Minio-Paluello et al. Molecular Autism (2020) 11:81 https://doi.org/10.1186/s13229-020-00371-0 RESEARCH Open Access Face individual identity recognition: a potential endophenotype in autism Ilaria Minio-Paluello1,2,3* , Giuseppina Porciello1,2, Alvaro Pascual-Leone4,5,6 and Simon Baron-Cohen7 Abstract Background: Face individual identity recognition skill is heritable and independent of intellectual ability. Difficulties in face individual identity recognition are present in autistic individuals and their family members and are possibly linked to oxytocin polymorphisms in families with an autistic child. While it is reported that developmental prosopagnosia (i.e., impaired face identity recognition) occurs in 2–3% of the general population, no prosopagnosia prevalence estimate is available for autism. Furthermore, an autism within-group approach has not been reported towards characterizing impaired face memory and to investigate its possible links to social and communication difficulties. Methods: The present study estimated the prevalence of prosopagnosia in 80 autistic adults with no intellectual disability, investigated its cognitive characteristics and links to autism symptoms’ severity, personality traits, and mental state understanding from the eye region by using standardized tests and questionnaires. Results: More than one third of autistic participants showed prosopagnosia. Their face memory skill was not associated with their symptom’s severity, empathy, alexithymia, or general intelligence. Face identity recognition was instead linked to mental state recognition from the eye region only in autistic individuals who had prosopagnosia, and this relationship did not depend on participants’ basic face perception skills. Importantly, we found that autistic participants were not aware of their face memory skills. Limitations: We did not test an epidemiological sample, and additional work is necessary to establish whether these results generalize to the entire autism spectrum. Conclusions: Impaired face individual identity recognition meets the criteria to be a potential endophenotype in autism. In the future, testing for face memory could be used to stratify autistic individuals into genetically meaningful subgroups and be translatable to autism animal models. Keywords: Autism, Individual identity recognition, Face memory, Prosopagnosia, Endophenotype, Heterogeneity, Social memory, Theory of mind, Emotion recognition Background Autism occurs in at least 1% of the population [2] and is Autism is a highly hereditable, lifelong, neurodevelopmen- associated with high levels of poor mental health which tal condition characterized by difficulties in social commu- could be reduced by a better and earlier intervention [3]. nication and interaction, alongside unusually restricted Clinical, etiological, and genetic heterogeneity in autism and repetitive behavior and interests, sensory hypersensi- poses challenges to the discovery of causes and the devel- tivity, and difficulties adjusting to unexpected change [1]. opment of effective interventions for autism [4–6]. Heterogeneity could be addressed via an endophenotype- based stratification approach which would accelerate the * Correspondence: ilaria.miniopaluello@istc.cnr.it 1 Department of Psychology, Sapienza University of Rome, Rome, Italy identification of genetic underpinnings and specific inter- 2 IRCCS Fondazione Santa Lucia, Rome, Italy ventions [7, 8]. An endophenotype approach in autism is Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 2 of 16 complicated by the fact that phenotypic expression in aut- focus of few well-controlled studies. In fact, studies that ism changes developmentally over time, depends on attempted to investigated face memory/IIR in autism often symptom severity, and varies in the presence of intellec- used tasks that suffered from (1) familiarity confounds tual disability (ID) which commonly co-occurs with aut- (i.e., the task could be solved by recognizing whether an ism [9, 10]. Researchers have proposed interesting individual is seen before or novel) [52–55], (2) no (or very behavioral and neural endophenotypes in autism [8, 11], short) retention time of the memory trace related to facial including language delay [12], gazing at social scenes [13], identity [56–58], (3) face stimuli that include non-facial scores on the Social Responsiveness Scale [14], and white features (e.g., hair, clothes) [56, 59, 60] which allow correct matter structure [15]. Many other proposed endopheno- performance even when facial internal features are cov- types in autism do not meet the definition of endopheno- ered [61], and (4) matching tasks with identical target and type they adopt [16] as, for example, they lack evidence of test face images, that therefore could be solved based on heritability [17–20]. Weaknesses of many endophenotypes feature matching strategies [54, 62]. proposed so far include the unknown prevalence in autism Overall, recent reviews of the available evidence have or relevance to a small subgroup of autistic individuals shown that, whenever memory is involved, autistic indi- [15, 21, 22], reduced proximity to gene action (i.e., being viduals as a group have reduced face identity recognition under genetic influences of unknown or small effect sizes skills compared to matched neurotypical controls [51, 63] that are comparable to those of autism itself) [23], un- and that their face memory difficulties seem both process known or high genetic complexity [14, 24], lack of known (e.g., memory vs. perception) and social-domain (e.g., faces or envisaged neurobiological bases [13], and limited trans- and bodies vs. houses) specific [64]. To the best of our latability to animal models [12, 14, 23]. Such weaknesses knowledge, no study has reported the prevalence of clin- are not specific to autism endophenotype literature but ical impairment in face memory (i.e., prosopagnosia) in apply to endophenotypes of psychiatric conditions in gen- autism. In fact, those studies that used the CFMT, in eral [25–27]. Individual identity recognition (IIR) could be many cases did not report group means and standard de- a potential new autism endophenotype devoid of most of viations but just case-control statistics [65–69]. When the weaknesses described above. they did report group means, however, they did not report The ability to recognize another individual is crucial the percentage of participants meeting the CFMT proso- for social interaction [28], emerges very early in develop- pagnosia cut-off [70–74]. An exception is the study by ment [29, 30], is conserved across species [28], and is Hedley and colleagues [75] who reported that 8 out of linked to the oxytocin (OXY) system [31, 32]. Humans, their 34 AUT adults (i.e., 24%) were prosopagnosic ac- like other primates, recognize other individuals mostly cording to the CFMT. by their face [33–35], which has evolved to signal indi- No study has tested face memory in a larger group of vidual identity [36]. Identity recognition is challenged by autistic adults, that is, at an age when difficulties in face a face’s intrinsic (e.g., age, facial expressions/movements) memory, if present, are most evident [70]. Further, no and extrinsic (e.g., visual perspective, luminosity) previous study has considered face memory difficulties identity-invariant changes and might benefit from plastic in terms of autism within-group variability rather than face representations [37–39]. Humans show large indi- mean case-control differences, addressing the heterogen- vidual differences in their face IIR ability [40], including eity in performance found in autism. 2–3% of individuals in the general population who re- Face IIR has the potential to be an autism endopheno- port severe difficulties recognizing identity from faces in type because it is highly heritable [76–78], independent everyday life [41, 42]. This face-blindness condition, of intellectual ability [78, 79], and difficulties in IIR are known as developmental prosopagnosia (DP), is not as- more common in family members of autistic individ- sociated with brain damage or deficits in low-level vi- uals [59, 80, 81]. Future autism research could benefit sion, can run in families, is likely polygenic [43], and the from the fact that IIR is measurable early in develop- OXY system seems to play a role in it [44, 45]. Although ment [30, 82] and in individuals with ID [83], may be there are no formal diagnostic criteria for DP, it is gen- linked to OXY polymorphisms in families with an erally agreed diagnostic assessment should primarily in- autistic child [60] (for different findings in neurotypi- volve objective measures of face IIR. The Cambridge cal participants, see [84]), and is translatable to mice Face Memory Test (CFMT) [46] is considered the gold [85, 86]. standard memory test for unfamiliar faces [47–49] with Here, we estimated the prevalence of prosopagnosia a clinical cut-off score for prosopagnosia (see the in autism, characterized it with respect to related face “Methods” section). perceptual processes, and investigated its links to aut- While face perception in general, ranging from face ism symptom severity, personality traits, and difficul- detection to emotion recognition, has been extensively ties in mental state understanding from the eye studied in autism [50, 51], face memory/IIR has been the region. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 3 of 16 Methods online). All participants completed the Cambridge Face Participants Memory Test (CFMT) [46] upright, and almost all autis- We tested 80 autistic (AUT) adults (16 females, mean tic participants completed the other tests and question- age = 31.3 ± 11.9 years, age range = 18–73 years) with naires (sample size is reported for each test). no ID (IQ- percentile = 78 ± 29, range = 4–100, N = 74 of which Raven’s SPM mean percentile = 91 ± 15, range Tests = 38–100, N = 31; Wechsler scales mean full-scale IQ The Cambridge Face Memory Test (CFMT) [46] upright and percentile = 69 ± 32, range = 4–100, N = 43), and 80 inverted neurotypical controls (NT) (16 females, mean age = 28.8 The CFMT is a computer-based test that uses a three- ± 9.1 years, age range = 18–56 years) with no ID (IQ alternative forced-choice paradigm with the unlimited re- percentile = 92 ± 13, range = 21–100, N = 69, of which sponse time. Participants have to memorize and recognize Raven’s SPM mean percentile = 93 ± 14, range = 21– 6 individuals. The test is divided into 3 parts of increasing 100, N = 47; Wechsler scales mean full-scale IQ percent- difficulty. In the first part (18 trials), participants view each ile = 89 ± 12, range = 50–100, N = 22) matched for age individual from three different angles and have to (t test with separate variance estimates, t(147.7) = 1.47, p recognize him in three consecutive trials (correct images = 0.14, 95% CI [− 5.75, 0.85]; Levene’s test F(1,158) = 4.43, are identical to the studied ones). In the second part (30 p = 0.04), sex, and country of residency (UK, USA, Italy). trials), the to-be-recognized individual can be any of the 6 Although we did not have IQ measures for 6 AUT and 11 previously learned ones, now presented with different an- NT participants, given that they all completed high school gles and/or lighting. The third part (24 trials) has the same and in some cases were enrolled in college-level educa- structure as the second one but Gaussian