)
    • Subjects (18-19)
    • Design (19-20)
    • Materials and Procedure (20-23)
    • Ethical considerations (23)
  • Results (23-28)
    • Change Blindness Manipulation Check (23-25)
  • Table 1. Number of Participants Detecting Change as a Function of Memory condition (24)
    • Confidence and Accuracy of Correct Identifications (25-26)
  • Table 2. Number of Participants that Correctly Identified the Actresses in the Stimulus as a Function of Memory Condition and Change Detection (26)
    • Additional Scales for Correct Identification (26-27)
  • Table 3. Participants Detecting Correct Suspect as a Function of Rating Systems (27)
    • Content Questionnaire (27-29)
  • Table 4. Number of Participants Detecting Change as a Function of Memory Condition (28)
    • Misidentification (29-30)
  • Table 5. Participants Rating Scores as a Function of Potential Suspect Misidentification (30)
  • Discussion (30-35)
  • Change Detection (30-31)
  • Correct Identification (31-32)
  • Conclusion of the Findings from the Content Test (32)
  • Misidentification (33-34)
  • Limitations and Improvements (34-35)
  • Conclusion (35-36)



Reasons For The Exploration Into The Topic


Change blindness is a perceptual phenomenon that refers to an observer’s inability to recognise the introduction of a change to a visual stimulus. This phenomenon can occur during everyday life and its effects range from individual to individual. The most common example of change blindness is in media, more specifically movies and television. The Oscar nominated ‘Apocalypse Now’ (1979), has a total of 559 continuity errors, the all time classis ‘The Wizard of Oz’ (1939) has 421. The majority of these continuity errors normally go unnoticed by the viewers, for a movie such as ‘Apocalypse Now’ with a running time of 153 minutes, there is a potential for change blindness 3.7 times every minute. Although the effects of change blindness can be deemed trivial knowledge when exploring it in relation to movies; the implications however, if the effects are consistent in all visual stimuli, it is then important to explore whether it has the same effects to a bystander’s eyewitness testimony.

Eyewitness testimonies have been a key instrument for the prosecution team and the police in the criminal justice system.  The over reliance of an eyewitness’ testimony in a court of law has come under scrutiny in psychological research, as some academics deem memory to be malleable and the retention abilities of the brain to be overestimated (Loftus & Palmer, 1974, 2011). As eyewitness testimonies have been analysed and evaluated further, differing effects have been explored to check its reliability. Change blindness is a perceptual phenomenon that has been explored in relation to eyewitness identification, with results suggesting that the phenomenon may play a key effect in the misidentification of a suspect. The experiment that will be conducted is an adaptation of the original work, ‘Change Blindness and Eye Witness Testimony’ (Davies & Hine, 2007). The difference being that a selection of photographs will be used to present the stimuli not a 2-minute video like the original.

Prior Research

Research into Memories Formation and its’ Questioned Reliability in the Courtroom

Psychologists’ since the 20th century have considered the mechanisms behind the formation of memory, as the depth of research that focused on memory subsequently increased. The heightened level of attention from the psychological scholastic community, due to the findings from a selection of studies and models proposed through the century (Atkinson & Shiffrin, 1969; Baddeley & Hitch, 1974; See also Münsterberg, 1908; Thorndike, 1971), the reliability of a human’s memory came under scrutiny. Memory as a topic since has been widely researched with a specific focus on its potential flaws and critiques, many of which had been previously suggested through the results of past studies into memory retention. Benedict et al. (1996) further delved into memories’ durability and accuracy, however the study’s findings highlighted potential limitations associated with specifically visuospatial memory. Benedict et al. (1996) research focused on revising the Brief Visuospatial Memory Test (BVMT), a very versatile test that can be applied to fields from the assessment of schizophrenia (Green et al., 2004; Nuechterlein, Green & Kern, 2010) to severe brain damage (Benedict & Groninger, 1995; Morey, Berry, Cilo & Cusick, 2003). Follow up studies have further tested the BVMT but with the exploration of the reliability of visuospatial memory being a key point of focus. The study provides an evaluation of the influence of processing speeds on BVMT learning, retention and memory (Benedict, 1997; Tam & Schmitter-Edgecombe, 2013). The results showed a significant level of impairment in visuospatial memory, thus causing increased perusal of specifically prior visuospatial memory retention related literature and its related effects to eye witness testimonies.

Work in the deterioration of visual short-term memory was further increased by the exploration of the phenomenon of perceptual blindness, with psychologist attempting to establish the link between perception, attention and memory. The focus of the psychological research looked at both inattentional and change blindness. Inattentional blindness was a term coined to describe the discovery of the phenomenon (Mack & Rock, 1998) and was defined as a temporary blindness effect that can cause the missing of an aspect of a stimulus due to the attention of the viewer being placed on another portion of the stimulus (Simons & Chabris, 1999). Change blindness is slightly different, as this phenomenon exists when a change central to the stimulus is introduced and the observer does not notice it. Change blindness is closely related to the research conducted in the 1970s within experimental psychology regarding the concerns about short-term visual memory deterioration (Haber, 1983).  Research into change blindness over the last two decades has greatly contributed to the psychological communities understanding of attention, consciousness and perception (Simons & Rensink, 2005). Both the abundance of studies into the flaws associated with memory retention and the introduction of perceptual phenomena casted doubts on the reliability of an eyewitness account by the start of the 21st century. Despite an eyewitness’ account being a pivotal factor in defining the truth in many of the worldwide judicial systems, disregarding much of the present research, an eyewitness testimony still holds a large bearing in the courtroom and is still assumed as ultimately reliable in the eyes of the law.

The limitations presented in memory studies and the exploration of specific phenomena during the 20th century helped in the understanding of the flaws to memory formation. The extent to which issues with memory recollection presented in the models and thesis’ (Atkinson & Shiffrin, 1968; Baddeley & Hitch, 1974; see also Peterson & Peterson, 1959; Bahrick, Bahrick & Wittlinger, 1975) possess enough ecological validity to label eyewitness testimonies as unreliable was however largely unexamined. Due to this by the end of the 20th century there was an over reliance on the validity of memory in areas of society such as the judicial system. Despite findings, the system remained unchanged and the level of research into perceptual blindness’ effect on an eyewitness’ testimony was limited due to a lack of backing by peers. Many still deemed eyewitness testimonies to be sufficiently reliable, negating the implications that due to the unreliability of memory a heavy weighting on the use of eyewitness identification in a court case can lead to a large level of false imprisonment. Ultimately the research into these phenomena was largely unnoticed by the world’s governments and media, until the advancement of DNA testing technology and the establishment of the Innocence project. DNA testing has led to the exoneration of over 2,000 wrongfully convicted individuals in the United States of America (Gross & Shaffer, 2012), with eyewitness misidentification playing a crucial role in more than 70% of the overturned convictions (Innocence Project, 2017).  The mass number of overturned convictions due to eyewitness misidentification has shown the potential issues with the overreliance on this method of information, in addition these figures cannot be solely accredited to just memory deterioration. Highlighting an increased need, specifically in the case of the formation of an eyewitness account, for research into the phenomena associated with perceptual blindness.

