Labelling faces as ‘Autistic’ reduces the inversion effect

Abstract

Does the belief that a face belongs to an individual with autism affect recognition of that face? To address this question, we used the inversion effect as a marker of face recognition. In Experiment 1, participants completed a recognition task involving upright and inverted faces labelled as either ‘regular’ or ‘autistic’. In reality, the faces presented in both conditions were identical. Results revealed a smaller inversion effect for faces labelled as autistic. Thus, simply labelling a face as ‘autistic’ disrupts recognition. Experiment 2 showed a larger inversion effect after the provision of humanizing versus dehumanizing information about faces labelled as ‘autistic’. We suggest changes in the inversion effect could be used as a measure to study stigma within the context of objectification and dehumanization.

Keywords

autism face inversion effect face recognition objectification stigma

Autism is one of the fastest growing and the most commonly diagnosed neurodevelopmental disorders (often associated with depression, obsessive compulsive and mood disorders). In 2018, the Centers for Disease Control and Prevention (CDC) released new data showing that in the United States 1 in 59 children is diagnosed with autism. Individuals with autism are often susceptible to stigmatization and its discriminatory consequences. For example, children with autism experience disproportionately high rates of bullying (Montes and Halterman, 2007) routinely struggle with loneliness, anxiety and depression and are under supported by academic institutions (Gelbar et al., 2014).

In a recent study, university students read a vignette describing a person with the symptoms of autism (without presenting the label) before completing a measure of openness towards that person. The majority of students stigmatized the target person by reporting a preference for having a distant relationship with them, rather than close engagement (Butler and Gillis, 2011; Gardiner and Iarocci, 2014).

Here, we investigated university students’ recognition of faces labelled as autistic. In Experiment 1, we manipulated the category label associated with a set of ‘regular’ faces presented to participants: one group was instructed to memorize, and later recognize, a set of faces labelled as ‘regular’, whereas another group was given the same instructions for faces labelled as belonging to individuals with ‘autism’.

We assessed the face inversion effect which refers to impaired performance when trying to recognize a face stimulus turned upside down. The explanation for this deficit is that inversion disrupts configural perceptual processing (i.e. spatial relationships among facial features). Critically, a significantly smaller inversion effect is found when stimuli are objects rather than faces, suggesting more featural perceptual processing for objects (see Maurer et al., 2002). This paradigm has been previously used as a cognitive marker of objectification/dehumanization of sexualized images of men and women (wearing under garments). Specifically, studies have shown the inversion effect to be smaller (i.e. more object-like featural processing) when participants are presented with sexualized images and objects (shoes) compared to non-sexualized images (people wearing a t-shirt and jeans). Furthermore, it has been shown that giving humanizing information about the sexualized images can reduce objectification by increasing the inversion effect (i.e. increased face-like configural processing; Bernard et al., 2012, 2015, 2018; Civile et al., 2016b; Civile and Obhi, 2016).

Here, we aimed to extend this paradigm to stigma and mental health. We specifically selected autism (future studies should examine other disorders) because of its increasing prevalence and the corresponding need to understand how people perceive individuals with autism. In Experiment 1, we engaged students in a recognition task involving upright and inverted face stimuli simply labelled as ‘autistic’. We anticipate a reduction of the inversion effect compared to the control group who were presented with the same stimuli labelled as ‘regular’. In Experiment 2, we tested an intervention providing humanizing versus dehumanizing information prior to exposure to the (autistic) faces in an attempt to determine whether top-down information would modulate the magnitude of the inversion effect.

Method

Participants

A total of 160 naïve participants took part in the two experiments (119 female, M_age = 18.62, SD = 1.51). Each experiment included 80 participants randomly assigned to either one of the two groups (40 in each group). All the participants were students from McMaster University. The experiment was approved by the research ethics committee at McMaster University. For Experiment 1, the sample size was determined from earlier studies that used the same face stimuli and behavioural paradigm (Civile et al., 2011, 2014, 2016a). At the end of the study, the participants in the autistic faces group were asked if they knew what autism was. All participants reported being familiar (some from direct experience and others from reading about it) with autism (for similar findings, see Tipton and Blacher (2014) who found that despite students, professors and university staff being knowledgeable about autism, misperceptions remain).

