Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Facial expression recognition as a candidate marker for autism spectrum disorder: how frequent and severe are deficits?

Facial expression recognition as a candidate marker for autism spectrum disorder: how frequent... Background: Impairments in social communication are a core feature of Autism Spectrum Disorder (ASD). Because the ability to infer other people’s emotions from their facial expressions is critical for many aspects of social communication, deficits in expression recognition are a plausible candidate marker for ASD. However, previous studies on facial expression recognition produced mixed results, which may be due to differences in the sensitivity of the many tests used and/or the heterogeneity among individuals with ASD. To ascertain whether expression recognition may serve as a diagnostic marker (which distinguishes people with ASD from a comparison group) or a stratification marker (which helps to divide ASD into more homogeneous subgroups), a crucial first step is to move beyond identification of mean group differences and to better understand the frequency and severity of impairments. Methods: This study tested 46 individuals with ASD and 52 age- and IQ-matched typically developing (TD) participants on the Films Expression Task, which combines three key features of real-life expression recognition: naturalistic facial expressions, a broad range of simple and complex emotions, and short presentation time. Test- retest reliability was assessed in 28 individuals who did not participate in the main study and revealed acceptable reliability (ICC r = .74). − 10 Results: Case-control comparisons showed highly significant mean group differences for accuracy (p = 1.1 × 10 ), with an effect size (Cohen’s d = 1.6), more than twice as large as the mean effect size reported in a previous meta- analysis (Uljarevic and Hamilton, 2012, J Autism Dev Disord). The ASD group also had significantly increased mean reaction times overall (p = .00015, d = .83) and on correct trials (p = .0002, d = .78). However, whereas 63% of people with ASD showed severe deficits (they performed below two standard deviations of the TD mean, a small subgroup (15.3%) performed normally (within one standard deviation of the mean). Conclusion: These findings indicate that the majority of people with ASD have severe expression recognition deficits and that the Films Expression Test is a sensitive task for biomarker research in ASD. Future work is needed to establish whether ASD subgroups with and without expression recognition deficits differ from one another in terms of their symptom profile or neurobiological underpinnings. Keywords: Autism Spectrum disorder, Facial expression recognition, Biomarker * Correspondence: [email protected] Sackler Institute for Translational Neurodevelopment, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, De Crespigny Park, Denmark Hill, London SE5 8AF, UK Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 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. Loth et al. Molecular Autism (2018) 9:7 Page 2 of 11 Background displayed with low intensities [14] using morphs [15, 16]. Autism spectrum disorder (ASD) is a life-long neurodeve- Furthermore, speed-accuracy trade-offs [17], abnormal lopmental disorder, behaviourally defined by impairments gaze fixation patterns [18], and differences in underlying in social communication and the presence of repetitive neural processes as measured by EEG [19] and fMRI have and restricted behaviours and interests [1]. The consider- also been reported [20, 21]. However, some inconsisten- able variability in the quality and severity of symptoms cies remain even within the different methodologies and between individuals with ASD is widely recognized [2]. sample types. For example, some studies have found mean For example, social communicative impairments can be group deficits in adult ASD samples with IQ in the normal manifested in a range of difficulties in social-emotional range on tasks that required labeling basic emotions with reciprocity, in non-verbal communicative behaviours, and unlimited presentation times [20, 22, 23], while others in developing and maintaining social relationships [1]. reported no deficits in the recognition of complex Recognition of this heterogeneity has begun to cast doubt emotions [24]. as to whether a truly diagnostic biological or behavioural Recently, a formal meta-analysis of 48 papers con- marker for ASD exists (which differentiates all or the cluded that there is an emotion recognition difficulty in majority of people with ASD from typically developing “autism”, with a mean effect size of Cohen’s D = 0.80, yet individuals or those with other neurodevelopmental/psy- this was estimated to decrease to around 0.40 after chiatric conditions) and prompted an increasing interest adjusting for publication biases [6]. Hence, it remains in the identification of stratification markers to parse ASD unclear whether differences in study findings reflect into more homogeneous subgroups [3]. Biomarkers variability in the nature or sensitivity of the many behav- have been defined as “a characteristic that is object- ioural tests used to assess expression recognition in ASD ively measured and evaluated as an indication of and/or differences in the severity of deficits among people normal biological processes, pathogenic processes, or with ASD. pharmacologic responses to a therapeutic interven- To ascertain the value of expression recognition defi- tion” [4]. Here we use the term “biomarker” in a broad cits as a candidate diagnostic or stratification biomarker sense to refer to measures of any modality, including we need, first, sensitive tests that incorporate several cognitive/behavioural tests [4]. Stratification bio- real-life characteristics. For example, in daily life, facial markers are then characteristics that vary between expressions are usually more subtly displayed than those ASD subgroups and that map onto differences in their depicted in most standard stimuli of “prototypical” facial symptom presentation, etiology, need for particular emotion expressions. People also often reveal their emo- treatments, and/or treatment response. tions for only a very brief time. The observer is then re- Because the ability to infer other people’semotions from quired to identify the facial emotion expression quickly their facial expressions is critical for many aspects of social to react appropriately to the person’s feelings. Second, communication, deficits in expression recognition have the field needs to move away from a sole focus on mean long been suggested to represent a core impairment in between-group differences to better understand the fre- ASD [5]. However, over the past three decades, behav- quency and severity of expression recognition deficits ioural studies of facial expression recognition in ASD have among individuals with ASD. For example, in case- produced mixed findings, ranging from reports of pro- control studies, a significant p value (especially in com- found deficits to apparently intact expression recognition bination with small or medium effect sizes) could either skills (see e.g., [6, 7] for recent reviews). A review of this reflect (small) deficits in most cases (say, around 1 SD literature suggests that in ASD, the presence and severity below the typically developing (TD) mean) or may be of expression recognition deficits on experimental tests is driven by a subgroup of individuals with severe deficits. influenced by both participant characteristics—age and Obviously these two scenarios have profoundly different ability level—and task requirements [7]. On the whole, implications for the potential utility of the test in clinical deficits in recognizing basic emotion expressions ap- practice. pear to be predominantly found in children and low- Hence, the aim of this study was to investigate the functioning individuals (i.e., people with ASD who also frequency and severity of deficits in the recognition of have intellectual disabilities) [5, 8]. By contrast, high- facial expressions of emotion using a task that is more functioning individuals with ASD (with IQ in the normal sensitive and naturalistic than previous tests. The range) usually perform well on tests that depict facial ex- Films Expression Task [25] uses still images captured pressions in a prototypical manner and that used relatively from movie scenes and combines three elements that long presentation times [9–11]. However, some studies appear to be challenging for people with ASD: depic- reported group-level deficits in recognizing expressions of tion of naturalistic facial expressions, inclusion of a complex emotions (e.g., guilt, defiance) [9, 12], expressions range of both basic and complex emotions, and brief that were presented only briefly [13] or subtle expressions presentation times. Loth et al. Molecular Autism (2018) 9:7 Page 3 of 11 Methods included ten new participants and employed the The Films Expression Task intended task procedure in which three pictures were The Films Expression Task [25] consists of 58 trials. In presented sequentially, one target and two distractors. each trial, participants were first presented with an ad- Only trials with equal or less than 30% errors were jective describing an emotional state (e.g., confident, carried forward, which resulted in 58 trials with tar- pleased). They were then briefly shown three images one gets taken from 15 films. after the other (500 ms each, with a 500 ms blank screen between images). The images were taken from films IQ measures made in non-English-speaking countries to decrease the Verbal, performance, and full-scale IQ were assessed probability that participants had seen them or were using the Wechsler Abbreviated Scale of Intelligence- familiar with the actors. 