Making a Spectacle of Yourself: The Effect of Glasses and Sunglasses on Face Perception

Abstract

We investigated the effect of wearing glasses and sunglasses on the perception of social traits from faces and on face matching. Participants rated images of people wearing no glasses, glasses and sunglasses on three social traits (trustworthiness, competence and attractiveness). Wearing sunglasses reduced ratings of trustworthiness. Participants also performed a matching task (telling whether two images show the same person or not) with pairs of images both wearing no glasses, glasses or sunglasses, and all combinations of eyewear. Incongruent eyewear conditions (e.g., one image wearing glasses and the other wearing sunglasses, etc.) reduced performance. Further analysis comparing performance on congruent and incongruent eyewear trials showed that our effects were driven by match trial performance, where differences in eyewear decreased accuracy. For same-eyewear-condition pairs, performance was poorer for pairs of images both wearing sunglasses than no glasses. Our results extend and update previous research on the effect of eyewear on face perception.

Keywords

face perception unfamiliar faces first impressions glasses sunglasses

Introduction

We form first impressions of people very quickly, and these first impressions can have real-world consequences. For example, people’s ratings of competence (Todorov, Mandisodza, Goren, & Hall, 2005) and attractiveness (Lutz, 2010) can predict election results, and first impressions from faces can predict whether or not someone is hired (Gilmore, Beehr, & Love, 1986). These findings may lead to the assumption that a person’s attractiveness is a stable property of their face, that is to say that a person is either attractive or not, irrespective of the photograph used. The majority of research on social trait judgements from faces has, therefore, used only one image of each identity. Recent research has shown, however, that different images of the same person can give rise to very different first impressions (Jenkins, White, Van Montfort, & Burton, 2011; Ritchie, Palermo, & Rhodes, 2017, Todorov & Porter, 2014). In fact, within-person variability in attractiveness has been shown in some cases to exceed between-person variability (Jenkins et al., 2011).

Research using multiple images of each identity often uses “ambient images,” or naturally occurring images which vary in many different aspects of the photograph, from person-specific variability such as facial expression, to variability in the world such as lighting. It is possible that many of these world- and person-specific differences between images may influence the first impression generated by any specific image. One simple change a person can make is whether they choose to wear glasses or sunglasses in a given photograph. An older body of literature has looked specifically at the effect of glasses on first impressions and suggested that glasses wearers are perceived as less attractive (e.g., Harris, 1991; Hasart & Hutchinson, 1993) and more competent (Terry & Krantz, 1993) than people without glasses. One study found that sunglasses wearers were perceived as less authoritative (Bartolini et al., 1988), but to date there has been very little research on the effect of sunglasses on the perception of social traits from faces.

It has been suggested that sunglasses may increase attractiveness because sunglasses increase symmetry in the face (Brown, 2015). More symmetrical faces are judged as more attractive (e.g., Jones et al., 2001; Perrett et al., 1999) and so by occluding the same portion of the face on each side, sunglasses may cover any asymmetries, and by themselves being symmetrical, may increase perceived attractiveness. Conversely, sunglasses may reduce perceived trustworthiness because they render the eyes invisible. The eyes have been shown to be important for making judgements of trustworthiness (Dotsch & Todorov, 2012) and so occluding the eye region may make this judgement more difficult, or lead to lower trustworthiness ratings.

In addition to forming first impressions from faces, we use face images to determine people’s identity. For example, photo-ID images are used in various security settings such as border control. Despite the widespread use of photo-ID, we are actually relatively poor at identifying unfamiliar people, even from videos (Bruce et al., 1999) or when comparing a photo to a live person (Kemp, Towell, & Pike, 1997; Ritchie, Mireku, & Kramer, in press). A task which is frequently used to test unfamiliar face processing is a face matching task wherein participants are shown pairs of images and asked to determine whether the two images show the same person or two different people. Participants are consistently more accurate at the task with familiar compared to unfamiliar faces (see Ritchie et al., 2015), and with unfamiliar faces, the addition of glasses has been shown to have a negative effect on performance. In face matching tasks when one image shows glasses and the other does not, performance is slower (Leder, Forster, & Gerger, 2011) and less accurate (Kramer & Ritchie, 2016). Recognition memory accuracy has also been shown to be poorer when a person seen initially wearing glasses is shown at the recognition phase without glasses (Leder et al., 2011). Research on unfamiliar face identification has not looked at the effect of sunglasses. This is a pertinent avenue for enquiry as security services and CCTV personnel are often tasked with searching for people in crowds where people may be outdoors and potentially wearing sunglasses.

