Beauty Is in the Eye of the Beholder: The Appraisal of Facial Attractiveness and Its Relation to Conscious Awareness

Abstract

Previous research suggests that facial attractiveness relies on features such as symmetry, averageness and above-average sexual dimorphic characteristics. Due to the evolutionary and sociobiological value of these characteristics, it has been suggested that attractiveness can be processed in the absence of conscious awareness. This raises the possibility that attractiveness can also be appraised without conscious awareness. In this study, we addressed this hypothesis. We presented neutral and emotional faces that were rated high, medium and low for attractiveness during a pilot experimental stage. We presented these faces for 33.33 ms with backwards masking to a black and white pattern for 116.67 ms and measured face-detection and emotion-discrimination performance, and attractiveness ratings. We found that high-attractiveness faces were detected and discriminated more accurately and rated higher for attractiveness compared with other appearance types. A Bayesian analysis of signal detection performance indicated that faces were not processed significantly at-chance. Further assessment revealed that correct detection (hits) of a presented face was a necessary condition for reporting higher ratings for high-attractiveness faces. These findings suggest that the appraisal of attractiveness requires conscious awareness.

Keywords

attractiveness masking awareness

Introduction

Attractiveness is considered a fundamental aspect of human interactions from infancy (Fisher & Ma, 2014; Thomas, De Bellis, Graham, & LaBar, 2007) to adulthood (Griffin & Langlois, 2006). It has been suggested to positively influence the quality of our peer relations, our behaviour and the behaviour of others towards us throughout our life (Langlois & Stephan, 1981), to influence and inform our mate choices (Saxton, Caryl, & Roberts, 2006), and even exert influence in our professional development and income status (Frieze, Olson, & Russell, 1991). In our modern society, where interpersonal communication, romantic-mate choice and even professional employability rely increasingly on on-line communication and posted photographic material (Hong, Tandoc, Kim, Kim, & Wise, 2012), the perception of facial attractiveness has been re-approached as a highly relevant psychological research subject (Swami & Farnham, 2008).

Facial attractiveness has been associated with certain perceived characteristics; these include – among others (Little, Jones, & DeBruine, 2011) – symmetry (Scheib, Gangestas, & Thornhill, 1999), averageness (Grammer & Thornhill, 1994) and above-average sexually dimorphic characteristics (Barber, 1995). Bilateral facial symmetry, for example, is assumed to function as a cue for developmental health and ontogenetic resistance to parasites (Fink, Neave, Manning, & Grammer, 2006). Facial averageness is also suggested to confer evolutionary important offspring-survival value due to indications of heterozygosity (Roberts et al., 2005) as well as a developmental propensity for familiarity due to average faces being overall more recognisable and prototypical exemplars of the category facial stimuli (Fink & Penton-Voak, 2002). In addition, pronounced facial characteristics that signify sexually dimorphic biological markers, such as testosterone in men and oestrogens in women, have been shown to increase gender-specific attractiveness ratings (Smith et al., 2006).

Although recent research has reported evidence that casts doubt on the role of attractiveness particularly as an indication of developmental health and immunocompetence (Cai et al., 2018; Foo, Simmons, & Rhodes, 2017; Jones et al., 2018; Scott, Clark, Boothroyd, & Penton-Voak, 2013), the association of attractiveness with characteristics that could confer evolutionary important sociobiological value was sufficient to prompt researchers to explore whether attractiveness can be processed, reported and appraised under conditions that do not necessarily involve conscious awareness (Mineka & Öhman, 2002; Öhman, 2009). This line of research has suggested that attractiveness is a highly salient visual characteristic (Barber, 1995) that can be appraised even from impoverished (pixelated) visual cues (Bachmann, 2007). Previous research has also suggested that – due to the evolutionary importance of high-attractiveness features – attractiveness can be processed and reported at a glance from faces, after only 13 ms of visual exposure to a high-attractiveness face (Olson & Marshuetz, 2005), and has reported evidence for inhibition of return to lateral Gabor patches preceded by high-attractiveness faces that were adjusted for visibility using staircase visual contrast manipulations (Hung, Nieh, & Hsieh, 2016). Based on these findings, the argument has been raised that attractiveness “can be processed in the absence of conscious awareness” (Hung et al., 2016; p. 6) and that it can be appraised and reported without conscious awareness (Olson & Marshuetz, 2005).

In the current study, we explored whether particularly the appraisal of attractiveness requires conscious awareness. We presented emotional and neutral faces that were rated high, medium and low in attractiveness. We presented these faces for 33.33 ms (Brooks et al., 2012) using backwards masking to a black and white pattern for 116.67 ms. We measured face-detection and emotion-discrimination performance as well as attractiveness ratings in response to these faces. To accurately assess these responses, we used the paradigm we have developed for the assessment of subliminality described in detail in previous research (Tsikandilakis & Chapman, 2018; Tsikandilakis, Chapman, & Peirce, 2018). This included response assessment using unbiased nonparametric signal detection theory criteria to measure face-detection and emotion-discrimination performance (Stanislaw & Todorov, 1999; Zhang & Mueller, 2005), Bayesian analysis (Dienes, 2015) for the assessment of chance-level significance (A = .5) that would indicate stimuli invisibility (Erdelyi, 2004) and hits (correct detection/discrimination) and misses (incorrect detection/discrimination) analysis of participant ratings (Fawcett, 2006).

The inclusion of neutral faces (Hung et al., 2016; Olson & Marshuetz, 2005) as well as emotional faces in the current study was necessitated by signal detection theory requirement for the exploration of different levels of conscious awareness (Stanislaw & Todorov, 1999). The inclusion of different emotions enabled us to explore whether the appraisal of attractiveness requires face detection as well as emotion discrimination of the presented face. This signal detection theory distinction could not have been performed without the inclusion of different emotions (Pessoa, 2005). This inclusion also meant that the current study would be the first to our knowledge study (Hung et al., 2016; Olson & Marshuetz, 2005; Ritchie, Palermo, & Rhodes, 2017) that included assessment of different emotional expressions and appearance types (high, medium and low-attractiveness faces) under conditions of backwards masking.

This experimental condition allowed us to formulate a secondary and – in the absence of previous research in the area (Fink & Penton-Voak, 2002) – exploratory hypothesis. We explored whether the interaction between emotional expressions and appearance types can influence face detection, emotion discrimination and attractiveness ratings under conditions of backwards masking, in the same manner that previous research has reported that it influences participant responses during supraliminal presentations (Calvo & Lundqvist, 2008; Lundqvist, Bruce, & Öhman, 2015; O’Doherty et al., 2002). The preliminary hypothesis for this exploratory objective was that faces that have high sociobiological value based on attractiveness ratings, that is, high-attractiveness faces (Bachmann, 2007), would reveal higher face-detection and emotion-discrimination scores when they expressed emotions that also have high sociobiological and survival-related evolutionary value (Brooks et al., 2012), such as fearful (Pessoa, 2005) and angry faces (Lundqvistet al., 2015). We also explored whether higher attractiveness ratings would be reported for high-attractiveness faces that expressed highly salient positive social signals, such as happy expressions, that have been associated with increased emotional-perceptual reward value in previous research (Calvo & Lundqvist, 2008; O’Doherty et al., 2002).