The Role of Eyewitness Testimonies in the Criminal Justice System


Criminal punitive proceedings are widely used throughout the world and the judicial system stems back to the 12th century. During the enforcement of the Magna Carta (1215) due to its promises of access to quick justice, the use of eyewitness testimonies became a crucial tool for the prosecution team within criminal court proceedings. Since its first use the reliance of eyewitness identification has exponentially grown. Devlin (1976) claims the identity parade had its first recorded use by the London Metropolitan Police in 1860. By the 21st century over 100,000 criminal identification line-ups were used in the United Kingdom alone (Flowe, Smith, Karoglu, Onwuegbusi & Rai, 2015). At that time eyewitness identification had become a vital tool for the prosecution in most criminal cases due to its large effect on the outcome of a jury or a judge’s decision. It has been utilised so often that cases that involve misidentification from an eyewitness has rapidly become the largest cause of wrongful convictions in the United States of America, which was discovered due to advancements in DNA testing (Borchard, 1932; See also Wells & Bradfield, 1998; Munsterberg, 1908). Eyewitness testimonies have gained popularity throughout the ever-increasing growth of human civilisation, but it only became subject of criticism during the mid-20th century due to an increasing number of psychological studies questioning the reliability and accuracy of memory formation (Atkinson & Shiffrin, 1969; Baddeley & Hitch, 1974; See also Münsterberg, 1908; Thorndike, 1914). The over-reliance of identity parades came about due to the advantages of its use in establishing identity, without the need for suggestive procedures such as confrontation. Unfortunately, in many cases proof of guilt has been based from a claimed positive identification (Davies & Griffiths, 2008). Under usual procedure police would pick individuals who somewhat resembles the suspect, and then stand them in a line whilst a victim squinted and tried to choose the real criminal.

Loftus and Palmer (1974) built on the ideas of memory and its fragility but within the contexts of law proceedings. Loftus et al. (1974) looked to investigate a pressing issue related to eyewitness testimonies; a courtroom tool that has been vital in the incarceration of countless accused criminals in recent centuries within westernised cultures (Fitzgerald, Price & Oriet, 2013; Steblay, 1997; See also Steblay, Wells & Douglas, 2014; Wells & Bradfield, 1998). Loftus et al. research into the validity of eye witness testimony has become more relevant in the 21st century due to the DNA exonerations of many wrongfully convicted individuals that were sentenced due to a false eye witness identification. The aim of the experiment was to alter the perception of the eyewitness to see how malleable memories truly are. This was tested by carefully selecting the language used by peers or individuals in a position of authority to see if it can alter memory. They were then asked a specific question, “about how fast were the cars going when they (hit/smashed/collided/bumped/contacted) each other?” The five verbs had a varying severity to its meaning, this allowed the researchers to test if the participants subconsciously adjusted their own memories of an event due to a leading word. They hypothesised that by manipulating the variable, which was displayed as a leading question one can potentially warp an individual’s recollection of the events making their account unreliable. The experiment was conducted by showing participants short clips of multiple car accidents and then asking them to describe what had just happened, as if it were an eyewitness testimony. The findings showed that the verb used affected what the participants said. On average those who were asked the ‘smashed’ question, thought that the cars were going faster than those who were asked the ‘hit’ question. Smashed recorded an average response of 41mph and hit received a 34mph average.

Loftus et al. showed that language can have a huge distorting effect on memory and it is possible due to interference that the initial memory has been reconstructed into a new memory that is still as factual in the eyes of the witness. Subsequent follow up studies have highlighted that it is not just language that can have a large effect on memory, many differing stimuli can cause a false memory. This study was conducted as a laboratory experiment, meaning there would have been limitations as it lacked ecological validity as an individual within this study could act differently in a real-life situation. Potential problems with participant cues and participant bias were also a potential flaw to the experimental methodology. Loftus et al. showed in their work that memory was in fact very susceptible to constructing false memories if it is presented will misleading information, despite viewing the crime themselves. The information obtained from this study was fundamental in understanding the potential flaws with human memory in context to an eyewitness testimony. Misleading questions however are not the only factor that can affect eyewitness identification

Perceptual Blindness’ Effect On Eyewitness Testimonies


Although the eye receives all the information it sees, the brain cannot process all the information whilst maintaining the fast level of cognitive function needed for day-to-day interaction. Some information is lost between the transfer of information of the stimulus from the sensory store through to the long-term store (Atkinson & Shiffrin, 1969, 1971; Brelsford, Keller, Shiffrin & Atkinson, 1966; Phillips, Shiffrin & Atkinson, 1967). Attention is a major part of memory processing; it plays a part in how the phenomenon of perceptual blindness comes to existence in eyewitness testimonies. Attention is the action where specific information from a stimulus will be focused upon and therefore the brain decides to process that information. When an individual’s attention is placed on a specific aspect of the stimuli the likelihood of it being encoded into both the short term and long-term stores subsequently increase. Inattentional blindness and change blindness are two phenomena that are both defined as a failure in visual awareness. The participant’s failure to notice an obvious change to the stimuli is defined as change blindness. The unexpected parts of a stimulus that the brain did not capture can be described as inattentional blindness (Jensen, Yao, Street & Simons, 2011). Both inattentional blindness and change blindness can be seen in a lot of day-to-day interactions with no detrimental effects. The two forms of blindness are everyday phenomena; however, it can sometimes have fatal consequences, especially in road traffic accidents or in witnessing specific details of a crime or crime scenes.

As change blindness and inattentional blindness share many similarities in research and findings, thus focus in one area can help to give the overall understanding of the effects of psychological blindness on a visual stimulus. To understand inattentional blindness, the relationship between perception and attention is essential; this can be explored in both abnormal circumstances as well as day-to-day interactions. Most individuals who drive would have witnessed or experienced “functional blindness” where clear pieces of the stimuli are missed, or if you have engaged in an absorbing conversation, “sighted blindness” may occur (Mack & Rock, 1998). “Although the inextricable link between perceiving and attending was noted long ago by Aristotle” (Mack, 2003), this phenomenon, now called inattentional blindness, only recently has been named and carefully studied. Mack and his peers’ work looked to deduce the key question posed to inattentional blindness research: Is “Inattentional blindness an instance of rapid forgetting, or is it a failure to perceive” (Mack, 2003). Mack and Rock’s (1998) standard procedure used a cross on a screen as the distraction task. Whilst participants were focused on the task of deciphering which arm of the cross was longest after several trials an unexpected object would appear on screen. Mack et al. findings showed that participants when their attention was placed on the cross, often failed to witness the unexpected stimuli, even if it was in the centre of the screen and brightly coloured.

Despite the similarities between the two phenomena, both types of blindness have its own unique background and distinct theoretical implications (Jensen Yao, Street & Simons, 2011). Even though the study of inattentional blindness has only come to fruition in the past few decades, there have been several studies into the phenomenon.

The most famous of these studies on inattentional blindness was the ‘Gorillas in our midst: sustained inattentional blindness for dynamic events’ (Simons & Chabris, 1999). A study which featured a task where the participants had to count the number of passes a group of basketball players make, whilst the passing drill goes on a fully grown adult male in a gorilla suit dances through the centre of the passers. The findings revealed that 46% of the participants did not see the gorilla. Despite the issues that Simons and Chabris (1999) encountered; “36 participants that were discarded for a variety of reasons” the results showed that it is possible and even common for a person to focus so hard on something that they become blind to the unexpected, even when staring right at it. When one develops inattentional blindness it becomes easy to miss details when one is not looking out for them and Simon et al. concluded that these findings suggest that we perceive and remember only the stimuli we decide the pay attention to. The experimental task was to count how many passes were made by the basketball team, therefore the man in a gorilla costume that walks directly through this drill was not deemed essential for the brain to remember and was discarded as irrelevant information. The information obtained from any stimulus where either attention was not applied or due to your perceived lowering of the importance of a stimulus, key facts can be missed. This can potentially have a large effect on the reliability of eyewitness testimonies and the validity of the information.