Here, we extend previous work on social impressions of faces wearing glasses to test the effect of glasses and sunglasses on perceptions of trustworthiness, competence and attractiveness. We also extend previous research on the effect of glasses on unfamiliar face matching performance to include sunglasses. It is possible that there is a relationship between the first impressions we have of faces and our ability to perceive two images as belonging to the same person. For example, if we perceive two images of the same person as very different in attractiveness (without knowing that those two photos show the same person), it is possible that we may be less inclined to perceive those two images as portraying the same person in a matching task. Therefore, in this study, we structured our task in such a way that the images participants saw in the face matching task were the exact images they had rated in the previous trait rating task. We predicted that photos showing glasses would be rated as less attractive (Harris, 1991; Hasart & Hutchinson, 1993) and more competent (Terry & Krantz, 1993) than images without glasses. We also predicted that photos showing sunglasses would be rated as more attractive (due to increased symmetry) than photos shown without eyewear. We also predicted that incongruencies in eyewear between two images would impair performance on a face matching task (following Kramer & Ritchie, 2016), and that image pairs including an image wearing sunglasses would result in lower performance due to the increased concealment of the face. Finally, we predicted a correlation between the difference in social trait ratings between two images of the same person and accuracy in the face matching task whereby larger differences in social trait ratings would correlate with poorer accuracy on match trials.

Method

Participants

Forty-seven participants (12 men; mean age: 20 years, range: 18–54 years) took part in both parts of the study. All were members of the University of Lincoln and took part voluntarily or in return for course credits. All participants gave informed consent, and the study was given ethical approval by the University of Lincoln School of Psychology Research Ethics Committee.

Stimuli

Our stimuli were 60 identities (30 women) chosen to be unfamiliar to participants in the United Kingdom. For each identity, we gathered two images without glasses, two with glasses, and two with sunglasses, as well as three images of a foil identity (someone who resembled that person), one image in each eyewear condition. Images were downloaded from Google Images (see Figure 1, e.g., stimuli). The images were ambient images as used in previous face matching research (e.g., Dowsett & Burton, 2015; Ritchie et al., 2015) and sampled natural variability in lighting, head angle, facial expression, and so forth. All images were cropped to 380 × 570 pixels to show the head and neck.

Figure 1.

Example stimuli. All photographs show the same person. (Copyright restrictions prevent publication of the face images used. The individual pictured in these images did not appear in the experiment and has given permission for their images to be reproduced here.)

Design and Procedure

Participants completed a rating task followed by a matching task. For ease of explanation, we describe the matching task first. In the matching task, participants saw pairs of images presented simultaneously and were asked to indicate whether the two images showed the same person or two different people. Each identity was shown once in one of six image-pair conditions: three congruent eyewear conditions (both images wore no glasses, both glasses, both sunglasses) and three incongruent eyewear conditions (no glasses-glasses, no glasses-sunglasses, glasses-sunglasses). Half of the trials in each eyewear condition were match trials (two photos of the same identity) and half mismatch (one photo of the identity and one of their foil). Each participant, therefore, completed 10 trials in each eyewear condition, 5 match trials and 5 mismatch trials (60 trials in total). The image pairs were shown in a random order, with the presentation of identities in each condition counterbalanced between participants.

In the preceding rating task, participants rated the exact images they would go on to see in the matching task (120 images in total). Participants rated two images of each identity (or the identity and the foil). Each participant rated each image for trustworthiness, competence and attractiveness on a 7-point scale. Participants saw each image once and gave all three trait ratings concurrently. Participants were not told that they would go on to see the identities again in a face matching task. The images were presented in a random order, one image at a time, not blocked by identity, and participants were not told that they would see multiple images of each person. We do not suspect that participants would have noticed that they were seeing two images of each person (or the person and their foil).