2nd edition (WASI-II; Wechsler, [26]) or a four-subtest Within each trial, images present the same actor or short-form of the Wechsler Adult Intelligence Scale- actress, but with different emotional expressions. Partici- Third Edition (WAIS-III; [27]). In both instances, the tests pants were asked to indicate, by key press, which of the included two verbal subscales (vocabulary, similarities) images best matched the target word. In 14 trials, the and two non-verbal subscales (block design, matrix rea- target emotion was a basic emotion (happy, angry, sad, soning). Standardized scores from the WASI-II and WAIS afraid, surprised, disgusted). In the remaining 44 trials, are comparable. the target emotion was complex (e.g., mocking, hurt, disappointed, resentful, see Additional file 1 for the list Autism Spectrum Quotient of target emotions). In trials with both basic and com- TheAutismSpectrumQuotient(AQ)[28]isa self- plex target emotions, the foils were selected to be similar report questionnaire to assess whether adults of to the targets in terms of perceptual features and inten- average intelligence have symptoms associated with sity of the expression (see supplementary information autism spectrum disorder. The test consists of 50 [25]). Basic vs. complex emotion trials were presented statements and participants indicate whether they interleaved, in a fixed-random order. Participants were “definitely agree”, “slightly agree”, “slightly disagree”, instructed to respond as quickly and accurately as or “definitely disagree” with each statement. Approxi- possible. mately half the items are worded to usually elicit an Immediately before the task, participants were pre- “agree” response from neurotypical individuals and sented with the definitions for each word describing a half to usually elicit a “disagree” response. The ques- target emotion (e.g., guilty). They were encouraged to tions cover five different domains associated with the ask questions if they were unfamiliar with or did not autism spectrum: social skills, communication skills, understand any of the words. Participants were also imagination, attention to detail, and attention switching/ allowed to review these definitions at the start of each tolerance of change. trial, after presentation of the target word. Task development and validation of the stimuli is Test-retest reliability of the films expression test described in [25] (supplementary information). Briefly, Thirty-one participants took part in a test-retest study to in pilot phase 1, three researchers suggested adjectives assess the reliability of the Films Expression Test. We to describe 122 pictures of facial expressions taken from aimed to recruit participants with a range of expression 18 films. In pilot phase 2, 32 native English speakers recognition abilities to ensure that test-retest reliability rated how well each adjective described the facial ex- on this task was stable at both high and low perform- pression on a scale from 1 to 5 (1 = “it doesn’t match at ance scores. Because we hypothesized that individuals all”,5= “it matches very well”). Adjectives were only with ASD would have lower accuracy scores than typically carried forward if they had received at least a mean developing individuals, four participants were recruited rating of 3.5 or a median rating of at least 4. Ten of from an autism support group. Hence, 27 participants these participants then also rated the intensity of the were typically developing, three had a formal diagnosis of target expression from 1 to 5 (1 = “not intense” to 5 = autism, and one was suspected of having autism. “very intense”) as well as the intensity of the expressions Two participants exceeded the intended maximum re- intended as foils. Targets and foils were then matched test interval of 6 weeks and did not complete the retest, based on their expression intensity. In pilot phase 3, 30 one participant’s retest data was not available after a different native English speakers were asked to choose technical error. This left 28 participants (20 females and one of two simultaneously presented expressions (the 8 males, mean age = 33.48 years; range = 20–57 years, target and a foil) that best matched an adjective. Distrac- four from the autism support group). Out of the 28 tors were only included if participants had selected them participants, English was a second language for six but less than 30% of the time. A fourth final pilot phase all participants described themselves as fluent English Loth et al. Molecular Autism (2018) 9:7 Page 4 of 11 speakers. The retest interval ranged from 14 to 34 days, (male, 32 years, VIQ = 92, first language Uzbek) and was with an average interval of 18.11 days. excluded from the analyses reported below. This left 46 All test-retest participants also completed a two-subtest people with ASD and 52 people with TD. The study was (vocabulary and matrix reasoning) short-form of the approved by the South London and Maudsley NHS Wechsler Abbreviated Scale of Intelligence (WASI-II; Trust ethics committee and the UCL Research Ethics [26]). The mean two-scale IQ was 113.07 (range 87–146). Committee. All participants or a parent in the case of Test-retest analysis revealed an intra-class correl- participants who were minors gave informed written ation coefficient of ICC = .74, indicating acceptable consent before study participation. reliability. The mean accuracy score at time 1 was 86.64% (SD = 7.49, range 62.07–98.28%) and at time 2 Results was also 86.64% (SD = 7.93, range 62.07–94.83%). Indi- Case-control comparisons vidual difference scores, calculated between time 1 Descriptive statistics are provided in Additional file 2. and time 2 performance, showed that 17 (60.71%) par- The ASD group accurately identified an average of ticipants had a difference score less than .5 SDs, 6 70.8% (SD = 13.5) of the briefly presented emotion ex- (21.43%) were between .5 and 1 SD, and 5 were pressions, compared to 87.5% (SD = 5.5) in the control (17.86%) greater than 1 SD. group (see Fig. 1a). The Welch test, which does not assume equal variances, showed that this mean group Participants difference was highly significant (t(58.1) = − 7.8, p = In the main study, participants consisted of 46 adoles- −10 1.1 × 10 ). The effect size (Cohen’s d) was 1.62. We cents or adults with ASD (34 male, 12 female) and 53 used bootstrapping in Matlab version 8.3.0.532 (Math- typically developing (TD) individuals (33 male, 20 works, Natick, MA) to estimate the 95% confidence interval female). Eighteen TD participants were tested as part of around this point estimate. We drew 10,000 resamples a previous study [25]. Although the ASD group included from the original sample using random sampling with a higher proportion of males than the TD group, the dif- replacement. We obtained a distribution of effect sizes for ference in the sex ratio between the groups was not sta- all the resamples, and using the percentile method (i.e. tistically significant (χ = 1.5, p = .21). Mean age of the using the 2.5 and 97.5 percentiles as lower and upper ASD group was 30.2 years (SD 9.4, 15–50 years) and of bounds of the CI), the 95% CI was [1.30 2.05]. the TD group was 27.5 years (SD 7.8, range 14–55 years). Analyses of reaction times (RT) revealed that the Two participants in the ASD group (15 and 17 years) ASD group also responded significantly more slowly and two participants in the TD group (15 and 14 years) (M =1459 ms, SD = 899) than the TD group (M =878 ms, were younger than 18 years. IQ data was available for 42 SD = 414) across all trials (t(61.5) = 4.02, p = .00015, out of 46 participants with ASD and all participants in the Cohen’s d = .83, 95% CI [0.49 1.19]) as well as on their cor- TD group. Mean verbal and full-scale IQ in the ASD group rect trials (ASD: M =1279 ms, SD =812; TD: M = 767 ms, were 113.9 (range 85–140) and 116.0 (range 87–135), re- SD =352) (t(df = 59.7) = 3.96, p =.0002, Cohen’s d =0.82, spectively, and in the TD group were 114.0 (range 74–146) 95% CI [0.49 1.17]). These findings remain unchanged and 115.5 (range 85–143). The groups did not significantly when adolescents below the age of 18 years were excluded differ from each other in terms of age (t(86) = 2.3, p =.12), (accuracy: ASD = 70.2%, TD = 87.7%; t(53.5) = 7.9, p = full-scale IQ (t(86) = .02, p = .87), verbal IQ (t(86) = .0004, −10 1.4 × 10 )(see Fig.1b). p =.98), or performance IQ (t(86) = .002, p = .96). The four individuals with ASD for whom no IQ data were available had attended mainstream schools, which suggests that Sex differences their IQ was within the normal range. Given previous reports of sex differences on several All participants with ASD except one were native social-cognitive tasks in the typical population and in English speakers. In the TD group, 37 of 53 participants ASD [29], we performed 2 (group) × 2 (sex) ANOVAs to were native English speakers. All TD participants for test whether accuracy, overall RT, or RT on correct trials whom English was not their first language described differed between males and females overall or in either their current English language level as “fluent”. Fourteen group. These tests confirmed the significant effects of −12 individuals performed within 1 SD of the TD mean or group on accuracy (F(1,97) = 65.0, p = 2.3 × 10 ) and above, one individual below 1 SD, and two individual RTs (all ps < .00007) but there were no significant effects below 2 SDs of the TD mean. Of these two TD partici- of sex (all ps > .5)) or group by sex interactions (all ps> pants, one had an above-average verbal IQ (VIQ) (124), .11). This finding indicates that the overall case-control indicative of adequate language comprehension and was difference cannot be attributed to differences in the therefore not excluded. The other TD participant had an slightly higher male:female ratio in the ASD than those accuracy score of more than 4 SDs below the TD mean in the control group. Loth et al. Molecular Autism (2018) 9:7 Page 5 of 11 Fig. 1 a Accuracy: percentage of correct trials, by group. b Mean reaction time of correct trials and mean reaction time overall, by group Basic vs. complex emotions significant relationship between accuracy scores and FIQ Next, we explored whether in the ASD group, the ex- (r(52) = .26, p = .057), such that participants with higher pression recognition impairments reported above were intelligence also had higher expression recognition driven by specific problems in identifying complex scores. emotional expressions. To do so, we separately ana- lysed trials with simple vs. complex target emotions Frequency and severity of expression recognition deficits (see Additional file 1). in ASD A 2 (group) × 2 (emotion category) repeated-measures To establish the frequency and severity of expression ANOVA revealed the above reported significant main ef- recognition deficits in the ASD group, we calculated −13 fect of group (F(1,96) = 70.7, p =3.8 ×10 ) and a signifi- how far the accuracy score for each individual with ASD cant effect of emotion category (F(1,96) = 5.9, p =.016) on deviated from the TD group mean. Analyses of the response accuracy, such that on average, both groups were scores showed that 63% of people with ASD performed better at recognizing simple than complex emotions. The more than 2 SDs below the TD mean. Of those, 23.9% group x emotion category interaction was not significant performed between 2 and 3 SDs below the TD mean F(1,96) = .14, p = .7) (see Fig. 2). and 39.1% performed more than 3 SDs below the TD mean (see Fig. 3). 