Running Head: IMPACT OF DISGUISE AND LINEUP TYPE 1 Impact of Disguise on Identification Decision and Confidence with Simultaneous and Sequential Lineups

In two studies (Ns = 87 and 91) we explored how varying the degree of two types of disguise (stocking mask, or sunglasses and toque [i.e., knitted hat]) affects lineup decisions and confidence from simultaneous and sequential lineups. Correct identifications decreased as the disguise covered more of the mock perpetrator’s face, supporting the conclusion that disguises reduce the amount of information available for encoding. Lineup selections decreased as the proportion of the face disguised increased, but more slowly than accuracy, indicating that witnesses are not attuned to encoding conditions. Correct identifications were most affected by sunglasses (ηp = .39), compared to a toque (ηp = .20) or stocking (ηp = .16). Correct Identifications were similarly low when faces were completely covered with a stocking versus covered to just below the nose, suggesting disguises which disrupt our ability to view faces as a whole may be as detrimental as ones that disrupt specific features. Correct rejections were influenced by sunglasses only (ηp = .08). Lineup type had typical effects on accuracy (i.e., higher correct identifications and lower correct rejections from simultaneous compared to sequential lineups), but did not interact with level of disguise for identification accuracy, and had unexpected effects on confidence. IMPACT OF DISGUISE AND LINEUP TYPE 3 Impact of Disguise on Identification Decision and Confidence with Simultaneous and Sequential Lineups Eyewitness identification evidence plays a pivotal role in many criminal cases. Although the police have no control over whether a perpetrator wears a disguise (i.e., it is an estimator variable; Wells, 1978), it clearly negatively impacts identification accuracy (e.g., Shapiro & Penrod, 1986). Thus, the impact of disguise on identification decisions and confidence warrants exploration. In addition, it is critical to understand the mechanisms through which disguise affects witness decisions. Brewer, Weber, and Semmler (2005) identify two ways in which disguise may affect identification accuracy. First, compared to an undisguised perpetrator, viewing a disguised perpetrator provides less identifying information for encoding. As a result, witnesses exposed to a disguised perpetrator must have a less complete memory of the perpetrator. A second, not mutually exclusive, explanation is that disguises may affect witnesses’ perceptions of the difficulty of the identification task. These meta-cognitions may change the decision criterion applied by a witness and subsequently decrease their likelihood of making a selection from a lineup. A third explanation for the impact of disguise on identification accuracy is encoding specificity—encoding to-be-remembered material in a way that facilitates accuracy at a particular recognition task (Davies & Flin, 1984; McKelvie, 1976; Tulving & Thomson, 1973). As indicated by Shapiro and Penrod's (1986) meta-analysis of face recognition and eyewitness identification studies, altering a person’s appearance between the time of encoding and a face recognition/identification task negatively affects identification accuracy. Davies and Flin (1984) found partial support for this mechanism. They found that correct identifications were highest for faces undisguised at both encoding and recognition, but worst for a face disguised at encoding IMPACT OF DISGUISE AND LINEUP TYPE 4 but not at recognition. The two intermediate conditions (undisguised at encoding, disguised at test and disguised at encoding and test) did not differ. Patterson and Baddeley (1977) also found partial support for this explanation: identification accuracy varied with whether faces were presented at test in the same way or differently than they were presented at encoding (e.g., with or without a beard). However, if a face was disguised at retrieval, there was no significant difference in accuracy if it was disguised at encoding or not. In summary, encoding specificity may play a role but on its own does not adequately explain the influence of disguise on identification. Finally, a fourth possibility is that the disguise may influence how a witness allocates their attention. A disguised perpetrator may be interpreted by witnesses as more dangerous. This perception may create additional cognitive load leaving witnesses with fewer processing resources for encoding (Deffenbacher, Bornstein, Penrod, & McGorty, 2004). However, in their meta-analysis looking at the effects of distress on identification, Deffenbacher and colleagues did not find that disguise accounted for significant variance. Similarly, a disguise may distract witnesses by drawing attention to the disguise itself and away from facial information required for later recognition. No research that we are aware of has examined this last possibility. To date, research has not tested which of these mechanisms—or combination thereof— best accounts for the effects of disguise. Most prior research focuses on how masking particular features influences face recognition (e.g., Sadr, Jarudi, & Sinha, 2003; Terry, 1993), or on the relative impact of disguise compared to other estimator variables (e.g., Cutler, Penrod & Martens, 1987a, 1987b; Shapiro & Penrod, 1986). One goal of the following two experiments was to explore Brewer et al.’s (2005) suggestion that disguise works by decreasing the amount of IMPACT OF DISGUISE AND LINEUP TYPE 5 information available for encoding by manipulating how much of a to-be-remembered face was covered with a disguise. Many parts of a face can be disguised, but one that is particularly easy to manipulate is hair. A perpetrator’s hairstyle and facial hair may be disguised and/or changed between a crime (encoding) and a lineup identification (test). Previous research shows that such changes result in a consistent decrease in recognition (Cutler et al., 1987a, 1987b; Patterson & Baddeley, 1977). Moreover, research suggests that the negative impact of disguise on accuracy is specifically related to obstruction of hair cues (e.g., Cutler et al., 1987a, 1987b; Narby, Cutler, & Penrod, 1996; O’Donnell, & Bruce, 2001). In particular, compared to normal exposure, Wright and Sladden (2003) found that viewing targets without hair cues impaired performance in subsequent facial recognition tasks. On the other hand, Yarmey (2004) found that obscuring hair with a baseball cap had no significant effect on accuracy in target-present or -absent lineups. Yarmey notes however, that hair cues may not have been appreciably obscured as his female target’s shoulder-length hair was still visible. Other research shows that disguising the eyes influences identification accuracy. McKelvie (1976) found that participants made more errors in recognizing faces when eyes were covered at either the encoding or recognition stage. He concluded that eyeglasses seemed to be encoded as part of the eyes. Likewise, the addition of eyeglasses at recognition hinders recognition (Hockley, Hemsworth, & Consoli, 1999; Terry, 1993), possibly because the eyes are a central area of focus (Janik, Wellens, Goldberg, & Dell’Osso, 1978; O’Donnell & Bruce, 2001). Conversely, Patterson and Baddeley (1977) found inconsistent effects of the presence of glasses on the accuracy of facial recognition though they found an overall main effect. Changing the presence or absence of glasses from encoding to test decreased identification accuracy IMPACT OF DISGUISE AND LINEUP TYPE 6 compared to no change. Changing from wearing glasses at exposure to not wearing glasses at recognition (and vice versa) in combination with a change in wig, beard, and both had a significant negative effect across pose changes. However, changing from wearing glasses at exposure to not wearing glasses at recognition (and vice versa) in combination with a change of wig had no effect if pose was changed to a profile view at test from a full face view at encoding. Head coverings and glasses are commonly worn as both accessories and disguises, and, as discussed above, are generally effective as disguises. A second goal of the current experiments is to examine whether one has a more detrimental effect than the other and whether the detrimental effects are additive when these disguise are worn together. Much research on facial recognition indicates that the eyes are the most important facial feature for identifying a face (e.g., Henderson, Williams, & Falk, 2005), though much of this research considers the role of either eyes (e.g., McKelvie, 1976) or hair (e.g., Wright & Sladden, 2003). Additionally, Janik et al. (1978) asked participants which portions of faces they looked at most after being asked to form an impression of each of a series of faces; participants reported looking primarily at the eye and mouth regions. Given that research indicates the eyes are the most important feature for identification and that people report that they use the eyes when reviewing faces, we predicted that masking a target’s eyes would be more detrimental to later identification than masking their