Introduction

Trisomy 21 (T21) or Down syndrome is the most frequent genetic disorder associated with intellectual disability, affecting between 1.0 and 2.2 of every 1000 live births according to statistics on prenatal testing and selective abortion [1]–[4]. Because this chromosomal disorder is also associated with various health problems (e.g., hypotonia, congenital heart defects, gastrointestinal diseases) and distinctive physical stigmata (e.g., round face, epicanthal fold, oblique lid axis, flat nasal bridge), persons with T21 are at high risk of being rejected socially. Not only are the common societal reactions unfavorable in terms of rates of fetal termination and abandonment [2], [5], but those who live with T21 are likely to be stigmatized by other people [6]. This is a critical issue, because stigmatization is one of the greatest obstacles to the successful integration and development of people with intellectual disabilities.

Although research in this area is very limited, it seems that persons with T21, especially children, are typically viewed as “friendly”, “affectionate”, and “happy” (e.g., [7]–[9]); with positive personality traits prevailing over the negative ones (e.g., “mentally retarded”). This positive stereotype, however, coexists with ambivalent attitudes about the integration of these children into regular schools (e.g., [8], [10], [11]). Based on a recent survey, Pace et al showed that whereas 45% of adult respondents (N = 5399) from the general (U.S.) population agreed with inclusive education for students with T21, 25% disagreed (considering for example that such students are “distracting”) and 29% abstained [11]. Likewise, about one-third of adolescent respondents (N = 1704) reported they were not willing to work on a class project or spend time outside of school with a student with T21. Although these negative attitudes are not majority rule, and can even be negligible in people who have relationships with persons with T21 [11]–[13], they may very be the trees that hide the forest.

Of particular interest here, all published studies on the social perception of intellectually disabled people, including those with T21, have measured attitudes and stereotyping solely at the explicit level. This is an important limitation, because social-desirability attitudes may interfere with the measurement of people's responses at this level. We are not suggesting that responses at the explicit level are deliberate lies that people use to cover up their implicit attitudes and convictions, but that it may not be sufficient to focus on consciously-accessible evaluations because people may be unaware of, or unwilling to report, all of their thoughts and feelings. There is ample evidence that attitudes and stereotypes may be used automatically, and thus quickly and effortlessly, without conscious awareness, and yet may influence perception, judgments, or actions even against a person's intentions [14]–[16]. For example, people who display explicit overt beliefs in the equality of the races may implicitly associate positive attributes (e.g., pleasant words) with White more than with Black persons (or more negative attributes with Black than with White persons). These automatic biases can be found even when pictures of equally unfamiliar Black and White people are used as stimuli, and when differences in stimulus familiarity are statistically controlled (e.g., [17]). There is also evidence that implicit evaluations explain variance in behavior over and above that explained by explicit evaluations (e.g., [18], [19]). Thus, by focusing exclusively on explicit (controlled) evaluations, past research on the social perception of intellectually disabled individuals has most certainly fallen short of revealing the whole picture.

Materials and Methods

Results

Explicit Stereotyping

A mixed ANOVA was conducted with trait valence (positive vs. negative) and target of evaluation (TD pictures, T21 pictures weakly typical of T21, T21 pictures strongly typical of T21) as repeated measures, and source of evaluation (students, non-student adults, and professional caregivers) as between-participant factor. Not surprisingly, this analysis yielded a huge effect of trait valence, F(1, 162) = 593.23, p<.001, η²_p = .79. Participants attributed more positive traits (M = 4.63, SE = .06) than negative traits (M = 2.17, SE = .05) to all faces (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Agreement score to the descriptive traits as a function of source of evaluation and target of evaluation.

https://doi.org/10.1371/journal.pone.0034369.t001

This difference favoring positive traits at the explicit level was modulated by both the target and the source of evaluation, F(4, 324) = 8.49, p<.001, η²_p = .09 (See Figure 1). This interaction was examined via two Helmert contrasts for each source of evaluation, with the difference in the trait valence score as the dependent variable. The first contrast compared the TD pictures with the average of the T21 pictures weakly and strongly typical, and thus tested for feature-based stereotyping between categories. The second compared the T21 pictures weakly typical with those strongly typical, and thus tested for feature-based stereotyping within the T21 category.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Explicit judgments.

Mean difference score (positive traits minus negative traits) on explicit judgments as a function of target of evaluation and source of evaluation. Error bars represent ±1 SEM.

https://doi.org/10.1371/journal.pone.0034369.g001

The first contrast revealed a significant difference for each source of evaluation (ps<.05), indicating that the difference favoring positive traits at the explicit level was always smaller for T21 pictures than for TD pictures. The second contrast was significant for every source of evaluation (ps<.05) except one. Both student and non-student adults made slightly, but significantly, less positive judgments for the pictures strongly typical of T21 than for the weakly typical pictures. This subtle bias was not found among caregivers, for whom the T21 pictures received the most positive ratings. The target by source of evaluation interaction for the positive and negative traits taken separately is described in Text S3.