Change Blindness and its Subsequent Effects on Visual Stimuli

Change blindness as a phenomenon has not been studied as excessively as inattentional blindness in recent psychological research, its effects on the validity of an eyewitness’ testimony however can be just as severe. The prior research conducted has been influential in explaining how change blindness can be a large contributing factor to eyewitness identification. Change blindness arguably first appears in McConkie and Currie (1996), in an experiment that measured eye movement. Although McConkie and Currie intentions for this experiment was not to investigate change blindness, however researchers found that surprisingly large changes, up to a fifth of the stimuli, was not noticed by the participants. The hypothesised reason behind these finding from an ophthalmologist perspective was initially an explanation of the mechanisms within the brain used to pair information with specific successive eye fixations, therefore the explanation of change blindness is merely the imperfection of this mechanism in action (McConkie & Currie, 1996). Follow up experiments further delved into the phenomenon, suggesting that the findings were not due to eye movements but instead was what is now known as change blindness, this differentiation was made by what is described as the “flicker” technique (Rensink, O’Regan & Clark, 1997).  The “flicker” technique utilised a similar experiment but instead of using eye movements a brief flicker was inputted between successive images. Still changes in the stimuli were not detected to a significant level. This experiment helped establish change blindness as its own phenomenon but also highlighted the need for attention and awareness for an individual to perceive change in a stimulus. It is believed to be necessary for attention to be maintained to consciously perceive change, but attention alone may not be sufficient. Attention and awareness do not always operate simultaneously, if attention alone is paid to a stimulus there is no assurance the change will be perceived (Williams & Simons, 2000; Triesch, Ballard, Hayhoe & Sullivan, 2003). Unexpected changes in a stimulus rarely go noticed by the eyewitness, suggesting that object perception is highly dynamic. When properties of a stimulus are consciously perceived it is for the sole purpose of the task at hand, therefore any changes will not be noticed unless directly relevant to the task (Triesch, Ballard, Hayhoe & Sullivan, 2003). Levin and Simons (1997) conducted an experiment to test an individual’s ‘Failure to detect changes to attended objects in motion pictures’. This examination looked at the role of attention in change detection, hypothesising that attention alone may not be sufficient to negate the effects of change blindness (Levin & Simons, 1997b). The experiment consisted of 10 Cornell University undergraduates; each participant was shown a short colour VHS video. Within the various camera angles adopted in the video a total of 9 changes were made to the stimuli, for example the change in the colour of the plates, which switched from red to white from one frame to the next. The subjects were informed to “pay close attention” but not told of any imminent changes. Despite the warning and given information that an eyewitness would not be reminded of in a real-life situation the results of this experiment shows that only one of the participants noticed any changes in the first viewing. The participant who reported a change however could only mention 1 out of 9 changes placed in the video. Upon the second viewing of the same video participants could only improve upon the recollection of 1 out of the 90 total changes to an average of only 2 out of the 9 changes perceived per participant. The most frequently noted change was the bright coloured scarf, which due to its bright garish colours stood out and could have formed an imprint in the participants’ memory. The findings showed the effects of change blindness on perception and memory with participants struggling to perceive the change in the stimuli even after viewing the same recording multiple times and learning of potential upcoming changes through active learning and participant observation. The conclusion of the results obtained was supported by similar findings in psychological research over the last two decades into change blindness (Aginsky & Tarr, 2000, Grimes, 1996; Rensink, O’Regan & Clark, 2000).




Convergence in Literature

The effect of change blindness on a subject’s accuracy during an eyewitness’ identification, in the context of a criminal proceeding, was examined by Davies and Hine (2007).  The research conducted looked at the relevance of former research into change blindness and facial memory in both experimental and real world situations (Angelone, Levin & Simons, 2003; Levin, Simons, Angelone & Chabris, 2002; see also Levin & Simons, 1997; Rensink, 2002), examining how the findings of its effects could be applied to eyewitness identification and recollection. In the early 21st century the research into change blindness (Levin, Simons, Angelone & Chabris, 2002) focused on the effects it played on the photo line-up examining the participants’ ability to retain residual visual representation of the original stimuli (Davies & Hine, 2007).  Findings presented prior to the study by Davies & Hine suggested that change blindness could have a significant effect on the participant’s ability to retain residual memories when viewing a police photographic line up. The study builds upon the thoughts presented in the early identity change studies (Levin & Simons, 1997). In Levin & Simons (1997) participants were shown a video clip of an actor getting up to answer a ringing phone, due to a change in the camera angle timed to simultaneously coincides with the identity change of the actor therefore the change is not immediately obvious, thus priming the video for change blindness. The actors had distinct differences in physical appearance 33% of the participants did not notice the change. The analysis and the subsequent follow up study (Simons & Levin, 1998) deduced that the data collected from the phone call experiment in the Levin & Simons’ study cannot be categorised under intentional blindness as the change was central to the video and the participant’s attention was not primarily engaged on another task, object or an event during the aforementioned stimuli (Levin, 2010).  The Levin & Simons’ stimulus was applied in a real-world context to test the ecological validity of the findings of change blindness. By using a marked target on a university campus, an actor would ask a pedestrian for directions, but during the conversation (defined as a staged reaction in the research paper), two males carrying a large door would go between the conversing individuals. Once the door is at the midpoint the switch is made and a new actor continues the conversation when the door passes by. The results in the real world setting however still showed significant effects of change blindness with only 50% of the pedestrians noticing the difference. This experiment incorporating a real-world setting has since been widely used to show the mundane realism of the phenomenon of change blindness. Levin et al. (2002) used the basis of Simons et al., research as one of the two conditions incorporated in the study, the other being a similar scenario but the actor asks instead for a photograph. Results showed consistent results to Levin et al. (1997) with 38% of the participants missing the change in the direction task and 53% oblivious to the change in the picture condition. Change blindness has increasing been focused on by the field of psychology due to investigations into its effects (NOVA, 2012) and a repeat of the photographic condition (Levin, Simons, Angelone & Chabris, 2002) within the national media during ABC’s game show ‘Would You Fall for That’. Both the results shown by previous research and its acclamation within the media has been influential in bringing awareness to the phenomenon of choice blindness awareness.

Davies and Hines (2007) incorporated several of the ideas presented in previous research into change blindness (Angelone, Levin & Simons, 2003; Davies et al., 2007; Levin, Simons, Angelone & Chabris, 2002; Simons & Levin, 1998) in ‘Change Blindness and Eyewitness Testimony’. The method of the experiment conducted within the paper consisted of 80 participants (40 men and 40 women). All were aged between 15 and 65 years old years old and volunteered to participate (Davies et al., 2007). The sample consisted of 68 individuals who were employed (85%), 8 students (10%) and 4 carers (5%) all of which were randomly assigned to either the “incidental condition” or the “intentional condition”, these conditions were the manipulation by the researcher within the experiment, labelling it as the independent variable. Participants were randomly assigned before they were presented with a 1:40 minute video of a robbery at a university accommodation; the identity of the burglar however changes at the halfway point of the video. The difference between the two conditions was the instructions given to them by the researchers. A participant in the “incidental condition” would be given the information that “this short video illustrates the ease and frequency of burglaries of student accommodation and the importance of keeping houses secure”. The “intentional condition” was told, “You are about to watch a short video. Pay careful attention to the content, as there will be a memory test later”, priming the subject to focus on specific details of the stimuli allowing for a potential increase in the chance of perceiving the upcoming change within the video. The results of the experiment showed 31 individuals out of the 90 detected the change, showing a 39% overall detection rate of change. 26 out of the 31 individuals (84%) who detected the change were placed in the “intentional condition” meaning that although 65% of the intentionally primed condition perceived the change, only 5 individuals (12.5%) perceived the change in the “incidental condition”. The limitation to this study included a lack of ecological validity due to it being a laboratory based study, in addition a criticism of the study was its over reliance on cued questions opposed to a mix of cued questions and free recall tasks. A combination of the two can provide a more accurate representation of the participants’ level of attention (Sammon & Bogue, 2015). The experiment however did not successfully explore the connection between change blindness and familiarity of the correct suspects. Despite the limitations these results were staggering due to the hypothesised ramifications it would have on eyewitness identification and testimonies.