visual noise is tion, we could assume they too did not have ID. added to the images in order to make participants rely Autism diagnosis was made by a professional expert in more on holistic (vs. feature-based) face processing [96]. A autism according to DSM-IV criteria and confirmed via number of correct answers at or below 42 (out of 72), the Autism Diagnostic Observation Schedule (ADOS) which corresponds to 2 standard deviations from the [87] and the Autism Diagnostic Interview Revised (ADI- mean, are indicative of prosopagnosia, while chance level R) [88] by a certified clinician (Table 1). Intellectual abil- corresponds to 24 correct responses [46, 47]. The CFMT ity was assessed via either the Wechsler Abbreviated has well-controlled stimuli (e.g., including only facial fea- Scale of Intelligence [89], the Wechsler Adult Scale of tures, the same individual is presented with identity invari- Intelligence IV [90], or the Raven’s Standard Progressive ant changes), is unidimensional, highly reliable, has high Matrices [91] (Table 1). NT participants did not have a discriminant and convergent validity, is precise over a neurological or psychiatric condition, autistic first-degree wide range of ability levels [47, 97], and can be reliably ad- relatives, or an autism spectrum quotient questionnaire ministered online [98, 99]. Performance on the CFMT is score above the autism cut-off [92]. Participants were re- heritable [100] and has little or no correlation with general cruited because they took part in other experimental stud- intelligence [78, 79]. The AUT group also took the CFMT ies [93–95]; therefore, they were not preselected based on with face stimuli presented upside down [47, 101]. AUT their interest or skills in identity recognition. The study participants took the upright and the inverted CFMT tests was approved by local Institutional Review Boards, ad- in counterbalanced order, one at the beginning and one at hered to the declaration of Helsinki, and all participants the end of the experiment, to reduce carry-over effects. gave informed consent before participation. Better performance for upright vs. inverted faces is known as the face inversion effect, which is interpreted in favor of Procedure typical holistic (vs. feature-based) processing of faces Participants performed computer-based versions of a [102]. In order to account for age-dependent decline in standardized battery of tests and questionnaires either performance, we used Bowles and colleagues [47] second- onsite or online (6% of NT and 14% of AUT participants order polynomial fit in conjunction with the standard de- completed between one and three questionnaires/tests viations of the residuals of that fit to calculate age- Table 1 Characteristics of the autistic sample ADOS CSS ADOS ADOS ADI Comm ADI ADI ADI IQ percentile AQ SA + RRB Comm + SocInt Soc Int RRB Abn Dev Cut-off 8 7 8 10 3 3 31 AUT 7±3 13 ± 5 10 ± 4 16 ± 5 19 ± 6 6±3 2±1 78 ± 29 31 ± 9 N = 67 N = 67 N = 69 N = 55 N = 55 N = 55 N = 55 N = 74 N = 78 ADOS Autism Diagnostic Observation Schedule, CSS Calibrated Severity Score, SA Social Affective, RRB Restricted Repetitive Behavior, Comm Communication, Soc Int Social Interaction, ADI Autism Diagnostic Interview, Abn Dev abnormal development, IQ intelligence quotient, AQ autism quotient Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 4 of 16 standardized prosopagnosia cut-off scores for participants online study of the AQ confirmed the sex difference (typ- over 49 years of age. ical males score higher on average than typical females) and the STEM effect (those working in Science, Technol- The Cambridge Face Perception Test (CFPT) [103] ogy, Engineering, and Math score higher on average than The CFPT is a computer-based test where each of its 16 those who do not) in half a million people [118]. A recent, trials involves limited time (i.e., 40 s in the version used even larger online study of over 600,000 people from the here) to sort 6 front-facing faces according to their simi- general population and 36,000 autistic people, confirmed larity to a target face presented with a ¾ profile. Each of the case-control difference, the sex difference, and the the 6 faces was created by morphing a different individ- STEM effect [119], using a short form of the AQ. ual with the target face by varying degrees (28%, 40%, 52%, 64%, 76%, and 88%). In half of the trials, faces are presented upright, while in the other half of the trials, The twenty-item Prosopagnosia Index (PI20) [120] faces are presented inverted. For each trial, the order in The PI20 is a 20-items self-report questionnaire proposed which the participant arranged the morphed faces is by its authors as a measure of prosopagnosic traits [121]. scored by summing the deviations of each morphed face Respondents indicate on a five-point Likert scale how from its correct position. The CFPT error score corre- much they agree with statements describing their face sponds to the sum of scores of upright trials. The higher identity recognition abilities and experience in everyday the error score, the worse the participants’ performance. life. Total score ranges from 20 to 100 and a score over 64 Perfect performance corresponds to an error score of 0, is considered indicative of prosopagnosia [120]. The PI20 while chance performance corresponds to an error score score correlates with performance on the CFMT in the of 93.3. The CFPT is considered a measure of face per- general population and in DP [120–122] and distinguishes ception skills with no memory demand, as the target and DPs from controls [120, 123, 124]. The need for DP diag- the morphed faces are visible during the task and, there- nostic assessment to include a self-report questionnaire fore, do not need to be memorized. Not all individuals assessing awareness of everyday difficulties in face mem- with DP perform worse than controls on the CFPT [47, ory is debated [48, 49, 125]. 104, 105], just those with apperceptive prosopagnosia. The Interpersonal Reactivity Index (IRI) [126] The Reading the Mind in the Eyes Test (RMET) [106] The IRI is a 28-item self-report questionnaire with four The RMET is a test where participants have to recognize subscales each said to be measuring an independent em- mental states (including complex emotions) from 36 pho- pathy component. Subscales include perspective taking, tographs of the eye region of individuals varying in sex which measures the ability to adopt another person’s and age. The RMET has good reliability [107], autistic in- view point; empathic concern, which measures the ten- dividuals consistently perform less well than matched neu- dency to respond with warm, compassionate feelings for rotypical controls [108–112] and do not show the others; fantasy, which measures the tendency to identify neurotypical advantage of female vs. male partici- with fictional characters; and personal distress, which pants [108]. A large online study of the RMET in over measures a self-oriented negative arousal/discomfort re- 80.000 individuals confirmed the neurotypical sex differ- sponse to another person’s distress/negative experience. ence (female advantage) and identified a single nucleotide Participants indicate on a five-point Likert scale how polymorphism associated with performance [113]. Individ- much they agree with each statement. uals with DP perform similarly to controls on the RMET [103, 114–116]. The Toronto Alexithymia Scale (TAS) [127] Questionnaires The TAS is a 20-item self-report questionnaire with The Autism Spectrum Quotient (AQ) [92] three subscales each tapping a component of alexithy- The AQ is a 50-item self-report questionnaire measuring mia: difficulty identifying feelings, difficulty describing the number of autistic traits across five domains: commu- feelings, and externally oriented thinking. Total scores nication, social skills, attention switching, imagination, range between 20 and 100, with higher scores indicating and attention to detail. The respondent rates how strongly more alexithymic traits. Sixty-one is the cut-off score for they agree or disagree with each statement, using a four- high alexithymia [128]. The TAS has good internal point scale. Total score ranges from 0 to 50 and scores consistency and good test-retest reliability [127]. Alex- above 31 are indicative of autism. Scores between 23 and ithymia seems highly prevalent in people on the autistic 28 are considered Broad Autism Phenotype (BAP), be- spectrum compared to the general population, and it tween 29 and 34 Medium Autism Phenotype (MAP), and seems to play a role in autism emotion recognition skills above 34 Narrow Autism Phenotype (NAP) [117]. A large and empathic response [129]. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 5 of 16 Statistical analysis expected, AUT (31.01 ± 9.02, N = 78) had higher AQ Before running each statistical test, we checked whether scores than NT (14.86 ± 6.01, N = 78) participants, t test its assumptions were met. Before running Student’s t with separate variance estimate t (134) = 13.17, p = tests, we checked for homogeneity of variances via 6.20e-26, 95% CI [13.73, 18.58]; Levene’s test F(1,154) = Levene’s test and when significant we considered separ- 9.21, p = 0.003. ate variances estimates. Before running moderation ana- lysis, we checked for multicollinearity (via variance Prosopagnosia is more common in autism than in inflation factor and tolerance), independence of residuals controls (via Durbin–Watson statistic), linearity and homoscedas- We found that prosopagnosia was more common in aut- ticity (via visual inspection of the standardized residual ism: 36% of AUT met the clinical cut-off for prosopagno- vs. standardized values scatter plot), and homogeneity of sia, while this was the case for only 6% of NT. Autism variances (via Levene’s test). We considered a dependent diagnosis was significantly associated with prosopagnosia variable to be normally distributed if Shapiro-Wilks test χ2(1) = 21.51, p < 0.001, 95% CI [0.18, 0.42], that is, based was not significant (p > 0.05) or if its absolute SME- on the odds ratio, the odds of being prosopagnosic were standardized Z-Skewness and Z-Kurtosis were consid- 8.5 times higher for AUT individuals than for NT (Fig. 1). ered normal (i.e., Z-Skewness and Z-Kurtosis < 1.96 at p Please see Table 2 for groups’ performance scores on the < 0.05) [130]. In case of normal distribution, we reported CFMT and note from Fig. 1 that all groups showed some parametric tests; otherwise, non-parametric tests were degree of variation in the number of correct responses on used. Specifically, when data distribution was not normal the CFMT. As evident in every figure, although one par- and/or sample sizes were different, we used robust sta- ticipant scored below the CFMT cut-off (i.e., 42), they tistics [131]. For each statistical test, we reported rele- vant statistical indices including the distribution’s parameters and their degrees of freedom (df), the sample size used (N), mean or median, standard deviation (SD), probability (p), confidence intervals (CI), and effect size. We did not randomize the recruitment selection of aut- istic (nor matched NT) participants as recruitment of autistic participants was not easy, and our main goal was to have the largest possible sample size. We did not find outliers in correlation and regression analyses as no participant met at least two between Mahalanobis