21.7% of people with ASD performed Correlations between accuracy and RT scores with age between 1 to 2 SDs below the TD mean while 15.3% had and IQ accuracy scores within the TD range. The 17-year-old Next, we tested whether age, verbal, performance, or full- with ASD performed within the 1 SD of the TD range, scale IQ were related to accuracy or reaction time (RT while the 15-year-old with ASD performed between 2 overall or on correct trials) on the Films Expressions Test. and 3 SDs below the TD range. As shown in Fig. 4, in In the ASD group, we found no significant relationships the ASD group, accuracy and RT scores were negatively between expression recognition variables and age or IQ correlated such that individuals with ASD who had (all p > .09). In the TD group, there was a marginally higher accuracy scores were also able to correctly recognize the emotion faster than those with low accur- acy scores. Therefore, there was no evidence of speed- accuracy trade-off for most of the ASD participants who performed accurately on the task. These results suggest that the majority (63%) of individuals with ASD had severe deficits, while a small subgroup of around 15% of people with ASD showed intact expression recognition skills on this test. Correlations between accuracy and RT scores and ASD symptoms Finally, we investigated correlations with ASD symptoms in a subset of N = 53 participants (ASD = 25, TD = 28) for whom AQ data was available (see Fig. 5). When both Fig. 2 Percentage of correct trials, by emotion category and group groups were collapsed, we found a significant negative Loth et al. Molecular Autism (2018) 9:7 Page 6 of 11 Fig. 3 Upper panel: Distribution of TD performance on percentage of accuracy. Lower panel: Distribution ofASD performance on percentage of accuracy. Grey bars denote − 2 to + 2 SDs of the TD means on percentage of correct expression recognition correlation between the percentage of correct responses pattern was found in the ASD group, it did not reach on the Films Expression Task and higher ASD symptoms statistical significance (ASD r(25) = .29, p = .16). r(53) = − .48, p = .0002. However, when the ASD and TD Likewise, for overall RTs, we found a significant rela- groups were considered separately, there were no signifi- tionship with ASD symptoms when both groups were cant relationships in either group (ASD r(25) = .09, p =.66, collapsed (RT : overall: r(53) = .54, p = .000026). overall TD r(28) = .03, p = .88). This suggests that the negative When the TD and ASD groups were analysed separately, correlation for the overall sample could be explained by similar but non-significant patterns were found (TD: the performance differences between the two groups. r(28) = .31, p = .10, ASD: (r(25) = .29, p = .16). A somewhat different picture emerged for RTs. We found that longer reaction times on correct trials were Discussion associated with more ASD symptoms. This was signifi- This study investigated the frequency and severity of im- cant overall, i.e., when both groups were collapsed pairments in facial expression recognition in ASD using (RT : overall r(53) = .53, p = .00003) and in the a test that captured several features of expression recog- correct trials TD group (TD r(28) = .48, p = .012). Although a similar nition in daily life. A test-retest study showed reliability of .74, which is considered adequate for a test to be of use in clinical settings [30, 31]. At the group level, we found highly significant differences in accuracy and re- sponse times between a group of high-functioning adults and adolescents with ASD and an age- and IQ-matched TD control group. Effect sizes of the accuracy scores were more than twice as large as those estimated in a re- cent meta-analysis of studies on emotion recognition in ASD and more than four times larger than estimates that accounted for publication biases [6]. As a first step towards ascertaining whether deficits in expression recognition may serve as a diagnostic or stratification marker for ASD, we investigated the fre- quency and severity of these deficits. This revealed that 63% of people with ASD had severe deficits, which would be expected to create substantial social communi- cative difficulties. For example, participants who per- Fig. 4 Scatterplot showing percentage of accuracy and RT on formed below 3 SDs of the TD mean failed to identify correct trials, by group. Grey bars denote − 2to + 2 SDs of the TD the briefly displayed target emotion expression in 3–7 means on percentage of correct expression recognition out of 10 trials. Translated into real-life settings, it is Loth et al. Molecular Autism (2018) 9:7 Page 7 of 11 Fig. 5 Scatterplot showing the relationship between AQ scores and a accuracy scores, b overall RTs, and c RTs for correct responses. Regression lines are plotted overall, and for the ASD and TD groups separately. Grey bars denote − 2 to + 2 SDs of the TD means on percentage of correct expression recognition, RTs, and RTs on correct trials easy to see how this may impact the ability to, for ex- recognition deficits and ASD symptom severity, as mea- ample, modulate a conversation or to respond empathic- sured by the AQ. When the ASD and TD groups were ally [32]. However, we also found considerable variability combined, we found both lower accuracy in expression rec- in that 21.7% performed between 1 to 2 SDs below the ognition and higher RTs to be moderately related to greater TD mean and 15.3% evidenced expression recognition severity of autism traits. However, whereas accuracy scores skills indistinguishable from TD individuals (within 1 SD were not sensitive to the severity of ASD symptoms within of the TD mean) as indexed by the combination of both the ASD and TD groups, respectively, longer reaction times accuracy and RT scores. This variability in expression on correct trials were significantly related to greater ASD recognition skills among individuals with ASD was unre- symptoms in the TD group. A similar relationship was lated to age or IQ (verbal or non-verbal). However, it found in the ASD group with moderate correlation coeffi- should be borne in mind that the IQ range in the cients which, however, did not reach statistical significance. current sample was restricted to the normal range, so These correlation analyses should be viewed as preliminary, that this finding may not generalize to individuals with becauseAQscoreswereonlyavailablein abouthalf ofthe ASD and intellectual disabilities. participants. As a consequence, the sample sizes were small The present findings are broadly consistent with other re- when split by group, which reduced our power to detect a cent efforts to parse heterogeneity in ASD. In particular, significant effect. In addition, the AQ is a composite meas- Lombardo and colleagues [33] used unsupervised hierarch- ureofarangeofautism-related traits, including features ical clustering approaches to identify ASD subgroups on such as intolerance of changes that may not be expected to the basis of item-level performance on the Reading the relate to expression recognition. This may have diluted po- Mind in theEyes Test(RMET).TheRMETisawidely used tentially higher associations with more specific social com- mentalising task that also involves a strong emotion recog- municative impairments. nition component [34] as it requires participants to identify In sum, our findings revealed that the majority of the mental or emotional state of a person from their eye re- individuals with ASD had severe deficits in expression gion only. The study (which comprised a discovery and recognition deficits but also that a sub-sample of indi- replication cohort) identified five ASD subgroups and four viduals with ASD had no behavioural impairments on TD subgroups. For individuals with ASD, 45–62% showed this task. This indicates that expression recognition may what the authors termed “clear to immediate impairments”, more likely serve as a stratification as opposed to diag- while two subgroups (19–36%) performed more than 2 and nostic marker, subject to further substantiation that sub- up to 11 SDs below the TD means - depending on the TD groups with/without expression recognition deficits also subgroup that was used as comparison. It would be valu- differ in symptom severity or adaptive behavior. One able for future studies to directly compare performance on implication of the current findings is that intervention the RMET and Films Expression Tasks or combine infor- programmes that specifically target expression recogni- mation from both measures to derive a composite score. tion may be valuable for a substantial proportion of A second requirement for a quantitative stratification people with this disorder. marker is to demonstrate clinical relevance, for example, that people who fall within a certain range of deficits differ Study limitations and future directions from those without abnormalities (normal range) in terms Several potential limitations of the current study as well of symptom severity. To begin to address this, we con- as implications for future research should be considered. ducted correlation analyses between severity of expression First, we deliberately chose a task that incorporated Loth et al. Molecular Autism (2018) 9:7 Page 8 of 11 several factors crucial for expression recognition in real- with ASD may also have deficits in theory of mind life situations, including a wide range of simple and [35]. Therefore, future studies should probe compre- complex target emotions, and relatively short presenta- hension of the different emotion words and account tion times. As a consequence, could factors unrelated to for individual differences in understanding particular expression recognition per se have contributed to the words and their emotional state—independent from deficits observed in some of the ASD participants, for recognition of their manifestation in facial example, comprehension of the target words or other expressions. cognitive impairments/ anomalies? Previous studies reported that expression recognition If difficulties in comprehending some of the emotion deficits varied with the type of emotion, such that recog- words accounted for performance impairments, one nition of fear has been found to be more impaired than would expect larger group differences in trials with happiness (see [6] for a review). The naturalistic charac- complex target emotions than basic emotions. However, ter of the task did not enable us to systematically analyse separate analyses, split by emotion target (simple vs com- recognition of different types of emotion expressions plex), showed significant impairments in the ASD group (fear, anger, disappointment, etc.) as 34 different emotion on trials with basic as well as complex emotions. Basic adjectives were included, and many of the target emo- emotion words are typically understood by the age of tions were only used once. However, descriptive trial-by- 6 years or earlier. Also most of the complex emotions trial analyses, split by group (see Fig. 6) suggest that the used here are typically understood by age 8 (see Add- mean percentages of correct response not only varied be- itional file 1), and the ASD participants had verbal tween target emotions but in some instances also within IQs within (and mostly above) the population average. target emotions. For example, among the ASD group However, recognition of complex emotions also re- “angry” and “happy” trials had the highest percentages of quires an understanding of the mental states they de- correct responses whereas trials with the target emotions scribe [34] and some high-functioning individuals “disappointed”, “tentative”,and “disbelieving” had the Fig. 