In the current research, it was extremely important that any potential influences of attractiveness could not be interpreted in terms of other evolutionarily relevant or low-level stimulus features. We thus included three pilot experimental stages (Study 1) to ensure that the stimuli we used did not confound attractiveness with gender, emotional expression or detectability through contrast changes. In the first experimental stage, we validated the facial stimuli for gender and emotional characteristics and used strict attractiveness criteria to preselect from an existing database (Gur et al., 2002) faces that were rated high, medium and low in attractiveness. In the second experimental stage, we assessed the preselected stimuli for emotionality to make sure that emotionality differences between appearance types would not bias signal detection performance in subsequent experimental stages (Calvo & Lundqvist, 2008). Finally, in Stage 3, we explored whether there were differences in visual contrast between the selected faces and their control condition during the main experiment (nonfacial pattern stimuli) that could artefactually impact signal detection and discrimination responses when using backwards masking (Bachmann & Francis, 2013).

Methods

Study 1

Stage 1: Stimuli preselection

Aims: The current stage had two aims. The first aim was to select from an existing database (Gur et al., 2002) the faces that were correctly recognised by participants and automatic facial recognition software (Noldus) for the emotion that they were expressing. The second aim was to select faces that were rated high, medium and low in attractiveness and test whether these faces produced significant differences in attractiveness that would make them appropriate stimuli for the inclusion in the following experimental stages.

Participants: A power calculation based on effect sizes (d = .81; f = .41) reported in previous research (Tsikandilakis et al., 2018) revealed that 15 participants were required for P_(1–β) ≥ .8 (Faul, Erdfelder, Buchner, & Lang, 2009). Eighteen participants (nine females) volunteered to participate in this experiment. All participants reported normal or corrected-to-normal vision. Participants gave informed consent to participate in the current study prior to the experiment. The participants were screened before the experiment with the Somatic and Psychological Health Report Questionnaire (Hickie et al., 2001); participants with scores at or below 1.0 were included. Participants were also screened using an online Alexithymia-Emotional Blindness questionnaire (Alexithymia, 2017) and participants with scores that indicated possible traits (P > 94) or diagnosis (P > 112) for alexithymia were excluded; data from a single participant were excluded from the study. Two participants were also excluded from the study due to neutral ratings on the attractiveness task (Hung et al., 2016; Olson & Marshuetz, 2005). The final population sample consisted of 15 participants (8 females) with a mean age of 32.87 (SD = 6.12). The experiment was approved by the Ethics Committee of the School of Psychology of the University of Nottingham.

Stimuli and procedures: The facial stimuli used were taken from the dataset created by Gur et al. (2002) and included faces with angry, happy, fearful, sad and neutral facial expressions. The stimuli were adjusted for interpupillary distance, transformed to grey scale and resized to a standard 1,024 × 768 pixels resolution. Their luminescence was averaged in SHINE, MATLAB Toolbox and Fourier Painter, and finally, they were spatially aligned and framed into pure white within a cropped circle (height: 6 cm, width: 4 cm).

A total of 300 faces were presented from 60 different actors. Ninety cropped nonfacial blurs patterns that were matched for luminescence (SHINE and MATLAB Toolbox) with the presented faces were also shown. The experimental trial started with a fixation cross for 2 s (±1 s). After the fixation cross in random order, a single-face or nonfacial pattern was presented at fixation for 1 s. After each target, a black and white pattern mask was also presented for 1 s. A blank screen interval was then presented for 2 s. After that participants were asked by an on-screen message to rate how attractive the presented stimulus was from 1 (not attractive at all) to 10 (very attractive) using the keyboard. Participants were also asked to decide from an on-screen list what kind of stimulus was presented during the trial using the keyboard. The list included (a) angry, (f) fearful, (h) happy, (s) sad, (n) neutral, (o) other and (i) nonfacial. The order of the two engagement tasks was randomised in each trial. A 2-s blank screen interval was presented before the next trial.

Stimuli preselection: We selected from the presented faces the ones that reported 100% accuracy in correct discrimination of emotional expression (n = 248). These stimuli were further analysed using Noldus Face Reader 7.0 to validate their emotion. We used the participant calibration module for emotional recognition that controlled for the action units that were present in the neutral expressions of each actor to accurately assess emotional expressions. We also used the cultural-background recognition module and specific cultural-background emotional recognition modules (e.g., General6₁, Asian, etc.) for each actor based on the cultural-background recognition assessment (Noldus, 2018). We set the emotional recognition certainty criteria for inclusion at >.99 for each facial stimulus; no stimuli were excluded (see Noldus Emotional Recognition Certainty Scores (%) section in Appendix). From the resulting dataset, we chose for each emotion (anger, fear, happiness, sadness and neutral) faces that were rated high, medium and low in attractiveness based on the following criteria. Six (three males and three females) high-, medium- and low-attractiveness faces were selected for each emotion. For high-attractiveness faces, the faces with a mean value that was >7 were preselected. For medium-attractiveness faces, we preselected faces that were rated between 4 and 6 on the attractiveness scale. For low-attractiveness faces, we preselected faces that were not rated higher than 3 on the attractiveness scale. Due to rating restrictions and to avoid identity priming due to uneven target repetition (Lander, Bruce, & Hill, 2001), in subsequent stages actors who met the required attractiveness criteria per emotion were selected three times each, resulting in a final sample of 30 actors and 90 emotional expressions.

Results and discussion: To confirm that attractiveness was different between different appearance types, we ran a repeated measures analysis of variance (ANOVA) with independent variables Appearance Type (high, medium and low attractiveness) and Type of Emotion (anger, fear, happiness, sadness, neutral) and dependent variable attractiveness ratings. Appearance Type was significant – F(2, 28) = 1,739.43, p < .01, η² = .99 – confirming that high-attractiveness faces (M = 7.8, SD = .23) were rated higher – t(14) = 30.01, p < .01; d = 9.24 – than medium-attractiveness faces (M = 5.17, SD = .33) and higher – t(14) = 74.08, p < .01; d = 26.55 – than low-attractiveness faces (M = 2.54, SD = .16) in attractiveness ratings. Medium-attractiveness faces were also rated higher than low-attractiveness faces – t(14) = 46.29, p < .01; d = 10.14 – in attractiveness ratings. The findings suggested that appearance types (high, medium and low) were significantly different in attractiveness ratings and that they were appropriate stimuli for their inclusion in the following experimental stages.

In accordance with previous literature in the area (Hung et al., 2016; Olson & Marshuetz, 2005), we did not find any differences for actor gender attractiveness ratings – F(1, 13) = .11; p = .75; η² = .01 – or with participant gender as a between-subjects variable – F(1, 13) = .37; p = .56; η² = .03. These findings were confirmed by a separate t-test analysis based on actor, t(14) = –.473; p = .64; d = .13, and participant gender attractiveness ratings, t(13) = .407; p = .69; d = .18. These findings suggested that male and female participants did not differ in their ratings for attractiveness (see also Gender Effects and Bayesian Analysis for Attractiveness in Stage 1 for Gender sections in Appendix; Figure 1).

Stage 2: Emotionality assessment

Aims: The aim of this stage was to assess the faces that were rated high, medium and low in attractiveness for emotionality differences to make sure that emotionality differences between appearance types would not bias signal detection performance in subsequent experimental stages.

Figure 1.

Attractiveness ratings per appearance type and expressed emotion. Mean attractiveness ratings (y-axis) per appearance type and expressed emotion (x-axis). Error bars indicate SEM (±2).