In a real-world example, an individual viewing a crime will under normal conditions have no prior knowledge of the crime about to be committed, this lack of preparation means that they would fall under the “incidental condition” in the experiment. In that condition only 12.5% of the participants noticed the change, therefore the hypothesised implications of the findings suggested that if a vital change happened to the stimuli during a crime sequence 87.5% of people would miss it, potentially making their testimony incorrect or incomplete.  Even though there was a low ecological validity in the experiment the results of the experiment must be considered valid due to the high levels of mundane realism change blindness and memory deterioration studies have shown prior to the release (Levin & Simons, 1997a, 1997b; Simons & Levin, 1998, Simons & Rensink, 2005) of ‘Change Blindness and Eyewitness Testimony’ (Davies & Hine, 2007). The experiment that will be conducted in this paper will be an adaption of the original Davies et al. paper with additional measures incorporated within the experiment and questionnaire that tests previously stated limitations of the initial study. Considering all the research presented it is hypothesised that change blindness will have a significant effect on the reliability of eyewitness identification and the validity of their testimony.

A study will be conducted to examine change blindness’ and its effects on an eyewitness testimony. The aim of this study will be to explore the effects of change blindness and establish if there is a significant causal link between the phenomenon and eyewitness misidentification. By focusing on this perceptual phenomenon and how it affects eyewitness testimonies or more specifically eyewitness identification, the results of this study can have a large effect in understanding the reliability of a major persecutory tool that is utilised around the world every day to decide if an individual is guilty or not of their accused crimes. It is hypothesised that less than half the subjects who participate in the study will notice the major change to the stimulus. As the independent variable will be divided into an incidental and intentional condition, due to the findings of Davis et al (2007) it is believed that participants within the intentional condition will have a higher rate of detection for the change. In addition participants who do in fact detect the change will be more likely to successfully identify the real perpetrators from an identity parade.








Subjects were 96 individuals from around the United Kingdom, with the most represented area of the country being the southeast, specifically the county of Kent. From this total, 8 individuals had their data omitted due to an incompletion of the study. A sample of 32 men and 56 women (N = 88) were used in this experiment. The overall sample had an average age of 22.3 years with an age range of 18-68 years old. The age of participants however was not directly proportional when separated by the age group categorisation shown in Petry (2003). The three categories under the age categorisation were; young adults (aged 18-35 years), which consisted of 81 individuals, middle aged (36-55 years) consisting of 3 individuals and old age (55+ years) which consisted of 1. The outlier to the categorisation method being the participant who labelled themselves 50+ years and the 2 participants who chose not to state their age. 87 out of the 88 participants labelled themselves as a student or stated that they have formerly studied a university degree, with 58 of the participants citing a form of psychology degree.

The study was only accessible on the Internet through either the selection of the study for credits via the Research Participation Scheme (n=56), or via a selected pool who accessed the study through a link sent to their email addresses (n=34). For the Research Participation Scheme (RPS) the target population was any individual that had access to the study. This was obtained by being a member of Kent University and studying under the school of psychology. The target population for the pool who accessed the study via email was not as limited by parameters as the RPS study. As the study was personally emailed, the target population was limited only by having prior contact with the researcher. In both cases the sampling technique used is an example of opportunity sampling as the participants’ involvement in the experiment was purely dependent on their convenience. This adds the benefits of negating the effects of sampling bias whilst remaining quick, with minimal additional effort and inexpensive. After participants were selected they were randomly assigned to one of two conditions with the constraint that equal proportions of each gender are represented in both conditions; from the functional N of 88, 45 were randomly assigned to the intentional condition and 43 to the incidental condition (Davies & Hines, 2007).



In a Between Subjects One Way (ANOVA) design, the two conditions (intentional vs. incidental) were manipulated as they are labelled as the independent variable for this experiment It was tested for its effects on the three dependent variables: Ability to successfully identify the first robber, ability to successfully identify the second robber and ability to detect the change between the first and the second robber at the halfway point (50 second mark). The experiment consisted of 5 sections; ethics and conditions, stimulus (1 minute 40 second slideshow), 2 simulated police line ups incorporating sliders to rate various relevant aspects of the suspects, 13 closed questions relevant to the eyewitness events of the robbery and 2 questions designed to detect change blindness. Each of the post stimuli sections was designed to test one of the three dependent variables or the baseline level of eyewitness reliability of facts.

Participants were randomly assigned to one of the two conditions, before viewing the 1 minute 40 second stimulus (slideshow).  The participants being assigned into one of two groups differed on the instruction given by the researcher prior to the start of the study. The standardised instructions received by all participants stated that they ‘will now be shown a series of photographs depicting a robbery at a university campus, it is essential to pay careful attention during the slideshow’. If a participant were placed in the incidental condition they would be given the information that there ‘ability to remember will be tested in a short questionnaire’. The intentional condition was instructed that their ‘ability to recognise the perpetrator will be tested in a short face identification task’. Similarly, through the slight difference in instructions used in the two conditions the increased priming effects for the change blindness depicted in the intentional condition was adapted from Davies and Hines’ (2007) study. Thus, allowing a testable independent variable to aid the detection of change blindness.

Materials and Procedure

After signing up for the study and agreeing to the ethical considerations, participants were into one of two conditions. From the functional N = 88, 45 (52 %) were randomly assigned to the intentional condition and 43 to the incidental condition (48%). The difference between the two conditions was the information given before the experiment. The difference between the conditions is the potential priming effect the intentional condition may have on the participant, due to the given information of facial recognition test after the stimulus. Condition A was told that the ‘ability to remember events and details will be tested in a short questionnaire’; this was referred to as the non-primed condition. Condition B instead informed the participants that their ‘ability to recognise the perpetrator will be tested in a short face identification task’ and referred to as the primed condition. The flow of the experiment consisted of the participant being assigned to one of the two conditions then they were shown the stimuli. Afterwards, they were made to complete the connected questionnaire, compiled of 13 closed questions, 2 open ended questions and 2 simulated line ups before their results are compiled and analysed.

Using the online survey software Qualtrics a study was created which consisted of an experimental section and a related compulsory questionnaire. The stimulus was also created through this software and compiled into one study investigating change blindness.

The average completion time for the experiment was 7 minutes and 58 seconds, however a large range was shown in the experiment with the shortest recorded time for a completed response being 3 minutes and 26 second but the longest being 25 minutes and 33 seconds.

The stimulus consisted of 20 photographs (a total of 1 minute 40 seconds) constructed into a continuous slideshow, with each slide having an automatic progression timer set on 5 seconds. The stimulus begins with an actress breaking into a flat using a lock pick. Over the next 9 slides the participant views the same woman proceeded to break into two rooms and steal a total of 5 items. Unannounced to the participant the actress committing the burglary changes at halfway point of the slideshow (50 seconds). Over the last 10 slides the second actress continues to steal items from a further 2 rooms before leaving the accommodation. All items shown in the stimulus were objects that would be found inside a student hall and the objects ranged from technological items such as a laptop to the more obscure choices like a teddy bear.

Both actress’ were chosen to participate because they were deemed to have a level of similarity between each other without looking near identical, a crucial factor needed for the experiment. This was displayed through a Likert scale (between 1-10) tested on 3 actresses’ all from South East Asian descent (see Figure 1) during pre-testing. This psychometric scale was presented to 10 individuals selected through opportunity sampling. Actresses 1 and 2 who were in the experiment received an average similarity score 7.1, with a 7 defined as ‘mostly similar’.  Actresses 1 and 3 received a score of 6.8 and Actress 2 and 3 received a score of 6.6 with 6 defined as ‘somewhat similar’.