6 Percentage of correct responses, by trial. a ASD group. b TD group Loth et al. Molecular Autism (2018) 9:7 Page 9 of 11 lowest percentages. For “sad” trials, correct recognition version of the Films Expression Test and a companion ranged from 56.5 to 73.9%. The latter finding may be due child version in the follow-up assessment battery of the to the fact that the difficulty level of a trial is influenced by EU-AIMS Longitudinal European Autism Project [47, 48] the nature of the target emotion and also by the expres- to test whether the current findings can be replicated in sions in the distractor stimuli used in a trial. an independent sample more diverse in age and ability Could more general problems in rapid visual process- level, to further determine the clinical usefulness of our ing have contributed to deficits in recognizing relatively findings, and to investigate potential underlying briefly presented (500 ms) facial expressions of emo- mechanisms. tions? This alternative explanation seems unlikely for two reasons. Several previous studies reported no defi- Conclusion cits in rapid visual processing in ASD [13, 36]. For Taken together, our findings show highly significant example, a study that tested recognition of micro-facial mean case-control differences in facial expression expressions, using much faster presentation times (15 recognition, with one of the largest effect sizes ever and 30 ms) than those employed here reported specific reported in the expression recognition literature of deficits in identifying facial expressions, but not objects ASD. This suggests that problems with expression rec- in an ASD group [37]. Moreover, as indicated above, our ognition are more widespreadthancurrently thought, trial-by-trial analyses (see Fig. 6) showed quite variable likely owing to the more naturalistic character of the performance scores between trials in the ASD group, tasks used here. Nevertheless, we also highlight im- with two trials being correctly identified by over 90% of portant variability in expression recognition skills the ASD participants. This pattern of largely intact ex- among individuals with this condition and showed pression recognition on some trials but impairments on that aminorityofpeoplewithASD hadnobehav- others would be inconsistent with general problems in ioural impairments. This finding indicates that the rapid visual processing. However, it remains a possibility Films Expression Test may serve as a valuable tool to that other cognitive abnormalities previously reported in study expression recognition as a candidate stratifica- (some people with) ASD, such as a detail-focused pro- tion marker for ASD. cessing style [38] abnormalities in top-down processing [39] or difficulties with other aspects of face perception [40], may have contributed to behavioural impairments Additional files in expression recognition. Additional file 1: Target emotion words, split by age of acquisition The present finding raises the question of what may ac- (AoA) norms. (DOCX 15 kb) count for the differences between people with ASD with/ Additional file 2: Descriptive statistics and contrasts for main variables without expression recognition deficits. This question can- of interest. (DOCX 19 kb) not be settled here but we discuss potential factors and outline some avenues for future research. A likely possibil- Abbreviations ity is that (some) people with ASD have problems in the AQ: Autism Spectrum Quotient; ASD: Autism Spectrum Disorder; neurocognitive mechanisms that represent facial expres- IQ: Intelligence Quotient; RT: Reaction time; SD: Standard deviation; sion information and/or multi-modal expression informa- TD: Typically developing; WAIS: Wechsler Adult Intelligence Scales; WASI: Wechsler Abbreviated Scales of Intelligence tion. One hypothesis that recently gained prominence is the notion that expression recognition deficits in ASD Acknowledgements may be linked to the high frequency of comorbid alexithy- We thank all participants and their families for their help with this study, as mia [41, 42], estimated to affect around 50% of people well as Raka Tavashmi, Hannah Hayward, Daisy Crawley, Jessica Sabet, Jessica Faulkner, Claire Ellis, and Antonia San Jose Caceres for their help with with ASD [43, 44]. Alexithymia is a sub-clinical trait char- volunteer recruitment and/or assessment. acterized by difficulties in identifying and describing one’s own emotional state [45]. Future work—ideally with larger Funding samples—will be needed to ascertain whether some or all This study was supported by an Economic and Social Research Council individuals with ASD who exhibit substantial expression (ESRC) project grant (RES-000- 22-1123) and by EU-AIMS (European Autism Interventions), which receives support from the Innovative Medicines Initia- recognition deficits also have higher rates of trait alexithy- tive Joint Undertaking under grant agreement no. 115300, the resources of mia. Other potential underlying mechanisms may include which are composed of financial contributions from the European Union’s differences in attention (e.g., to the eye region [11, 46]), or Seventh Framework Programme (grant FP7/2007-2013), from the European Federation of Pharmaceutical Industries and Associations companies’ in-kind brain structural/functional anomalies [20]. To identify the contributions, and from Autism Speaks. factors that underpin expression recognition impairments at an individual level will require multi-modal studies Availability of data and materials and/or designs that assess a range of cognitive functions The data set generated and analysed during this study is available from the within each individual. We have included an abridged corresponding author upon request. Loth et al. Molecular Autism (2018) 9:7 Page 10 of 11 Authors’ contributions 12. Capps L, Yirmiya N, Sigman M. Understanding of simple and complex EL designed the study and wrote the first and final draft. LG designed the emotions in non-retarded children with autism. J Child Psychol Psychiatry. study materials and contributed to the writing of the manuscript. JA carried 1992;33:1169–82. out the test-retest reliability study, contributed to data collection, prepared 13. Clark TF, Winkielman P, McIntosh DN. Autism and the extraction of emotion all figures, and contributed to the writing of the manuscript. EW contributed from briefly presented facial expressions: stumbling at the first step of to data collection and commented on the manuscript. ACD contributed to empathy. Emotion. 2008;8:803–9. the data collection and commented on the manuscript. BD designed the 14. Wingenbach TSH, Ashwin C. Brosnan: diminished sensitivity and specificity study and contributed to the writing of the manuscript. All authors read and at recognising facial emotion expressions of varying intensity underlie approved the final manuscript. emotion-specific recognition deficits in autism spectrum disorders. Res Autism Spectr Disord. 2017;34:52–61. 15. Kennedy DP, Adolphs R. Perception of emotions from facial expressions in Ethics approval and consent to participate high-functioning adults with autism. Neuropsychologia. 2012;50:3313–9. The study was approved by the South London and Maudsley NHS Trust 16. Wang S, Adolphs R. Reduced specificity in emotion judgment in people ethics committee and the UCL Research Ethics Committee. All with autism spectrum disorder. Neuropsychologia. 2017;99:286–95. participants—and in the case of minors—a parent gave informed written 17. Sucksmith E, Allison C, Baron-Cohen S, Chakrabarti B, Hoekstra RA. Empathy consent. and emotion recognition in people with autism, first-degree relatives, and controls. Neuropsychologia. 2013;51:98–105. Consent for publication 18. Pelphrey KA, Sasson NJ, Reznick JS, Paul G, Goldman BD, Piven J. Visual Not applicable scanning of faces in autism. J Autism Dev Disord. 2002;32:249–61. 19. Dawson G, Webb SJ, Carver L, Panagiotides H, McPartland J. Young children Competing interests with autism show atypical brain responses to fearful versus neutral facial The authors declare that they have no competing interests. expressions of emotion. Dev Sci. 2004;7:340–59. 20. Ashwin C, Baron-Cohen S, Wheelwright S, O'Riordan M, Bullmore ET. Differential activation of the amygdala and the ‘social brain’ during fearful Publisher’sNote face-processing in Asperger syndrome. Neuropsychologia. 2007;45:2–14. Springer Nature remains neutral with regard to jurisdictional claims in 21. Wang AT, Dapretto M, Hariri AR, Sigman M, Bookheimer SY. Neural published maps and institutional affiliations. correlates of facial affect processing in children and adolescents with autism spectrum disorder. J Am Acad Child Adolesc Psychiatry. 2004;43:481–90. Author details 22. Macdonald H, Rutter M, Howlin P, Rios P, Le Conteur A, Evered C, Folstein S. Sackler Institute for Translational Neurodevelopment, Institute of Psychiatry, Recognition and expression of emotional cues by autistic and normal Psychology and Neuroscience, King’s College London, De Crespigny Park, adults. J Child Psychol Psychiatry. 1989;30:865–77. Denmark Hill, London SE5 8AF, UK. Department of Forensic and 23. Boraston Z, Blakemore SJ, Chilvers R, Skuse D. Impaired sadness Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and recognition is linked to social interaction deficit in autism. Neuroscience, King’s College London, London, UK. Division of Psychology, Neuropsychologia. 2007;45:1501–10. Department of Life Sciences, Brunel University, London, UK. Department of 24. Buitelaar JK, van der Wees M, Swaab-Barneveld H, van der Gaag RJ. Theory of Psychological and Brain Sciences, Dartmouth College, Hanover, USA. mind and emotion-recognition functioning in autistic spectrum disorders and in psychiatric control and normal children. Dev Psychopathol. 1999;11:39–58. Received: 26 August 2017 Accepted: 10 January 2018 25. Garrido L, Furl N, Draganski B, Weiskopf N, Stevens J, Tan GC, Driver J, Dolan RJ, Duchaine B. Voxel-based morphometry reveals reduced grey matter volume in the temporal cortex of developmental prosopagnosics. Brain. References 2009;132:3443–55. 1. American Psychiatric Association. Diagnostic and statistical manual of 26. Wechsler D. Wechsler Abbreviated Scale of Intelligence - Second Edition mental disorders. Fifth ed. Arlington: American Psychiatric Publishing; 2013. (WASI-II). London: Pearson; 2011. 2. Charman T. Variability in neurodevelopmental disorders: evidence from 27. Wechsler D. Wechsler Adult Intelligence Scale. Third ed. London: Pearson; 1997. autism. In: Riby D, Van Herwegen J, editors. Applied research and key 28. Baron-Cohen S, Wheelwright S, Skinner R, Martin J, Clubley E. The autism- issues in neurodevelopmental disorders. Hove: Routledge Psychology spectrum quotient (AQ): evidence from Asperger syndrome/high- Press; 2015. p. 177–40. functioning autism, males and females, scientists and mathematicians. 3. Loth E, Spooren W, Ham LM, Isaac MB, Auriche-Benichou C, Banaschewski T, J Autism Dev Disord. 