distance, Cook’s distance and Leverage cut- offs [130]. Statistical analysis was run via SPSS (IBM) and PROCESS plugin [132], STATISTICA (StatSoft, Inc. 2007), and R (R Development Core Team 2013, packages WRS2, RVAideMemoire, ggplot2). Results Intellectual ability and autistic traits in autistic and neurotypical participants Comparison of available autistic (74 out of 80) and neuro- typical (69 out of 80) IQ percentile scores via robust Yuen’s test (WRS2, R-package) with default trimmed value of 0.2 provided Ty(54.53) = 2.42, p = 0.02, trimmed mean difference = 7.74, 95% CI [1.33, 14.15], and explanatory measure of effect size = 0.45, 95% CI [0.18, 0.71]. There- Fig. 1 The prevalence of prosopagnosia in autism. Performance fore, based on the available data, NT (92 ± 13) had higher (number of correct responses) at the Cambridge Face Memory Test (CFMT) for 160 participants plotted as a function of their diagnostic IQ percentiles than AUT (78 ± 29) participants. group (autistic-AUT vs. neurotypical-NT) and prosopagnosia Results of the AQ showed that 41% of AUT partici- (prosopagnosic-P vs. non-prosopagnosic-NP). Thicker horizontal lines pants scored within the narrow autism phenotype range represent the medians, boxes the interquartile ranges, and whiskers (vs. 0% of NT), 27% within the medium autism pheno- the maximum and minimum values. The solid horizontal black line is type range (vs. 1% of NT), 13% within the broad autism the CFMT clinical cut-off for prosopagnosia (i.e., 42). AUT-P: dark blue, AUT-NP: light blue, NT-P: dark red, NT-NP: light red boxes. Black phenotype range (vs. 8% of NT) and 19% outside the dots represent individual data points broad autism phenotype (vs. 91% of NT) [117]. As Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 6 of 16 Table 2 Group performance at the Cambridge Face Memory Test (CFMT) of prosopagnosic (P) and non-prosopagnosic (NP) autistic (AUT) and neurotypical (NT) participants CFMT (prosopagnosia cut-off = 42) N mean number of correct responses (over a total of 72) SD Range NT (NT-NP + NT-P) 80 56.20 9.76 25–72 95% CI [54.03, 58.37] 95% CI [8.45, 11.56] NT-NP 75 57.45 8.57 43–72 95% CI [55.48, 59.42] 95% CI [7.38, 10.21] NT-P 5 37.40 7.23 25–42 95% CI [28.42, 46.38] 95% CI [4.33, 20.78] AUT (AUT-NP + AUT-P) 80 46.85 11.20 18–72 95% CI [44.36, 49.34] 95% CI [9.69, 13.26] AUT-NP* 51 53.64 7.55 43–72 95% CI [51.49, 55.79] 95% CI [6.31, 9.41] AUT-P 29 35.52 5.87 18–42 95% CI [33.28, 37.75] 95% CI [4.66, 7.94] NT neurotypical participants, AUT autistic participants, P prosopagnosic, NP non-prosopagnosic, CFMT Cambridge Face Memory Test, SD standard deviation, CI confidence interval *AUT-NP mean, SD, and range do not include a participant who scored 36 and was assigned to the AUT-NP group as they did not meet their age-standardized prosopagnosia cut-off Table 3 Comparisons between autistic participants with (AUT-P) and without (AUT-NP) prosopagnosia Group N Mean SD SE t (df)/Ty(df) p 95% CI Cohen’s d/Yuen’s effect size ADOS CSS AUT-NP 43 6.40 2.55 0.39 AUT-P 24 7.08 2.60 0.53 − 1.05(65) 0.30 − 1.99,0.62 − 0.27 ADI TOT AUT-NP 38 43.16 11.28 1.83 AUT-P 17 45.77 12.95 3.14 − 0.76 (53) 0.45 − 9.52, 4.31 − 0.22 AQ AUT-NP 51 30.86 9.53 1.33 AUT-P 27 31.30 8.12 1.56 0.13 (39) 0.89 − 3.54, 4.04 0.07 IQ AUT-NP 49 80.56 26.86 3.84 AUT-P 25 73.68 31.73 6.35 0.87 (17.94) 0.35 − 8.52, 23.09 0.21 RMET AUT-NP 50 0.65 0.16 0.02 AUT-P 26 0.60 0.19 0.04 0.95 (22.3) 0.35 − 0.07, 0.17 0.18 PI20 AUT-NP 42 55.00 15.51 2.39 AUT-P 21 56.57 16.07 3.51 0.40 0.69 − 12.11, 819 0.1 (23.59) PT AUT-NP 50 13.22 6.07 0.86 AUT-P 26 12.46 5.57 1.09 0.69 0.49 − 1.84, 3.79 0.17 (74) EC AUT-NP 50 16.08 5.30 0.75 AUT-P 26 16.27 6.27 1.23 − 0.004 (74) 1.00 − 2.70, 2.69 − 9.37e−4 TAS AUT-NP 44 50.32 10.60 1.60 AUT-P 23 59.09 10.99 2.29 1.85 0.07 − 0.41, 10.57 − 0.44 (65) Cohen’s d effect size is interpreted as 0.2 (small), 0.5 (medium), 0.8 (large); (*) Yuen’s effect size is interpreted as 0.10 (small), 0.30 (medium), and 0.50 (large) t = Student t, (*) Ty = Yuen’s T, SD standard deviation, SE standard error, df degrees of freedom, CI confidence interval, AUT-P autistic prosopagnosic, AUT-NP autistic non-prosopagnosic, ADOS Autism Diagnostic Observation Schedule, ADI TOT Autism Diagnostic Interview Total score (i.e., Communication + Social Interaction + Restricted Repetitive Behaviour + Developmental Abnormalities), AQ* autism quotient, IQ* Intelligence Quotient, RMET* Reading the Mind in the Eyes Test, PI20 Twenty-item Prosopagnosia Index, PT Perspective Taking, EC Empathic Concern, TAS Toronto Alexithymia Scale Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 7 of 16 were assigned to the AUT-NP group as they did not meet their age-standardized prosopagnosia cut-off. Clinical and personality trait measures do not distinguish between autistic individuals with and without prosopagnosia Autistic prosopagnosic (AUT-P) and autistic non- prosopagnosic (AUT-NP) individuals did not differ in their diagnostic symptom severity, assessed via the ADOS Calibrated Severity Score (CSS) [133] and the ADI-R total score (i.e., Communication, Social Inter- action, Restricted Repetitive Behaviors, and Develop- mental Abnormalities sum of scores), number of autistic traits assessed via the AQ questionnaire [92], level of general intelligence assessed via Raven’s progressive matrices or Wechsler adult intelligence scales, mental state recognition from the eye region assessed via the RMET [106], perspective taking (PT) and empathic con- cern (EC) assessed via the IRI [126], alexithymia assessed Fig. 2 Prosopagnosia moderates the influence of identity recognition on mental state understanding. In autistic participants via the TAS [127], and self-report prosopagnosia who are prosopagnosic (AUT-P, dark blue circles), face memory skill assessed via the PI20 questionnaire [120] (see Table 3). at the CFMT predicted their ability to understand another person’s mental states at the RMET, while this was not the case for autistic participants who are not prosopagnosic (AUT-NP, light blue circles). Face individual identity recognition is linked to mental The dark blue solid line represents a significant regression line for state recognition only in autistic individuals with the AUT-P group, while the light blue line represents nonsignificant prosopagnosia regression line for the AUT-NP group. The black solid line represents We completed a moderation analysis to investigate the CFMT cut-off score (i.e., 42) role of face IIR over mental state recognition in the two groups of prosopagnosic and non-prosopagnosic AUT participants. Moderation analysis, R2 = 0.16, F (3, 72) = 4.48, p = 0.006, revealed that the interaction between identity recognition, assessed via the CFMT, and whether AUT participants were prosopagnosic or non Table 4 Group performance at the CFPT N Error score SD Range CFPT upright AUT 62 51.97 25.02 18–116 95% CI [45.61, 58.32] 95% CI [21.26, 30.41] AUT-NP 41 44.44 22.97 20–96 95% CI [37.19, 51.69] 95% CI [18.86, 29.39] AUT-P 21 66.67 22.61 18–116 95% CI [56.37, 76.96] 95% CI [17.30, 32.65] CFPT inverted AUT 62 73.94 15.61 36–102 95% CI [69.97, 77.90] 95% CI [13.27, 18.98] AUT-NP 41 71.61 16.48 36–100 95% CI [66.41, 76.81] 95% CI [13.53, 21.08] AUT-P 21 78.48 12.96 50–102 95% CI [72.58, 84.38] 95% CI [9.82, 18.72] CFPT inversion effect (Inv-Up) AUT-NP 41 27.17 16.77 − 10–58 95% CI [21.88, 32.47] AUT-P 21 11.81 19.37 − 32–44 95% CI [2.99, 20.62] Controls from Bowles et al. 2009 118 26.43 14.41 CFPT Cambridge Face Perception Test, AUT autistic participants, P prosopagnosic, NP nonprosopagnosic, SD standard deviation, CI confidence interval Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 8 of 16 interaction between identity recognition and prosopag- nosia over participants’ ability to recognize another per- son’s mental state from the eye region. The CFPT was not normally distributed (significant Shapiro-Wilks test) due to a positive skew, which was resolved by square root transformation. None of the main effects nor inter- actions were significant. In particular, the three-way interaction between identity recognition × face percep- tion × prosopagnosia was not significant, b = − 0.003, t (54) = − 0.41, p = 0.68, 95% CI [− 0.02, 0.01], N = 62. Autistic individuals with prosopagnosia do not show face memory inversion effect and have no general memory difficulties To further investigate differences in face processing and memory skills, we compared AUT-P and AUT-NP per- formance on the CFMT with upright and inverted faces (Fig. 3). Type III mixed ANOVA revealed significant main effects of prosopagnosia, F(1,62) = 54.30, p = 4.94e-10, ηp2 = 0.47, and face orientation, F(1,62) = 98.82, p = 1.88e-14, ηp2 = 0.61, which were explained by a sig- nificant prosopagnosia × face orientation interaction, F(1,62) = 41.01, p = 2.29e-08; ηp2 = 0.40. Tukey HSD post hoc tests evidenced that while AUT-NP (N = 42) Fig. 3 Autistic prosopagnosics do not show face inversion effect showed a face inversion effect, AUT-NP upright = 53.14 and have no general memory impairment. Non-prosopagnosic ± 7.90, AUT-NP inverted = 35.53 ± 5.55, p = 0.0002, autistic participants (AUT-NP, N = 42, light blue dots) showed a face inversion effect, that is they performed better on upright (gray bars) AUT-P (N = 21) did not perform better on upright vs. inverted (white bars) faces, while prosopagnosic autistic (35.14 ± 6.28) than inverted faces (31.33 ± 8.06, p = participants (AUT-P, N = 22, dark blue dots) did not. AUT-P did not 0.15). AUT-P performed worse than AUT-NP on up- show a general memory impairment, that is, although they right faces (p = 0.0001), but they did not differ on performed worse than AUT-NP on upright faces, they did not differ inverted faces (p = 0.11). All other post hoc comparisons from AUT-NP on inverted faces. The red asterisk indicates that AUT- NP performance at the upright CFMT differs from all other were not significant (ps > 0.14). To test whether group conditions (ps < 0.05). The dashed horizontal line indicates the CFMT performance differed from the chance level of respond- chance level which corresponds to 24 correct responses ing, for each condition, we ran one-sample t tests against the chance level and found that both groups performed differently from the chance level in all conditions: up- prosopagnosic, predicted participants’ ability to infer an- right CFMT, AUT-P: t (20) = 8.13, p = 9.04e-8, 95% CI other person's mental states by looking at their eye re- [32.28, 38.00]; AUT-NP: t(42) = 24.20, p = 2.76e-26, 95% gion, assessed via the RMET, b = 0.02, SE = 0.006, t = CI [50.71, 55.57] and inverted CFMT, AUT-P: t(20) = 3.00, p = 0.004, 95% CI [0.006, 0.030], N = 76 (Fig. 2). 