Participants viewed the same images in the rating and matching task so that we could directly determine whether the difference in ratings given to the two images of each identity correlated with accuracy on the matching task with the same images. Although this means that each participant saw all the images twice (once in the ratings task and once in the matching task), we do not have reason to believe that seeing each face in isolation prior to the matching task would have influenced matching performance.

Results

Social Trait Ratings

We carried out three repeated measures analyses of variance (ANOVAs; one per trait) to investigate the effect of eyewear on trait ratings. We found a significant effect of eyewear on ratings of trustworthiness, F(2, 92) = 29.20, p < .001, η_p² = .39, and follow-up paired samples t tests (Bonferroni corrected) showed higher ratings for images wearing no glasses (M = 4.17) and glasses (M = 4.25) than sunglasses (M = 3.70); no glasses versus sunglasses: t(46) = 5.30, p < .01, d = .77 and glasses versus sunglasses: t(46) = 6.90, p < .01, d = 1.01. There was a nonsignificant difference between trustworthiness ratings for the no glasses versus glasses conditions: t(46) = 1.30, p = .606, d = .19. There was a nonsignificant effect of eyewear on both competence, F(2, 92) = .446, p = .642, η_p² = .01, and attractiveness ratings, F(2, 92) = 1.85, p = .162, η_p² = .04 (see Figure 2).

Figure 2.

Data from the trait rating task. Error bars denote SEM.

Previous studies have shown that the degree of smiling in an image can influence trait ratings of both trustworthiness (Hehman, Flake, & Freeman, 2015; Oosterhof & Todorov, 2008; Todorov & Porter, 2014) and attractiveness (Todorov & Porter, 2014). Therefore, in order to determine whether our observed effects were due to the degree of smiling in the images as opposed to the eyewear conditions, we ran a small secondary study. Twenty new participants (four men; mean age: 21 years, range: 19–34 years) rated all of the images used in the main experiment for degree of smiling from not smiling at all to extremely smiling on a 7-point scale. Cronbach’s alpha for each participant’s mean smiling rating for each of the three conditions was .88. A repeated measures ANOVA showed a significant effect of eyewear condition on the perceived degree of smiling, F(2, 38) = 8.16, p < .01, η_p² = .30. Bonferroni comparisons showed that faces wearing no glasses (M = 3.15) were judged as smiling more than both faces wearing glasses (M = 3.06) and faces wearing sunglasses (M = 2.87, both ps < .05). There was a nonsignificant difference between degree of smiling for faces wearing glasses and sunglasses (p = .091). This result does not entirely rule out the possibility that the degree of smiling played a role in participants’ trait ratings of the sunglasses images compared with other images. The sunglasses images were judged to be numerically the least smiling images, but the difference between glasses and sunglasses images was nonsignificant. It is possible that the combination of wearing sunglasses and smiling (numerically but not significantly) least made the sunglasses images look the least trustworthy. However, we would expect the degree of smiling to influence perceptions of attractiveness as well as trustworthiness. We did not find a difference in attractiveness ratings for sunglasses images compared with other images. This could again be due to smiling less than other images; however, the pattern of results for the smiling experiment (with no glasses images being judged as smiling more than the glasses and sunglasses images) does not follow the pattern of data from the trait ratings experiment. We, therefore, suggest that although it may play a role, the degree of smiling does not fully explain the trait rating data.

Face Matching

To investigate the effect of eyewear on unfamiliar face matching, we first compared performance across the congruent (same eyewear) conditions to performance across the incongruent (different eyewear) conditions using a paired samples t test. Accuracy was higher for the congruent conditions (M = 78.72%) than the incongruent conditions (M = 74.96%), t(46) = 2.15, p = .037, d = 0.31. We also used signal detection measures calculating d ′ (sensitivity) and criterion (c, a measure of response bias) where hits are correct responses when both images show the same identity and false alarms are incorrect responses to different identity pairs. As with percent correct, d′ was significantly higher for the congruent conditions (M = 1.55) than the incongruent conditions (M = 1.35), t(46) = 2.22, p = .031, d = 0.32. We explored this effect further by carrying out a repeated measures ANOVA on d′ values for the congruent conditions, which showed an effect of eyewear, F(2, 88) = 4.85, p = .010, η_p² = .10, and follow-up paired samples t tests (Bonferroni corrected) showed higher d′ values for image pairs where both photos wore no glasses (M = 1.67) compared to sunglasses (M = 1.32), t(46) = 2.61, p = .036, d = .38. There was a marginal effect (of similar effect size to the no glasses vs. sunglasses effect) for image pairs where both photos wore glasses (M = 1.65) compared with sunglasses, t(46) = 2.43, p = .057, d = .35. There was a nonsignificant difference in d′ values for the no glasses and glasses conditions, t(46) = .19, p = 1, d = .03. A repeated measures ANOVA on d′ values for the incongruent conditions showed a nonsignificant effect of eyewear, F(2, 84) = .17, p = .848, η_p² < .01.