2001;31:5–17. Baron-Cohen S, Broich K, Bolte S, Bourgeron T, et al. Identification and 29. Lai MC, Lombardo MV, Auyeung B, Chakrabarti B, Baron-Cohen S. Sex/ validation of biomarkers for autism spectrum disorders. Nat Rev Drug gender differences and autism: setting the scene for future research. J Am Discov. 2016;15:70–3. Acad Child Adolesc Psychiatry. 2015;54:11–24. 4. Biomarkers Definition Working Group. Biomarkers and surrogate endpoints: 30. George D, Mallery P. SPSS for Windows step by step: a simple guide and preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001; reference. Boston: Allyn & Bacon; 2003. 69:89–95. 31. Kaplan R, Saccuzzo D. Psychological testing: principles, applications and 5. Hobson RP. The autistic child's appraisal of expressions of emotion. J Child issues (5th ed). Belmont: Wadsworth; 2001. Psychol Psychiatry. 1986;27:321–42. 32. Baron-Cohen S. Autism: the empathizing-systemizing (E-S) theory. Ann N Y 6. Uljarevic M, Hamilton A. Recognition of emotions in autism: a formal meta- Acad Sci. 2009;1156:68–80. analysis. J Autism Dev Disord. 2012;43(7):1517-26. 33. Lombardo MV, Lai M-C, Auyeung B, Holt RJ, Allison C, Smith P, Chakrabarti 7. Harms MB, Martin A, Wallace GL. Facial emotion recognition in autism B, Ruigrok ANV, Suckling J, Bullmore E, et al. Unsupervised data-driven spectrum disorders: a review of behavioral and neuroimaging studies. stratification of mentalizing heterogeneity in autism. Sci Rep Oct. 2016;18:18. Neuropsychol Rev. 2010;20:290–322. 34. Oakley BF, Brewer R, Bird G, Catmur C. Theory of mind is not theory of 8. Celani G, Battacchi MW, Arcidiacono L. The understanding of the emotional emotion: a cautionary note on the reading the mind in the eyes test. meaning of facial expressions in people with autism. J Autism Dev Disord. J Abnorm Psychol. 2016;125:818–23. 1999;29:57–66. 35. Castelli F, Frith C, Happe F, Frith U. Autism, Asperger syndrome and brain 9. Baron-Cohen S, Jolliffe T, Mortimore C, Robertson M. Another advanced test mechanisms for the attribution of mental states to animated shapes. Brain. of theory of mind: evidence from very high functioning adults with autism 2002;125:1839–49. or Asperger syndrome. J Child Psychol Psychiatry. 1997;38:813–22. 10. Castelli F. Understanding emotions from standardized facial expressions in 36. Rinehart N, Tonge B, Brereton A, Bradshaw J. Attentional blink in young autism and normal development. Autism. 2005;9:428–49. people with high-functioning autism and Asperger’s disorder. Autism. 2010; 11. Rutherford MD, Towns AM. Scan path differences and similarities during 14:47–66. emotion perception in those with and without autism spectrum disorders. 37. Wright B, Clarke N, Jordan J, Young AW, Clarke P, Miles J, Nation K, Clarke L, J Autism Dev Disord. 2008;38:1371–81. Williams C. Emotion recognition in faces and the use of visual context in Loth et al. Molecular Autism (2018) 9:7 Page 11 of 11 young people with high-functioning autism spectrum disorders. Autism. 2008;12:607–26. 38. Happe F, Frith U. The weak coherence account: detail-focused cognitive style in autism spectrum disorders. J Autism Dev Disord. 2006;36:5–25. 39. Loth E, Gómez JC, Happé F. When seeing depends on knowing: adults with autism spectrum conditions show diminished top-down processes in the visual perception of degraded faces but not degraded objects. Neuropsychologia. 2010;48:1227–36. 40. Verhallen RJ, Bosten JM, Goodbourn PT, Lawrance-Owen AJ, Bargary G, Mollon JD. General and specific factors in the processing of faces. Vis Res. 2017;141: 217-27. 41. Bird G, Cook R. Mixed emotions: the contribution of alexithymia to the emotional symptoms of autism. Transl Psychiatry. 2013;3:e285. 42. Cook R, Brewer R, Shah P, Bird G. Alexithymia, not autism, predicts poor recognition of emotional facial expressions. Psychol Sci. 2013;24:723–32. 43. Berthoz S, Haviland MG, Riggs ML, Perdereau F, Bungener C. Assessing alexithymia in French-speaking samples: psychometric properties of the Observer Alexithymia Scale-French translation. Eur Psychiatry. 2005;20:497–502. 44. Lombardo MV, Barnes JL, Wheelwright SJ, Baron-Cohen S. Self-referential cognition and empathy in autism. PLoS One. 2007;2:e883. 45. Nemiah JC, Freyberger H, Sifneos PE. Alexithymia: a view of the psychosomatic process. In: Hill OW, editor. Modern trends in psychosomatic medicine. London: Butterworths; 1976. 46. Klin A, Jones W, Schultz R, Volkmar F, Cohen D. Visual fixation patterns during viewing of naturalistic social situations as predictors of social competence in individuals with autism. Arch Gen Psychiatry. 2002;59:809–16. 47. Loth E, Charman T, Mason L, Tillmann J, Jones EJH, Wooldridge C, Ahmad J, Auyeung B, Brogna C, Ambrosino S, et al. The EU-AIMS Longitudinal European Autism Project (LEAP): design and methodologies to identify and validate stratification biomarkers for autism spectrum disorders. Mol Autism. 2017;8:24. https://doi.org/10. 1186/s13229-017-0146-8. 48. Charman T, Loth E, Tillmann J, Crawley D, Wooldridge C, Goyard D, Ahmad J, Auyeung B, Ambrosino S, Banaschewski T, et al. The EU-AIMS Longitudinal European Autism Project (LEAP): clinical characterisation. Mol Autism. 2017; 8:27. https://doi.org/10.1186/s13229-017-0145-9. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries � Our selector tool helps you to find the most relevant journal � We provide round the clock customer support � Convenient online submission � Thorough peer review � Inclusion in PubMed and all major indexing services � Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Molecular Autism Springer Journals

Facial expression recognition as a candidate marker for autism spectrum disorder: how frequent and severe are deficits?

Download PDF
11 pages

Loading next page...
 
/lp/springer-journals/facial-expression-recognition-as-a-candidate-marker-for-autism-VK2jh57i6t

References (59)

Publisher
Springer Journals
Copyright
2018 The Author(s).
eISSN
2040-2392
DOI
10.1186/s13229-018-0187-7
Publisher site
See Article on Publisher Site

Abstract

Background: Impairments in social communication are a core feature of Autism Spectrum Disorder (ASD). Because the ability to infer other people’s emotions from their facial expressions is critical for many aspects of social communication, deficits in expression recognition are a plausible candidate marker for ASD. However, previous studies on facial expression recognition produced mixed results, which may be due to differences in the sensitivity of the many tests used and/or the heterogeneity among individuals with ASD. To ascertain whether expression recognition may serve as a diagnostic marker (which distinguishes people with ASD from a comparison group) or a stratification marker (which helps to divide ASD into more homogeneous subgroups), a crucial first step is to move beyond identification of mean group differences and to better understand the frequency and severity of impairments. Methods: This study tested 46 individuals with ASD and 52 age- and IQ-matched typically developing (TD) participants on the Films Expression Task, which combines three key features of real-life expression recognition: naturalistic facial expressions, a broad range of simple and complex emotions, and short presentation time. Test- retest reliability was assessed in 28 individuals who did not participate in the main study and revealed acceptable reliability (ICC r = .74). − 10 Results: Case-control comparisons showed highly significant mean group differences for accuracy (p = 1.1 × 10 ), with an effect size (Cohen’s d = 1.6), more than twice as large as the mean effect size reported in a previous meta- analysis (Uljarevic and Hamilton, 2012, J Autism Dev Disord). The ASD group also had significantly increased mean reaction times overall (p = .00015, d = .83) and on correct trials (p = .0002, d = .78). However, whereas 63% of people with ASD showed severe deficits (they performed below two standard deviations of the TD mean, a small subgroup (15.3%) performed normally (within one standard deviation of the mean). Conclusion: These findings indicate that the majority of people with ASD have severe expression recognition deficits and that the Films Expression Test is a sensitive task for biomarker research in ASD. Future work is needed to establish whether ASD subgroups with and without expression recognition deficits differ from one another in terms of their symptom profile or neurobiological underpinnings. Keywords: Autism Spectrum disorder, Facial expression recognition, Biomarker * Correspondence: [email protected] Sackler Institute for Translational Neurodevelopment, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, De Crespigny Park, Denmark Hill, London SE5 8AF, UK Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 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. Loth et al. Molecular Autism (2018) 9:7 Page 2 of 11 Background displayed with low intensities [14] using morphs [15, 16]. Autism spectrum disorder (ASD) is a life-long neurodeve- Furthermore, speed-accuracy trade-offs [17], abnormal lopmental disorder, behaviourally defined by impairments gaze fixation patterns [18], and differences in underlying in social communication and the presence of repetitive neural processes as measured by EEG [19] and fMRI have and restricted behaviours and interests [1]. The consider- also been reported [20, 21]. However, some inconsisten- able variability in the quality and severity of symptoms cies remain even within the different methodologies and between individuals with ASD is widely recognized [2]. sample types. For example, some studies have found mean For example, social communicative impairments can be group deficits in adult ASD samples with IQ in the normal manifested in a range of difficulties in social-emotional range on tasks that required labeling basic emotions with reciprocity, in non-verbal communicative behaviours, and unlimited presentation times [20, 22, 23], while others in developing and maintaining social relationships [1]. reported no deficits in the recognition of complex Recognition of this heterogeneity has begun to cast doubt emotions [24]. as to whether a truly diagnostic biological or behavioural Recently, a formal meta-analysis of 48 papers con- marker for ASD exists (which differentiates all or the cluded that there is an emotion recognition difficulty in majority of people with ASD from typically developing “autism”, with a mean effect size of Cohen’s D = 0.80, yet individuals or those with other neurodevelopmental/psy- this was estimated to decrease to around 0.40 after chiatric conditions) and prompted an increasing interest adjusting for publication biases [6]. Hence, it remains in the identification of stratification markers to parse ASD unclear whether differences in study findings reflect into more homogeneous subgroups [3]. Biomarkers variability in the nature or sensitivity of the many behav- have been defined as “a characteristic that is object- ioural tests used to assess expression recognition in ASD ively measured and evaluated as an indication of and/or differences in