Toward a Systematic Understanding of Eyewitness Evidence
Eyewitness identification research has typically been focused either on isolated specific factors or on more broadly defined and more ecologically valid contexts that occur in more realistic but less-controlled crime simulations. However, a paucity of studies have addressed both controlled and realistic contexts, investigating specific factors systematically in context. In the present study, researchers have addressed the effects of the following circumstances on memory: stimulus complexity, gender of assailant, gender of respondent, presence of weapons, clothing and physical characteristics of assailants, peripheral sources of hazard, and seeing perpetrators in lineups. Standardized contexts were used to strengthen experimental controls. Results indicated that eyewitness memory, even under idealized conditions, was highly unreliable, both for perpetrator characteristics and for other aspects of the crime scene, including weapons and peripheral sources of hazard. These results provide important information about the elements of eyewitness testimony that are most likely to yield difficulties in both investigative and courtroom settings.
THE FORENSIC EXAMINER Fall 2007
Fall 2007 THE FORENSIC EXAMINER
This article is approved by the following for continuing education credit:
(ACFEI) The American College of Forensic Examiners International provides this continuing education credit for Diplomates.
(CFC) The American College of Forensic Examiners International provides this continuing education credit for Certified Forensic Consultants.
(APA) The American College of Forensic Examiners International is approved by the American Psychological Association to sponsor continuing education for psychologists. ACFEI maintains responsibility for this program and its content.
By Matthew J. Sharps, PhD, DABPS, FACFEI; Adam B. Hess, MA; Hilary Casner; Bethany Ranes, BA; and Jenna Jones, BA
Key Words: eyewitness memory, recognition memory
Over the past century, numerous studies have addressed eyewitness identification. Many sophisticated studies have addressed individual factors of importance, such as exposure time, suspect race, witness gender, or weapon presence (e.g., Narby, Cutler, & Penrod, 1996), typically in isolation. Such studies of isolated variables have been designed to improve experimental control. However, isolating such critical factors from their contexts, while increasing experimental control, has typically reduced ecological validity (for review, see Sporer, Malpass, & Koehnken, 1996; also, see Kassin, Tubb, Hosch, & Memon, 2001).
Other research has been directed at field studies, video, or live simulations of crime events. However, these studies, which improve on ecological validity, are necessarily less controlled (Morgan et al., 2004) and, consequently, are more difficult to interpret. There is a lack of systematic research on eyewitness testimony that bridges the gaps between isolated variables with better experimental controls and more realistic ecological variables with better validity. Additionally, whether an isolated-variable focus or a more ecological approach, there has been little research on eyewitness memory for forensically important elements other than identification of persons (elements such as recall of weapons and peripheral sources of hazard) (e.g., Sharps, Barber, Stahl, & Villegas, 2003; Villegas, Sharps, Satterthwaite, & Chisholm, 2005).
The integration of experimental control with more realistic environmental validity formed the foundation for the studies reported in this research. The experiments described in this research were designed to enhance the development of a systematic contextual understanding of eyewitness memory, which would help bridge the gap between isolated variables in experimentally controlled designs and environmental variables that are more realistic, but less controlled. Also, the present research was designed to address the broader context of eyewitness memory to include important inanimate elements of crime scenes while identifying the narrower specifics for each individual to get the highest degree of experimental precision possible.
In previous studies, Sharps et al. (2003) and Villegas et al. (2005) utilized this integrated tradition of addressing relatively ecologically valid contexts in experimentally controlled environments—in the important, but hitherto neglected, areas of weapon and vehicle identification. These studies were conducted under conditions of ideal lighting, exposure time, and freedom from contextual distraction. Even under these ideal conditions, which would be considerably diminished under the real-world conditions of a typical crime-scene, eyewitness memory was shockingly poor.
In our study of weapon identification (Sharps et al., 2003), response rates for semi-automatic pistol identification were correct in fewer than half the cases. More familiar handgun types were better recognized than less familiar types, but even familiar revolvers were recognized correctly only 43% of the time. Even assault weapons, which are used less frequently in crimes but are larger in size and have greater complexity of features and, consequently, are subject to a higher level of feature-intensive processing (Sharps, 2003; Sharps & Nunes, 2002), were correctly recognized less than three-quarters of the time. These data strongly suggest appreciable reconfiguration of memory for weapons after initial encoding at the given crime scene.