In addition to d′, we calculated criterion values, a measure of bias. We compared criterion values for the congruent and incongruent conditions and found more bias in the incongruent (M = −0.28) than the congruent conditions (M = −0.09), t(46) = 4.46, p < .001, d = 0.65. This shows that participants were more biased to respond that the two images showed two different people in the incongruent conditions where the two images showed two different eyewear conditions.

In addition, we can look at performance on match (both photos show the same person) and mismatch trials (photos of two different people) for congruent and incongruent eyewear conditions. A 2 (eyewear congruency: congruent and incongruent) × 2 (trial type: match and mismatch) ANOVA showed a significant main effect of eyewear, F(1, 46) = 4.60, p = .037, η_p² < .09, a significant main effect of trial type, F(1, 46) = 25.05, p < .001, η_p² < .35, and a significant interaction, F(1, 46) = 19.82, p < .001, η_p² < .30. Follow-up paired samples t tests (Bonferroni corrected) showed poorer performance on match trials for incongruent eyewear conditions (M = 63.26%) than congruent conditions (M = 74.61%), t(46) = 4.20, p < .001, d = 0.61, and a nonsignificant difference between congruent (M = 82.84%) and incongruent eyewear (M = 86.67%) for mismatch trials, t(46) = 1.78, p = .164, d = 0.26 (see Figure 3, and see Supplementary Table S1 for all data).

Figure 3.

Data from the matching task. Error bars denote SEM.

We constructed the task in such a way that participants would rate each image that they would subsequently see in the matching task. This allows us to examine whether the difference in trait ratings of two images of the same person correlates with performance on the matching task. To assess this, we took responses to only match trials (two images showing the same identity). For each participant, we calculated the mean difference given to all matching image pairs on each social trait. We then correlated these difference scores for each trait with overall accuracy on match trials in the face matching task. We found no significant correlations (trustworthiness: r = .25, p = .088; competence: r = .19, p = .196; attractiveness: r = −.12, p = .409). Therefore, in this case, differences in social trait ratings of two images of the same person do not influence performance on a face matching task using those same images.

Discussion

Here, we have shown that wearing sunglasses makes a person look less trustworthy but did not support our predictions that images wearing glasses would be rated as less attractive and more competent and images wearing sunglasses as more attractive. Where older research showed lower attractiveness ratings for faces wearing glasses, more recent research showed a more nuanced effect whereby only rimmed but not rimless glasses produced this effect (Leder et al., 2011). Here, we found no negative effect of glasses wearing on attractiveness. This may be due to the fact that glasses wearing has become more fashionable in recent years and so no longer bears the stigma attached in the past. Previous research has shown that symmetry increases attractiveness (e.g., Jones et al., 2001; Perrett et al., 1999), and sunglasses are both symmetrical and may occlude any asymmetries in the face. We, therefore, predicted that sunglasses would increase attractiveness; however, we did not find support for this idea. Instead, we have shown that sunglasses decrease perceived trustworthiness, which may be explained by the importance of the eyes in the perception of trustworthiness (Dotsch & Todorov, 2012) such that occluding the eyes decreases the perception of trustworthiness in the face.

Here, we also sought to combine the areas of first impression formation and face matching predicted that two images of the same person which are rated as less similar on social traits are more difficult to match in a face matching task. We did not find a correlation between the mean difference in trait ratings of two images of the same person and mean accuracy with those images in the match trials of a face matching task; therefore, here we have not found a relationship between differences in trait ratings and matching accuracy.