Participants: A power calculation based on effect sizes (d = .86; f = .43) reported in previous research (Tsikandilakis et al., 2018) revealed that 13 participants were required for P_(1–β) ≥ .8. Fifteen participants (seven females) who were not part of Stage 1 volunteered to participate in this stage. All participants reported normal or corrected-to-normal vision and gave informed consent to participate in the current study prior to the experiment. The participants were screened with the Somatic and Psychological Health Report Questionnaire and an online Alexithymia-Emotional Blindness questionnaire; no participants were excluded. Data from two participants were excluded due to neutral ratings on the emotionality task (Hung et al., 2016; Olson & Marshuetz, 2005). The final population sample consisted of 13 participants (8 females) with a mean age of 24.85 (SD = 3.95). This stage was approved by the Ethics Committee of the School of Psychology of the University of Nottingham.

Stimuli and procedures: A total of 90 faces were shown during this stage from 30 actors. An equal number of female and male faces (n = 45) and actors (n = 15) were presented. Six faces (three males and three females) were shown per emotion (anger, fear, happiness, sadness and neutral) for each appearance type (high, medium and low attractiveness). The faces presented during this stage were the preselected stimuli from Stage 1. The 90 nonfacial pattern blurs that were shown during Stage 1 were also shown during this stage. The experimental trial started with a fixation cross for 3 s (±1 s). After the fixation cross in random order, a single-face or nonfacial pattern was presented at fixation for 1 s. After each target, a black and white pattern mask was also presented for 1 s. A blank screen interval was then presented for 2 s. Participants were then asked by an on-screen message to rate how emotional the presented stimulus was from 1 (not emotional at all) to 10 (very emotional) using the keyboard. A 2-s blank screen interval was presented before the next trial.

Results and discussion: A repeated measures ANOVA tested the effects of Appearance Type (high, medium and low attractiveness) and Type of Emotion (angry, fearful, happy, sad and neutral) on emotional ratings. The analysis revealed that there were no significant differences in emotional ratings between high, medium and low-attractiveness faces – F(2, 24) = 2.94, p = .12; η² = .16 (see also Bayesian Analysis of Emotionality in Stage 2 Between Appearance Types section in Appendix). These findings suggested that high, medium and low-attractiveness faces were not overall different in emotional ratings and were appropriate stimuli for their inclusion in subsequent experimental stages.

Figure 2.

Emotionality ratings per appearance type and expressed emotion. Mean emotionality ratings (y-axis) per appearance type and expressed emotion (x-axis). Error bars indicate SEM (±2).

An ANOVA revealed that there were no differences for actor gender in emotionality ratings – F(1, 11) = .64; p = .44; η² = .06 – or with participant gender as a between-subjects variable – F(1, 11) = .1; p = .75; η² = .01. This was confirmed by a separate t-test analysis based on actor – t(12) = –.77; p = .12; d = .05 – and participant gender emotionality ratings – t(11) = 1.41; p = .19; d = .73. These findings suggested that male and female participants did not differ in their ratings for emotionality (see also Bayesian Analysis for Emotionality in Stage 2 for Gender section in Appendix; Figure 2).

Stage 3: Subjective contrast discrimination

Aims: Study 2 was designed to include a 116.67 ms black and white pattern mask and 33.33 ms nonfacial blurs as a signal detection control condition for facial stimuli. For such a control condition to work, it is important that low-level changes in contrast cannot be used by participants to detect the presence of faces. The aim of the current stage was to ensure that brief nonfacial blurs did not have differences in visual contrast in respect to the black and white pattern mask when compared with the facial stimuli.

Participants: A power calculation based on effect sizes (d = .05; f = .03; see Schmider, Ziegler, Danay, Beyer, & Bühner, 2010) reported in previous research (Tsikandilakis et al., 2018) revealed that 14 participants were required for β < .2 (Soper, 2018; Wessa, 2016). Fifteen participants (seven females) who were not part of Experiment 1 or Experiment 2 volunteered to participate in this experimental stage. All participants reported normal or corrected-to-normal vision and provided informed consent prior to the study. The participants were screened with the Somatic and Psychological Health Report Questionnaire and an online Alexithymia-Emotional Blindness questionnaire; no participants were excluded based on this assessment. One participant was excluded due to noncompliance with the study procedures. The final population sample consisted of 14 participants (seven females) with a mean age of 23.07 (SD = 2.70). Participants were briefed in writing concerning the experimental task and were asked to respond in the consent form whether they understood the instructions (yes or no). All participants responded positively. This stage was approved by the Ethics Committee of the School of Psychology of the University of Nottingham.

Stimuli and procedures: Due to experimental time restrictions, a subset of the selected faces including 30 faces and 30 nonfacial blurs were presented during this stage. The presented faces were randomly chosen from the preselected stimuli in Stages 1 and 2. The selected set was assigned two faces per emotion (neutral, angry, fearful, happy and sad) for each appearance type (high, medium and low attractiveness). An equal number of male and female faces (n = 15) were presented to participants and no actor was repeated more than once at this stage. The experimental trial started with a fixation cross for 3 s (±1 s). After the fixation cross in random order, a single-face or nonfacial blur was presented for 33.33 ms followed by a black and white patterned mask for 116.67 ms. After the presentation of the black and white pattern mask, participants were shown a blank screen for 2 s and were then asked by an on-screen message to rate their subjective experience of visual contrast from 1 (not at all) to 10 (intense). A 2-s blank screen interval was presented before the next trial.

Results and discussion: To test if the nonfacial blurs had significantly different ratings for subjective experience of contrast compared with the presented faces in respect to the pattern mask a paired samples t-test was run. Subjective experience of contrast for the nonfacial blurs (M = 4.95, SD = .49) was not rated higher than contrast in the face condition – M = 4.96, SD = .21; t(13) = –.119, p = .907; d = .03 (see also Bayesian Analysis in Stage 3 for Visual Contrast section in Appendix). These findings suggested that differences of visual contrast between the nonfacial blurs and the presented faces, and the pattern mask would not artefactually impact signal detection and discrimination performance (Bachmann & Francis, 2013) in subsequent experimental stages.

Study 2

Aims: The primary aims of this study were twofold. First, we wanted to test whether high-attractiveness faces would be detected and discriminated more accurately than other appearance types under conditions of backwards masking. Second, we wanted to test whether the appraisal of attractiveness in high-attractiveness faces requires conscious awareness. Finally, an exploratory aim of the current study was to test if high, medium and low-attractiveness faces interact with different types of emotional expressions (fearful, angry, happy, sad and neutral) to influence face-detection and emotion-discrimination performance as well attractiveness ratings under conditions of backwards masking.

Participants: A power calculation based on medium effect sizes (η² = .06; f = .25) revealed that 23 participants would be required for P_(1–β) ≥ .8. Twenty-six participants (13 females) that were not part of Study 1 volunteered to participate in this experiment. All participants reported normal or corrected-to-normal vision and gave informed consent to participate in the current study prior to the experiment. The participants were screened with the Somatic and Psychological Health Report Questionnaire (Hickie et al., 2001) and an online Alexithymia-Emotional Blindness questionnaire (Alexithymia, 2017); data from one participant were excluded due to possible Alexithymia traits (>94). Data from two additional participants were also excluded due to neutral ratings on the preexperimental emotionality task and attractiveness tasks (Hung et al., 2016; Olson & Marshuetz, 2005; van der Ploeg, Brosschot, Versluis, & Verkuil, 2017). The final population sample consisted of 23 participants (13 females) with a mean age of 32.13 (SD = 7.47). This experiment was approved by the Ethics Committee of the School of Psychology of the University of Nottingham.