4.17, p = 4.72e-4, 95% CI [27.67, 35.00]; AUT-NP: t(42) The increase in R2 due to the interaction was 0.10, = 13.63, p = 5.08e-17, 95% CI [33.83, 37.24]. The latter F(1,72) = 8.86, p = 0.004. While in AUT-P, face result, together with the fact that only 3 AUT-P and 1 memory skill on the CFMT predicted their ability to AUT-NP participants performed below the chance level understand another person’s mental states on the on the inverted CFMT and only 1 AUT-P on the upright RMET (b = 0.02, SE = 0.005, t = 3.45, p = 0.001, CFMT (Fig. 3), excluded the presence of a floor effect in 95% CI [0.008, 0.029], N = 26), this was not the case performance. for AUT-NP (b = 0.0004, SE = 0.003, t = 0.15, p = AUT participants completed both the CFMT upright 0.88, 95% CI [− 0.005, 0.006], N = 50). and inverted (in counterbalanced order) while NT par- ticipants completed only the CFMT Upright. Results of The relation between identity and mental state recognition a 2 × 2 mixed ANOVA showed a non-significant main does not depend on individuals’ basic face perception skills effect of order, F(1,62) = 0.18, p = 0.67, ηp2 = 0.003, and We ran a moderated moderation model in order to in- a significant main effect of orientation, F(1,62) = 102.5, p vestigate the role of face perception, assessed via the = 9.21e-15, ηp2 = 0.62, with a more accurate performance CFPT (see Table 4 for performance scores), on the for upright compared to inverted faces. Notably, the Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 9 of 16 Fig. 4 a General intelligence does not correlate with face memory skill in autism. In autistic participants, IQ level (percentile) does not significantly correlate with face memory skill (number of correct responses) at the Cambridge Face Memory Test (CFMT). Light blue dots represent nonprosopagnosic autistic participants (AUT-NP) while the dark blue dots represent prosopagnosic autistic participants (AUT-P). The dashed black line represents the nonsignificant regression line, while the surrounding gray-shaded area represents a 95% confidence interval. b Subjective face memory awareness does not predict objective face memory performance in autistic individuals. In autistic participants self-reported prosopagnosia traits at the prosopagnosia index questionnaires (PI20) do not correlate with their objective performance at the Cambridge Face Memory Test (CFMT). The light blue dots represent non-prosopagnosic autistic participants (AUT-NP) while the dark blue dots represent prosopagnosic autistic participants (AUT-P). The dashed black line represents the nonsignificant regression line, while the surrounding gray- shaded area represents a 95% confidence interval order × orientation interaction was non-significant, F(1, Autistic individuals with and without prosopagnosia show 62) = 1.62, p = 0.29, ηp2 = 0.02, assuring that the order face perception inversion effect in which AUT participants completed the upright or To extend our investigation about differences in identity inverted CFMT did not influence their performance at processing between AUT-P and AUT-NP to their per- the upright CFMT. ceptual abilities, we compared AUT-P and AUT-NP Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 10 of 16 performance on the CFPT with upright and inverted faces. such a schematic model, the association between genetic Type III mixed ANOVA on the CFPT error scores re- risk factors of autism and those of prosopagnosia could vealed significant main effects of prosopagnosia, F(1,60) = vary from reflecting mere spatial proximity of the impli- 9.75, p = 0.003, ηp2 = 0.14, and face orientation, F(1,60) = cated alleles, which therefore remain linked over genera- 67.49, p = 2.11e−11, ηp2 = 0.53, which were explained by a tions, to functional impact on the same neural circuits significant prosopagnosia × face orientation interaction, (e.g., the OXY system). Face IIR might be valuable to help F(1,60) = 10.48, p = 0.002, ηp2 = 0.15. Tukey HSD post focus research on new genetically meaningful autistic sub- hoc tests showed that both groups performed better with groups, neurobiological pathways, and neural systems. Its upright vs. inverted faces (AUT-NP: p = 0.0002; role may be similar to what proposed by Constantino AUT-P: p = 0.02) and that only in the case of upright [138] for other symptoms which are not specific to autism faces, AUT-P performed worse than AUT-NP (up- but which are highly prevalent in autism and are strongly right: p = 0.0004; inverted: p = 0.56). Finally, AUT-P genetically influenced. showed a smaller perceptual face inversion effect than We investigated the relationship between face memory AUT-NP, t(60) = − 3.24, p = 0.004, 95% CI [− 24.85, and mental state recognition from the eye region as both − 5.87] (see Table 4). are impaired in autism [51, 108] and rely on extracting in- formation from the same (eye) region [139] which has Face individual identity recognition is not linked to strong diagnostic [87, 88] and predictive [140, 141] value general intelligence to autism. Our data show that identity recognition was as- Level of intelligence (IQ percentile) did not significantly sociated with the ability to recognize another person’s correlate with face memory (number of correct responses mental state by looking at their eyes, both essential skills on the CFMT) both in NT (Spearman ρ = 0.11, p = 0.36, to navigate the social world. In particular, face identity 95% CI [− 0.12, 0.34], N = 69) and AUT (Spearman ρ = recognition was linked to mental states understanding ex- 0.20, p = 0.09, 95% CI [− 0.06, 0.42], N = 74) participants clusively in autistic participants who were prosopagnosic. (Fig. 4a). Such association was unlikely due to reduced expertise with faces, as non-autistic individuals with DP are not im- Subjective face memory awareness does not predict paired in mental state recognition from the eye region objective face memory performance in autistic individuals [103, 114–116]. Alternatively, reduced attention to the Regression analysis showed that AUT participants self- eyes may affect both identity [139, 142–144] and mental reported prosopagnosic traits (assessed via the PI20 state recognition and may be linked to altered OXY- questionnaire) did not predict their objective face mem- mediated social processing of identity sensory cues [31, ory/individual identity recognition skills (assessed via the 45]. The chances of a common neurobiological mechan- CFMT), b = − 0.16, t(61) = − 1.29, p = 0.20, R2 = 0.03, ism underlying both face identity and mental state recog- F(1,61) = 1.66, p = 0.20, N = 63 (Fig. 4b). nition in AUT-P increase the potential relevance of IIR as an endophenotype, as it stratifies autistic individuals in a Discussion way that is meaningful to distinctive difficulties in social Our results show that prosopagnosia potentially occurs interaction. The differential relationship we found in in more than one third (i.e., 36%) of autistic adults with- AUT-P and AUT-NP between face IIR and mental state out ID. Importantly, our prosopagnosia-based stratifica- recognition from the eye region supports the idea that tion into two groups, AUT-P and AUT-NP, was AUT-P and AUT-NP might be two separate subgroups, independent of symptom severity, autistic traits, IQ, gen- and not just that AUT-NP were autistic individuals with eral memory skills, empathy, and alexithymia. This is in poor identity recognition skills [125]. We included per- keeping with our idea that face memory difficulties do formance on the Cambridge Face Perception Test as an not interact with autism at a necessary and sufficient additional moderator to investigate whether difficulties in aetiological level, an idea that is also supported by the face identity and emotion recognition in AUT-P were fact that DP individuals do not have difficulties in social both due to perceptual failure in face processing. We skills [134], nor, to the best of our knowledge, they are found that the relationship between identity and mental known to have higher rates of autism. We speculate that state recognition was not moderated by altered basic face difficulties in face IIR may therefore not lie on the causal perception skills that impair face recognition even when pathway from genes to autism. In keeping with the poly- face memorization is not needed (i.e., apperceptive genic nature of autism [135] and probably of DP [43], dif- prosopagnosia). ficulties in IIR may rather contribute to autism’s genetic To further investigate between-group differences in face background liability [136] in a manner resembling a multi- processing and general memory skills, we tested autistic variate correlated liability model [137] including multiple participants’ memory for inverted faces, which, unlike from genes, endophenotypes, and environmental factors. In upright faces, are typically not processed holistically [145]. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 11 of 16 In their review, Weigelt and colleagues [63] reported that Limitations only 2 out of 14 case-control studies showed no face inver- This was not an epidemiological sample and therefore sion effect in autistic participants, and so, the majority of may not be representative of the prevalence of prosopag- studies suggest better face recognition for upright vs. nosia in autism. However, it is the largest sample avail- inverted faces [102] also in autistic individuals. However, able to date and it was not biased with respect to face none of these studies controlled for prosopagnosia. Our re- processing skills, so it may indeed reflect the true preva- sults show that while AUT-NP remembered faces better lence of prosopagnosia in autistic adults with no ID. We when presented upright vs. inverted, AUT-P did not show do not know whether our results generalize to the entire such face inversion effect. This suggests that upright faces autism spectrum. Future work is needed to determine may not be a special class of stimuli to remember for AUT- whether prosopagnosia is equally prevalent and similarly P and that presence of prosopagnosia, not autism, drove associated with mental state recognition skills also in the lack of face memory inversion effect as both groups had other autistic individuals not represented in the current an autism diagnosis. Interestingly instead, both AUT-NP sample, such as in children, individuals with ID and fe- and AUT-P showed face inversion effect when their face males (our sample included only 20% of females). perception skills were assessed with no memory demands Our proposal that prosopagnosia might be a potential (i.e., on the CFPT), again suggesting that retention of facial endophenotype in autism would benefit from additional identity information is crucial in differentiating AUT-P findings supporting co-segregation of autism and proso- from AUT-NP. Nonetheless, interpretation is not so pagnosia within families and higher rate of prosopagno- straight forward as results on the presence of face inver- sia in non-autistic family members compared to the sion effects in DP are mixed, evidencing holistic face general population. processing is not always impaired in DP [43, 101]. The absence of face memory inversion effect and a Future perspectives positive correlation between face memory and mental Autism research benefits from parallel human and ani- state understanding are instances in which AUT-P mal studies and our results, taken in the context of the differed from DP. It remains an open question current literature, open various avenues of research. Fu- whether prosopagnosia in AUT shares similar charac- ture human studies could investigate whether face mem- teristics and neurobiological correlates with prosopag- ory difficulties can subgroup autism high-risk infants in nosia in non-autistic, DP individuals. In addition, we prospective meaningful ways. Given that intranasal OXY found that, while AUT-P performed worse than AUT- (INOXY) was shown to normalize identity recognition NP with upright faces (at both perceptual and mem- in DP [44], correctly increase familiar judgments of pre- ory levels), they did not differ with inverted faces, viously seen faces in controls [148], and improve eye