A similar study of vehicle identification (Villegas et al., 2005) yielded even poorer results. Again under idealized conditions, vehicles were identified correctly, on average, only one quarter of the time.
In Villegas et al.’s study (2005), vehicle color was more important in the generation of errors than vehicle model. This result was consistent with the Gestalt/Feature-Intensive Processing (G/FI) theory of visual cognition, which has proven useful in several areas of eyewitness memory research (Sharps, 2003; Sharps et al., 2003; Sharps & Nunes, 2002). Color, an identifiable feature of an object that may readily be processed linguistically, was more important in the generation of errors than was the vehicle model, which depends on the less linguistically-accessible shapes of the various structural elements of the given vehicle body type (see Sharps & Nunes).
The results of this research may be helpful in explaining such apparent anomalies as occurred in the Washington sniper case, in which the suspect vehicle, a dark-blue Chevrolet Caprice, was apparently misidentified by multiple witnesses as a white or cream-colored van (Blades, 2005). In this case, of course, neither color nor vehicle model were particularly helpful to witness memory. The repetition of the error is explicable in view of the relatively poor performance of witnesses in our laboratory situation, in which factors of darkness, brief exposure, occlusion, movement, and perceived danger had not been present. In these studies, memory for vehicles and for weapons was reconfigured in young, healthy witnesses to an extent that reliable eyewitness identification was unlikely.
The fact that such reconfiguration of memory occurs has been known for more than 70 years. Bartlett (1932) showed that memory is not the static, videotape-like system of popular imagination. Rather, memories become reconfigured in three primary ways. First, memories become shorter and more abbreviated. Second, details are lost, leaving a general representation upon which information obtained after the fact may act to alter elements of initial memory that may or may not have been correct in the first place. Finally, personal belief can alter memory significantly in both the visual and verbal realms. Bartlett showed that custom and cultural beliefs significantly reconfigured British subjects’ memories of Native American stories and that relatively abstract figures were recalled by respondents as specific, meaningful pictures of everyday objects if suggested to be such by the experimenter. Modern work in several venues has confirmed the major points originally described by Bartlett (e.g., Ahlberg & Sharps, 2002; Bergman & Roediger, 1999).
Research in G/FI theory has shown that these various phenomena may be encapsulated along a single processing dimension (Sharps, 2003; Sharps & Nunes, 2002). People in eyewitness situations typically encode a given original memory in a gestalt manner, with little attention to details that might assist in making a correct identification.
However, as the witness reviews the given memories over time, especially under circumstances in which others such as attorneys and investigators are directing and framing the viewpoint from which the events are considered, reviewed, and reported, the resultant feature-intensive analysis will process both actual/true details as well as details that have been derived from a reconfigurative process or from post-event suggestions. As Bartlett demonstrated, such details may readily be fabricated wholly unconsciously on the part of the witness from vague impressions, post-event information, or other sources. Thus, researchers can now explain the respondents’ tendencies to focus on feature-intensive factors such as vehicle color or on the greater number of identifiable features presented by a large assault weapon than by a small, simple handgun. Witnesses may search their memories for veridical features of the given crime situation, but then focus instead on inconsequential, inaccurate, or even accidentally fabricated elements that alter their memories further from the underlying genuine memory stimuli.
What happens to eyewitness memory in a context where features and details must be sorted and separated from visual situations with differing elements such as weapon presence and complexity? The experiments reported herein are initial attempts to cohere variable-by-variable studies of the variable components of eyewitness memory typically addressed solely in isolation, with more ecologically valid but less controlled studies employing simulated live or video-based crime scenes. In this initial effort, researchers employed high-quality digital photographs of ecologically valid crime scenes developed under the advice and supervision of expert senior field-training officers of the Fresno, California, Police Department. These officers were highly experienced in tactical realities and in the sorts of situations encountered by witnesses and officers on the street.