A simulated police line-up consisting of 6 girls was shown immediately after the stimuli and then repeated at the end of the experiment.  Both line-ups included the same 6 girls the order of the facial shots was randomised and the photos in the first and last police line-up were consistent, this allowed for counterbalancing. The first line up asked the participants per suspect if they recognised the individual from the slideshow and to rate out of 10 how confident they are of their answer. Each line up used 6 different facial shots of girls, all of which were of Asian descent (including the two actresses). To be placed within the line-up, during pretesting each of the 4 new individuals must been given an average likeness rating to the original actresses that fell between 6.6 and 7.6 (± 0.5 from the original value of similarity between the actresses). This was achieved by using another 20 participants in the pretesting stage; using the same Likert rating scale the two actresses within the stimuli were chosen from. The 4 women who most closely fit the specification level of similarity, for the photographic line-ups, are chosen to participate in the police line-up. The options the participants were rating was a mixture of official mug shot photos and staged mug shot photos, to negate choices being affected by non-facial differences or different facial poses.

Then followed 13 questions designed to test the participants basic recall abilities in an eyewitness situation. 13 mandatory-closed questions separated into two pages one of 7 questions, one of 6 questions and both pages were internally randomized for question order. The reason for the separation of the 13 questions was due to the information given out by of some of the question. A selection of 12 of the 13 was questions focused on testing the participant’s ability to recall specific events and items that happened during the slide show (see Figure 2). The question “Did you see a male or female committing the crime?” was added with the sole intention to test if the participant had been paying attention to the stimulus, followed the instruction and understood the task, by answering a simple related question. Followed by a post recall test with two questions designed to detect the effects change blindness, these two questions appeared as a closed question with the options “Yes” or “No”. There was a hidden open-ended drop-down option. The restriction was dependent however on the participants who responded, both open ended questions were only visible as an additional ‘please explain your answer’ if the participant selected yes to the first part of the question (see Figure 3). Therefore, the accessibility the open-ended part was dependent if the participated detected a change or perceived something as unusual within the stimuli.

Finally, another 6-girl simulated line up, this time asking two questions both measured by sliders, firstly a familiarity rating (out of 10) and a probability rating that the face was in the stimuli (out of 100). The simulated police line-ups were presented at the beginning and the end the questionnaire section. Due to the 15 filler questions, in between the two police line-ups, both immediate recall of the assailant or in this case assailants and a delayed recall could be tested.

The final section, consisted of an identical 6-girl simulated line up, with the order the faces appear being randomised. This time two questions both measured by sliders were asked (see Figure 4). Participants were firstly instructed to complete a familiarity rating (out of 10) and a probability rating that the face was in the stimuli (out of 100).


Ethical Considerations


Before the experiment could be conducted, an ethical application to the Research Participation Scheme (RPS) and Kent University had to be approved. Every participant who completed the study was briefed prior to the experiment and debriefed at the end, whilst abiding by the ethical guidelines throughout. Ethical consideration was placed on the risk the experiment can have on the participant, allowing participants to answer anonymously followed their right to confidentiality.  At no point was the participant intentionally deceived unless deemed necessary to avoid the true aim of the experiment being found. All participants were informed that they could withdraw at any time and a contact address was left in both the introduction and debrief section (RPS, 2010).




Change Blindness Manipulation Check


The first examination of the results tested the manipulation check by analysing the interaction between the major independent and dependent variables. The results presented as a function of the conditions (incidental vs. intentional) and change (change detected or change not detected) is shown in Table 1. In total 20 participants selected that they noticed a change in the stimuli.

TABLE 1. Number of Participants Detecting Change as a Function of Memory Condition 

Condition Yes No
Incidental 9 34
Intentional 7 38
Total 16 72

Change detected

31% (n = 28) of individuals claimed they had noticed something unusual or perceived a change in the stimuli, however only 16 individuals correctly reported a change between the burglars, out of the 88 (18.18%) participants that were deemed eligible during the results collation process. The participants who highlighted the change within the stimuli were drawn disproportionately from the incidental group (n = 9) producing a large significant effect

X2(1, N= 88) = 19.62, p < .001. There was no gender difference between men and women, as both recorded equal amounts of detection of change (8 vs. 8, respectively), however with fewer male participants present in the sample their rate of detection was higher in both conditions. Although to allow the best comparison between the experiment and the original (Davies & Hine, 2007), a hierarchal log linear analysis would be the preferred analytic technique when the interaction of gender was added to the factors of the conditions and change detection. Due to the low abundance of significant data this however was proven impractical as a method. An ANOVA statistical test was again used to test the significance of gender on the change detection. On this occasion however a 2×2 analysis of variance was utilised with condition and awareness of change labelled as the main effects and gender used as a covariate. The analysis showed gender exerted a significant effect; men were more likely to notice the change then females,

F(1, 88) = 2564.33, p < .001 d = .66. Through examining the Tukey test it showed that males had a significantly higher detection of change rating (p < .05). Additionally it allowed the significance between the conditions and change detection to once again be analysed showing the incidental condition was more likely to perceive change.


Confidence and Accuracy of Correct Identifications

Participants were shown a series of 6 randomised photographs, consisting of Burglar 1, 2 and 4 filler choices. Participants were divided into three separate groups, detection of no suspects, the detection of one suspect and the detection of both burglars. The data shown in Table 2 portrays this with the figures presented as a function of change blindness detection and experimental condition. As aforementioned in this case a log linear analytical method would be preferred to allow for a direct comparison between the results of the original study (Davies & Hine, 2007), however this was not possible due to the data set. However a Spearman’s Rank test was used to test the relationship between individuals who detected the change and their abilities to correctly identify the correct suspects. The results showed a trivial significant effect between the two variables on the 0.1 level of analysis,

X2(1, N=88), p < .1, indicating that if the change was detected a participant would be more likely to select either one or both of the burglars.

52 participants in total chose one of either Burglar 1 or Burglar 2, however this was not evenly distributed. Of the participants in both conditions who selected one of the two actresses, Burglar 1 (n = 17) was identified a lot less (51.4%) than Burglar 2 (n = 35). 19.32% (n = 17) selected both actors in the stimulus.  The significance between the frequencies of choices of correct identification was deemed partially significant through a binomial statistical test (p ≥ .1). Both Burglar 1 (6.74, σ = 1.85) and Burglar 2 (6.77, σ = 1.77) received the highest confidence ratings out of all suspects in the identity parade. Individuals who selected that Burglar 1 was not the correct suspect however received a higher average incorrect identification confidence rating of 6.78 (σ = 2.24). The incorrect identification confidence rating for Burglar 2 was slightly lower than the correct confidence rating with an average rating of 6.72 (σ = 1.96).

TABLE 2. Number of Participants that Correctly Identified the Actresses in the Stimulus as a Function of Memory Condition and Change Detection 



Identification of actresses


Incidental Condition
Change Not Detected 9 20 10
Total 12 24 12
Intentional Condition
Change Not Detected 5 25 3
Total 7 28 5

Additional Scales for Correct Identification


In addition to a sliding scale (slider) out of 10 being added to test the confidence of the participant’s choice in their selected perpetrator, at the end of the study two more sliders were added.  These sliders were used to test the participants’ familiarity out of 10 and the probability that individual was in the study out of 100%. The results of the two perpetrators displayed as a function of correct suspect identification and rating systems (familiarity and probability) are displayed in Table 3. Results showed that Burglar 1 received a far lower familiarity average (∆ = 1.57) and probability (∆ = 16.44%) within the slideshow. This coincided with results of the correct identification of both suspects, however the difference in the familiarity and probability levels were far higher than the identification results would suggest. The results of the familiarity scale showed a similarity between the standard deviation (σ) of the data sets between the first and second burglar (2.62 and 2.69 respectively). The same applied for the data set acquired from the probability scale results between Burglar 1 and Burglar 2 (24.80 and 25.83 respectively).