Implicit Stereotyping (IAT)

IAT scores were calculated following the scoring algorithm recommended by Greenwald et al., including the following features: (a) error trials were removed and replaced with the block mean +600 ms, (b) response latencies >10,000 ms were removed, (c) standard deviations were calculated on all correct response trials, (d) participants who had >10% of trials with responses below 300 ms were removed, and (e) participants who had an error rate of >40% in any of the four combined sorting blocks were also removed [32]. Overall 6 participants were excluded (their inclusion in the statistical analyses did not change the results). Likewise, measures of association strength based on the two IATs were computed using the classic D measure [32]. D was the difference between the mean latencies for the “T21 pictures+positive traits” and “TD pictures+negative traits” block on one hand, and the “T21 pictures+negative traits” and “TD pictures+positive traits” block on the other hand, divided by the inclusive standard deviation of the latencies in the two blocks.

The D scores were then analyzed using a mixed ANOVA with the two IATs (based on faces either weakly or strongly typical of T21) as repeated measures, and source of evaluation (students, non-student adults, professionals) as a between-participants factor. This analysis yielded a main effect of source of evaluation, F(2, 162) = 6.00, p = .003, η²_p = .07 (see Figure 2); the target effect, and the source-by-target interaction were not significant (ps>.40). The global IAT effect (two IAT-D scores averaged) was smaller among the caregivers than among students and non-student adults (Tukey contrast, p<.05). As indicated by one-sample t-tests against zero, the global IAT effect was still clearly significant for each source of evaluation: students, t(54) = 10.44; non-student adults, t(54) = 13.08; and caregivers, t(54) = 5.91 (all ps<.001). Correlations between implicit and explicit evaluations were rather weak (see Text S4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Implicit judgments.

Mean IAT scores (±1 SEM) based on pictures weakly and strongly typical of T21, as a function of source of evaluation (students, non-student adults, and professional caregivers).

https://doi.org/10.1371/journal.pone.0034369.g002

Discussion

Research into the social perception of intellectually disabled persons has been largely one-sided, focusing predominantly on thoughts and feelings, which may be distorted by social desirability concerns. In the current research, we found evidence of subtle forms of prejudice toward children with special needs at both the explicit and implicit levels, in this case children with T21.

In line with previous research, students, non-student adults, and professional caregivers all attributed more positive than negative personality traits to children with T21, as they also did for TD children. However, in accordance with standard models of stereotyping operating between categories [20]–[23], this difference in favor of positive traits was lower for children with T21 than for TD children. Perhaps more importantly, as expected from research on feature-based stereotyping operating within categories [25], [26], the difference favoring positive traits proved smaller for pictures strongly typical of T21 than for weakly typical ones, at least among students and non-student adults. This is the first evidence that facial features distinctive of a genetic disorder, in this case T21, can lead to stereotyping in two ways, between and within categories. Feature-based stereotyping within the category of T21 pictures did not operate in the professional caregivers, who were also the most positive participants toward those pictures. This is consistent with the idea that relationships with intellectually disabled persons promote positive attitudes toward them (e.g., [11], [29]).

The IAT findings indicated that, in each group of participants, the T21 pictures were automatically associated with a negative valence. Participants, professional caregivers included, were indeed faster at categorizing T21 pictures with negative traits than T21 pictures with positive traits (and faster at categorizing TD pictures with positive traits than TD pictures with negative traits). The IAT effect also occurred regardless of whether the children's faces were strongly or weakly distinctive of T21. Taken together, these findings can be regarded as evidence that negative attitudes about, and stereotyping of, children with T21 in fact prevail at the implicit level.

Finally, the fact that the overall IAT effect was clearly significant in the three groups of participants suggests that professional caregivers' implicit mind-set about children with T21 may not differ from that of non-professionals. One important point must be made here, however. Although each group of participants exhibited a large IAT effect, this effect was significantly smaller in the caregivers than in the students and non-student adults (with equal statistical power). This suggests that caregivers' sustained contact with individuals with T21 may operate somewhat at the implicit level. Consistent with this idea, the caregivers' overall IAT effect decreased significantly with years of professional experience (which ranged from 1 to 34 years), ß = −.32, t(51) = −2.36, p<.03. This additional finding strengthens the conclusion of a few recent studies indicating a beneficial effect of intergroup contact on implicit evaluations (e.g., [33]). It offers a new reason to believe that even implicit stereotypes (not just explicit ones) can be reduced under the influence of repeated contact with the stigmatized. This critical issue is attracting more and more attention in the struggle against intergroup prejudice (e.g.; [34], [35]). Henry et al. showed that intergroup contact is more likely to have a beneficial impact at the implicit level in low-status groups (e.g., Blacks toward Whites but not Whites toward Blacks) [34]. Our own findings suggest that this impact can also be found in high-status groups, at least when rivalry with the low-status group is not an issue (here non-disabled people with intellectually disabled people).