Stimuli and procedures: The experiment involved two phases. In Phase 1, participants were presented with 20 faces and 20 nonfacial pattern blurs that were not part of the preselected stimuli. These stimuli were chosen based on discrimination, emotionality, intensity, expression-ambiguity ratings and physiological responses (skin conductance and heart rate) in a previous study (Tsikandilakis et al., 2018). These facial stimuli included four neutral, angry, happy, sad and fearful stimuli from different actors with an equal number (n = 10) of males and females. The experimental trial started with a fixation cross for 3 s (±1 s). After the fixation cross in random order, a single-face or nonfacial pattern was presented at fixation for 1 s. After each target, a black and white pattern mask was also presented for 1 s. A blank screen interval was then presented for 2 s. After that participants were asked by an on-screen message to rate how emotional the presented stimulus was from 1 (not emotional at all) to 10 (very emotional) using the keyboard. The participants were also asked by an on-screen message to rate how attractive the presented stimulus was from 1 (not attractive at all) to 10 (very attractive) using the keyboard. The participants were also asked to decide from an on-screen list what kind of stimulus was presented during the trial using the keyboard. The list included (a) angry, (f) fearful, (h) happy, (s) sad, (n) neutral, (o) other and (i) nonfacial. The order of the engagement tasks was randomised in each trial. A blank screen for 2 s was presented before each next trial. Phase 1 was conducted to ensure that participants were familiar with the tasks and stimuli under conditions in which the target stimuli were clearly visible for all participants. In Phase 2, we used the same procedure with brief backwardly masked stimuli presented at durations where they would not necessarily be available to conscious awareness (Dehaene, Changeux, Naccache, Sackur, & Sergent, 2006).

After Phase 1, participants were allowed a 5-min break. After the break, the participants were presented with the 90 preselected faces and 90 nonfacial pattern blurs. The experimental trial started with a fixation cross for 3 s (±1 s). After the fixation cross in random order, a single-face or nonfacial pattern was presented at fixation for 33.33 ms. After each target, a black and white pattern mask was also presented for 116.67 ms. A blank screen interval was then presented for 2 s. After that participants were asked to reply to a set of engagement tasks with order randomised. They were asked by an on-screen message to press E if they saw a facial stimuli or W if the presented target was nonfacial; the assignment of the keyboard responses was randomly counterbalanced in each trial. After this initial task, we used conditional branching to present the participants with additional engagement tasks. If the participant responded having seen a facial stimulus, an on-screen message asked participants to decide from a list what kind of emotion the facial stimulus was expressing using the keyboard. The list included (a) angry, (f) fearful, (h) happy, (s) sad, (n) neutral and (o) other. If the participants replied not having seen facial stimulus, an on-screen message asked them to decide what kind of emotion best described their experience during the presentation using the keyboard. The list included (a) anger, (f) fear, (h) happiness, (s) sadness, (n) neutral and (o) other. Participants were also asked by an on-screen message to rate how attractive the presentation was from 1 (not attractive at all) to 10 (very attractive) using the keyboard (Figure 3). A blank screen for 2 s was presented before the next trial.

Figure 3.

Example stimuli sequence with high-attractiveness happy masked face. During the main experimental stage, participants were presented with 30 high-attractiveness, 30 medium-attractiveness and 30 low-attractiveness emotional faces (angry, fearful, happy, sad and neutral) and 90 pattern blurs for 33.33 ms. Subsequently, they were asked to make face-detection and emotion-discrimination, and attractiveness-rating responses.

Apparatus and presentation testing: All experiments were generated using the coder and builder components of Psychopy version 1.90d (Peirce, 2007). All stimuli for all experimental stages were presented on a standard 60 Hz Toshiba monitor in the same quiet laboratory space. To ensure that particularly brief stimuli (33.33 and 116.67 ms) were correctly presented, an IPAD PRO camera with 120 Hz refresh rate (8.33 ms) recorded two pilot runs for Study 1 (Stage 3) and Study 2. The stimuli presentation was assessed frame by frame; no instances of dropped frames were detected. Subsequently, a self-developed dropped frame report script with one frame (16.67 ms) tolerance threshold was coded in Python and two pilot experimental diagnostic sessions were run. The presenting monitor reported no dropped frames; prognostic dropped frame rate was estimated at 1/5,000 trials. Experimental stages were, subsequently, run using dropped frames diagnostics; no instances of dropped frames were reported.

Results and discussion: Does attractiveness influence face detection under conditions of backwards of masking?: To test whether high-attractiveness faces are detected and discriminated more accurately than other appearance types, the participants’ face detection responses were transformed to nonparametric sensitivity index A (Zhang & Mueller, 2005). An ANOVA with independent variables Appearance Type (high, medium and low attractiveness) and Type of Emotion (angry, fearful, happy, sad and neutral) and dependent variable face-detection performance (A) was run. The analysis revealed a significant effect of Appearance Type – F(2, 44) = 9.49, p < .01; η² = .3 – and a significant effect of Type of Emotion – F(2.81, 61.79) = 5.27, p < .01; η² = .19; Greenhouse-Geisser corrected. Further, Bonferonni-corrected pairwise comparisons revealed that high-attractiveness faces – M (A) = .797, SD (A) = .088; M (hit rates [HR]) = 74.07%, SD (HR) = 14.64% – were detected more accurately than medium-attractiveness faces – M (A) = .768, SD (A) = .093; M (HR) = 69.72%, SD (HR) = 20.77%; p < .01; d = .32. We did not observe any gender effects – F(2, 42) = .71, p = .49; η² = .03.

Emotion discrimination was also calculated using nonparametric sensitivity index A (Zhang & Mueller, 2005). An ANOVA was run to assess the effects of Appearance Type and Type of Emotion with dependent variable emotion-discrimination performance (A). The analysis revealed a significant effect of Appearance Type – F(2, 44) = 4.97, p = .01; η² = .18 – a significant effect of Type of Emotion – F(40.6, 2.57) = 12.58, p < .01; η² = .36 – and a significant interaction – F(8, 176) = 10.81, p < .01; η² = .32. Further, Bonferonni-corrected pairwise comparisons revealed that high-attractiveness faces – M (A) = .756, SD (A) = .082; M (H.R.) = 66.82% SD (H.R.) = 20.37% – were discriminated better than low-attractiveness faces – M (A) = .734, SD (A) = .087; M (H.R.) = 63.78% SD (H.R.) = 20.51%; p < .01; d = .26. We did not observe any gender effects – F(2, 42) = 1.65, p = .2; η² = .07. These results suggested that attractiveness influenced face detection and emotion discrimination under conditions of backwards masking and more specifically that high-attractiveness faces were detected more accurately than medium-attractiveness faces and discriminated more acutely than low attractiveness-faces.

Exploratory analysis: The exploratory analysis in the current stage tested whether appearance type and emotion interact under conditions of backwards masking to influence face-detection and emotion-discrimination performance. The analysis revealed a significant interaction between Appearance Type and Emotion – F(8, 176) = 10.81, p < .01; η² = .32. Further, Bonferonni-corrected pairwise comparisons revealed that fearful (p < .01; d = .36) and sad (p < .01; d = 42) high-attractiveness faces were detected more accurately than other appearance type to emotion combinations, and that fearful (p < .01; d = 36) and angry (p < .001; d = .65) high-attractiveness faces, and also neutral low-attractiveness faces (p < .001; d = .99) were discriminated more accurately than other appearance type to emotion combinations. These results suggested that a significant interaction between attractiveness and emotion influenced face detection and emotion discrimination for the presented faces (Table 1).