TABLE 3. Participants Detecting Correct Suspect as a Function of Rating Systems 

Perpetrator Familiarity Probability (%)
Burglar 1 4.76 46.23
Burglar 2 6.33 62.67
Mean 5.55 54.45

Rating system




Content Questionnaire


Performance on the content questionnaire is displayed as a function of condition (incidental vs. intentional) and change (detected or not detected) in Table 4. To test the relationship between the participants who noticed the change and their results in the content questionnaire, a two-way Analysis of Variance test was conducted. Detection of the change within the stimulus was found to be significantly correlated with the results of the participants in the content questionnaire,

F(1, 88) = 3.73, p < 0.1, d= .5. On average individuals who did not perceive the change, in both conditions, scored higher than individuals who did not perceive the change.

TABLE 4. Number of Participants Detecting Change as a Function of Memory Condition 

Change detected

Condition  M SD
Incidental 8.00  1.32
Intentional 8.14 1.07
Total 8.07 1.45
Condition  M SD
Incidental 8.68  2.36
Intentional 8.60 1.92
Total 8.64 2.14

Change not detected

The content questionnaire was marked out of a total of 12, as the question “was the perpetrator male or female”, was omitted from the results due to the question being a test of the participants concentration on the initial rules, which stated the gender of the perpetrator, and not a question regarding the content of the stimulus. In addition, all 88 participants answered this correctly, thus not affecting the statistical analysis in any condition. Participants in the intentional condition who noticed the change received and average score of 8.14 (σ =  1.07), however perceiving the change but being placed in the incidental condition gave an average content questionnaire score of 8.00 (σ =  1.32). This gave an average of 8.07 for participants who perceived the change. The average score for participants that didn’t notice the change was 8.64. Of these participants, the ones placed within the incidental condition scored an average of 8.68 (σ =  2.36); the participants in the placed in the intentional condition scored a combined average of 8.60 (σ =  1.92). The conditions of incidental vs. intentional however has no significant effect on the likelihood of a high ratio on the content questionnaire.  By being placed in the incidental group, a participant was no more likely to score higher in the content section,

F(1, 88) = .013, p > 0.1, d = .5. During the content section females on average scored higher (8.45, σ = 2.01) than males (7.96 σ = 1.49).


In total there were 4 additional filler suspects added to the 2 perpetrators mug shots during the eyewitness identification parade. Each participant would be asked to give the 4 incorrect suspects (one at a time) an identification confidence rating (out of 10), a familiarity rating (out of 10) and a probability rating (out of 100%). The results of this experiment are displayed as a function of potential suspect misidentification (filler suspects 1-4) and rating systems (confidence, familiarity and probability) in Table 5. As a participant was not limited to picking one or two suspects as they were individually presented in the parade, an individual could potentially select all 6 suspected criminals. In total there were 60 misidentifications between the 4 filler suspects. The performance of participants whose choices resulted in a misidentification is shown as a function of the displayed potential perpetrator and recorded identification score is displayed in Table 5. 11 (12.50%) individuals incorrectly identified the first filler suspect as the perpetrator of the crime.  23 (26.13%) individuals misidentified the second filler suspect, meaning that it was the most misidentified suspect during the identity parade. The third filler suspect was select by 7 (7.95%) participants in total making it the least selected misidentification within the study.  The final filler was the penultimate highest misidentified suspect with a total of 19 (21.59%) participants selecting the woman as committing the crime.

TABLE 5. Participants Rating Scores as a Function of Potential Suspect Misidentification 

Filler Suspect Confidence Familiarity Probability (%)
1 5.45 2.38 23.82
2 6.70 3.67 38.73
3 6.00 2.49 26.50
4 6.05 4.18 43.13




Change Detection


When placed side by side the visual differences between Burglar 1 and Burglar 2 shows dissimilarities in the build of the body, the height and the length of the hair between the two women.  Although the two actresses appeared to look similar, the facial differences are clear when analysed and compared. Despite the various differences only 16 individuals detected the change out of the 88 who completed the experimental phase of the study. Giving an overall detection percentage of 18.18% the results showed a significantly low detection ratio compatible to the recent research into change blindness that utilised visual stimulus (Davies & Hine, 2007; Levin & Simons, 1997; Nelson et al., 2011). It was thus predicted from findings by Davies & Hine (2007) that a high level of participants (>50%) would not recognise the central change to the stimuli. It was first predicted that the placement of a participant in the intentional condition would prime the detection of change blindness. The results however did not correspond with the findings of Davies et al. as 56.25% of the participants who detected the change were in the incidental condition. The incidental condition however would be the condition present in a real world application of the findings of this experiment. As a bystander would most likely witness a crime without prior information of the upcoming event, the significant effect on the independent variable of the incidental condition shows significance for its effects.

Although the detection showed an equal number of males and females noticed the change. The smaller ratio of male to female participants (32 to 56 respectively) showed a significantly higher proportion of men (25%) noticed the change. This finding contradicts suggestions by past research that there is either no gender difference between male and female eyewitness detection (Yarmey, Yarmey & Yarmey, 1996) or that a woman can remember faces better than men (Clifford & Bull, 1980). Recent supporting literature however has put forward for consideration the idea that males and females may process different visual information when viewing a strangers face (Hall, Hutton & Morgan, 2010), which may have applied to the burglars present in the stimulus.

Correct Identification


Burglar 1 was selected by 19.32% of the participants compared to the 39.77% that selected Burglar 2. The reason for this may be the fact that Burglar 2 appeared last in the slideshow leaving it in the participant’s immediate memory.  Memory that passes through the short-term storage suffers from deterioration due to time decay (Broadbent, 1958) and the capacity difference, which suggests that information in quick succession, can be overwritten by new information (Cowan, 2008). This could not apply for all participants however because 19.32% (n = 17) of the participants selected both actors in the stimulus. As predicted the participants present in the study who detected the change were significantly more likely to then be able to identify either one or both of the correct suspects. A sliding scale ranking the self-confidence rating of the participant’s choice on a range of 0 (Not confident at al) to 10 (Very confident) was added.  This was a major modification from the original study by Davies and Hine (2007), added to see what effect the change had on people’s perception of the suspects. Allowing a more in depth look than the yes or no response to the identity parade that was displayed in prior research (Davies & Hine, 2007; Nelson et al, 2011)

The analysis of this data shows that participants who did choose either Burglar 1 (6.74) or Burglar 2 (6.77) was given the highest confidence ratings out of all 6 suspects in the identity parade. The first and the second correct suspects also received the highest confidence ratings when incorrectly selected as not being recognised from the slideshow (6.78 and 6.72 respectively).  An additional identity parade was added at the end of the study to allow for counterbalancing and a comparison of results immediately and after a delayed period. Additional sliders were added in this phase to test the familiarity of each suspect (scale from 0 – 10) and to measure the probability that each face was in the stimulus (scale 0 – 100%). The results of these sliders supported the first set of data obtained from the identity parade as Burglar 1 received a lower familiarity average (1.57) and a probability score (6.44%) within the slideshow.

Conclusion of the Findings from the Content Test


The content questionnaire section showed that the mean participant score was 8.36 out of 12. There was a negative correlation shown between change detection and the score of the content test. The participants who perceived the change received and average score of 8.07 compared to the 8.64 for the participants who did not. It was predicted that the conditions the participant was placed into would have an effect on the content scores. As the incidental condition informed the observer to pay specific attention to the surrounding areas it was believed that the scores would be higher, as the questionnaire tested the memory of items and events during the stimulus. Participants who detected the change in the incidental condition scored lower (8.00) than the intentional condition (8.14). For the participant who did not detect the change in the incidental condition (8.68) followed the prediction with a higher score than the intentional condition (8.60). When separated solely into conditions, intentional (8.37) had a slightly higher average than incidental (8.34). The difference between the conditions the participants were placed into (intentional and incidental) was too minute and therefore did not play a significant effect on the results.