The results of our face matching task supported our prediction that when two images are presented in different eyewear conditions, unfamiliar face matching is more difficult than when the two images show the same eyewear condition. Moreover, when both images are pictured wearing sunglasses, performance is poorer than when they both show no glasses. This result extends previous research which found slower (Leder et al., 2011) and poorer performance (Kramer & Ritchie, 2016) for image pairs showing one photo with glasses and one without. Our effect is driven by match trials, where incongruent eyewear conditions produce poorer performance than congruent eyewear conditions (see Figure 3). This effect is not present for mismatch trials. This provides an interesting theoretical point whereby small changes in appearance seem to lead to the perception of a change in identity. This effect of congruency on the stable perception of identity ties in with the encoding specificity principle (Tulving & Thomson, 1973) which has been shown previously in face perception research (Leder & Carbon, 2005). The authors argued that whole-to-part superiority in face learning (whereby recognising facial parts presented in the context of a full face is easier than recognising parts presented alone) can be explained by the encoding specificity principle as only the precise encoded information is retrieved. In an adaptation of the standard task, participants saw facial parts at encoding, and recognition of parts was disrupted when shown in the context of a full face. The encoding specificity principle can also be used as a framework to explain our results whereby small changes between images lead to the perception of identity change. Our task does not rely on memory, but the same principle could be applied to the perceptual task of face matching.

Our results are important for face matching in forensic settings such as identifying faces in crowds where people may be wearing glasses or sunglasses. Current standard tests of face matching ability such as the Glasgow Face Matching Test (Burton, White, & McNeill, 2010) only include images with no eyewear, and so forensic training tasks based on such tests may not reflect the real-world problem of glasses and sunglasses in face matching. In one unfamiliar face matching study, mismatch trials were incorrectly accepted as a match more often (by both students and police officers) when both images wore identical glasses, indicating that fraudsters who adjust their appearance to match the person they are impersonating may be more successful than those who do not (Wirth & Carbon, 2017). In fact, deliberate disguise has been shown to reduce performance in a face matching task for both viewers who are unfamiliar and those who are familiar with the targets (Noyes & Jenkins, in press). In their evasion condition, Noyes and Jenkins (in press) gave participants a previously taken image of themselves and asked participants to make themselves look as different from that image as possible. In a face matching task, unfamiliar viewers responded correctly on only 60% of trials in this condition, and familiar viewers were only 86% correct. That study did not use sunglasses, and so it is possible that with the addition of sunglasses, both familiar and unfamiliar viewers would have performed even more poorly. Future research should investigate the limits of familiar and unfamiliar face recognition under conditions of disguise including the use of sunglasses.

Our results show that not only does wearing sunglasses make you look less trustworthy, they also make you more difficult to recognise. In fact, any change in eyewear between two images leads to reduced face matching accuracy.

Supplemental Material

Supplemental material for Making a Spectacle of Yourself: The Effect of Glasses and Sunglasses on Face Perception

Supplemental Material for Making a Spectacle of Yourself: The Effect of Glasses and Sunglasses on Face Perception by Daisy L. Graham and Kay L. Ritchie in Perception

Footnotes

Acknowledgements

The authors would like to thank Louis Kernahan and Samile A. Escobar Abadia for data collection on the smiling rating task.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplementary Material

Supplementary material for this article is available online.

References

Bartolini

Kresge

McLennan

Windham

Buhr

T. A.

Pryor

(1988). Perceptions of personal characteristics of men and women under three conditions of eyewear. Perceptual and Motor Skills, 67, 779–782.

Brown

(2015). Cool shades: The history and meaning of sunglasses. London: Bloomsbury Publishing.

Bruce

Henderson

Greenwood

Hancock

P. J. B.

Burton

A. M.

Miller

(1999). Verification of face identities from images captured on video. Journal of Experimental Psychology: Applied, 5, 339–360.

Burton

A. M.

White

McNeill

(2010). The Glasgow Face Matching Test. Behavior Research Methods, 42, 286–291.

Dotsch

Todorov

(2011). Reverse correlating social face perception. Social Psychological and Personality Science, 3, 562–571.

Dowsett

A. J.

Burton

A. M.

(2015). Unfamiliar face matching: Pairs out-perform individuals and provide a route to training. British Journal of Psychology, 106, 433–445.