Materials

The study used 120 images of faces (60 males and 60 females) standardized in grey-scale format and cropped around the hairline (pics.stir.ac.uk).

Procedure

The old/new recognition task consisted of two parts: a ‘study phase’ and an ‘old/new recognition phase’ (Civile et al., 2011, 2014, 2016a). In the study phase, each participant was shown upright and inverted faces with 30 images for each type (60 images overall). Faces were presented one at a time in random order. In each trial, participants saw first a fixation cross in the centre of the screen presented for 1 s. After this, one of the faces was presented on screen for 3 s. The next trial started with the presentation of a fixation cross again. After the 60 faces had been presented, the programme displayed another set of instructions, explaining the recognition task. Hence, 60 novel faces split into the same stimulus types were added to the 60 faces seen in the study phase, and all 120 images were presented one at a time in random order. No face ever appeared in more than one condition during the experiment for the same participant. In the recognition task, participants were asked to press the ‘.’ key if they recognized the stimulus as having been shown in the study phase on any given trial or press ‘x’ if they did not (the keys were counterbalanced). The faces were shown for 3 s during which time participants had to respond (Figure 1, Panel a).

Figure 1.

Panel a shows a schematic representation of the procedure adopted in the study. The experiment was implemented using SuperLab 4.5 installed on a PC. The dimensions of the stimuli were 6.95 cm × 5.80 cm. Panel b shows the results for the old/new recognition task in Experiment 1. Panel c shows the results for the old/new recognition task in Experiment 2. Error bars represent SEM.

The critical (and only) manipulation in both experiments was applied as part of the study phase instructions by providing category label information about the faces being presented.

In Experiment 1, participants assigned to the regular face condition read the following instruction:

You will see a set of faces presented one at a time

These are faces of REGULAR people

Try to remember as many as you can

Critically, participants assigned to the autistic faces condition read the following instruction:

You will see a set of faces presented one at a time

These are faces of people diagnosed with AUTISM

Try to remember as many as you can

In Experiment 2, participants assigned to the humanizing condition read the following instruction:

You will see a set of faces presented one a time

These are faces of individuals diagnosed with Autism

Some of the traits of people diagnosed with Autism are:

-Trustworthiness. Individuals with Autism don’t have hidden agendas.

-High Integrity. The idea of trying to cheat on the job, or to slack in their work, does not occur to individuals with autism. They are conscientious, diligent workers.

-Little/No prejudice. Because they know what it is like to be different, people with Autism tend to be more accepting of others.

Try to remember as many faces as you can

In contrast, participants assigned to the dehumanizing condition read the following instruction:

You will see a set of faces presented one at a time

These are faces of individuals diagnosed with Autism

Some of the traits of people diagnosed with Autism are:

-Significant problems developing nonverbal communication skills, such as eye-to-eye gazing, facial expressions, and body posture.

-Lack of interest in sharing enjoyment, interests, or achievements with other people.

-Lack of empathy. People with autism may have difficulty understanding another person’s feelings, such as pain or sorrow.

Try to remember as many faces as you can

Data analysis

Our primary measure was performance accuracy. The data from all the participants were used to extract d′ sensitivity in the recognition task where a d′ = 0.00 indicates chance-level performance (Stanislaw and Todorov, 1999). Each p value reported is two-tailed, and we also report the F or t value along with effect size $(η_{p}^{2})$ . We also analysed data from the accuracy scores which confirmed the effects obtained from our d′ sensitivity analysis. Finally, we analysed the RTs data which did not add anything to our interpretation of the results.

Experiment 1

We computed a 2 × 2 mixed-model analysis of variance (ANOVA) using, as a within-subjects factor, Face Orientation (upright or inverted), and the between-subjects factor Face Label (autistic or regular). There was a significant two-way interaction, F(1, 78) = 5.87, p = 0.018 and $η_{p}^{2} = 0.07$ , which reflected the fact that the inversion effect in the autistic faces group was smaller than in the regular faces group. There was a significant main effect of Orientation F(1, 78) = 61.08, p < 0.001 and $η_{p}^{2} = 0.43$ , which simply confirmed that upright faces were better responded to than inverted ones. Follow-up paired t test analyses were conducted to compare performance on upright and inverted face stimuli (the inversion effect) in each Face Label group (regular, autistic). Based on previous studies that used the same stimuli and experimental paradigm (Civile et al., 2014, 2016a; Civile and Obhi, 2016), our primary measure was the face inversion effect given by comparing performance on upright and inverted faces in each Face Label group. A significant inversion effect was found in the regular faces group, t(39) = 7.13, p < 0.001 and $η_{p}^{2} = 0.56$ , and a reduced (but still significant) inversion effect was found in the autistic faces group, t(39) = 3.87, p < 0.001 and $η_{p}^{2} = 0.27$ (see Figure 1, Panel b).