The same four additional potential suspects were added to both identity parades, the order however was completely randomised from participant to participant. The appearance of each of the filler suspects during this stage was dependent on a score of similarity to the perpetrators during pretesting. It was understood that there was a possibility for a participant to select that they recognised all 6 suspects. The most chosen by any one participant however was the recognition of 2 different individuals.  Although there were a total of 60 misidentifications in the study, this result cannot be analysed as 60 participants misidentifying the target, as two incorrect identifications can occur from one participant. The high levels of misidentification nevertheless cannot be ignored and a discussion of the analysis of the data for the misidentified individuals must be noted.  The mean level of identification showed that any filler suspect placed in the line up, assuming a ± 0.5 rating from the original value of similarity between the burglars, the suspect would have 17.04% chance of being misidentified by the participants. The highest filler suspect received a misidentification from 26.13% of the participants (n=23). The lowest received a misidentification from 7.95% (n=7) of the total 88 participants. The participants who misidentified the suspects display a marginally lower average confidence rating (6.05) than the participants who were correct in their identification (6.76). The results of the familiarity and the probability scale however show a much lower score in comparison, highlighting a potential cognitive dissonance over the true offenders of the crime in some participants.

The finding are proved to have relevance because at least 1.61 million individuals in the last year dealt with the criminal justice system, with 1.43 million being prosecuted and an average conviction ratio of 84%, this issue has a huge relevance (Ministry of Justice, 2016). This is due to the fact that eyewitness testimonies still play a major part in the United Kingdom criminal justice system as sufficient DNA evidence is found at less than 1% of crime scenes (Rowley, 2009). Concluding from this data it is clear that if a change occurred in a stimulus, there could be major ramifications for the eyewitness identification abilities of the bystander. It is shown that not only does the successful identification of both correct individuals appear challenging, with only 19.32% of the subjects correctly identifying Burglar 1 and Burglar 2; it is also highly more likely that under these conditions an innocent with a resemblance to the burglar will be misidentified.


Limitations and Improvements


Change blindness was detected by only a small proportion of the participants (18.18%), a statistic considerably lower than the 39% detection rate that was stated in Davies & Hine (2007). A series of photos was utilised as opposed to the preferred medium of video recordings of the event seen in past literature (Nelson et al., 2011; Triesch, Ballard, Hayhoe, Sullivan, 2003; See also Davies & Hine, 2007; Williams & Simons, 2000). In light of the experimental results, it is essential that in future research it be tested that the presentation of the stimulus as an automated slideshow did not have a significant effect on the results.  Any replication of the study however would allow for an increase in the reliability of the results. The difference in sample size between Davies et al. (2007) and this current study was small (80 vs. 88 respectively), but by further increasing this sample, there will be a subsequent expected increase in the accuracy of the results.

The use of this study on an online platform ensured the researcher and the subjects had minimal contact, negating potential effects of participant bias, a factor that has been a large criticism of studies relating to eyewitness identification (Dysart, Lindsay, Hammond & Dupuis, 2001). Gender was equally represented in relation to the placement in the intentional and incidental condition. However the number of males (n=32) present was less than females (n=56), therefore to allow for a reduction in gender specific biases, future research should focus on an evenly gendered sample. The actresses who played Burglar 1 and 2 were also both female consequently a similar experiment with two male participants would be advised.  Current research has started to focus on race effects in relation to change blindness (Humphreys, Hodsoll & Campbell, 2005). An exploration into the effects of race on change blindness within an eye witness testimony would also be advised in future research. The sample was obtained through two means: Through an email link sent by the researcher or for credits on the RPS website. This caused the participant pool to be heavily dominated by individuals from the South East of England. The low geographical variety of the sample could potentially have had an effect on the results, to test this any future research with results specific to the United Kingdom should base a sample exploring a larger percentage of the country. The age range was also limited through the methods of obtaining participants; therefore a large proportion of the U.K’s population would not be represented. Finally a further exploration of the results indicating that the incidental condition as opposed to the intentional condition was significant in the detection of change blindness would be beneficial, as that was a major unexpected result.



Regarding the results of the experiment, the low detection of the change in the stimulus supports the findings of Davies and Hine (2007). The poor accuracy by the eyewitnesses’ in the detection of change and the reduced ability to successfully identify the perpetrators, demonstrate a clear relationship between the two. Upon analysis of the results of this experiment and prior research, the suggestion that change blindness strongly affects the participant’s ability to adequately function as a good eyewitness is warranted. The implications of these results could potentially be a large attributing factor to the high levels of overturned convictions that have came to light due to DNA testing. Eyewitness identification has played a major role in over 70% of all the criminal sentences that have been successfully appealed by the foundation, innocence project. When this is examined in relation to change blindness literature, a movement towards the ideas presented in the Devlin Report (1976) is suggested. As the report states due to the problems of inaccuracy of an eyewitness testimony, the account alone should not cause a conviction. As it is possible to imagine a situation where change blindness may occur, such as a bystander who may confuse a criminal entering a crime scene and an innocent member of the public who leaves at a later time. Or specifically relevant to the study, the confusion of seeing two burglars committing a crime and believing it as only one (Davies & Hine, 2007), an example of unconscious transference (Davis, Loftus, Vanous & Cuccaire, 2008). The overreliance on the reliability of eyewitness testimonies in legal systems around the world must be questioned.  To conclude due to the findings presented prior to this experiment and the results of this experiment, it is important to note that any eyewitness identification that has taken place in the past, present or future, must be analysed and utilised with caution.





Reference list


Aginsky, V., & Tarr, M. (2000). How Are Different Properties of a Scene Encoded in Visual Memory?. Visual Cognition, 7(1-3), 147-162.

Angelone, B., Levin, D., & Simons, D. (2003). The Relationship between Change Detection and Recognition of Centrally Attended Objects in Motion Pictures. Perception, 32(8), 947-962.

Atkinson, R., & Shiffrin, R. (1969). Storage and retrieval processes in long-term memory. Psychological Review, 76(2), 179-193.

Atkinson, R., & Shiffrin, R. (1971). The Control of Short-Term Memory. Scientific American, 225(2), 82-90.

Baddeley, A., & Hitch, G. (1974). The psychology of learning and motivation: Advances in research and theory. New York: Academic Press., 8, 47-89.

Bahrick, H., Bahrick, P., & Wittlinger, R. (1975). Fifty years of memory for names and faces: A cross-sectional approach. Journal Of Experimental Psychology: General, 104(1), 54-75.

Benedict, R. (1997). Brief visuospatial memory test – revised: Professional manual. Psychological Assessment Resources, Inc.

Benedict, R., & Groninger, L. (1995). Preliminary standardization of a new visuospatial memory test with six alternate forms. The Clinical Neuropsychologist, 9(1), 11-16.

Benedict, R., Schretlen, D., Groninger, L., Dobraski, M., & Shpritz, B. (1996). Revision of the Brief Visuospatial Memory Test: Studies of normal performance, reliability, and validity. Psychological Assessment, 8(2), 145-153.

Borchard, E. (1932). America and a court of justice for the nations (1st ed.). Washington, D.C.: American Peace Society.

Brelsford, J., Keller, L., Shiffrin, R., & Atkinson, R. (1966). Short-term recall of paired-associates as a function of the number of interpolated pairs. Psychonomic Science, 4(1), 73-74.

Broadbent, D. (1958). Perception and Communication. London: Pergamon Press..

Clifford, B., & Bull, R. (1980). The psychology of person identification. European Journal Of Social Psychology, 10(1), 97-97.

Cowan, N. (2008). Chapter 20 What are the differences between long-term, short-term, and working memory?. Progress In Brain Research, 323-338.