The photographs employed in these studies depicted a potentially violent crime scene in which either a male or a female perpetrator appeared armed with a Beretta handgun. The scene was either simple, sparse in terms of potentially distracting objects, or complex, including a victim being threatened by an armed perpetrator and typical street clutter. Peripheral sources of hazard were also included in the complex scenes. Researchers included in the complex scene an explicit explosive device (i.e., a disarmed and hollowed-out surplus hand grenade, which, from outward appearance, could not be distinguished from a live weapon) and a potentially implicit explosive device (i.e., a military-surplus ammunition box which may or may not have contained live explosives, but which was clearly out of place in the street scene and was placed next to the hand grenade, near the perpetrator). These crime elements were included because such weapons and ambush tactics have become increasingly important in real-world criminal and terrorist operations (Gelles, 2006).
For the sake of control in the case of weapon identification, in one full set of conditions (either a simple or complex context, and each with either male or female perpetrator), the Beretta was replaced by a power screwdriver, which the perpetrator might have been pointing or waving for emphasis in a non-lethal argument with the victim.
The conditions for all scenes involved uniformly excellent lighting (strong sunlight), extended exposure time, and the relative comfort of witnesses being seated 10–20 feet from a standard white movie screen, which they faced and on which the given scene was projected. There was no movement or occlusion of important elements of the scene, and of course, there was an absence of personal danger. As in our previous work (Sharps et al., 2003; Villegas et al., 2005), these idealized conditions were selected with the knowledge that an actual crime scene might well involve uncertain lighting and potentially occluded conditions and would typically result in poorer performance of eyewitnesses because of established dynamics of human visual cognition (e.g., Narby et al., 1996; Spoehr & Lehmkule, 1982).
The participants in this study were 149 women (mean age 19.86 years, SD = 2.48) and 49 men (mean age = 21.10 years, SD = 4.59) recruited from freshman psychology classes at California State University, Fresno. Gender proportions reflect the proportions of the classes. Also, this population was generally younger and probably in better health than the population at large. A Snellen test of visual acuity was performed, and all respondents possessed at least 20/40 eyesight corrected or uncorrected. Each respondent viewed only one scene (simple or complex, with a male or female perpetrator either armed or holding the power tool).
In real-world eyewitness identification situations in general, exposures to critical stimuli are frequently brief and fragmented. Witnesses may be under stress and may be operating under elevated levels of arousal. Also, elements that might be important for later identification may actually be obscured in real crime situations by vehicles, buildings, curbs, or other contextual features. Further research should address eyewitness memory under systematically varied conditions of these critical parameters, and efforts along these lines are currently in progress in this laboratory. However, in this preliminary research, researchers sought to evaluate identification under ideal conditions and to establish baseline data as outlined above. Therefore, the condition of an extended exposure time was imposed—a full 5 seconds. Law enforcement experts suggest that, generally, a firearm assault situation, such as the one depicted in this study, tends to develop rapidly and may result in a violent conclusion in literally less than a second (e.g., Moore, 2006; Montejano, 2005; Tietjen, 2005). Therefore, respondents in this research had substantially more than the typical period of observation and processing time.
In order to evaluate witness performance, the aid of senior, experienced Fresno Police field training officers were enlisted to produce a realistic police interview focused on the scenes viewed by respondents. The interview administered requested information on the perpetrator’s dress, physical characteristics (as both general verbal descriptions and as specific numerical estimates of height, weight, and age), weapon type if any, and the presence of other sources of hazard in the scene. This information was collected via multiple methods for each topic. Respondents were scored on each item in an objective scoring system from 1 to 3: 3 reflected a correct answer, 1 reflected an incorrect answer, and 2 reflected uncertainty or an answer too vague to be of use.
These scores were combined and summed into specific indices: a clothing index (composed of 3-point scales for hat, coat, shirt, and shoes), and a physical characteristics index (composed of 3-point scales for complexion, facial hair, marks or tattoos, height, weight, and age). Numerical estimates for the latter three elements of this index were also obtained, although these were not entered into the combined score to avoid potential redundancy. Finally, there was an index of weapon characteristics (3-point scales on whether the respondent was armed, with what type of weapon generally, and then requesting a specific description of the firearm in question).
These combined indices were subjected to analysis of variance against four independent variables: the complexity of the scene; the gender of the perpetrator; whether or not the perpetrator was actually armed or carried the power screwdriver; and the gender of the witness.