suggesting our results were not due to unspecific contact in autistic individuals [149], researchers could memory impairments. investigate its effects on face memory in autism. Re- As IQ varies greatly along the autistic spectrum [146], searchers could also investigate, as a downstream cas- we checked whether IQ correlated with IIR and found cade, the effects of INOXY on other face memory- that this was not the case both for our AUT and NT related social behaviors relevant to autism, such as social participants, a result similar to that found in the general anxiety and attentional preference for faces [150, 151]. population [78, 79]. Face memory may therefore be po- Further, since face memory could be a proxy to autistic tentially relevant to the entire autistic spectrum, across participants’ OXY-relevant genetic background, re- all levels of intellectual ability. It remains to be tested searchers could investigate whether it predicts re- whether the independence between face memory skills sponders to INOXY. This face IIR-based stratification of and IQ holds true also with individuals with ID. participants may help addressing the failure of many Lastly, we explored autistic participants’ awareness of [152, 153], often underpowered [154], INOXY interven- their face IIR skills and whether a self-report question- tion studies with autistic participants. naire, here the PI20 questionnaire, could be used for Recent evidence showing that assessment of IIR in autis- screening autistic individuals with difficulties in face mem- tic individuals may extend beyond visual-face to other sen- ory. Contrary to the general population and to DP individ- sory systems and identity-conveying cues such as auditory- uals [120–122], subjective scores of autistic participants voice [155] and olfactory-body odor [156], increases the on the PI20 did not predict their objective performance possibility to directly translate experimental paradigms and on the CFMT. Therefore, a self-report questionnaire does research questions to autism rodent models [85]. Animal not seem a reliable prosopagnosia screening tool for aut- studies, overcoming limitations intrinsic to human research, ism, which is at odds with previous findings showing that could examine if and which genetic autism mouse models adult autistic participants had similar face memory aware- show IIR deficits, to then uncover their neuro-biological ness compared to neurotypical individuals [147]. correlates with a focus on OXY’s possible modulatory role Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 12 of 16 in generating states for optimized information extraction Availability of data and materials [157] and in attributing salience and reward to identity- The anonymized datasets analyzed in the current study are available from the corresponding author on request. relevant sensory cues [158]. In conclusion, we found that difficulties in face individ- Ethics approval and consent to participate ual identity recognition are highly prevalent in autism and Ethical clearance was granted by the local ethics committees at IRCCS are linked to difficulties in mental state understanding Fondazione Santa Lucia, Rome, Italy; University College London, London, UK; Beth Israel Deaconess Medical Center, Boston, MA, USA, and the study was from the eye region independently from face perception conducted in accordance with the Declaration of Helsinki. skills. Further, they stratify autistic individuals irrespective of intellectual ability, diagnostic symptoms, and personal- Consent for publication ity traits. Because of its potential role as an endopheno- Participants gave informed consent to participate in the study and to publish their anonymized individual data prior to participation in the study. type, we believe individual identity recognition may be Participants were paid for their time. important to advance our understanding of autism within a translational framework informed by, and informative Competing interests to, neurobiological non-human animal research. A.P.L. serves on the scientific advisory boards for the Starlab Neuroscience, Neuroelectrics, Neosync, NovaVision, Magstim, and Cognito and is listed as an inventor on several issued and pending patents on the real-time integration of Abbreviations transcranial magnetic stimulation with electroencephalography and magnetic ADI-R: Autism Diagnostic Interview Revised; ADOS: Autism Diagnostic resonance imaging. The other authors declare no competing interests. Observation Schedule; AQ: Autism spectrum quotient; AUT: Autism; AUT- NP: Autistic non-prosopagnosic; AUT-P: Autistic prosopagnosic; Author details 1 CFMT: Cambridge Face Memory Test; CFPT: Cambridge Face Perception Test; Department of Psychology, Sapienza University of Rome, Rome, Italy. 2IRCCS CI: Confidence interval; CSS: Calibrated Severity Score; EC: Empathic concern; Fondazione Santa Lucia, Rome, Italy. 3Institute of Cognitive Sciences and ID: Intellectual disability; IIR: Individual identity recognition; INOXY: Intranasal Technologies, National Research Council, Rome, Italy. 4Hinda and Arthur oxytocin; IQ: Intelligence quotient; IRI: Interpersonal Reactivity Index; Marcus Institute for Aging Research and Center for Memory Health, Hebrew NT: Neurotypical; OXY: Oxytocin; PI20: Twenty-item Prosopagnosia Index; SeniorLife, Boston, MA, USA. 5Department of Neurology, Harvard Medical PT: Perspective taking; RRB: Restricted Repetitive Behaviors; SA: Social School, Boston, MA, USA. 6Guttmann Brain Health Institute, Institut Guttmann affective; SD: Standard deviation; SE: Standard error; TAS: Twenty-item de Neurorehabilitació, Universitat Autonoma de Barcelona, Badalona, Spain. 7 Toronto Alexithymia Scale; VIF: Variance inflation factor Autism Research Centre, Department of Psychiatry, University of Cambridge, Cambridge, UK. Acknowledgements Received: 16 February 2019 Accepted: 11 August 2020 We thank all participants who took part in this study. Particular thanks go to autistic participants and their relatives who, with their interest and dedication, make autism research possible. References 1. American Psychiatric Association. Diagnostic and statistical manual of Authors’ contributions mental disorders : DSM-5. Fifth edition. Arlington: American Psychiatric IMP designed the study, run clinical assessments, collected and analyzed the Association, [2013]; 2013. Available from: https://search.library.wisc.edu/ data, and wrote the manuscript. GP analyzed the data and wrote the manuscript. catalog/9910187853902121. APL and SBC supervised the project and wrote the manuscript. The authors read 2. Elsabbagh M, Divan G, Koh Y, Kim YS, Kauchali S, Marcín C, et al. Global and approved the manuscript. prevalence of autism and other pervasive developmental disorders. Autism Res. 2012;5(3):160–79 Available from: http://doi.wiley.com/10.1002/aur.239. 3. Lai M, Lombardo MV, Baron-Cohen S. Autism. Lancet. 2013;6736(13):1–15. Funding 4. Geschwind DH, State MW. Gene hunting in autism spectrum disorder: on IMP was supported by the USA—Italy Fulbright Commission and the Italian the path to precision medicine. Lancet Neurol. 2015;14(11):1109–20 Ministry of Health (GR-2009-1607360) during the period of this work. Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4694565/. APL was supported by grants from the National Institutes of Health (NIH, 5. Lai MC, Lombardo MV, Auyeung B, Chakrabarti B, Baron-Cohen S. Sex/ primarily RO1MH100186); the National Science Foundation, Sidney R. Baer, Jr. Gender Differences and autism: setting the scene for future research. Foundation, Harvard Catalyst The Harvard Clinical and Translational Science Journal of the American Academy of Child and Adolescent Psychiatry. 2015; Center (NCRR and the NCATS NIH, UL1 RR025758), and by the Defense 54:11–24 Available from: https://linkinghub.elsevier.com/retrieve/pii/ Advanced Research Projects Agency (DARPA) via HR001117S0030. The S0890856714007254. content is solely the responsibility of the authors and does not necessarily 6. Szatmari P, Georgiades S, Duku E, Bennett TA, Bryson S, Fombonne E, et al. represent the official views of Harvard Catalyst, Harvard University, and its Developmental trajectories of symptom severity and adaptive functioning affiliated academic health care centers, the National Institutes of Health, in an inception cohort of preschool children with autism spectrum disorder. National Science Foundation, the Sidney R. Baer Jr. Foundation, or DARPA. JAMA Psychiatry. 2015;72(3):276–83 Available from: http://archpsyc. SBC was supported by the MRC (grant number RNAG/128, award number jamanetwork.com/article.aspx?doi=10.1001/jamapsychiatry.2014.2463. RG46450), the Wellcome Trust (grant number RNAG/235, award number 7. Kapur S, Phillips AG, Insel TR. Why has it taken so long for biological RG58828), and the Autism Research Trust, during the period of this work. He psychiatry to develop clinical tests and what to do about it? Mol Psychiatry. was also supported by the Innovative Medicines Initiative 2 Joint 2012;17(12):1174–9. Available from:. https://doi.org/10.1038/mp.2012.105. Undertaking (JU) under grant agreement No 777394. The JU receives 8. Viding E, Blakemore S-J. Endophenotype approach to developmental support from the European Union’s Horizon 2020 research and innovation psychopathology: implications for autism research. Behav Genet. 2007;37(1): program and EFPIA and AUTISM SPEAKS, Autistica, SFARI. He was also 51–60 Available from: http://link.springer.com/10.1007/s10519-006-9105-4. supported by the National Institute for Health Research (NIHR) Collaboration 9. Christensen DL, Braun KVN, Baio J, Bilder D, Charles J, Constantino JN, et al. for Leadership in Applied Health Research and Care East of England at Prevalence and characteristics of autism spectrum disorder among children Cambridgeshire and Peterborough NHS Foundation Trust. The views aged 8 years — autism and developmental disabilities monitoring network, expressed are those of the authors and not necessarily those of the NHS, 11 Sites, United States, 2012. MMWR Surveill Summ. 2018;65(13):1–23 NIHR, or Department of Health and Social Care’. Finally, he received support Available from: http://www.cdc.gov/mmwr/volumes/65/ss/ss6513a1.htm?s_ from the NIHR Biomedical Research Centre (BRC). cid=ss6513a1_w. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 13 of 16 10. Volkmar FR, McPartland JC. From Kanner to DSM-5: autism as an evolving 30. Turati C, Bulf H, Simion F. Newborns’ face recognition over changes in diagnostic concept. Annu Rev Clin Psychol. 2014;10(1):193–212 Available viewpoint. Cognition. 2008;106(3):1300–21 Available from: https:// from: http://www.annualreviews.org/doi/10.1146/annurev-clinpsy-032 linkinghub.elsevier.com/retrieve/pii/S0010027707001746. 813-153710. 31. Johnson ZV, Young LJ. Oxytocin and vasopressin neural networks: 11. DiLalla LF, McCrary M, Diaz E. A review of endophenotypes in schizophrenia implications for social behavioral diversity and translational neuroscience. and autism: the next phase for understanding genetic etiologies. Am J Med Neurosci Biobehav Rev. 2017;76:87–98. Available from:. https://doi.org/10. Genet Part C Semin Med Genet. 2017;175(3):354–61 Available from: http:// 1016/j.neubiorev.2017.01.034. doi.wiley.com/10.1002/ajmg.c.31566. 