Davies, G., & Griffiths, L. (2008). Eyewitness Identification and the English Courts: A Century of Trial and Error*. Psychiatry, Psychology And Law, 15(3), 435-449.

Davis, D., Loftus, E., Vanous, S., & Cucciare, M. (2008). ‘Unconscious transference’ can be an instance of ‘change blindness’. Applied Cognitive Psychology, 22(5), 605-623.

Devlin, P. (1976). Report to the Secretary of State for the Home Department on the Departmental Committee on Evidence of Identification in Criminal Cases. London: HMSO, 63-78.

Dysart, J., Lindsay, R., Hammond, R., & Dupuis, P. (2001). Mug shot exposure prior to lineup identification: Interference, transference, and commitment effects. Journal Of Applied Psychology, 86(6), 1280-1284.

Eyewitness Misidentification – Innocence Project. (2017). Innocence Project. Retrieved 2 May 2017, from

Fitzgerald, R., Price, H., & Oriet, C. (2013). Intentionally forgetting other-race faces: Costs and benefits?. Journal Of Experimental Psychology: Applied, 19(2), 130-142.

Flowe, H., Smith, H., Karoğlu, N., Onwuegbusi, T., & Rai, L. (2015). Configural and component processing in simultaneous and sequential lineup procedures. Memory, 24(3), 306-314.

Green, M., Nuechterlein, K., Gold, J., Barch, D., Cohen, J., & Essock, S. et al. (2004). Approaching a consensus cognitive battery for clinical trials in schizophrenia: The NIMH-MATRICS conference to select cognitive domains and test criteria. Biological Psychiatry, 56(5), 301-307.

Grimes, J. (1996). On the failure to detect changes in scenes across saccades. New York: Oxford University Press., 2, 89-110.

Gross, S., & Shaffer, M. (2012). Exonerations in the United States, 1989 – 2012 Report by the National Registry of Exonerations (1st ed., pp. 24-62). Michigan: University of Michigan Law School. Retrieved from

Haber, R. (1983). The impending demise of the icon: A critique of the concept of iconic storage in visual information processing. Behavioral And Brain Sciences, 6(01), 1.

Hall, J., Hutton, S., & Morgan, M. (2010). Sex differences in scanning faces: Does attention to the eyes explain female superiority in facial expression recognition?. Cognition & Emotion, 24(4), 629-637.

Hine, S., & Davies, G. (2007). Change blindness and eyewitness testimony. The Journal Of Psychology, 141(4), 423-434.

Individuality. E. L. Thorndike. (1914). The School Review, 22(7), 497-497.

Jensen, M., Yao, R., Street, W., & Simons, D. (2011). Change blindness and inattentional blindness. Wiley Interdisciplinary Reviews: Cognitive Science, 2(5), 529-546.

Levin, D. (2010). Visual metacognitions underlying change blindness blindness and estimates of picture memory. Journal Of Vision, 1(3), 10-10.

Levin, D., & Simons, D. (1997). Change blindness. Trends In Cognitive Sciences, 1(7), 261-267.

Levin, D., & Simons, D. (1997). Failure to detect changes to attended objects in motion pictures. Psychonomic Bulletin & Review, 4(4), 501-506.

Levin, D., Simons, D., Angelone, B., & Chabris, C. (2002). Memory for centrally attended changing objects in an incidental real-world change detection paradigm. British Journal Of Psychology, 93(3), 289-302.

Loftus, E., & Palmer, J. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal Of Verbal Learning And Verbal Behavior, 13(5), 585-589.

Loftus, E., Davis, D., Knowles, E., Fowler, N., Laney, C., & Nelson, K. (2011). Change blindness can cause mistaken eyewitness identification. Legal And Criminological Psychology, 16(1), 62-74.×482625

Mack, A. (2003). Inattentional Blindness. Current Directions In Psychological Science, 12(5), 180-184.

Mack, A., & Rock, I. (1998). Inattentional blindness. Cambridge, MA: MIT Press..

McConkie, G., & Currie, C. (1996). Visual stability across saccades while viewing complex pictures. Journal Of Experimental Psychology: Human Perception And Performance, 22(3), 563-581.

Ministry of Justice – Statistics at MOJ – GOV.UK. (2016). Retrieved 3 May 2017, from

Morey, C., Cilo, M., Berry, J., & Cusick, C. (2003). The effect of Aricept® in persons with persistent memory disorder following traumatic brain injury: a pilot study. Brain Injury, 17(9), 809-815.

Munsterberg, H. (1908). On the witness stand. Harvard Law Review, 22(1), 73.

Nelson, K., Laney, C., Fowler, N., Knowles, E., Davis, D., & Loftus, E. (2011). Change blindness can cause mistaken eyewitness identification. Legal And Criminological Psychology, 16(1), 62-74.×482625

NOVA | PBS. (2012). Change Blindness. Retrieved 2 May 2017, from


Peterson, L., & Peterson, M. (1959). Short-term retention of individual verbal items. Journal Of Experimental Psychology, 58(3), 193-198.

Phillips, J., Shiffrin, R., & Atkinson, R. (1967). Effects of list length on short-term memory. Journal Of Verbal Learning And Verbal Behavior, 6(3), 303-311.

Rensink, R. (2002). Change Detection. Annual Review Of Psychology, 53(1), 245-277.

Rensink, R., Kevin O’Regan, J., & Clark, J. (2000). On the Failure to Detect Changes in Scenes Across Brief Interruptions. Visual Cognition, 7(1-3), 127-145.

Rensink, R., O’Regan, J., & Clark, J. (1997). To See or not to See: The Need for Attention to Perceive Changes in Scenes. Psychological Science, 8(5), 368-373.

Rowley, S. (2017). Wrongful conviction throws spotlight on unreliability of eyewitness evidence. the Guardian. Retrieved 3 May 2017, from

Sammon, N., & Bogue, J. (2015). The Impact of Attention on Eyewitness Identification and Change Blindness. Journal Of European Psychology Students, 6(2), 95-103.

Simons, D., & Chabris, C. (1999). Gorillas in Our Midst: Sustained Inattentional Blindness for Dynamic Events. Perception, 28(9), 1059-1074.

Simons, D., & Levin, D. (1998). Failure to detect changes to people during a real-world interaction. Psychonomic Bulletin & Review, 5(4), 644-649.

Simons, D., & Rensink, R. (2005). Change blindness: past, present, and future. Trends In Cognitive Sciences, 9(1), 16-20.

Steblay, N. (1997). Social influence in eyewitness recall: A meta-analytic review of lineup instruction effects. Law And Human Behavior, 21(3), 283-297.

Steblay, N., Wells, G., & Douglass, A. (2014). The eyewitness post identification feedback effect 15 years later: Theoretical and policy implications. Psychology, Public Policy, And Law, 20(1), 1-18.

Tam, J., & Schmitter-Edgecombe, M. (2013). The Role of Processing Speed in the Brief Visuospatial Memory Test – Revised. Retrieved 2 May 2017, from

Thorndike, E. (1971). The fundamentals of learning (1st ed.). New York: AMS Press.

Triesch, J., Ballard, D., Hayhoe, M., & Sullivan, B. (2003). What you see is what you need. Journal Of Vision, 3(1), 9-9.

Wells, G., & Bradfield, A. (1998). “Good, you identified the suspect”: Feedback to eyewitnesses distorts their reports of the witnessing experience. Journal Of Applied Psychology, 83(3), 360-376.

Williams, P., & Simons, D. (2000). Detecting Changes in Novel, Complex Three-dimensional Objects. Visual Cognition, 7(1-3), 297-322.

Yarmey, A., Yarmey, M., & Yarmey, A. (1996). Accuracy of eyewitness identifications in showups and lineups. Law And Human Behavior, 20(4), 459-477.



Leave a Reply