A second study employing a sample of 47 college students (thirty-one females, mean age 25.68, years, SD = 7.44, and sixteen males, mean age 23.38 years, SD =3.14), was conducted to investigate standard line-up identification under the conditions provided by the same scenes described above. To replicate typical and realistic crime scenes, only the scenes involving the armed male perpetrator were employed in the second study. Witnesses were exposed to the given scene for the usual 5-second exposure, followed by administration of a “six-pack” photographic lineup recognition test, to which the sample responded between 10 and 15 minutes after termination of the exposure.
The lineup was administered according to Department of Justice (DOJ) guidelines (1999). In the lineups, the perpetrator was placed at either position 3 or 5. A simultaneous lineup procedure was chosen because this remains the most commonly employed technique in actual criminal investigations (Wogalter, Malpass, & Burger, 1993), despite research that indicates sequential lineups are typically more accurate (e.g., MacLin, Zimmerman, & Malpass, 2005; Wells, 1993; Wogalther, Malpass, & McQuiston, 2004).
Finally, all respondents in both experiments completed the Conners Adult ADHD Rating Scales (CAARS) (Conners, Erhardt, & Sparrow, 1999) and the Dissociative Experiences Scale (DES) (Carlson et al., 1991). These scales were included to explore the possibility that symptoms of attention deficit and dissociation at levels found in a general, non-diagnosed population and falling below the threshold of identifiable clinical significance might influence performance in eyewitness identification. These specific factors were analyzed, as they have been shown in previous research to influence performance in other areas of G/FI processing (e.g., Sharps, Matthews, & Asten, 2006; Sharps, Price-Sharps, Day, Villegas, & Nunes, 2005).
Results and Discussion
Neither DES scores nor scores on any subscales of the CAARS were significantly predictive of performance on any index or measurement in either of these experiments. Within the present research framework, and with symptomatology at clinically non-significant levels, eyewitness abilities were influenced minimally, if at all, by attention deficit or dissociative anomalies. These results do not, of course, speak to clinically significant disorders or syndromes or to other possible frameworks of research.
Specific memory results for each index described above are as follows:
Memory for Clothing
The average combined score for witness recall of perpetrator clothing was a respectable 9.64 out of 12 (SD = 1.67), an accuracy of 80.3%. The effect of scene complexity was significant, F (1,182) = 4.60, p = .033, but modest; complex scenes resulted in only 3.3% less accuracy than did simple (sparse) scenes. The effect of perpetrator gender was also significant, F (1,182) = 12.55, p < .001, with the female perpetrator’s clothing being described 12.6% better than that of the male on average. This effect interacted significantly with the effect of scene complexity, F (1,182) = 3.93, p = .049. The presence or absence of a weapon was not significant on clothing description, but the gender of the witness was, F (1,182) = 4.08, p = .045. This effect should be interpreted with caution because of the disparity in numbers of the two sexes, but within this research framework, women were 7.46% better than men on average at describing perpetrator clothing.
Memory for Physical Characteristics
The average combined score for witness recall of perpetrator physical appearance was less impressive than that for clothing—only 12.71 out of a possible 18 points (SD = 1.81), an accuracy of 70.6%. For this index, the effects of scene complexity, witness gender, and weapon were non-significant; the only significant variable was gender of perpetrator, F (1,182) = 12.64, p < .001, with the appearance description scores for the female perpetrator 11.97% higher on average than those of the male.
As noted above, in addition to the questions requesting verbal descriptions of height, weight, and age, numerical estimates were sought from witnesses as well. There was no identifiable correlation, significant or otherwise, between numerical or verbal estimates. In other words, a descriptive term like heavy, thin, or tall had no identifiable relationship with numerical estimates of the same quantities. Further research will be needed to understand the mechanisms by which witnesses arrived at these apparently unrelated qualitative and quantitative descriptions. However, if confirmed in other research, this finding may have important implications for police interview techniques. For example, respondents may be better at making general, qualitative descriptions of persons than they are at attempting to convert their observations into realistic quantitative estimates. How law enforcement officers elicit such information (qualitatively or quantitatively) may result in differences in the utility of such information in searches, investigations, and courtroom proceedings.
Memory for Weapons and for Peripheral Sources of Hazard.