Experiment 2

The same statistical analysis was conducted for Experiment 2. The 2 × 2 ANOVA (Face Orientation × Face Label) revealed a significant interaction, F(1, 78) = 5.26, p = 0.025 and $η_{p}^{2} = 0.06$ , indicating a greater inversion effect for faces associated with positive traits compared to that for faces associated with negative traits. There was a significant main effect of Orientation F(1, 78) = 30.63, p < 0.001 and $η_{p}^{2} = 0.28$ , confirming that upright faces responded better than inverted ones. A significant inversion effect was found in the positive traits group, t(39) = 5.58, p < 0.001 and $η_{p}^{2} = 0.44$ , and a reduced (but still significant) inversion effect was found in the negative traits group, t(39) = 2.27, p = 0.029 and $η_{p}^{2} = 0.11$ (see Figure 1, Panel c).

General discussion

In Experiment 1, we investigated how being exposed to a set of faces labelled as ‘autistic’ can affect the way that university students recognize faces. Critically, the robust inversion effect typical of face recognition was significantly reduced for faces labelled as belonging to individuals with autism compared to (identical) faces labelled as belonging to ‘regular’ non-autistic individuals. This finding suggests that when people think they are presented with a face of someone with autism, they show poorer recognition performance. In Experiment 2, we examined the effects of humanizing versus dehumanizing information on recognition of the same set of autistic faces used in Experiment 1. Importantly, we found a larger inversion effect when humanizing versus dehumanizing information was presented.

Our results are reminiscent of previous studies that adopted the inversion effect paradigm as an index of ‘objectification’ (Bernard et al., 2012, 2015, 2018; Civile et al., 2016b; Civile and Obhi, 2016). This paradigm makes use of the consensus that faces are processed more configurally, whereas objects are processed more featurally (Maurer et al., 2002). In Experiment 1, we suggest that the autistic faces resulted in a smaller inversion effect because they were processed more featurally – in a manner more similar to processing of objects rather than faces.

Objectification is often linked to dehumanization – perceiving others as lacking core human characteristics such as a sense of agency and warmth (Cameron et al., 2016). Hugenberg et al. (2016, Experiment 1 and 2) provided direct evidence of how inversion disrupts activation and categorization of humanness in human faces compared to controls (machine or animal comparisons). Importantly, Fincher and Tetlock (2016) demonstrated that the inversion effect is attenuated when faces are paired with negative social information (e.g. faces of norm violators). Moreover, Bernard et al. (2015, Experiment 3) showed that inversion effect for sexualized images of women increased (i.e. more face-like processing vs object-like processing) after the provision of humanizing information about the women depicted in the images. The results from our Experiment 2 are consistent with these findings and provide some evidence in support of the de-stigmatization effects of pairing autism with humanizing information.

Future research exploring the cognitive bases of stigma could seek to further inform what responses are elicited when such a label (i.e. ‘autistic’) is presented. For instance, will participants feel pity or contempt when presented with the ‘autistic’ label, and how might individual difference factors influence these feelings? Does the provision of humanizing information about autism make participants more compassionate?

In sum, our research demonstrates that the mere labelling of faces as belonging to individuals with autism reduces the magnitude of the inversion effect that normally occurs for face stimuli (Experiment 1). We suggest that stigma-laden labelling induces a processing style akin to the typical processing deployed in object recognition. Crucially, this effect can then be modulated by providing humanizing versus dehumanizing information about autism (Experiment 2). We suggest that changes in the inversion effect may be a useful index of the objectification and dehumanization of individuals diagnosed with autism. Research investigating the cognitive bases of stigma could be instrumental as a first step towards eliminating it altogether. If we can better understand why and how individuals are stigmatized, perhaps, we can devise more efficient and effective solutions for combating stigma (e.g. providing humanizing information).