32. Yokoi S, Naruse K, Kamei Y, Ansai S, Kinoshita M, Mito M, et al. Sexually 12. Alarcón M, Yonan AL, Gilliam TC, Cantor RM, Geschwind DH. Quantitative dimorphic role of oxytocin in medaka mate choice. Proc Natl Acad Sci U S genome scan and ordered-subsets analysis of autism endophenotypes A. 2020;117(9):4802–8. support language QTLs. Mol Psychiatry. 2005;10(8):747–57 Available from: 33. Perrodin C, Kayser C, Abel TJ, Logothetis NK, Petkov CI. Who is that ? http://www.nature.com/articles/4001666. Brain networks and mechanisms for identifying individuals. Trends 13. Constantino JN, Kennon-McGill S, Weichselbaum C, Marrus N, Haider A, Cogn Sci. 2015;xx:1–14. Available from:. https://doi.org/10.1016/j.tics.2015. Glowinski AL, et al. Infant viewing of social scenes is under genetic control 09.002. and is atypical in autism. Nature. 2017;547(7663):340–4. Available from:. 34. Chang L, Tsao DY. The code for facial identity in the primate brain. Cell. https://doi.org/10.1038/nature22999. 2017;169(6):1013–1028.e14. Available from:. https://doi.org/10.1016/j.cell. 14. Lowe JK, Werling DM, Constantino JN, Cantor RM, Geschwind DH. Social 2017.05.011. responsiveness, an autism endophenotype: genomewide significant linkage 35. Stevens CF. Conserved features of the primate face code. Proc Natl Acad to two regions on chromosome 8. Am J Psychiatry. 2015;172(3):266–75 Sci. 2018;201716341 Available from: http://www.pnas.org/lookup/doi/10.1 Available from: http://psychiatryonline.org/doi/10.1176/appi.ajp.2014.14050576. 073/pnas.1716341115. 15. Lisiecka DM, Holt R, Tait R, Ford M, Lai MC, Chura LR, et al. Developmental 36. Sheehan MJ, Nachman MW. Morphological and population genomic white matter microstructure in autism phenotype and corresponding evidence that human faces have evolved to signal individual identity. Nat endophenotype during adolescence. Transl Psychiatry. 2015;5(3):e529. Commun. 2015;5(4800):1–21. Available from:. https://doi.org/10.1038/tp.2015.23. 37. Porciello G, Bufalari I, Minio-Paluello I, Di Pace E, Aglioti SM. The 16. Gottesman II, Gould TD. The endophenotype concept in psychiatry: ‘enfacement’ illusion: a window on the plasticity of the self. Cortex. 2018; etymology and strategic intentions. Am J Psychiatry. 2003;160(4):636–45 104:261–75. Available from: http://doi.wiley.com/10.1111/j.1365-4632.2005.02252.x. 38. Bufalari I, Porciello G, Sperduti M, Minio-Paluello I. Self-identification with 17. Spencer MD, Holt RJ, Chura LR, Suckling J, Calder AJ, Bullmore ET, et al. A another person’s face: the time relevant role of multimodal brain areas in novel functional brain imaging endophenotype of autism: the neural the enfacement illusion. J Neurophysiol. 2015;113(7):1959–62 Available from: response to facial expression of emotion. Transl Psychiatry. 2011;1(7):e19–7. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4416605/. Available from:. https://doi.org/10.1038/tp.2011.18. 39. Minio-Paluello I, Porciello G, Gandolfo M, Boukarras S, Aglioti SM. The 18. Spencer MD, Holt RJ, Chura LR, Calder AJ, Suckling J, Bullmore ET, et al. enfacement illusion boosts facial mimicry. Cortex. 2019;123:113–23 Available Atypical activation during the embedded figures task as a functional from: https://www.ncbi.nlm.nih.gov/pubmed/31765877. magnetic resonance imaging endophenotype of autism. Brain. 2012;135(11): 40. Tardif J, Morin Duchesne X, Cohan S, Royer J, Blais C, Fiset D, et al. Use of 3469–80. face information varies systematically from developmental prosopagnosics 19. Gabard-Durnam L, Tierney AL, Vogel-Farley V, Tager-Flusberg H, Nelson CA. to super-recognizers. Psychol Sci. 2018:1–9 Available from: http://journals. Alpha asymmetry in infants at risk for autism spectrum disorders. J Autism sagepub.com/doi/10.1177/0956797618811338. Dev Disord. 2015;45(2):473–80. 41. Kennerknecht I, Grueter T, Welling B, Wentzek S. First report of prevalence 20. Tierney AL, Gabard-Durnam L, Vogel-Farley V, Tager-Flusberg H, Nelson CA. of non-syndromic hereditary prosopagnosia (HPA). Am J Med Genet Part A. Developmental trajectories of resting EEG power: an endophenotype of 2006;140A:1617–22. autism spectrum disorder. PLoS One. 2012;7(6). 42. Kennerknecht I, Ho NY, Wong VCN. Prevalence of hereditary 21. Kaiser MD, Hudac CM, Shultz S, Su Mei Lee, Cheung C, Berkena AM, et al. prosopagnosia (HPA) in Hong Kong Chinese population. Am J Med Neural signatures of autism. Proc Natl Acad Sci U S A. 2010;107(49):21223– Genet Part A. 2008;146A(22):2863–70 Available from: http://doi.wiley. 21228. com/10.1002/ajmg.a.32552. 22. Moseley RL, Ypma RJF, Holt RJ, Floris D, Chura LR, Spencer MD, et al. Whole- 43. Susilo T, Duchaine B. Advances in developmental prosopagnosia research. brain functional hypoconnectivity as an endophenotype of autism in Curr Opin Neurobiol. 2013;23(3):423–9. Available from:. https://doi.org/10. adolescents. NeuroImage Clin. 2015;9:140–52. Available from:. https://doi. 1016/j.conb.2012.12.011. org/10.1016/j.nicl.2015.07.015. 44. Bate S, Cook SJ, Duchaine B, Tree JJ, Burns EJ, Hodgson TL. Intranasal 23. Liu XQ, Paterson AD, Szatmari P. Genome-wide linkage analyses of inhalation of oxytocin improves face processing in developmental quantitative and categorical autism subphenotypes. Biol Psychiatry. 2008; prosopagnosia. Cortex. 2014;50:55–63. 64(7):561–70. 45. Cattaneo Z, Daini R, Malaspina M, Manai F, Lillo M, Fermi V, et al. Congenital 24. Chaste P, Klei L, Sanders SJ, Hus V, Murtha MT, Lowe JK, et al. A genome- prosopagnosia is associated with a genetic variation in the oxytocin wide association study of autism using the Simons simplex collection: does receptor (OXTR) gene: an exploratory study. Neuroscience. 2016;339:162–73. reducing phenotypic heterogeneity in autism increase genetic Available from:. https://doi.org/10.1016/j.neuroscience.2016.09.040. homogeneity? Biol Psychiatry. 2015;77(9):775–84 Available from: https:// 46. Duchaine B, Nakayama K. The Cambridge Face Memory Test: results for linkinghub.elsevier.com/retrieve/pii/S0006322314007161. neurologically intact individuals and an investigation of its validity using 25. Geschwind DH, Flint J. Genetics and genomics of psychiatric disease. inverted face stimuli and prosopagnosic participants. Neuropsychologia. Science (80- ). 2015;349(6255):1489–94 Available from: http:// 2006;44(4):576–85 Available from: https://linkinghub.elsevier.com/retrieve/ psychiatryonline.org/doi/10.1176/appi.ajp.2014.14050576. pii/S0028393205002496. 26. Flint J, Timpson N, Munafò M. Assessing the utility of intermediate 47. Bowles DC, McKone E, Dawel A, Duchaine B, Palermo R, Schmalzl L, et al. phenotypes for genetic mapping of psychiatric disease. Trends Neurosci. Diagnosing prosopagnosia: effects of ageing, sex, and participant-stimulus 2014;37(12):733–41. ethnic match on the Cambridge face memory test and Cambridge face 27. Krol A, Wimmer RD, Halassa MM, Feng G. Thalamic reticular dysfunction as a perception test. Cogn Neuropsychol. 2009;26(5):423–55. circuit endophenotype in neurodevelopmental disorders. Neuron. 2018; 48. Bate S, Tree JJ. The definition and diagnosis of developmental 98(2):282–95 Available from: https://linkinghub.elsevier.com/retrieve/pii/ prosopagnosia. Q J Exp Psychol. 2017;70(2):193–200. S0896627318301983. 49. Dalrymple KA, Palermo R. Guidelines for studying developmental 28. Tibbetts EA, Dale J. Individual recognition: it is good to be different. Trends prosopagnosia in adults and children. Wiley Interdiscip Rev Cogn Sci. 2016; Ecol Evol. 2007;22(10):529–37 Available from: https://linkinghub.elsevier.com/ 7(1):73–87. retrieve/pii/S0169534707002376. 50. Nomi JS, Uddin LQ. Neuropsychologia face processing in autism spectrum 29. Bushnell IWR. Mother’s Face Recognition in Newborn Infants: Learning and disorders: from brain regions to brain networks. Neuropsychologia. 2015;71: Memory. Infant Child Dev. 2001;10(1–2):67–74. 201–16. Available from:. https://doi.org/10.1016/j.neuropsychologia.2015.03.029. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 14 of 16 51. Webb SJ, Neuhaus E, Faja S. Face perception and learning in autism 71. Kirchner JC, Hatri A, Heekeren HR. Dziobek I. Autistic symptomatology, face spectrum disorders. Q J Exp Psychol. 2017;70(5):970–86 Available from: processing abilities, and eye fixation patterns. 2011:158–67. http://journals.sagepub.com/doi/10.1080/17470218.2016.1151059. 72. Fedor J, Lynn A, Foran W, DiCicco-Bloom J, Luna B, O’Hearn K. Patterns of 52. Dawson G, Carver L, Meltzoff AN, Panagiotides H, McPartland J, Webb SJ. fixation during face recognition: differences in autism across age. Autism Neural correlates of face and object recognition in young children with [Internet]. 2017;136236131771498. Available from: http://journals.sagepub. autism spectrum disorder, developmental delay, and typical development. com/doi/10.1177/1362361317714989. Child Dev. 2002;73(3):700–17. 73. Dwyer P, Xu B, Tanaka JW. Investigating the perception of face identity in 53. Sterling L, Dawson ÆG, Webb ÆS, Murias ÆM, Munson J, Panagiotides ÆH, adults on the autism spectrum using behavioural and electrophysiological et al. The role of face familiarity in eye tracking of faces by individuals with measures. Vision Res [Internet]. 2018;(February):0–1. Available from: https:// autism spectrum disorders; 2008. p. 1666–75. doi.org/10.1016/j.visres.2018.02.013. 54. Chawarska K, Shic F. Looking but not seeing: atypical visual scanning and 74. Schelinski S, Roswandowitz C, von Kriegstein K. Voice identity processing in recognition of faces in 2 and 4-year-old children with autism spectrum autism spectrum disorder. Autism Res [Internet]. 2017;10(1):155–68 Available disorder. J Autism Dev Disord [Internet]. 2009;39(12):1663–72 Available from: from: http://doi.wiley.com/10.1002/aur.1639. http://link.springer.com/10.1007/s10803-009-0803-7. 75. Hedley D, Brewer N, Young R. Face recognition performance of individuals 55. Eussen MLJM, Louwerse A, Herba CM, Van Gool AR, Verheij F. Verhulst FC, with Asperger syndrome on the Cambridge face memory test. Autism Res. et al. Childhood facial recognition predicts adolescent symptom severity in 2011;4(6):449–55. autism spectrum disorder. 2015:1–11. 76. Zhu Q, Song Y, Hu S, Li X, Tian M, Zhen Z, et al. Heritability of the specific 56. Barton JJS, Cherkasova MV, Hefter R, Cox TA, O’Connor M, Manoach DS. Are cognitive ability of face perception. Curr Biol [Internet]. 2010;20(2):137–42 patients with social developmental disorders prosopagnosic? Perceptual Available from: 10150.1016/j.cub.2009.11.067. heterogeneity in the Asperger and socio-emotional processing disorders. 77. Wilmer JB, Germine L, Chabris CF, Chatterjee G, Williams M, Loken E, et al. Brain. 2004;127(8):1706–16. Human face recognition ability is specific and highly heritable. Proc Natl 57. Wolf JM, Tanaka JW, Klaiman C, Cockburn J, Herlihy L, Brown C, et al. Acad Sci [Internet]. 2010;107(11):5238–41 Available from: http://www.pnas. Specific impairment of face-processing abilities in children with autism org/cgi/doi/10.1073/pnas.0913053107. spectrum disorder using the Let’s Face It! skills battery. Autism Res [Internet]. 78. Shakeshaft NG, Plomin R. Genetic specificity of face recognition. Proc Natl 2008;1(6):329–40 Available from: http://doi.wiley.com/10.1002/aur.56. Acad Sci [Internet]. 2015;112(41):12887–92 Available from: http://www.pnas. 58. Wilson CE, Palermo R, Burton AM, Brock J. Recognition of own- and other- org/lookup/doi/10.1073/pnas.1421881112. race faces in autism spectrum disorders. Q J Exp Psychol. 2011;64(10):1939–54. 