In contrast to the data concerning the physical description of the perpetrator, the independent variables of scene complexity, perpetrator gender, and gun-versus-screwdriver all resulted in significant effects on identification of the weapon carried by the perpetrator. Also significant were the interaction of complexity and perpetrator gender and the interaction of these factors with the gun/screwdriver variable [F values, respectively, were F (1,182) = 4.18, p = .042; F (1,182) = 4.053, p = .046; F (1,182) = 242.69, p < .001; F (1,182) = 7.50, p = .007; and F (1,182) = 8.17, p = .005].
The effect of witness gender was not significant. Average performance across conditions was not impressive, with a mean score of 6.20 (SD = 2.63) out of a possible 9 points, producing an average witness accuracy of 68.9%. The simple scene yielded better performance than the complex scene by 4.12%. The male perpetrator resulted in better performance than the female by 22.28%. Further, the gun condition was massively superior to the screwdriver condition, by 211%.
This gun condition results warranted further investigation. In the screwdriver condition, out of 103 respondents, 92 were in error, believing the screwdriver to have been, in fact, a firearm. Only five witnesses had been uncertain as to what they had seen, and a mere six correctly identified the power tool. It should be reiterated that this result was obtained under conditions of ideal lighting and exposure, as well as an exposure period several times longer than typical for real-world criminal activity.
Even with enhanced light and time, most witnesses identified the tool as a gun. This result may prove useful in explaining the large numbers of cases in which civilians and police mistake an innocuous object for a firearm. Rather than having anything to do with witness or law enforcement duplicity, such errors may frequently derive from typical characteristics of human visual cognition when witnesses are faced with this type of situation (see Sharps et al., 2003). It is important to develop a better empirical understanding of the types of conditions under which such errors are most prevalent.
Identification of peripheral sources of hazard was also unexpectedly poor. Only a single respondent, out of 94, identified the hand grenade, even though the grenade was in plain sight and viewed under the optimized observation conditions described above. Not a single individual identified the ammunition box as a potential source of hazard, even though it was placed close to the hand grenade and was clearly an anomalous object in a normal street scene.
In view of the terrorist threats currently confronting law enforcement and military authorities in much of the world and the relative reliance of terrorists on ambush tactics and pre-placed improvised explosive devices (e.g., Gelles, 2006), the failure to detect a grenade hazard clearly highlights the need for a better understanding of witness, bystander, and professional perceptual and cognitive responses to such devices. This is especially important in view of the fact that such devices are not normally placed in plain sight, as they were in the present study.
On a train or in a crowded street, observation of such peripheral sources of explosive hazard prior to their detonation and the recognition of an anomaly—something that simply should not be there—is likely to come from a law enforcement officer, a security officer, or a particularly alert civilian. Given the poor performance of respondents in this study to such hazards placed in plain sight and under optimal viewing conditions, more research on the topic of observation, detection, and interpretation of anomalous threatening devices is clearly needed.
This experiment required witnesses to respond to a typical police lineup, only 10 to 15 minutes after viewing the simple or the complex scene of the armed male perpetrator under ideal conditions. The sole visual transformation required of the respondents involved viewing the perpetrator in profile in the crime scenes and viewing the perpetrator and foils full-face in the lineup photographs. Though there appear to be no current statistics on this subject, it seems that in a typical, actual crime situation, witnesses rarely have the opportunity to gaze full-face on a suspect. Actual perpetrators are typically in motion, frequently with their faces occluded by a partial disguise, such as a hood, cap, or even stocking mask, and conditions are seldom optimal (e.g., Narby et al., 1996). Thus, this simple transformation of the view of the alleged perpetrator under optimized laboratory conditions would seem to pose little challenge compared to the use of lineups in real-world forensic practice (e.g., DOJ, 1999).
However, the results of this experiment led to little cause for optimism in the realm of alleged perpetrator identification. Out of 47 respondents, only 5 were able to identify the perpetrator correctly from a standard simultaneous lineup. Position 5 (central, bottom row) was slightly more likely to result in a correct identification than Position 3 (more visually peripheral). However, the level of identification observed here was sufficiently poor as to render miniscule the position effect, and, essentially, render moot this picture-placement finding. Of the five correct identifications, four of the five came from the simple condition. Only one individual in the complex and more realistic condition was able to make the identification correctly.