the severity of deficits among people normal biological processes, pathogenic processes, or with ASD. pharmacologic responses to a therapeutic interven- To ascertain the value of expression recognition defi- tion” [4]. Here we use the term “biomarker” in a broad cits as a candidate diagnostic or stratification biomarker sense to refer to measures of any modality, including we need, first, sensitive tests that incorporate several cognitive/behavioural tests [4]. Stratification bio- real-life characteristics. For example, in daily life, facial markers are then characteristics that vary between expressions are usually more subtly displayed than those ASD subgroups and that map onto differences in their depicted in most standard stimuli of “prototypical” facial symptom presentation, etiology, need for particular emotion expressions. People also often reveal their emo- treatments, and/or treatment response. tions for only a very brief time. The observer is then re- Because the ability to infer other people’semotions from quired to identify the facial emotion expression quickly their facial expressions is critical for many aspects of social to react appropriately to the person’s feelings. Second, communication, deficits in expression recognition have the field needs to move away from a sole focus on mean long been suggested to represent a core impairment in between-group differences to better understand the fre- ASD [5]. However, over the past three decades, behav- quency and severity of expression recognition deficits ioural studies of facial expression recognition in ASD have among individuals with ASD. For example, in case- produced mixed findings, ranging from reports of pro- control studies, a significant p value (especially in com- found deficits to apparently intact expression recognition bination with small or medium effect sizes) could either skills (see e.g., [6, 7] for recent reviews). A review of this reflect (small) deficits in most cases (say, around 1 SD literature suggests that in ASD, the presence and severity below the typically developing (TD) mean) or may be of expression recognition deficits on experimental tests is driven by a subgroup of individuals with severe deficits. influenced by both participant characteristics—age and Obviously these two scenarios have profoundly different ability level—and task requirements [7]. On the whole, implications for the potential utility of the test in clinical deficits in recognizing basic emotion expressions ap- practice. pear to be predominantly found in children and low- Hence, the aim of this study was to investigate the functioning individuals (i.e., people with ASD who also frequency and severity of deficits in the recognition of have intellectual disabilities) [5, 8]. By contrast, high- facial expressions of emotion using a task that is more functioning individuals with ASD (with IQ in the normal sensitive and naturalistic than previous tests. The range) usually perform well on tests that depict facial ex- Films Expression Task [25] uses still images captured pressions in a prototypical manner and that used relatively from movie scenes and combines three elements that long presentation times [9–11]. However, some studies appear to be challenging for people with ASD: depic- reported group-level deficits in recognizing expressions of tion of naturalistic facial expressions, inclusion of a complex emotions (e.g., guilt, defiance) [9, 12], expressions range of both basic and complex emotions, and brief that were presented only briefly [13] or subtle expressions presentation times. Loth et al. Molecular Autism (2018) 9:7 Page 3 of 11 Methods included ten new participants and employed the The Films Expression Task intended task procedure in which three pictures were The Films Expression Task [25] consists of 58 trials. In presented sequentially, one target and two distractors. each trial, participants were first presented with an ad- Only trials with equal or less than 30% errors were jective describing an emotional state (e.g., confident, carried forward, which resulted in 58 trials with tar- pleased). They were then briefly shown three images one gets taken from 15 films. after the other (500 ms each, with a 500 ms blank screen between images). The images were taken from films IQ measures made in non-English-speaking countries to decrease the Verbal, performance, and full-scale IQ were assessed probability that participants had seen them or were using the Wechsler Abbreviated Scale of Intelligence- familiar with the actors. 2nd edition (WASI-II; Wechsler, [26]) or a four-subtest Within each trial, images present the same actor or short-form of the Wechsler Adult Intelligence Scale- actress, but with different emotional expressions. Partici- Third Edition (WAIS-III; [27]). In both instances, the tests pants were asked to indicate, by key press, which of the included two verbal subscales (vocabulary, similarities) images best matched the target word. In 14 trials, the and two non-verbal subscales (block design, matrix rea- target emotion was a basic emotion (happy, angry, sad, soning). Standardized scores from the WASI-II and WAIS afraid, surprised, disgusted). In the remaining 44 trials, are comparable. the target emotion was complex (e.g., mocking, hurt, disappointed, resentful, see Additional file 1 for the list Autism Spectrum Quotient of target emotions). In trials with both basic and com- TheAutismSpectrumQuotient(AQ)[28]isa self- plex target emotions, the foils were selected to be similar report questionnaire to assess whether adults of to the targets in terms of perceptual features and inten- average intelligence have symptoms associated with sity of the expression (see supplementary information autism spectrum disorder. The test consists of 50 [25]). Basic vs. complex emotion trials were presented statements and participants indicate whether they interleaved, in a fixed-random order. Participants were “definitely agree”, “slightly agree”, “slightly disagree”, instructed to respond as quickly and accurately as or “definitely disagree” with each statement. Approxi- possible. mately half the items are worded to usually elicit an Immediately before the task, participants were pre- “agree” response from neurotypical individuals and sented with the definitions for each word describing a half to usually elicit a “disagree” response. The ques- target emotion (e.g., guilty). They were encouraged to tions cover five different domains associated with the ask questions if they were unfamiliar with or did not autism spectrum: social skills, communication skills, understand any of the words. Participants were also imagination, attention to detail, and attention switching/ allowed to review these definitions at the start of each tolerance of change. trial, after presentation of the target word. Task development and validation of the stimuli is Test-retest reliability of the films expression test described in [25] (supplementary information). Briefly, Thirty-one participants took part in a test-retest study to in pilot phase 1, three researchers suggested adjectives assess the reliability of the Films Expression Test. We to describe 122 pictures of facial expressions taken from aimed to recruit participants with a range of expression 18 films. In pilot phase 2, 32 native English speakers recognition abilities to ensure that test-retest reliability rated how well each adjective described the facial ex- on this task was stable at both high and low perform- pression on a scale from 1 to 5 (1 = “it doesn’t match at ance scores. Because we hypothesized that individuals all”,5= “it matches very well”). Adjectives were only with ASD would have lower accuracy scores than typically carried forward if they had received at least a mean developing individuals, four participants were recruited rating of 3.5 or a median rating of at least 4. Ten of from an autism support group. Hence, 27 participants these participants then also rated the intensity of the were typically developing, three had a formal diagnosis of target expression from 1 to 5 (1 = “not intense” to 5 = autism, and one was suspected of having autism. “very intense”) as well as the intensity of the expressions Two participants exceeded the intended maximum re- intended as foils. Targets and foils were then matched test interval of 6 weeks and did not complete the retest, based on their expression intensity. In pilot phase 3, 30 one participant’s retest data was not available after a different native English speakers were asked to choose technical error. This left 28 participants (20 females and one of two simultaneously presented expressions (the 8 males, mean age = 33.48 years; range = 20–57 years, target and a foil) that best matched an adjective. Distrac- four from the autism support group). Out of the 28 tors were only included if participants had selected them participants, English was a second language for six but less than 30% of the time. A fourth final pilot phase all participants described themselves as fluent English Loth et al. Molecular Autism (2018) 9:7 Page 4 of 11 speakers. The retest interval ranged from 14 to 34 days, (male, 32 years, VIQ = 92, first language Uzbek) and was with an average interval of 18.11 days. excluded from the analyses reported below. This left 46 All test-retest participants also completed a two-subtest people with ASD and 52 people with TD. The study was (vocabulary and matrix reasoning) short-form of the approved by the South London and Maudsley NHS Wechsler Abbreviated Scale of Intelligence (WASI-II; Trust ethics committee and the UCL Research Ethics [26]). The mean two-scale IQ was 113.07 (range 87–146). Committee. All participants or a parent in the case of Test-retest analysis revealed an intra-class correl- participants who were minors gave informed written ation coefficient of ICC = .74, indicating acceptable consent before study participation. reliability. The mean accuracy score at time 1 was 86.64% (SD = 7.49, range 62.07–98.28%) and at time 2 Results was also 86.64% (SD = 7.93, range 62.07–94.83%). Indi- Case-control comparisons vidual difference scores, calculated between time 1 Descriptive statistics are provided in Additional file 2. and time 2 performance, showed that 17 (60.71%) par- The ASD group accurately identified an average of ticipants had a difference score less than .5 SDs, 6 70.8% (SD = 13.5) of the briefly presented emotion ex- (21.43%) were between .5 and 1 SD, and 5 were pressions, compared to 87.5% (SD = 5.5) in the control (17.86%) greater than 1 SD. group (see Fig. 1a). The Welch test, which does not assume equal variances, showed that this mean group Participants difference was highly significant (t(58.1) = − 7.8, p = In the main study, participants consisted of 46 adoles- −10 1.1 × 10 ). The effect size (Cohen’s d) was 1.62. We cents or adults with ASD (34 male, 12 female) and 53 used bootstrapping in Matlab version 8.3.0.532 (Math- typically developing (TD) individuals (33 male, 20 works, Natick, MA) to estimate the 95% confidence