Table 1.

Detection and Discrimination Performance per Appearance Type and Emotion.

Appearance type	Type of emotion	Detection (A)		Discrimination (A)		Cohen’s d
Appearance type	Type of emotion	Mean	SD	Mean	SD	Detection	Discrimination
High attractiveness	Fearful	.818	.132	.782	.101	.357*	.355*
	Angry	.798	.094	.815	.102	.161	.654**
	Happy	.752	.127	.687	.102	–.281	–.535**
	Sad	.825	.094	.747	.115	.416*	.026
	Neutral	.79	.08	.75	.09	.085	.05
Medium attractiveness	Fearful	.756	.112	.725	.11	–.242	–.184
	Angry	.788	.095	.76	.102	.067	.14
	Happy	.751	.113	.701	.098	–.287	–.409**
	Sad	.777	.139	.771	.11	–.044	.245
	Neutral	.766	.093	.716	.099	–.151	–.266
Low attractiveness	Fearful	.77	.098	.717	.128	–.108	–.256
	Angry	.809	.094	.734	.125	.267	–.099
	Happy	.742	.102	.683	.102	–.378*	–.576**
	Sad	.809	.1	.729	.139	.262	–.143
	Neutral	.768	.090	.851	.086	–.125	.996**

Note. The two end-right columns show the standardised effect size (Cohen’s d) per stimuli type in units of standard deviations from the overall mean for detection (M = .781, SD = .104) and discrimination performance (M = .745, SD = .107). Asterisk (*) indicates significance at <.01 level. Double asterisk (**) indicates significance at <.001 level.

Results and discussion: Does the appraisal of Attractiveness require conscious awareness?: To test if high-attractiveness faces can be processed without conscious awareness, a Bayesian analysis (Dienes, 2015) with corrected degrees of freedom (df < 30; SE = $(S E \times (1 + (\frac{20}{d f \times d f})))$ (Berry, 1996)) and higher and lower bounds set at .6 and .4 with .5 indicating absolute chance-level performance was run. The analysis revealed that face detection (A) for high-attractiveness (M = .796, SE = .019; B > 3), medium-attractiveness (M = .768, SE = .017; B > 3) and low-attractiveness (M = .779, SE = .02; B > 3) faces, and emotion-discrimination performance for high-attractiveness (M = .756, SE = .018; B > 3), medium-attractiveness (M = .743, SE = .021; B > 3) and low-attractiveness (M =. 734, SE = .019; B > 3) faces were above chance-level (Dienes, 2015) that would indicate stimuli invisibility (Erdelyi, 2004). An analysis of overall attractiveness ratings was not appropriate for the exploration of whether the appraisal of attractiveness requires conscious awareness (Figure 5(a) and (b)).

To further explore whether attractiveness could influence ratings without conscious awareness, we ran an analysis of hits and misses per appearance type for face-detection and emotion-discrimination performance (Pessoa, 2005). For face-detection performance, a factorial ANOVA with independent variables Appearance Type (high, medium and low attractiveness), Emotional Type (fearful, angry, happy, sad and neutral) and Detection Response (hits, misses) and dependent variable attractiveness ratings was performed. The analysis revealed a significant effect of Appearance Type – F(2, 22) = 122.14, p < .01; η² = .92 – a significant effect of Emotion – F(4, 44) = 3.37, p = .02; η² = .23 – and a significant effect of Detection Response – F(1, 11) = 51.47, p < .01; η² = .63. Critically, an Appearance Type by Detection Response interaction was reported – F(2, 22) = 167.08, p < .01; η² = .94 – suggesting that there were attractiveness-rating differences between hits and misses for different appearance types.

To further explore these findings, Bonferonni-corrected pairwise comparisons were run. The comparisons revealed that for face detection high-attractiveness facial-hits (M = 7.21, SD = .42) were rated higher than medium-attractiveness facial-hits – M = 4.79, SD = .24; t(22) = 30.55, p < .01; d = 7.07 – and low-attractiveness facial-hits – M = 3.78, SD = .54; t(22) = 22.29, p < .01; d = 7.09. No significant differences in attractiveness ratings were reported for misses for face detection – F(1.08, 20.5) = 1.10, p = .38; η² = .06; Greenhouse-Geisser corrected. For detection-misses, high-attractiveness faces were not significantly different than medium-attractiveness (p = 1; d = .22) and low-attractiveness faces (p = .96; d = .19) and medium-attractiveness faces were not significantly different than low-attractiveness faces (p = .85; d = .1).

A similar pattern was reported for emotion-discrimination responses. An ANOVA was run with independent variables Appearance Type (high, medium and low attractiveness), Emotion (fearful, angry, happy, sad and neutral) and Discrimination Response (hits and misses) with attractiveness ratings as the dependent variable. The analysis revealed a significant effect of Appearance Type – F(2, 14) = 74.01, p < .01; η² = .91 – a significant effect of Emotion – F(4, 28) = 2.98, p = .04; η² = .29 – and a significant effect of Discrimination Response – F(1, 7) = 132.55, p < .01; η² = .95. Critically, a significant Appearance Type by Discrimination Response interaction was revealed –F(2, 14) = 39.78, p < .01; η² = .85.

To further explore these findings, Bonferonni-corrected pairwise comparisons were run. For emotion-discrimination performance, high-attractiveness facial-hits were rated higher (M = 7.44, SD = .45) than medium-attractiveness facial-hits – M = 4.83, SD = .25; t(22) = 26.46, p < .01; d = 7.17 – and low-attractiveness facial-hits – M = 3.58, SD = .53; t(22) = 23.12, p < .01; d = 7.85. Medium-attractiveness facial-hits were also rated higher for attractiveness ratings than low-attractiveness facial-hits (p < .01; d = 3.02). For emotion discrimination, high-attractiveness facial-misses (M = 6.23, SD = .68) reported significantly higher attractiveness ratings compared with medium-attractiveness facial-misses – M = 4.23, SD = 1.14; t(15) = 4.91, p < .01; d = 2.13 – and low-attractiveness facial-misses – M = 4.59, SD = .49; t(11) = 9.44, p < .01; d = 2.77.

These results suggested that the appraisal of attractiveness could not be performed in the absence of conscious awareness (face-detection misses) and that correct face detection (hits) was a necessary condition for the appraisal of attractiveness (see Figure 4). Interestingly, although correct emotion discrimination (hits) enhanced the acuity of the appraisal for attractiveness, incorrect emotion discrimination (misses) reported significant differences between different appearance types, suggesting that emotion discrimination was not necessary for the appraisal of attractiveness from faces.

Figure 4.

Attractiveness ratings for hits and misses. Mean attractiveness ratings per appearance type for hits and misses for face-detection and emotion-discrimination performance in Study 2. Error bars indicate SEM (±2). Asterisks (*) indicate significance at p < .01 level.

Exploratory analysis: The exploratory analysis in the current stage tested whether appearance type and emotion interact under conditions of backwards masking to influence attractiveness ratings. The analysis revealed a trend for significance for an interaction between Appearance Type and Emotion – F(4, 28) = 2.01, p = .06; η² = .22; no pairwise comparisons survived the Bonferonni corrections (Figure 5). These findings suggested that the interaction between emotion and attractiveness was associated with minor differences in attractiveness ratings. Further hits and misses analysis for this interaction could not be performed because of insufficient face-detection and emotion-discrimination hits and misses responses for several appearance type and emotion combinations (e.g., high-attractiveness angry faces, low-attractiveness neutral faces) and because the available responses did not meet – P_(1–β) = .34 – the minimum statistical power requirement criteria – P_(1–β) ≥ .8.