79. Wilmer JB, Germine LT, Nakayama K. Face recognition: a model specific 59. Oerlemans AM, Droste K, Van Steijn DJ, De Sonneville LMJ, Buitelaar JK, ability. Front Hum Neurosci [Internet]. 2014;8(October):1–5 Available from: Rommelse NNJ. Co-segregation of social cognition, executive function and http://journal.frontiersin.org/article/10.3389/fnhum.2014.00769/abstract. local processing style in children with ASD, their siblings and normal 80. Wilson CE, Freeman P, Brock J, Burton AM, Palermo R. Facial identity recognition controls. J Autism Dev Disord. 2013;43(12):2764–78. in the broader autism phenotype. Miles J, editor. PLoS One [Internet]. 2010 22; 60. Skuse DH, Lori A, Cubells JF, Lee I, Conneely KN, Puura K, et al. Common 5(9):e12876. Available from: https://dx.plos.org/10.1371/journal.pone.0012876. polymorphism in the oxytocin receptor gene ( OXTR ) is associated with 81. de Klerk CCJM, Gliga T, Charman T, Johnson MH. Face engagement during human social recognition skills. Proc Natl Acad Sci [Internet]. 2014;111(5): infancy predicts later face recognition ability in younger siblings of children 1987–92 Available from: http://www.pnas.org/lookup/doi/10.1073/pnas.1302 with autism. Dev Sci [Internet]. 2014;17(4):596–611 Available from: http://doi. 985111. wiley.com/10.1111/desc.12141. 61. Duchaine BC, Weidenfeld A. An evaluation of two commonly used tests of 82. de Haan M, Nelson CA. Brain activity differentiates face and object processing unfamiliar face recognition. Neuropsychologia. 2003;41(6):713–20. in 6-month-old infants. Dev Psychol [Internet]. 1999;35(4):1113–21 Available 62. Wilson CE, Palermo R, Brock J. Visual scan paths and recognition of facial from: http://doi.apa.org/getdoi.cfm?doi=10.1037/0012-1649.35.4.1113. identity in autism spectrum disorder and typical development. PLoS One. 83. Key AP, Dykens EM. Incidental memory for faces in children with different 2012 Jan;7(5):e37681. genetic subtypes of Prader-Willi syndrome. Soc Cogn Affect Neurosci 63. Weigelt S, Koldewyn K, Kanwisher N. Face identity recognition in autism [Internet]. 2017;12(6):918–27 Available from: https://academic.oup.com/scan/ spectrum disorders: a review of behavioral studies [Internet]. Vol. 36, article/12/6/918/3002819. Neuroscience and Biobehavioral Reviews. Elsevier Ltd; 2012. p. 1060–1084. 84. Verhallen RJ, Bosten JM, Goodbourn PT, Lawrance-Owen AJ, Bargary G, Available from: https://doi.org/10.1016/j.neubiorev.2011.12.008. Mollon JD. The oxytocin receptor gene (OXTR) and face recognition. 64. Weigelt S, Koldewyn K, Kanwisher N. Face recognition deficits in autism Psychol Sci [Internet]. 2017;28(1):47–55. Available from: http://journals. spectrum disorders are both domain specific and process specific. Pavlova sagepub.com/doi/10.1177/0956797616672269. M, editor. PLoS One [Internet]. 2013 Sep 11;8(9):e74541. Available from: 85. Minio-Paluello I, Porciello G, Mandillo S, Golini E, Aglioti SM, Gross CT, http://dx.plos.org/10.1371/journal.pone.0074541. D'Amato FR. Poster at the workshop: understanding the neuroregulatory 65. Whyte EM, Behrmann M, Minshew NJ, Garcia NV, Scherf KS. Animal, but not actions of oxytocin and its potential clinical applications. Erice, Italy. 22-27/ human, faces engage the distributed face network in adolescents with 05/ 2018. Poster title: Identity Recognition A Promising Oxytocin-relevant autism. Dev Sci. 2016;19(2):306–17. Endophenotype in Autism. 2018. 66. O’Hearn K, Tanaka J, Lynn A, Fedor J, Minshew N, Luna B. Brain and 86. Cheetham SA, Thom MD, Jury F, Ollier WER, Beynon RJ. The genetic basis of cognition developmental plateau in visual object processing from individual-recognition signals in the mouse. Curr Biol. 2007;17(October 23): adolescence to adulthood in autism. BRAIN Cogn [Internet]. 2014;90:124–34. 1771–7. Available from:. https://doi.org/10.1016/j.bandc.2014.06.004. 87. Lord C, Risi S, Lambrecht L, Cook EH, Leventhal BL, DiLavore PC, et al. The 67. Scherf KS, Elbich D, Minshew N, Behrmann M. Individual differences in autism diagnostic observation schedule-generic: a standard measure of symptom severity and behavior predict neural activation during face social and communication deficits associated with the spectrum of autism. processing in adolescents with autism. NeuroImage Clin [Internet]. 2015;7: J Autism Dev Disord [Internet]. 2000 Jun;30(3):205–23. Available from: http:// 53–67. Available from:. https://doi.org/10.1016/j.nicl.2014.11.003. www.ncbi.nlm.nih.gov/pubmed/11055457. 68. Ewbank MP, Pell PJ, Powell TE, Von EAH, Baron-cohen S, Calder AJ. Repetition 88. Lord C, Rutter M, Le Couteur A. Autism diagnostic interview-revised: a revised suppression and memory for faces is reduced in adults with autism spectrum version of a diagnostic interview for caregivers of individuals with possible conditions. Cereb Cortex [Internet]. 2016;1–12. Available from: https:// pervasive developmental disorders. J Autism Dev Disord. 1994;24(5):659–85. academic.oup.com/cercor/article-lookup/doi/10.1093/cercor/bhw385. 89. Wechsler D. Wechsler Abbreviated Scale of Intelligence WASI: Manual 69. Lynn AC, Padmanabhan A, Simmonds D, Foran W, Hallquist MN. Luna B, [Internet]. Pearson/PsychCorpl; 1999. Available from: https://books.google.it/ et al. Functional connectivity differences in autism during face and car books?id = adTXtwAACAAJ. recognition: underconnectivity and atypical age-related changes. 2016:1–18. 90. Wechsler D. Wechsler Adult Intelligence Scale (WAIS-IV). [Internet]. Pearson, 70. O’hearn K, Schroer E, Minshew N, Luna B. Lack of developmental The Psychological Corporation; 2008. Available from: https://books.google.it/ improvement on a face memory task during adolescence in autism. books?id = aZhltAEACAAJ. Neuropsychologia [Internet]. 2010;48(13):3955–60 Available from: https:// 91. Raven JC. Standard Progressive Matrices Sets A, B, C, D & E [Internet]. Harcourt; linkinghub.elsevier.com/retrieve/pii/S0028393210003787. 2006. Available from: https://books.google.it/books?id = V3HxPgAACAAJ. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 15 of 16 92. Baron-Cohen S, Wheelwright S, Skinner R, Martin J, Clubley E. The autism- and their siblings. Psychol Med [Internet]. 2014;44(15):3215–27 Available spectrum quotient (AQ): evidence from Asperger syndrome/high- from: https://www.cambridge.org/core/product/identifier/S0033291714 functioning autism, males and females, scientists and mathematicians. J 000233/type/journal_article. Autism Dev Disord [Internet]. 2001;31(1):5–17 Available from: https://link. 111. Lai M-C, Lombardo M V., Ruigrok AN V., Chakrabarti B, Wheelwright SJ, springer.com/content/pdf/10.1023/A:1005653411471.pdf. Auyeung B, et al. Cognition in males and females with autism: similarities 93. Minio-Paluello I, Baron-Cohen S, Avenanti A, Walsh V, Aglioti SM. Absence of and differences. Botbol M, editor. PLoS One [Internet]. 2012 17;7(10):e47198. embodied empathy during pain observation in Asperger syndrome. Biol Available from: http://dx.plos.org/10.1371/journal.pone.0047198. Psychiatry [Internet]. 2009;65(1):55–62. Available from:. https://doi.org/10. 112. Wilson CE, Happé F, Wheelwright SJ, Ecker C, Lombardo M V., Johnston P, 1016/j.biopsych.2008.08.006. et al. The neuropsychology of male adults with high-functioning autism or 94. Minio-Paluello I, Lombardo M V., Chakrabarti B, Wheelwright S, Baron-Cohen Asperger syndrome. Autism Res [Internet]. 2014 Oct;7(5):568–81. Available S. Response to smith’s letter to the editor “Emotional empathy in Autism from: http://doi.wiley.com/10.1002/aur.1394. spectrum conditions: Weak, Intact, or heightened?” J Autism Dev Disord 113. Warrier V, Grasby KL, Uzefovsky F, Toro R, Smith P, Chakrabarti B, et al. [Internet]. 2009;39(12):1749–54. Available from: https://link.springer.com/ Genome-wide meta-analysis of cognitive empathy: heritability, and article/10.1007%2Fs10803-009-0800-x. correlates with sex, neuropsychiatric conditions and cognition. Mol 95. Curioni A, Minio-Paluello I, Sacheli LM, Candidi M, Aglioti SM. Autistic traits Psychiatry [Internet]. 2018;23(6):1402–9 Available from: http://www.nature. affect interpersonal motor coordination by modulating strategic use of role- com/articles/mp2017122. based behavior. Mol Autism [Internet]. 2017;8(1):23. Available from: http:// 114. Duchaine BC, Parker H, Nakayama K. Normal recognition of emotion in a molecularautism.biomedcentral.com/articles/10.1186/s13229-017-0141-0. prosopagnosic. Perception [Internet]. 2003;32(7):827–38 Available from: 96. McKone E, Martini P, Nakayama K. Categorical perception of face identity in http://journals.sagepub.com/doi/10.1068/p5067. noise isolates configural processing. J Exp Psychol Hum Percept Perform. 115. Palermo R, Willis ML, Rivolta D, McKone E, Wilson CE, Calder AJ. Impaired 2001;27(3):573–99. holistic coding of facial expression and facial identity in congenital 97. Cho S, Wilmer J, Fiset D, Gulick AE Van, Ryan KF. Item Response Theory prosopagnosia. Neuropsychologia [Internet]. 2011;49(5):1226–35 Available Analyses of the Cambridge Face Memory Test (CFMT). 2015;27(2):552–566. from: https://linkinghub.elsevier.com/retrieve/pii/S0028393211000832. 98. Germine L, Nakayama K, Duchaine BC, Chabris CF, Chatterjee G, Wilmer 116. Lee Y, Duchaine B, Wilson HR, Nakayama K. Three cases of developmental JB. Is the Web as good as the lab? Comparable performance from Web prosopagnosia from one family: detailed neuropsychological and and lab in cognitive/perceptual experiments. Psychon Bull Rev. 2012; psychophysical investigation of face processing. Cortex [Internet]. 2010;46(8): 19(5):847–57. 949–64. Available from:. https://doi.org/10.1016/j.cortex.2009.07.012. 99. Bate S, Frowd C, Bennetts R, Hasshim N, Murray E, Bobak AK, et al. Applied 117. Wheelwright S, Auyeung B, Allison C, Baron-Cohen S. Defining the broader, screening tests for the detection of superior face recognition. Cogn Res Princ medium and narrow autism phenotype among parents using the Autism Implic [Internet]. 2018 27;3(1):22. Available from: https:// Spectrum Quotient (AQ). Mol Autism [Internet]. 2010;1(1):10. Available from: cognitiveresearchjournal.springeropen.com/articles/10.1186/s41235-018-0116-5. http://molecularautism.biomedcentral.com/articles/10.1186/2040-2392-1-10. 100. Wilmer JB, Germine L, Chabris CF, Chatterjee G, Williams M, Loken E, et al. 118. Ruzich E, Allison C, Chakrabarti B, Smith P, Musto H, Ring H, et al. Sex and Human face recognition ability is specific and highly heritable. Proc Natl STEM occupation predict Autism-Spectrum Quotient (AQ) Scores in half a Acad Sci. 2010;107(11):5238–41. million people. Taniike M, editor. PLoS One [Internet]. 2015 21;10(10): 101. Duchaine B, Yovel G, Nakayama K. No global processing deficit in the e0141229. Available from: https://doi.org/10.1371/journal.pone.0141229. Navon task in 14 developmental prosopagnosics. Soc Cogn Affect Neurosci. 119. Greenberg DM, Warrier V, Allison C, Baron-Cohen S. Testing the 2007;2(2):104–13. empathizing–systemizing theory of sex differences and the extreme male 102. Valentine T. Upside-down faces: a review of the effect of inversion upon brain theory of autism in half a million people. Proc Natl Acad Sci [Internet]. face recognition. Br J Psychol [Internet]. 1988;79(4):471–491. Available from: 2018;115(48):12152–7 Available from: http://www.pnas.org/lookup/doi/10.1 http://doi.wiley.com/10.1111/j.2044-8295.1988.tb02747.x. 073/pnas.1811032115. 