Gilmore

D. C.

Beehr

T. A.

Love

K. G.

(1986). Effects of applicant sex, applicant physical attractiveness, type of rater and type of job on interview decisions. Journal of Occupational Psychology, 59, 103–109.

Harris

M. B.

(1991). Sex differences in stereotypes of spectacles. Journal of Applied Social Psychology, 21, 1659–1680.

Hasart

J. K.

Hutchinson

K. L.

(1993). The effects of eyeglasses on perceptions of interpersonal-attraction. Journal of Social Behavior and Personality, 8, 521–528.

10.

Hehman

Flake

J. K.

Freeman

J. B.

(2015). Static and dynamic facial cues differentially affect the consistency of social evaluations. Personality and Social Psychology Bulletin, 41, 1123–1134.

11.

Jenkins

White

Van Montfort

Burton

A. M.

(2011). Variability in photos of the same face. Cognition, 121, 313–323.

12.

Jones

B. C.

Little

A. C.

Penton-Voak

I. S.

Tiddeman

B. P.

Burt

D. M.

Perrett

D. I.

(2001). Facial symmetry and judgements of apparent health: Support for a “good genes” explanation of the attractiveness-symmetry relationship. Evolution and Human Behavior, 22, 417–429.

13.

Kemp

Towell

Pike

(1997). When seeing should not be believing: Photographs, credit cards and fraud. Applied Cognitive Psychology, 11, 211–222.

14.

Kramer

R. S. S.

Ritchie

K. L.

(2016). Disguising Superman: How glasses affect unfamiliar face matching. Applied Cognitive Psychology, 30, 841–845.

15.

Lutz

(2010). The electoral success of beauties and beasts. Swiss Political Science Review, 16, 457–480.

16.

Leder

Forster

Gerger

(2011). The glasses stereotype revisited. Swiss Journal of Psychology, 70, 211–222.

17.

Leder

Carbon

C. C.

(2005). When context hinders! Learn-test compatibility in face recognition. Quarterly Journal of Experimental Psychology, 58A, 235–250.

18.

Noyes

Jenkins

(in press). Deliberate disguise in facial image comparison. Journal of Vision. Advance online publication. doi: 10.1037/xap0000213.

19.

Oosterhof

N. N.

Todorov

(2008). The functional basis of face evaluation. Proceedings of the National Academy of Sciences of the United States of America, 105, 11087–11092.

20.

Perrett

D. I.

Burt

D. M.

Penton-Voak

I. S.

Lee

K. J.

Rowland

D. A.

Edwards

(1999). Symmetry and human facial attractiveness. Evolution and Human Behavior, 20, 295–307.

21.

Ritchie

K. L.

Mireku

M. O.

Kramer

R. S. S.

(in press). Face averages and multiple images in a live matching task. British Journal of Psychology. Advance online publication. doi:10.1111/bjop.12388

22.

Ritchie

K. L.

Palermo

Rhodes

(2017). Forming impressions of facial attractiveness is mandatory. Scientific Reports, 7, 469.

23.

Ritchie

K. L.

Smith

F. G.

Jenkins

Bindemann

White

Burton

A. M.

(2015). Viewers base estimates of face matching accuracy on their own familiarity: Explaining the photo-ID paradox. Cognition, 141, 161–169.

24.

Terry

R. L.

Krantz

J. H.

(1993). Dimensions of trait attributions associated with eyeglasses, men's facial hair, and women's hair length. Journal of Applied Social Psychology, 23, 1757–1769.

25.

Todorov

Mandisodza

A. N.

Goren

Hall

C. C.

(2005). Inferences of competence from faces predict election outcomes. Science, 308, 1623–1626.

26.

Todorov

Porter

J. M.

(2014). Misleading first impressions: Different for different facial images of the same person. Psychological Science, 25, 1404–1417.

27.

Tulving

Thomson

D. M.

(1973). Encoding specificity and retrieval responses in episodic memory. Psychological Review, 80, 359–380.

28.

Wirth

B. E.

Carbon

C. C.

(2017). An easy game for frauds? Effects of professional experience and time pressure on passport-matching performance. Journal of Experimental Psychology: Applied, 23, 138–157.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.27 MB