interval female). Eighteen TD participants were tested as part of around this point estimate. We drew 10,000 resamples a previous study [25]. Although the ASD group included from the original sample using random sampling with a higher proportion of males than the TD group, the dif- replacement. We obtained a distribution of effect sizes for ference in the sex ratio between the groups was not sta- all the resamples, and using the percentile method (i.e. tistically significant (χ = 1.5, p = .21). Mean age of the using the 2.5 and 97.5 percentiles as lower and upper ASD group was 30.2 years (SD 9.4, 15–50 years) and of bounds of the CI), the 95% CI was [1.30 2.05]. the TD group was 27.5 years (SD 7.8, range 14–55 years). Analyses of reaction times (RT) revealed that the Two participants in the ASD group (15 and 17 years) ASD group also responded significantly more slowly and two participants in the TD group (15 and 14 years) (M =1459 ms, SD = 899) than the TD group (M =878 ms, were younger than 18 years. IQ data was available for 42 SD = 414) across all trials (t(61.5) = 4.02, p = .00015, out of 46 participants with ASD and all participants in the Cohen’s d = .83, 95% CI [0.49 1.19]) as well as on their cor- TD group. Mean verbal and full-scale IQ in the ASD group rect trials (ASD: M =1279 ms, SD =812; TD: M = 767 ms, were 113.9 (range 85–140) and 116.0 (range 87–135), re- SD =352) (t(df = 59.7) = 3.96, p =.0002, Cohen’s d =0.82, spectively, and in the TD group were 114.0 (range 74–146) 95% CI [0.49 1.17]). These findings remain unchanged and 115.5 (range 85–143). The groups did not significantly when adolescents below the age of 18 years were excluded differ from each other in terms of age (t(86) = 2.3, p =.12), (accuracy: ASD = 70.2%, TD = 87.7%; t(53.5) = 7.9, p = full-scale IQ (t(86) = .02, p = .87), verbal IQ (t(86) = .0004, −10 1.4 × 10 )(see Fig.1b). p =.98), or performance IQ (t(86) = .002, p = .96). The four individuals with ASD for whom no IQ data were available had attended mainstream schools, which suggests that Sex differences their IQ was within the normal range. Given previous reports of sex differences on several All participants with ASD except one were native social-cognitive tasks in the typical population and in English speakers. In the TD group, 37 of 53 participants ASD [29], we performed 2 (group) × 2 (sex) ANOVAs to were native English speakers. All TD participants for test whether accuracy, overall RT, or RT on correct trials whom English was not their first language described differed between males and females overall or in either their current English language level as “fluent”. Fourteen group. These tests confirmed the significant effects of −12 individuals performed within 1 SD of the TD mean or group on accuracy (F(1,97) = 65.0, p = 2.3 × 10 ) and above, one individual below 1 SD, and two individual RTs (all ps < .00007) but there were no significant effects below 2 SDs of the TD mean. Of these two TD partici- of sex (all ps > .5)) or group by sex interactions (all ps> pants, one had an above-average verbal IQ (VIQ) (124), .11). This finding indicates that the overall case-control indicative of adequate language comprehension and was difference cannot be attributed to differences in the therefore not excluded. The other TD participant had an slightly higher male:female ratio in the ASD than those accuracy score of more than 4 SDs below the TD mean in the control group. Loth et al. Molecular Autism (2018) 9:7 Page 5 of 11 Fig. 1 a Accuracy: percentage of correct trials, by group. b Mean reaction time of correct trials and mean reaction time overall, by group Basic vs. complex emotions significant relationship between accuracy scores and FIQ Next, we explored whether in the ASD group, the ex- (r(52) = .26, p = .057), such that participants with higher pression recognition impairments reported above were intelligence also had higher expression recognition driven by specific problems in identifying complex scores. emotional expressions. To do so, we separately ana- lysed trials with simple vs. complex target emotions Frequency and severity of expression recognition deficits (see Additional file 1). in ASD A 2 (group) × 2 (emotion category) repeated-measures To establish the frequency and severity of expression ANOVA revealed the above reported significant main ef- recognition deficits in the ASD group, we calculated −13 fect of group (F(1,96) = 70.7, p =3.8 ×10 ) and a signifi- how far the accuracy score for each individual with ASD cant effect of emotion category (F(1,96) = 5.9, p =.016) on deviated from the TD group mean. Analyses of the response accuracy, such that on average, both groups were scores showed that 63% of people with ASD performed better at recognizing simple than complex emotions. The more than 2 SDs below the TD mean. Of those, 23.9% group x emotion category interaction was not significant performed between 2 and 3 SDs below the TD mean F(1,96) = .14, p = .7) (see Fig. 2). and 39.1% performed more than 3 SDs below the TD mean (see Fig. 3). 21.7% of people with ASD performed Correlations between accuracy and RT scores with age between 1 to 2 SDs below the TD mean while 15.3% had and IQ accuracy scores within the TD range. The 17-year-old Next, we tested whether age, verbal, performance, or full- with ASD performed within the 1 SD of the TD range, scale IQ were related to accuracy or reaction time (RT while the 15-year-old with ASD performed between 2 overall or on correct trials) on the Films Expressions Test. and 3 SDs below the TD range. As shown in Fig. 4, in In the ASD group, we found no significant relationships the ASD group, accuracy and RT scores were negatively between expression recognition variables and age or IQ correlated such that individuals with ASD who had (all p > .09). In the TD group, there was a marginally higher accuracy scores were also able to correctly recognize the emotion faster than those with low accur- acy scores. Therefore, there was no evidence of speed- accuracy trade-off for most of the ASD participants who performed accurately on the task. These results suggest that the majority (63%) of individuals with ASD had severe deficits, while a small subgroup of around 15% of people with ASD showed intact expression recognition skills on this test. Correlations between accuracy and RT scores and ASD symptoms Finally, we investigated correlations with ASD symptoms in a subset of N = 53 participants (ASD = 25, TD = 28) for whom AQ data was available (see Fig. 5). When both Fig. 2 Percentage of correct trials, by emotion category and group groups were collapsed, we found a significant negative Loth et al. Molecular Autism (2018) 9:7 Page 6 of 11 Fig. 3 Upper panel: Distribution of TD performance on percentage of accuracy. Lower panel: Distribution ofASD performance on percentage of accuracy. Grey bars denote − 2 to + 2 SDs of the TD means on percentage of correct expression recognition correlation between the percentage of correct responses pattern was found in the ASD group, it did not reach on the Films Expression Task and higher ASD symptoms statistical significance (ASD r(25) = .29, p = .16). r(53) = − .48, p = .0002. However, when the ASD and TD Likewise, for overall RTs, we found a significant rela- groups were considered separately, there were no signifi- tionship with ASD symptoms when both groups were cant relationships in either group (ASD r(25) = .09, p =.66, collapsed (RT : overall: r(53) = .54, p = .000026). overall TD r(28) = .03, p = .88). This suggests that the negative When the TD and ASD groups were analysed separately, correlation for the overall sample could be explained by similar but non-significant patterns were found (TD: the performance differences between the two groups. r(28) = .31, p = .10, ASD: (r(25) = .29, p = .16). A somewhat different picture emerged for RTs. We found that longer reaction times on correct trials were Discussion associated with more ASD symptoms. This was signifi- This study investigated the frequency and severity of im- cant overall, i.e., when both groups were collapsed pairments in facial expression recognition in ASD using (RT : overall r(53) = .53, p = .00003) and in the a test that captured several features of expression recog- correct trials TD group (TD r(28) = .48, p = .012). Although a similar nition in daily life. A test-retest study showed reliability of .74, which is considered adequate for a test to be of use in clinical settings [30, 31]. At the group level, we found highly significant differences in accuracy and re- sponse times between a group of high-functioning adults and adolescents with ASD and an age- and IQ-matched TD control group. Effect sizes of the accuracy scores were more than twice as large as those estimated in a re- cent meta-analysis of studies on emotion recognition in ASD and more than four times larger than estimates that accounted for publication biases [6]. As a first step towards ascertaining whether deficits in expression recognition may serve as a diagnostic or stratification marker for ASD, we investigated the fre- quency and severity of these deficits. This revealed that 63% of people with ASD had severe deficits, which would be expected to create substantial social communi- cative difficulties. For example, participants who per- Fig. 4 Scatterplot showing percentage of accuracy and RT on formed below 3 SDs of the TD mean failed to identify correct trials, by group. Grey bars denote − 2to + 2 SDs of the TD the briefly displayed target emotion expression in 3–7 means on percentage of correct expression recognition out of 10 trials. Translated into real-life settings, it is Loth et al. Molecular Autism (2018) 9:7 Page 7 of 11 Fig. 5 Scatterplot showing the relationship between AQ scores and a accuracy scores, b overall RTs, and c RTs for correct responses. Regression lines are plotted overall, and for the ASD and TD groups separately. Grey bars denote − 2 to + 2 SDs of the TD means on percentage of correct expression recognition, RTs, and RTs on correct trials easy to see how this may impact the ability to, for ex- recognition deficits and ASD symptom severity, as mea- ample, modulate a conversation or to respond empathic- sured by the AQ. When the ASD and TD groups were ally [32]. However, we also found considerable variability combined, we found both lower accuracy in expression rec- in that 21.7% performed between 1 to 2 SDs below the ognition and higher RTs to be moderately related to greater TD mean and 15.3% evidenced expression recognition severity of autism traits. However, whereas accuracy scores skills indistinguishable from TD individuals (within 1 SD were not sensitive to the severity of ASD symptoms within of the TD mean) as indexed by the combination of both the ASD and TD groups, respectively, longer reaction times accuracy and RT scores. This variability in expression on correct trials were significantly related to greater ASD recognition skills among individuals with ASD was unre- symptoms in the TD group. A similar relationship