103. Duchaine B, Germine L, Nakayama K. Family resemblance: ten family 120. Shah P, Gaule A, Sowden S, Bird G, Cook R, Cook R. The 20-item prosopagnosia members with prosopagnosia and within-class object agnosia. Cogn index (PI20): a self-report instrument for identifying developmental Neuropsychol [Internet]. 2007;24(4):419–30 Available from: http://www. prosopagnosia. R Soc Open Sci [Internet]. 2015;2(140343):1–11. Available from:. tandfonline.com/doi/abs/10.1080/02643290701380491. https://doi.org/10.1098/rsos.140343. 104. Bobak AK, Parris BA, Gregory NJ, Bennetts RJ, Bate S. Eye-movement 121. Gray KLH, Bird G, Cook R. Robust associations between the 20-item strategies in developmental prosopagnosia and “super” face recognition. Q prosopagnosia index and the Cambridge Face Memory Test in the general J Exp Psychol [Internet]. 2017;70(2):201–17 Available from: http://journals. population. R Soc Open Sci [Internet]. 2017 Mar;4(3):160923. Available from: sagepub.com/doi/10.1080/17470218.2016.1161059. https://royalsocietypublishing.org/doi/10.1098/rsos.160923. 105. Garrido L, Duchaine B, Nakayama K. Face detection in normal and 122. Livingston LA, Shah P. People with and without prosopagnosia have insight prosopagnosic individuals. J Neuropsychol [Internet]. 2008;2(1):119–40 into their face recognition ability. Q J Exp Psychol. 2017:1–3. Available from: http://doi.wiley.com/10.1348/174866407X246843. 123. Rubino C, Corrow SL, Corrow JC, Duchaine B, Barton JJS. Word and text 106. Baron-Cohen S, Wheelwright S, Hill J, Raste Y, Plumb I. The “Reading the processing in developmental prosopagnosia. Cogn Neuropsychol [Internet]. Mind in the Eyes” test revised version: a study with normal adults, and 2016;33(5–6):315–28 Available from: https://www.tandfonline.com/doi/full/1 adults with Asperger syndrome or high-functioning autism. J Child Psychol 0.1080/02643294.2016.1204281. Psychiatry [Internet]. 2001;42(2):241–51 Available from: https://onlinelibrary. 124. Biotti F, Cook R. Impaired perception of facial emotion in developmental wiley.com/doi/abs/10.1111/1469-7610.00715. prosopagnosia. Cortex [Internet]. 2016;81:126–36. Available from: https://doi. 107. Fernández-Abascal EG, Cabello R, Fernández-Berrocal P, Baron-Cohen S. org/10.1016/j.cortex.2016.04.008. Test-retest reliability of the ‘Reading the Mind in the Eyes’ test: a one-year 125. Barton JJS, Corrow SL. The problem of being bad at faces. follow-up study. Mol Autism [Internet]. 2013;4(1):33. Available from: http:// Neuropsychologia [Internet]. 2016;89:119–24. Available from:. https://doi.org/ molecularautism.biomedcentral.com/articles/10.1186/2040-2392-4-33. 10.1016/j.neuropsychologia.2016.06.008. 108. Baron-Cohen S, Bowen DC, Holt RJ, Allison C, Auyeung B, Lombardo M V, 126. Davis MH. A multidimensional approach to individual differences in et al. The “Reading the Mind in the Eyes” test: complete absence of typical empathy. J Pers Soc Psychol [Internet]. 1983;44(1):113–26. Available from: sex difference in ~400 men and women with autism. Yamasue H, editor. http://www.uv.es/~friasnav/Davis_1980.pdf%5Cn. http://www.uv.es/~ PLoS One [Internet]. 2015 27;10(8):e0136521. Available from: http://dx.plos. friasnav/Davis_1980.pdfhttp://content.apa.org/journals/psp/44/1/113%5Cn. org/10.1371/journal.pone.0136521. 127. Bagby RM, Parker JDA, Taylor GJ. The twenty-item Toronto Alexithymia 109. Losh M, Adolphs R, Poe MD, Couture S, Penn D, Baranek GT, et al. scale—I. Item selection and cross-validation of the factor structure. Neuropsychological profile of autism and the broad autism phenotype. Arch J Psychosom Res [Internet]. 1994 Jan;38(1):23–32. Available from: https:// Gen Psychiatry [Internet]. 2009 1;66(5):518. Available from: http://archpsyc. linkinghub.elsevier.com/retrieve/pii/0022399994900051. jamanetwork.com/article.aspx?doi=10.1001/archgenpsychiatry.2009.34. 128. Parker JD, Taylor GJ, Bagby RM. The 20-Item Toronto Alexithymia Scale. 110. Holt RJ, Chura LR, Lai MC, Suckling J, Von Dem Hagen E, Calder AJ, et al. J Psychosom Res [Internet]. 2003;55(3):269–75 Available from: https:// “Reading the Mind in the Eyes”: an fMRI study of adolescents with autism linkinghub.elsevier.com/retrieve/pii/S0022399902005780. Minio-Paluello et al. Molecular Autism (2020) 11:81 Page 16 of 16 129. Bird G, Cook R. Mixed emotions: the contribution of alexithymia to the 151. Kanat M, Spenthof I, Riedel A, Elst LT Van, Heinrichs M, Domes G. Restoring emotional symptoms of autism. Transl Psychiatry [Internet]. 2013;3(7):e285. effects of oxytocin on the attentional preference for faces in autism. 2017; Available from:. https://doi.org/10.1038/tp.2013.61. 7(4):e1097-e1098. Available from: https://doi.org/10.1038/tp.2017.67. 130. Field A. Discovering Statistics Using SPSS. Third. Los Angeles: SAGE; 2009. v–821. 152. Alvares GA, Quintana DS, Whitehouse AJO. Beyond the hype and hope: critical 131. Field AP, Wilcox RR. Robust statistical methods: a primer for clinical psychology considerations for intranasal oxytocin research in autism spectrum disorder. Autism and experimental psychopathology researchers. Behav Res Ther [Internet]. Res [Internet]. 2017;10(1):25–41 Available from: http://doi.wiley.com/10.1002/aur.1692. 2017;98:19–38. Available from:. https://doi.org/10.1016/j.brat.2017.05.013. 153. Bakermans-kranenburg MJ, Van Ijzendoorn MH. Sniffing around oxytocin: 132. Hayes AF. PROCESS: a versatile computational tool for observed variable review and meta-analyses of trials in healthy and clinical groups with mediation, moderation, and conditional process modeling. Unpubl Manuscr implications for pharmacotherapy. Transl Psychiatry [Internet]. 2013;3(5): [Internet]. 2012;1–39. Available from: http://www.afhayes.com/public/ e258–14 Available from: https://doi.org/10.1038/tp.2013.34. process2012.pdf. 154. Walum H, Waldman ID, Young LJ. Statistical and methodological 133. Gotham K, Pickles A, Lord C. Standardizing ADOS Scores for a measure of severity considerations for the interpretation of intranasal oxytocin studies. Biol in autism spectrum disorders. J Autism Dev Disord [Internet]. 2009;39(5):693–705 Psychiatry [Internet]. 2016;79(3):251–7 Available from: https://linkinghub. Available from: http://link.springer.com/10.1007/s10803-008-0674-3. elsevier.com/retrieve/pii/S0006322315005223. 134. Duchaine B, Murray H, Turner M, White S, Garrido L. Normal social cognition 155. Schelinski S, Borowiak K, von Kriegstein K. Temporal voice areas exist in in developmental prosopagnosia. Cogn Neuropsychol. 2009;26(7):620–34. autism spectrum disorder but are dysfunctional for voice identity 135. de la Torre-Ubieta L, Won H, Stein JL, Geschwind DH. Advancing the understanding recognition. Soc Cogn Affect Neurosci [Internet]. 2016;11(11):1812–22 of autism disease mechanisms through genetics. Nat Med [Internet]. 2016;22(4):345– Available from: https://academic.oup.com/scan/article/11/11/1812/2514636. 61 Available from: http://www.nature.com/articles/nm.4071. 156. Endevelt-Shapira Y, Perl O, Ravia A, Amir D, Eisen A, Bezalel V, et al. Altered responses 136. Bourgeron T. From the genetic architecture to synaptic plasticity in autism spectrum to social chemosignals in autism spectrum disorder. Nat Neurosci [Internet]. 2018; disorder [Internet]. Vol. 16, Nature Reviews Neuroscience. Nature Publishing Group; 21(1):111–22 Available from: https://doi.org/10.1038/s41593-017-0024-x. 2015. p. 551–563. Available from: https://doi.org/10.1038/nrn3992. 157. Oettl L-L, Ravi N, Schneider M, Scheller MF, Schneider P, Mitre M, et al. 137. Kendler KS, Neale MC. Endophenotype: a conceptual analysis. Mol Psychiatry Oxytocin enhances social recognition by modulating cortical control of [Internet]. 2010;15(8):789–97 Available from: http://www.nature.com/articles/ early olfactory processing. Neuron [Internet]. 2016;90(3):609–21 Available mp20108. from: https://linkinghub.elsevier.com/retrieve/pii/S0896627316300241. 138. Constantino JN. Deconstructing autism: from unitary syndrome to 158. Grinevich V, Stoop R. Interplay between oxytocin and sensory systems in contributory developmental endophenotypes. Int Rev Psychiatry [Internet]. the orchestration of socio-emotional behaviors. Neuron [Internet]. 2018 Sep; 2018;30(1):18–24 Available from: https://www.tandfonline.com/doi/full/10.1 99(5):887–904. Available from: https://doi.org/10.1016/j.neuron.2018.07.016. 080/09540261.2018.1433133. 139. Peterson MF, Eckstein MP. Looking just below the eyes is optimal across Publisher’s Note face recognition tasks. Proc Natl Acad Sci [Internet]. 2012;109(48):E3314–23 Springer Nature remains neutral with regard to jurisdictional claims in Available from: http://www.pnas.org/cgi/doi/10.1073/pnas.1214269109. published maps and institutional affiliations. 140. Jones W, Klin A. Attention to eyes is present but in decline in 2–6-month- old infants later diagnosed with autism. Nature [Internet]. 2013;504(7480): 427–31 Available from: http://www.nature.com/articles/nature12715. 141. Elsabbagh M, Mercure E, Hudry K, Chandler S, Pasco G, Charman T, et al. Infant neural sensitivity to dynamic eye gaze is associated with later emerging autism. Curr Biol [Internet]. 2012;22(4):338–42 Available from: https://linkinghub.elsevier.com/retrieve/pii/S0960982211014692. 142. DeGutis J, Cohan S, Mercado RJ, Wilmer J, Nakayama K. Holistic processing of the mouth but not the eyes in developmental prosopagnosia. Cogn Neuropsychol [Internet]. 2012;29(5–6):419–46 Available from: https://www. tandfonline.com/doi/full/10.1080/02643294.2012.754745. 143. Royer J, Blais C, Charbonneau I, Déry K, Tardif J, Duchaine B, et al. Greater reliance on the eye region predicts better face recognition ability. Cognition [Internet]. 2018;181(August):12–20. Available from. https://doi.org/10.1016/j. cognition.2018.08.004. 144. Davis J, Mckone E, Zirnsak M, Moore T, Kearney RO, Apthorp D, et al. Social and attention-to-detail subclusters of autistic traits differentially predict looking at eyes and face identity recognition ability. 2017:191–219. 145. Maurer D, Le Grand R, Mondloch CJ. The many faces of configural processing. Trends Cogn Sci. 2002;6(6):255–60. 146. Christensen DL, Braun KVN, Baio J, Bilder D, Charles J, Constantino JN, et al. Prevalence and characteristics of autism spectrum disorder among children aged 8 years — autism and developmental disabilities monitoring network, 11 Sites, United States, 2012. MMWR Surveill Summ [Internet]. 2018;65(13):1–23 Available from: http://www.cdc.gov/mmwr/volumes/65/ss/ss6513a1.htm?s_cid=ss6513a1_w. 147. Wilkinson DA, Best CA, Minshew NJ, Strauss MS. Memory awareness for faces in individuals with autism. J Autism Dev Disord [Internet]. 2010;40(11): 1371–7 Available from: http://link.springer.com/10.1007/s10803-010-0995-x. 148. Rimmele U, Hediger K, Heinrichs M, Klaver P. Oxytocin makes a face in memory familiar. J Neurosci [Internet]. 2009;29(1):38–42 Available from: http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.4260-08.2009. 149. Auyeung B, Lombardo M V, Heinrichs M, Chakrabarti B, Sule A, Deakin JB, et al. Oxytocin increases eye contact during a real-time , naturalistic social interaction in males with and without autism. 2015;5(2):e507-e506. Available from: https://doi.org/10.1038/tp.2014.146. 150. Davis JM, McKone E, Dennett H, O’Connor KB, O’Kearney R, Palermo R. Individual differences in the ability to recognise facial identity are associated with social anxiety. Baune BT, editor. PLoS One [Internet]. 2011 14;6(12): e28800. Available from: https://dx.plos.org/10.1371/journal.pone.0028800.