Figure 5.

Attractiveness and emotion under conditions of backwards masking. Attractiveness ratings for each Appearance type (high, medium and low attractiveness) for each Emotion (neutral, fearful, angry, happy and sad). A significance trend (p = .06) for an Appearance to Emotion Interaction was reported although no pairwise comparisons survived the Bonferonni corrections. Asterisks (*) indicate significance at p < .01 level.

Summary of Findings

The primary aim of the current article was to explore whether attractiveness influences face-detection and emotion-discrimination performance and whether self-reports for the appraisal of facial attractiveness require conscious awareness. We implemented several methodological developments to explore these hypotheses such as extensive pilot experimental stimuli controls, signal detection analysis using sensitivity index A, Bayesian assessment for chance-level significance that would indicate stimuli invisibility and analysis of hits and misses in face-detection and emotion-discrimination performance for attractiveness-rating responses. We found that attractiveness influences face-detection and emotion-discrimination performance, and more specifically that high-attractiveness faces can be detected and discriminated more accurately than other appearance types. Our analysis also revealed that specific appearance type and emotion combinations such as high-attractiveness fearful, angry and sad faces, and low-attractiveness neutral faces were detected and discriminated more accurately than other appearance type to emotion combinations. Critically, we found that face detection (hits) was a necessary condition for the appraisal of attractiveness for high-attractiveness faces and that when participants had absence of conscious awareness of the presented face (face-detection misses) they did not rate high-attractiveness faces higher than other appearance types. Interestingly, although correct emotion discrimination (hits) enhanced the acuity of the appraisal for high-attractiveness faces, incorrect emotion discrimination (misses) also reported higher ratings for high-attractiveness faces compared with other appearance types, suggesting that emotion discrimination is not necessary for the appraisal of attractiveness from faces. Finally, our exploratory analysis revealed that the interaction of emotion and appearance type impacts face-detection and emotion-discrimination performance but has only a minor effect in ratings for attractiveness.

General Discussion

Previous research has reported that attractiveness can be appraised from minimal information, such as pixelated images (Bachmann, 2007) and brief presentations (Olson & Marshuetz, 2005), and processed despite interocular suppression (Hung et al., 2016). For example, Olson and Marshuetz (2005) reported that high-attractiveness faces presented for as little as 13 ms, preceded by a high-frequency contour scrambled face mask for 39 ms and followed by a carton mask for 39 ms, were associated with higher attractiveness ratings than low-attractiveness faces presented under the same conditions. They also reported that 13 ms high-attractiveness faces elicited shorter reaction times for the appraisal of subsequently presented positively valanced words than low-attractiveness faces (Olson & Marshuetz, 2005; p. 500). Similarly, Hung et al. (2016) ran a series of experiments using interocular suppression (Sengpiel, Blakemore, & Harrad, 1995). They used staircase reduction of contrast visibility when bilaterally presenting high-attractiveness and low-attractiveness faces followed by Gabor patches. They reported a reduction in Gabor lines orientation discrimination performance for lateral high-attractiveness faces due to perceptual inhibition of return, suggesting that high-attractiveness faces were attended despite the staircase reduction in visual contrast (Klein, 2000). These findings have been used to propose that attractiveness can be appraised from minimal information and have also been used to suggest that attractiveness can be processed and reported in the absence of conscious awareness (see also Kleckner et al., 2018).

The current findings refer to whether attractiveness can be appraised (Olson & Marshuetz, 2005) without conscious awareness. Our findings support previous research (Bachmann, 2007; Hung et al., 2016; Olson & Marshuetz, 2005) in that – like other evolutionary important stimuli (Mineka & Öhman, 2002; Öhman, 2009) that confer “sociobiological value” (Bachmann, 2007; p. 848) – high-attractiveness faces convey highly salient cues (Fink & Penton-Voak, 2002) that render them more accurately detectable during brief presentations (Calvo & Lundqvist, 2008; Lähteenmäki, Hyönä, Koivisto, & Nummenmaa, 2015). The current findings also add to the existing literature by suggesting that high-attractiveness faces are discriminated more accurately (Adolphs, 2008) and that correct discrimination of the expressed emotion enhances the acuity but is not a necessary condition for the appraisal of the attractiveness from faces (Sergent & Dehaene, 2004).

Despite this partial consensus with previous research, the current article employed different methodological assessment and statistical applications (Tsikandilakis & Chapman, 2018; Tsikandilakis et al., 2018) compared with previous publications. These included the application of signal detection theory (Zhang & Mueller, 2005), Bayesian analysis of chance-level significance (Dienes, 2015) that would indicate stimuli invisibility (Erdelyi, 2004) and separate analysis for hits and misses for attractiveness ratings (Pessoa, 2005). Based on the current methods, our results also partly disagree with previous findings and we propose that the appraisal of attractiveness does require conscious target-detection meta-awareness of facial characteristics (Tsikandilakis et al., 2018). This finding does not per se oppose previous findings that have suggested that attractiveness can be processed in the absence of awareness (Hung et al., 2016) because ratings for attractiveness were explicitly measured in the current study. Instead, the key finding of the current report is that the appraisal of attractiveness cannot be reported in the absence of awareness and that correct detection (hits) of a masked face was a required condition for the appraisal of attractiveness. Therefore, we suggest that to make conscious judgements (see also Lau, 2008) in relation to the attractiveness of a presented face, conscious detection of that face is a necessary condition.

Finally, in respect to our exploratory analysis, we reported that high-attractiveness angry, fearful and sad faces as well as low-attractiveness neutral faces were detected and discriminated more accurately than other appearance type to emotion combinations. We also reported a trend for an interaction between emotion and appearance type in relation to attractiveness ratings under conditions of backwards masking. These findings are a novel contribution to the field and support that high-attractiveness angry and fearful faces confer a face-detection superiority effect (Lundqvist et al., 2015) compared with other stimuli, possibly due to their high-salience and sociobiological value (Brooks et al., 2012). The finding that emotion and appearance type revealed only a trend for an interaction in relation to attractiveness ratings means that we could not provide solid support for the idea that particularly positive emotional expressions influence attractiveness ratings when these are presented under conditions of backwards masking (O’Doherty et al., 2002). This could relate to low power in the current experiment or relate to participants’ inability to process and integrate multiple informational input when the signal strength is reduced under conditions of backwards masking (Baars, 2002).