Thus, the present research reveals that facial features associated with a genetic disorder such as T21 can lead to between- and within-category stereotyping at the explicit level. Although positive evaluations are especially likely at this level, they were reduced when the individuals being rated had facial features highly distinctive of this chromosomal disorder. Our research also reveals that the positive evaluations of children with T21 that can be found at the explicit level can coexist with negative evaluations at the implicit level, which helps explain why the general population is so ambivalent about practical issues such as inclusive education for these children [11], [36]. Taken together, the present findings show how important it is to pay attention to feature-based stereotyping at the explicit level. They also suggest going beyond explicit evaluations when attempting to estimate the true extent of stigmatization of intellectually disabled people. Explaining why implicit stereotyping did not operate within the T21 category is beyond the scope of this paper. However, we suggest that within-category stereotyping cannot be ruled out at this level. As noted earlier, participants in Blair et al research seemed unaware of, and unable to control, stereotyping based on Afrocentric features [26]. Thus, future research is needed to clarify whether feature-based stereotyping can occur at both the explicit and implicit levels.

A limitation of the present study (as well as many other IAT studies) is that the negative associations found at the implicit level do not necessarily mean that children with T21 are the object of implicit prejudice. As suggested earlier in this paper, however, our IAT findings are consistent with the ambivalence of the general population toward inclusive education for students with T21 [11]. Likewise, there is evidence that implicit attitudes can unintentionally lead to discriminatory behaviors such as reducing one's interaction time with the stigmatized [37], [38]. There is also evidence of discriminatory behaviors based on facial features [25]. Additional research, therefore, is also needed to determine whether the implicit attitudes and/or the visibility of the physical stigmata associated with T21 or other disorders influence the way the stigmatized are treated during social interaction. Whether those factors can raise the awareness of the disorder in the stigmatized themselves also merits special attention. People with T21 who have mental ages of about 8 years or more engage in social comparisons with other people, and start to form complex social categories of T21/disability [39]–[41]. Not only may these individuals be aware of their differences with other people, but their stigmatized identity may actually be chronically accessible, especially when their physical stigmata cannot be easily hidden. Like many other people living with a visible stigmatized identity, persons with salient phenotypic T21 features may also suffer from specific threats, for example, the threat of being judged and treated stereotypically on the basis of their facial features (at least those with mental ages above 7–8 years). This threat may contribute to lowering their performance in test situations, in the same way children and adults may suffer from stereotype threat in test situations due to their gender (e.g., [42]–[45]), age (e.g., [46]), social or academic background (e.g. [47]–[49]), or ethnic identities (e.g., [50]). Future research on these issues could help increase our understanding of stigmatization and its related consequences among people with intellectual disabilities.

Finally, one may wonder whether the negative associations found here at the implicit level are specific about T21 or could as well be explained by general implicit attitudes towards disabilities. The very few IAT studies in this area indeed indicate that people tend to associate words or pictures related to various physical disabilities [51]–[54] and mental illness [53], [55], with negatively connoted words. So far, however, we just did not know whether negative implicit associations could also be found towards people with intellectual disabilities. Here, we offer first evidence that persons (children) with T21, the most representative exemplar of the intellectually disabled, are the target of negative associations at the implicit level despite the positive stereotype that characterize them at the explicit level. As noted by Menolascino, the persons with T21 have long been regarded as “the Prince charming of the mentally retarded population” [56]. This positive view about people with T21 at the explicit level made the capture of negative associations at the implicit level especially informative. In addition, contrary to other exemplars of the intellectually disabled with known aetiology (e.g., Williams syndrome, Fragile X), persons with T21 are the only ones who can easily and automatically be identified as intellectually disabled by the general population, because they share well-known physical stigmata. This additional characteristic offered the possibility (also neglected to date) to test whether intellectually disabled people are subject to stereotyping based on their facial features (i.e., featured-based stereotyping operating within the category of the intellectually disabled). Future research based on direct comparisons between physical and intellectual disabilities might help clarify whether featured-based stereotyping can also be found for people with physical disabilities (using several degrees of physical disability), and whether the type of disability (physical vs. intellectual) makes any difference on the size of explicit and implicit attitudes.

Supporting Information

Acknowledgments

We are thankful to all participants and especially to families and children who made this work possible. We are indebted to the clinician colleague Catherine Ayoun and heads of the institutions who referred families and/or professional participants.