Individuals with a mental disorder experience two sets of problems – problems associated with the disorder itself and problems associated with the stigma – and this report has made strides towards better understanding the latter.

Supplemental Material

AUT807158__Lay_Abstract – Supplemental material for Labelling faces as ‘Autistic’ reduces the inversion effect

Supplemental material, AUT807158__Lay_Abstract for Labelling faces as ‘Autistic’ reduces the inversion effect by Ciro Civile, Eamon Colvin, Hasan Siddiqui and Sukhvinder S Obhi in Autism

Footnotes

Declaration of conflicting interests

S.S.O. is supported by a discovery grant from the Natural Sciences and Engineering Research council of Canada. C.C. is supported by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie (grant agreement No. 743702) and by the Economic and Social Research Council (ESRC) New Investigator Grant (Ref. ES/R005532).

Funding

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

ORCID iD

Ciro Civile

References

Bernard

Gervais

Allen

et al . (2012) Integrating sexual objectification with object versus person recognition: the sexualized-body-inversion hypothesis. Psychological Science 23: 469–471.

Bernard

Gervais

Allen

et al . (2015) From sex objects to human beings: masking sexual body parts and humanization as moderators to women’s objectification. Psychology of Women Quarterly 39: 432–446.

Bernard

Rizzo

Hoonhorst

et al . (2018) The neural correlates of cognitive objectification: an ERP study on the body inversion effect associated with sexualized bodies. Social Psychological and Personality Science 9(5): 550–559.

Butler

Gillis

(2011) The impact of labels and behaviors on the stigmatization of adults with Asperger’s disorder. Journal of Autism and Developmental Disorders 41: 741–749.

Cameron

Harris

Payne

(2016) The emotional cost of humanity: anticipated exhaustion motivates dehumanization of stigmatized targets. Social Psychological and Personality Science 7: 105–112.

Civile

Obhi

(2016) Power, objectification, and recognition of sexualized women and men. Psychology of Women Quarterly 40: 199–212.

Civile

McLaren

IPL

(2011) Perceptual learning and face recognition: disruption of second order relational information reduces the face inversion effect. In: Carlson

Hoelscher

Shipley

(eds) Proceedings of the 33rd annual conference of the Cognitive Science Society. Austin, TX: Cognitive Science Society, pp.2083–2088.

Civile

McLaren

IPL

(2014) The face inversion effect – parts and wholes: individual features and their configuration. The Quarterly Journal of Experimental Psychology 67: 728–746.

Civile

McLaren

IPL

(2016a) The face inversion effect: roles of first- and second-order configural information. The American Journal of Psychology 129: 23–35.

10.

Civile

Rajagobal

Obhi

(2016b) Power, ethnic origin, and sexual objectification. SAGE Open 6(2): 1–13.

11.

Corrigan

(2000) Mental health stigma as social attribution: implications for research methods and attitude change. Clinical Psychology: Science and Practice 7(1): 48–67.

12.

Fincher

Tetlock

(2016) Perceptual dehumanization of faces is activated by norm violations and facilitates norm enforcement. Journal of Experimental Psychology: General 145: 131–146.

13.

Gardiner

Iarocci

(2014) Students with autism spectrum disorder in the university context: peer acceptance predicts intention to volunteer. Journal of Autism and Developmental Disorders 44: 1008–1017.

14.

Gelbar

Smith

Reichow

(2014) Systematic review of articles describing experience and supports of individuals with autism enrolled in college and university programs. Journal of Autism and Developmental Disorders 44(10): 2593–2601.

15.

Hugenberg

Young

Rydell

et al . (2016) The face of humanity: configural face processing influences ascriptions of humanness. Social Psychological and Personality Science 7: 167–175.

16.

Maurer

LeGrand

Mondloch

(2002) The many faces of configural processing. Trends in Cognitive Sciences 6: 255–260.

17.

Montes

Halterman

(2007) Bullying among children with autism and the influence of comorbidity with ADHD: a population-based study. Ambulatory Pediatrics 7: 253–257.

18.

Stanislaw

Todorov

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

19.

Tipton

Blacher

(2014) Brief report: autism awareness: views from a campus community. Journal of Autism and Developmental Disorders 44: 477–483.

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

0.43 MB