These results are sufficiently close to the statistical “floor” as to defy inferential statistical testing. The data clearly indicate that, at least within the conditions of this research, at best, only about 10% of witnesses can be anticipated to be correct in their identifications. Furthermore, accuracy is probably likely to be lower in more complex, and hence more realistic, visual situations. These results may at first seem to fly in the face of some schools of conventional wisdom on this subject. For example, human visual recognition memory was shown to be quite efficient in a series of classic studies, especially for detailed pictures (e.g., Nelson, Metzler, & Reed, 1974), but performance declined with shorter exposure times (e.g., Loftus & Bell, 1975). However, these studies required recognition of pictures in their entirety, rather than the recognition of abstracted elements of original encoding situations involving visual transformations, which is typical of lineup procedures.
Other studies have, of course, found recall or recognition-enhancing effects of, for example, instructional manipulations (e.g., Cutler, Penrod, & Martens, 1987), which may enhance lineup memory (in the case of the Cutler et al. study, to 44% accuracy). However, in the absence of such manipulations, and with only a single visual change in the aspect of the individual to be identified, the present results add to the growing body of information suggesting that person identification, even when obtained from properly conducted police lineups (DOJ, 1999), must be viewed with extreme caution by law enforcement and judicial personnel.
Summary and Conclusions
As anticipated from previous research (e.g., Narby et al., 1996; Sporer, Malpass, & Koehnken, 1996; Sharps et al.; Villegas et al., 2005), levels of eyewitness performance in person description, weapon description, and person identification were generally low. When a contextual systematic assessment of specific factors involved in these effects was conducted, more complex (and hence more realistic) scenes resulted in diminished performance compared to simpler situations. Witnesses generally reported an innocuous object (i.e., the screwdriver) to have been a handgun, even under the ideal viewing conditions employed here. Under typical, approved lineup conditions, recognition of a perpetrator seen in context was effectively at a statistical floor. Peripheral sources of hazard, even when placed in plain view, generally went unnoticed and unreported. Finally, the viewing of a female perpetrator, as opposed to a male, improved person description significantly but reduced weapon recognition.
An immediate limitation on the present research is noted. The use of a single male and female perpetrator in these studies obviously does not allow broad general conclusions about perpetrator gender and identification. A more exhaustive accounting of this effect would require depictions of a large number of individuals of both sexes, systematically varied according to race, physique, perceived attractiveness, and other characteristics. Such an effort would require a substantial long-term research project devoted to this question alone, well beyond the scope of this preliminary research. There must be, however, limitations to the generalized results of such a broad effort, as any given real-world perpetrator might possess characteristics that differed significantly from those evaluated.
For the present, there is a reasonable, parsimonious working explanation for the significant effects of perpetrator gender observed here. Although women do commit violent crimes, the vast majority are committed by men. Confronting a female perpetrator may reduce processing and performance, even in seasoned combatants and law-enforcement officers (e.g., Grossman, 1996; Kessler, 2002). The authors suggest that the witnesses’ patterns of attention and cognitive resources may have been altered by the surprise or the violation of expectations, engendered by the depiction of the female perpetrator. The higher average reportage of the female perpetrator’s clothing and physical characteristics than those of the male may speak to this effect. Attention and processing resources were drawn to her as an unexpected stimulus, resulting in the superior description of her clothing and body per se.
However, with better identification of the female perpetrator, resources were reciprocally drawn away from other elements of the scene, such as the gun, which might have been more expected in the hands of a male. As a result, the study suggests that weapon identification was superior with the male perpetrator.
The role of expectations may also help explain the tendency of witnesses to report the innocuous power screwdriver as a firearm. There may have been a tacit expectation that a given object, pointed by one individual at another, was in fact a weapon. This expectation-based reconfiguration (see Bartlett, 1932; Ahlberg & Sharps, 2002) could have resulted in the potentially vast witness misperception observed in this research. However, these impressions are conjectural at present. A significant body of future research will be needed to elucidate the mechanisms of these effects.
The gender of the witness had a significant effect on clothing description. Women’s performance exceeded that of men on average. However, and perhaps contrary to expectations, no gender differences were observed in weapon identification, physical description, or lineup recognition of persons. Because of the extremely poor performance of witnesses of both sexes, lineup recognition could not even be analyzed statistically.