Limitations

To our knowledge, the current study is the first attempt to explore the appraisal of attractiveness using backwards masking and angry, fearful, sad and happy faces in addition to neutral faces that have been employed in previous research (Hung et al., 2016; Olson & Marshuetz, 2005). Despite previous research suggesting a happy face detection superiority effect during low-level visual processing (Miyazawa & Iwasaki, 2010) and higher attractiveness ratings for happy faces during supraliminal presentations (O’Doherty et al., 2003), we were not able to find higher face-detection or emotion-discrimination responses for happy faces for any appearance type. In addition, emotional expression and appearance type provided only an overall trend for an interaction with attractiveness ratings as the dependent variable. Future research could benefit from a dedicated and appropriately powered exploration of the interaction between emotional expressions and appearance types (Penton-Voak & Chang, 2008) under conditions of backwards masking, and particularly the exploration of the possibility that different emotional type (fearful, angry, happy, sad and neutral) to appearance type (high, medium and low attractiveness) combinations could result in differentiating patterns of attractiveness reports (Rhodes, 2006). In the same context, it is possible, although outside the scope of the current article, that low-attractiveness faces could also report face-detection and emotion-discrimination differences compared with other appearance types, such as mid-attractiveness faces, because they confer health- and fitness-related perceptual cues with avoidance response value (Jaeger, Wagemans, Evans, & van Beest, 2018; Pazda, Thorstenson, Elliot, & Perrett, 2016). This possibility is also supported by the data in the current article (see Table 1), and future research could benefit from a dedicated exploration of face-detection and emotion-discrimination performance of low-attractiveness faces under conditions of backwards masking.

Conclusions

The current study assessed whether attractiveness has an impact on face detection and emotion discrimination using backwards masking, and whether attractiveness ratings require conscious awareness. Our results revealed that high-attractiveness faces were detected and discriminated more accurately than other appearance types and also that detection of a face was a necessary condition for the appraisal of attractiveness. This is a novel contribution to the field and suggests that appraisal of attractiveness requires conscious awareness of a presented face.

Footnotes

Acknowledgements

The primary author would like to thank Sevim Hodge, Harry Purser, Martyn Quigley, Alison Mostyn, Jon Peirce, Jan Derfuss, and Anthony Watkins for their contribution in this project personally and academically. The data for the current report have been made open access at .

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The current research project was funded by the Economic and Social Research Council of the University of Nottingham.

References

Adolphs

(2008). Fear, faces, and the human amygdala. Current Opinion in Neurobiology, 18, 166–172.

Alexithymia. (2017, September 17). Alexithymia: Emotional blindness. Retrieved from http://www.alexithymia.us/test-alex.html

Baars

B. J.

(2002). The conscious access hypothesis: Origins and recent evidence. Trends in Cognitive Sciences, 6, 47–52.

Bachmann

(2007). When beauty breaks down: Investigation of the effect of spatial quantisation on aesthetic evaluation of facial images. Perception, 36, 840–849.

Bachmann

Francis

(2013). Visual masking: Studying perception, attention, and consciousness. Academic Press.

Barber

(1995). The evolutionary psychology of physical attractiveness: Sexual selection and human morphology. Ethology and Sociobiology, 16, 395–424.

Berry

D. A.

(1996). Statistics: A Bayesian perspective. London, England: Duxbury Press.

Brooks

S. J.

Savov

Allzén

Benedict

Fredriksson

Schiöth

H. B.

(2012). Exposure to subliminal arousing stimuli induces robust activation in the amygdala, hippocampus, anterior cingulate, insular cortex and primary visual cortex: A systematic meta-analysis of fMRI studies. NeuroImage, 59, 2962–2973.

Cai

Hahn

A. C.

Zhang

Holzleitner

I. J.

Lee

A. J.

DeBruine

L. M.

Jones

B. C.

(2018, September 12). No evidence that women’s facial attractiveness, femininity, or averageness are valid health cues. Retrieved from https://osf.io/f9tu2/

10.

Calvo

M. G.

Lundqvist

(2008). Facial expressions of emotion (KDEF): Identification under different display-duration conditions. Behavior Research Methods, 40, 109–115.

11.

Dehaene

Changeux

J. P.

Naccache

Sackur

Sergent

(2006). Conscious, preconscious, and subliminal processing: A testable taxonomy. Trends in Cognitive Sciences, 10, 204–211.

12.

Dienes

(2015). How Bayesian statistics are needed to determine whether mental states are unconscious. Behavioural Methods in Consciousness Research, 2015, 199–220.

13.

Erdelyi

M. H.

(2004). Subliminal perception and its cognates: Theory, indeterminacy, and time. Consciousness and Cognition, 13, 73–91.

14.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using G* Power 3.1: Tests for correlation and regression analyses. Behavior research methods, 41(4), 1149--1160.

15.

Fawcett, T. (2006). An introduction to ROC analysis. Pattern recognition letters, 27(8), 861--874.

16.

Fink

Penton-Voak

(2002). Evolutionary psychology of facial attractiveness. Current Directions in Psychological Science, 11, 154–158.

17.

Fink

Neave

Manning

J. T.

Grammer

(2006). Facial symmetry and judgements of attractiveness, health and personality. Personality and Individual Differences, 41, 491–499.

18.

Fisher

R. J.

(2014). The price of being beautiful: Negative effects of attractiveness on empathy for children in need. Journal of Consumer Research, 41, 436–450.

19.

Foo

Y. Z.

Simmons

L. W.

Rhodes

(2017). Predictors of facial attractiveness and health in humans. Scientific Reports, 7, 1–12.

20.

Frieze

I. H.

Olson

J. E.

Russell

(1991). Attractiveness and income for men and women in management. Journal of Applied Social Psychology, 21, 1039–1057.

21.

Grammer

Thornhill

(1994). Human (Homo sapiens) facial attractiveness and sexual selection: The role of symmetry and averageness. Journal of Comparative Psychology, 108, 233.

22.

Griffin

A. M.

Langlois

J. H.

(2006). Stereotype directionality and attractiveness stereotyping: Is beauty good or is ugly bad? Social Cognition, 24, 187–206.

23.

Gur

R. C.

Sara

Hagendoorn

Marom

Hughett

Macy

, . . . Gur

R. E.

(2002). A method for obtaining 3-dimensional facial expressions and its standardization for use in neurocognitive studies. Journal of Neuroscience Methods, 115, 137–143.

24.

Hickie

I. B.

Davenport

T. A.

Hadzi-Pavlovic

Koschera

Naismith

S. L.

Scott

E. M.

Wilhelm

K. A.

(2001). Development of a simple screening tool for common mental disorders in general practice. The Medical Journal of Australia, 175, S10–S17.

25.

Hong

Tandoc

Jr. Kim

E. A.

Kim

Wise

(2012). The real you? The role of visual cues and comment congruence in perceptions of social attractiveness from Facebook profiles. Cyberpsychology, Behavior, and Social Networking, 15, 339–344.

26.

Hung

S. M.

Nieh

C. H.

Hsieh

P. J.

(2016). Unconscious processing of facial attractiveness: Invisible attractive faces orient visual attention. Scientific Reports, 6, 37117.

27.

Jaeger

Wagemans

F. M.

Evans

A. M.

Beest

V.I.

(2018). Effects of facial skin smoothness and blemishes on trait impressions. Perception, 47(6), 608–625. doi:10.11770301006618767258

28.

Jones

B. C.

Hahn

A. C.

Fisher

C. I.

Wang

Kandrik

Lao

, . . . DeBruine

L. M.

(2018). No evidence that more physically attractive women have higher estradiol or progesterone. bioRxiv, 136515.

29.

Klein

R. M.

(2000). Inhibition of return. Trends in Cognitive Sciences, 4(4), 138–147.

30.

Kleckner

I. R.

Anderson

E. C.

Betz

N. J.

Wormwood

J. B.

Eskew

R. T.

Barrett

L. F.

(2018). Conscious awareness is necessary for affective faces to influence social judgments. Journal of Experimental Social Psychology, 79, 181–187.

31.