Also, there was also no evidence of an effect, in this civilian witness population, of familiarity with weapons on witnesses’ ability to identify them. One might, however, expect such an effect with trained law-enforcement or military personnel, who are familiar with the idiosyncratic features of many weapons, including the standard police Beretta used in these studies. Finally, tendencies toward attention deficit and dissociation at subclinical levels did not influence eyewitness performance.
In aggregate, these studies attest to the importance of the empirical evaluation of eyewitness memory in systematically-varied context. The combined effects observed here further add to the growing body of evidence indicating the necessity of viewing eyewitness evidence with extreme caution. Contrary to beliefs often observed to prevail in courtroom proceedings, eyewitness identifications of persons, weapons, and peripheral sources of hazard in crime scenes are unlikely to be highly accurate, especially under the relatively brief conditions of exposure, the high levels of arousal, and the poor viewing conditions typical of real-world crime scenes.
Ahlberg, S. W., & Sharps, M. J. (2002). Bartlett revisited: Reconfiguration of long-term memory in young and older adults. Journal of Genetic Psychology, 163, 211–218.
Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology. Cambridge, UK: Cambridge University Press.
Bergman, E. T., & Roediger, H.L. (1999). Can Bartlett’s repeated reproduction experiments be replicated? Memory and Cognition, 27, 937–947.
Blades, H. B. (2005). The Washington, D.C. sniper case: A case study in how eyewitness identification of vehicles can go wrong. The Forensic Examiner, 14(3), 26.
Carlson, E. B., Putnam, F. W., Ross, C. A., Torem, M., Coons, P., Dill, D., et al. (1991). Factor analysis of the Dissociative Experiences Scale: A multicenter study. In B. G. Braun & E. B. Carlson (Eds.), Proceedings of the Eighth International Conference on Multiple Personality and Dissociative States. Chicago: Rush.
Conners, C. K., Erhardt, D., & Sparrow, M. A. (1999). Conners’ CAARS adult ADHD rating scales. North Tonawanda, NY: Multi-Health Systems.
Cutler, B. L., Penrod, S. D., & Martens, T. K. (1987). The reliability of eyewitness identifications: The role of system and estimator variables. Law and Human Behavior, 11(3), 23–258,
Department of Justice (1999). Eyewitness evidence: A guide for law enforcement. Washington, DC: National Institute of Justice.
Gelles, M. (2006, October). Behavioral threat assessment: Law enforcement considerations. Washington DC: Society for Police and Criminal Psychology.
Grossman, D. (1996). On killing. Boston: Little, Brown.
Kassin, S. M., Tubb, V. A., Hosch, H. M., & Memon, A. (2001). On the “general acceptance” of eyewitness testimony research: A new survey of the experts. American Psychologist, 56(3), 405–416.
Kessler, R. (2002). The bureau. New York: St. Martin’s.
Loftus, G. R., & Bell, S. M. (1975). Two types of information in picture memory. Journal of Experimental Psychology: Human Learning and Memory, 104(2), 103–113.
MacLin, O. H., Zimmerman, L. A., & Malpass, R. S. (2005). PC_Eyewitness and the Sequential Superiority Effect: Computer Based lineup administration. Law and Human Behavior, 29(3), 303–321.
Morgan, C. A., Hazlett, G., Doran, A., Gattett, S., Hoyt, G., Thomas, P., et al. (2004). Accuracy of eyewitness memory for persons encountered during exposure to highly intense stress. International Journal of Law and Psychiatry, 27(3), 265–279.
Montejano, D. (2005). The impact of job related stress on individuals in law enforcement. Frenso, CA: Fresno Police Department, Sierra Education and Research Institute, and Alliant International University.
Moore, L. (2006, November). Conference on the use of force in law enforcement. Office of the United States Marshal, Fresno, CA.
Narby, D. J., Cutler, B. L., & Penrod, S. D. (1996). The effects of witness, target, and situational factors on eyewitness identifications. In S. L. Sporer, R. S. Malpass, & G. Koehnken (Eds.), Psychological issues in eyewitness identification (pp. 23–52). Mahwah, NJ: Erlbaum.
Nelson, T. O., Metzler, J., & Reed, D. A. (1974). Role of details in the long-term recognition of pictures and verbal descriptions. Journal of Experimental Psychology, 102(1), 184–186.
Sharps, M. J. (2003). Aging, representation, and thought: Gestalt and feature-intensive processing. Piscataway, NJ: Transaction.
Sharps, M. J., Barber, T. L., Stahl, H., & Villegas, A. B. (2003). Eyewitness memory for weapons. The Forensic Examiner, 12(5), 34–37.
Sharps, M. J., Matthews, J., & Asten, J. (2006). Cognition, affect, and beliefs in paranormal phenomena: Gestalt/Feature Intensive Processing Theory and tendencies toward ADHD, depression, and dissociation. Journal of Psychology, 140(6), 579–590.
Sharps, M. J., & Nunes, M. A. (2002). Gestalt and feature-intensive processing: Toward a unified model of human information processing. Current Psychology, 21(1), 68–84.
Sharps, M. J., Price-Sharps, J. L., Day, S. S., Villegas, A. B., & Nunes, M. A. (2005). Cognitive predisposition to substance abuse in adult attention deficit hyperactivity disorder. Addictive Behaviors, 30(2), 355–359.
Spoehr, K. T., & Lehmkule, S. W. (1982). Visual information processing. San Francisco: Freeman.
Sporer, S. L., Malpass, R. S., & Koehnken, G. (1996). Psychological issues in eyewitness identification. Mahwah, NJ: Erlbaum.
Tietjen, T. (2005, January). The impact of job related stress on individuals in law enforcement. Fresno Police Department, Sierra Education and Research Institute, and Alliant International University, Fresno, CA.
Villegas, A. B., Sharps, M. J., Satterthwaite, B., & Chisholm, S. (2005). Eyewitness memory for vehicles. The Forensic Examiner, 14(3), 24–28.
Wells, G. L. (1993). What do we know about eyewitness identification? American Psychologist, 48(5), 553–571.
Wogalter, M. S., Malpass, R. S., & Burger, M. A. (1993). How police officers construct lineups: A national survey. In Proceedings of the Human Factors and Ergonomics Society (pp. 640–644). Santa Monica, CA: Human Factors and Ergonomics Society.
Wogalter, M. S., Malpass, R. S., & McQuiston, D. E. (2004). A national survey of police on preparation and conduct of identification lineups. Psychology, Crime and Law, 10(1), 69–82.
The authors wish to thank Chief of Police Jerry Dyer of the Fresno Police Department and the staff and field training officers of the Fresno Police Department for giving of their time and expertise and for their continued and very generous assistance, advice, and support of this research. They would also like to thank Officer Scott Larson, Los Angeles Police Department and Amy Neff, Morgan Goodwin, and Randy Vaughn-Dotta for their excellent help in the preparation of the materials for this research. The research, views, and opinions presented in this article are those of the authors and do not necessarily reflect the views or opinions of the Fresno or Los Angeles Police Departments or their chiefs, staffs, officers, or employees.
About the Authors
Matthew J. Sharps, PhD, DABPS, FACFEI, is professor of psychology at California State University, Fresno, and adjunct faculty member at Alliant International University. He received his MA (clinical psychology) from UCLA and MA and PhD (psychology) from the University of Colorado. He is the author of numerous articles and papers on visual cognition and related topics as well as the book Aging, Representation, and Thought: Gestalt and Feature-Intensive Processing (2003, Transaction Publishers). He has consulted on issues of eyewitness identification in more than 160 criminal cases.
Adam B. Hess, MA, is a lecturer in Criminology at California State University, Fresno, and a doctoral candidate in forensic clincical psychology at Alliant International University, Fresno.
Bethany Ranes, BA, is a doctoral candidate in Forensic Clinical Psychology at Alliant International University, Fresno.
Jenna Jones, BA, is a doctoral candidate in Forensic Clinical Psychology at Alliant International University, Fresno.
Hilary Casner, BA, is a graduate student in psychology at the University of Arkansas.
Factors Leading to Wrongful Convictions
The most common factors leading to wrongful convictions that were found in the first 130 DNA exonerations.
Mistaken I.D. (101)
False Confessions (35)
Microscopic Hair Comparison Matches (21)
DNA Inclusions at Time of Trial (3)