Lähteenmäki

Hyönä

Koivisto

Nummenmaa

(2015). Affective processing requires awareness. Journal of Experimental Psychology: General, 144(2), 339.

32.

Lander

Bruce

Hill

(2001). Evaluating the effectiveness of pixelation and blurring on masking the identity of familiar faces. Applied Cognitive Psychology, 15, 101–116.

33.

Langlois, J. H., & Stephan, C. W. (1981). Beauty and the beast: The role of physical attractiveness in the development of peer relations and social behavior. Developmental social psychology: Theory and research, 152--168.

34.

Lau

(2008). Are we studying consciousness yet? In Weiskrantz

Davies

(Eds.), Frontiers of consciousness: Chichele lectures (pp. 245–258). Oxford, England: Oxford University Press.

35.

Little

A. C.

Jones

B. C.

DeBruine

L. M.

(2011). Facial attractiveness: Evolutionary based research. Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1571), 1638–1659.

36.

Lyons

M. J.

Campbell

Plante

Coleman

Kamachi

Akamatsu

, (2000). The Noh mask effect: Vertical viewpoint dependence of facial expression perception. Proceedings of the Royal Society of London B: Biological Sciences, 2(1459), 2239–2245.

37.

Lundqvist

Bruce

Öhman

(2015). Finding an emotional face in a crowd: Emotional and perceptual stimulus factors influence visual search efficiency. Cognition and Emotion, 29(4), 621–633.

38.

Mineka

Öhman

(2002). Phobias and preparedness: The selective, automatic, and encapsulated nature of fear. Biological Psychiatry, 52(10), 927–937.

39.

Miyazawa

Iwasaki

(2010). Do happy faces capture attention? The happiness superiority effect in attentional blink. Emotion, 10, 712–716.

40.

Noldus. (2018, June 3). Reference manual. Retrieved from http://www.noldus.com/human-behaviorresearch/products/facereader

41.

O'Doherty, J. P., Deichmann, R., Critchley, H. D., & Dolan, R. J. (2002). Neural responses during anticipation of a primary taste reward. Neuron, 33(5), 815--826.

42.

O’Doherty

Winston

Critchley

Perrett

Burt

D. M.

Dolan

R. J.

(2003). Beauty in a smile: The role of medial orbitofrontal cortex in facial attractiveness. Neuropsychologia, 41, 147–155.

43.

Öhman

(2009). Of snakes and faces: An evolutionary perspective on the psychology of fear. Scandinavian Journal of Psychology, 50, 543–552.

44.

Olson

I. R.

Marshuetz

(2005). Facial attractiveness is appraised in a glance. Emotion, 5, 498–502.

45.

Pazda

A. D.

Thorstenson

C. A.

Elliot

A. J.

Perrett

D. I.

(2016). Women’s facial redness increases their perceived attractiveness: Mediation through perceived healthiness. Perception, 45, 739–754. doi: 10.1177/0301006616633386

46.

Peirce

J. W.

(2007). PsychoPy—Psychophysics software in Python. Journal of Neuroscience Methods, 162, 8–13.

47.

Penton-Voak

I. S.

Chang

H. Y.

(2008). Attractiveness judgements of individuals vary across emotional expression and movement conditions. Journal of Evolutionary Psychology, 6, 89–100.

48.

Pessoa

(2005). To what extent are emotional visual stimuli processed without attention and awareness? Current Opinion in Neurobiology, 15, 188–196.

49.

Rhodes

(2006). The evolutionary psychology of facial beauty. Annual Review of Psychology, 57, 199–226.

50.

Ritchie

K. L.

Palermo

Rhodes

(2017). Forming impressions of facial attractiveness is mandatory. Scientific Reports, 7, 469–468.

51.

Roberts

S. C.

Little

A. C.

Gosling

L. M.

Perrett

D. I.

Carter

Jones

B. C.

, . . . Petrie

(2005). MHC-heterozygosity and human facial attractiveness. Evolution and Human Behavior, 26, 213–226.

52.

Saegusa

Watanabe

(2016). Judgments of facial attractiveness as a combination of facial parts information over time: Social and aesthetic factors. Journal of Experimental Psychology: Human Perception and Performance, 42, 173.

53.

Saxton

T. K.

Caryl

P. G.

Roberts

S. C.

(2006). Vocal and facial attractiveness judgments of children, adolescents and adults: The ontogeny of mate choice. Ethology, 112, 1179–1185.

54.

Scheib

J. E.

Gangestad

S. W.

Thornhill

(1999). Facial attractiveness, symmetry and cues of good genes. Proceedings of the Royal Society of London B: Biological Sciences, 266, 1913–1917.

55.

Scott

I. M. L.

Clark

A. P.

Boothroyd

L. G.

Penton-Voak

I. S.

(2013). Do men’s faces really signal heritable immunocompetence? Behavioral Ecology, 24, 579–589.

56.

Sengpiel

Blakemore

Harrad

(1995). Interocular suppression in the primary visual cortex: A possible neural basis of binocular rivalry. Vision Research, 35, 179–195.

57.

Sergent

Dehaene

(2004). Is consciousness a gradual phenomenon? Evidence for an all-or-none bifurcation during the attentional blink. Psychological Science, 15, 720–728.

58.

Schmider

Ziegler

Danay

Beyer

Bühner

(2010). Is it really robust? Methodology, 6, 147–151.

59.

Smith

M. L.

Perrett

D. I.

Jones

B. C.

Cornwell

R. E.

Moore

F. R.

Feinberg

D. R.

, . . . Pitman

R. M.

(2006). Facial appearance is a cue to oestrogen levels in women. Proceedings of the Royal Society of London B: Biological Sciences, 273, 135–140.

60.

Soper

(2018, September 17). Type II error. Retrieved from https://www.danielsoper.com/statcalc/calculator.aspx?id=50

61.

Stanislaw

Todorov

(1999). Calculation of signal detection theory measures. Behavior Research Methods, Instruments, & Computers, 31, 137–149.

62.

Swami

Furnham

(2008). The psychology of physical attraction. Hove, England: Psychology Press.

63.

Thomas

L. A.

De Bellis

M. D.

Graham

LaBar

K. S.

(2007). Development of emotional facial recognition in late childhood and adolescence. Developmental Science, 10, 547–558.

64.

Tsikandilakis

Chapman

Peirce

(2018). Target meta-awareness is a necessary condition for physiological responses to masked emotional faces: Evidence from combined skin conductance and heart rate assessment. Consciousness and Cognition, 58, 75–89.

65.

Tsikandilakis

Chapman

(2018). Skin conductance responses to masked emotional faces are modulated by hit rate but not signal detection theory adjustments for subjective differences in the detection threshold. Perception, 47, 432–450. doi: 10.1177/0301006618760738

66.

van der Ploeg

M. M.

Brosschot

J. F.

Versluis

Verkuil

(2017). Peripheral physiological responses to subliminally presented negative affective stimuli: A systematic review. Biological Psychology, 129, 131–153.

67.

Wessa

(2016). Testing mean with known variance (type II error) (v1.0.5) in Free Statistics Software (v1.2.1). Office for Research Development and Education. Retrieved from http://www.wessa.net/rwasp_hypothesismean3.wasp/

68.

Winkler

Rhodes

(2005). Perceptual adaptation affects attractiveness of female bodies. British Journal of Psychology, 96, 141–154.

69.

Zhang

Mueller

S. T.

(2005). A note on ROC analysis and non-parametric estimate of sensitivity. Psychometrika, 70, 203–212.