was lated to age or IQ (verbal or non-verbal). However, it found in the ASD group with moderate correlation coeffi- should be borne in mind that the IQ range in the cients which, however, did not reach statistical significance. current sample was restricted to the normal range, so These correlation analyses should be viewed as preliminary, that this finding may not generalize to individuals with becauseAQscoreswereonlyavailablein abouthalf ofthe ASD and intellectual disabilities. participants. As a consequence, the sample sizes were small The present findings are broadly consistent with other re- when split by group, which reduced our power to detect a cent efforts to parse heterogeneity in ASD. In particular, significant effect. In addition, the AQ is a composite meas- Lombardo and colleagues [33] used unsupervised hierarch- ureofarangeofautism-related traits, including features ical clustering approaches to identify ASD subgroups on such as intolerance of changes that may not be expected to the basis of item-level performance on the Reading the relate to expression recognition. This may have diluted po- Mind in theEyes Test(RMET).TheRMETisawidely used tentially higher associations with more specific social com- mentalising task that also involves a strong emotion recog- municative impairments. nition component [34] as it requires participants to identify In sum, our findings revealed that the majority of the mental or emotional state of a person from their eye re- individuals with ASD had severe deficits in expression gion only. The study (which comprised a discovery and recognition deficits but also that a sub-sample of indi- replication cohort) identified five ASD subgroups and four viduals with ASD had no behavioural impairments on TD subgroups. For individuals with ASD, 45–62% showed this task. This indicates that expression recognition may what the authors termed “clear to immediate impairments”, more likely serve as a stratification as opposed to diag- while two subgroups (19–36%) performed more than 2 and nostic marker, subject to further substantiation that sub- up to 11 SDs below the TD means - depending on the TD groups with/without expression recognition deficits also subgroup that was used as comparison. It would be valu- differ in symptom severity or adaptive behavior. One able for future studies to directly compare performance on implication of the current findings is that intervention the RMET and Films Expression Tasks or combine infor- programmes that specifically target expression recogni- mation from both measures to derive a composite score. tion may be valuable for a substantial proportion of A second requirement for a quantitative stratification people with this disorder. marker is to demonstrate clinical relevance, for example, that people who fall within a certain range of deficits differ Study limitations and future directions from those without abnormalities (normal range) in terms Several potential limitations of the current study as well of symptom severity. To begin to address this, we con- as implications for future research should be considered. ducted correlation analyses between severity of expression First, we deliberately chose a task that incorporated Loth et al. Molecular Autism (2018) 9:7 Page 8 of 11 several factors crucial for expression recognition in real- with ASD may also have deficits in theory of mind life situations, including a wide range of simple and [35]. Therefore, future studies should probe compre- complex target emotions, and relatively short presenta- hension of the different emotion words and account tion times. As a consequence, could factors unrelated to for individual differences in understanding particular expression recognition per se have contributed to the words and their emotional state—independent from deficits observed in some of the ASD participants, for recognition of their manifestation in facial example, comprehension of the target words or other expressions. cognitive impairments/ anomalies? Previous studies reported that expression recognition If difficulties in comprehending some of the emotion deficits varied with the type of emotion, such that recog- words accounted for performance impairments, one nition of fear has been found to be more impaired than would expect larger group differences in trials with happiness (see [6] for a review). The naturalistic charac- complex target emotions than basic emotions. However, ter of the task did not enable us to systematically analyse separate analyses, split by emotion target (simple vs com- recognition of different types of emotion expressions plex), showed significant impairments in the ASD group (fear, anger, disappointment, etc.) as 34 different emotion on trials with basic as well as complex emotions. Basic adjectives were included, and many of the target emo- emotion words are typically understood by the age of tions were only used once. However, descriptive trial-by- 6 years or earlier. Also most of the complex emotions trial analyses, split by group (see Fig. 6) suggest that the used here are typically understood by age 8 (see Add- mean percentages of correct response not only varied be- itional file 1), and the ASD participants had verbal tween target emotions but in some instances also within IQs within (and mostly above) the population average. target emotions. For example, among the ASD group However, recognition of complex emotions also re- “angry” and “happy” trials had the highest percentages of quires an understanding of the mental states they de- correct responses whereas trials with the target emotions scribe [34] and some high-functioning individuals “disappointed”, “tentative”,and “disbelieving” had the Fig. 6 Percentage of correct responses, by trial. a ASD group. b TD group Loth et al. Molecular Autism (2018) 9:7 Page 9 of 11 lowest percentages. For “sad” trials, correct recognition version of the Films Expression Test and a companion ranged from 56.5 to 73.9%. The latter finding may be due child version in the follow-up assessment battery of the to the fact that the difficulty level of a trial is influenced by EU-AIMS Longitudinal European Autism Project [47, 48] the nature of the target emotion and also by the expres- to test whether the current findings can be replicated in sions in the distractor stimuli used in a trial. an independent sample more diverse in age and ability Could more general problems in rapid visual process- level, to further determine the clinical usefulness of our ing have contributed to deficits in recognizing relatively findings, and to investigate potential underlying briefly presented (500 ms) facial expressions of emo- mechanisms. tions? This alternative explanation seems unlikely for two reasons. Several previous studies reported no defi- Conclusion cits in rapid visual processing in ASD [13, 36]. For Taken together, our findings show highly significant example, a study that tested recognition of micro-facial mean case-control differences in facial expression expressions, using much faster presentation times (15 recognition, with one of the largest effect sizes ever and 30 ms) than those employed here reported specific reported in the expression recognition literature of deficits in identifying facial expressions, but not objects ASD. This suggests that problems with expression rec- in an ASD group [37]. Moreover, as indicated above, our ognition are more widespreadthancurrently thought, trial-by-trial analyses (see Fig. 6) showed quite variable likely owing to the more naturalistic character of the performance scores between trials in the ASD group, tasks used here. Nevertheless, we also highlight im- with two trials being correctly identified by over 90% of portant variability in expression recognition skills the ASD participants. This pattern of largely intact ex- among individuals with this condition and showed pression recognition on some trials but impairments on that aminorityofpeoplewithASD hadnobehav- others would be inconsistent with general problems in ioural impairments. This finding indicates that the rapid visual processing. However, it remains a possibility Films Expression Test may serve as a valuable tool to that other cognitive abnormalities previously reported in study expression recognition as a candidate stratifica- (some people with) ASD, such as a detail-focused pro- tion marker for ASD. cessing style [38] abnormalities in top-down processing [39] or difficulties with other aspects of face perception [40], may have contributed to behavioural impairments Additional files in expression recognition. Additional file 1: Target emotion words, split by age of acquisition The present finding raises the question of what may ac- (AoA) norms. (DOCX 15 kb) count for the differences between people with ASD with/ Additional file 2: Descriptive statistics and contrasts for main variables without expression recognition deficits. This question can- of interest. (DOCX 19 kb) not be settled here but we discuss potential factors and outline some avenues for future research. A likely possibil- Abbreviations ity is that (some) people with ASD have problems in the AQ: Autism Spectrum Quotient; ASD: Autism Spectrum Disorder; neurocognitive mechanisms that represent facial expres- IQ: Intelligence Quotient; RT: Reaction time; SD: Standard deviation; sion information and/or multi-modal expression informa- TD: Typically developing; WAIS: Wechsler Adult Intelligence Scales; WASI: Wechsler Abbreviated Scales of Intelligence tion. One hypothesis that recently gained prominence is the notion that expression recognition deficits in ASD Acknowledgements may be linked to the high frequency of comorbid alexithy- We thank all participants and their families for their help with this study, as mia [41, 42], estimated to affect around 50% of people well as Raka Tavashmi, Hannah Hayward, Daisy Crawley, Jessica Sabet, Jessica Faulkner, Claire Ellis, and Antonia San Jose Caceres for their help with with ASD [43, 44]. Alexithymia is a sub-clinical trait char- volunteer recruitment and/or assessment. acterized by difficulties in identifying and describing one’s own emotional state [45]. Future work—ideally with larger Funding samples—will be needed to ascertain whether some or all This study was supported by an Economic and Social Research Council individuals with ASD who exhibit substantial expression (ESRC) project grant (RES-000- 22-1123) and by EU-AIMS (European Autism Interventions), which receives support from the Innovative Medicines Initia- recognition deficits also have higher rates of trait alexithy- tive Joint Undertaking under grant agreement no. 115300, the resources of mia. Other potential underlying mechanisms may include which are composed of financial contributions from the European Union’s differences in attention (e.g., to the eye region [11, 46]), or Seventh Framework Programme (grant FP7/2007-2013), from the European Federation of Pharmaceutical Industries and Associations companies’ in-kind brain structural/functional anomalies [20]. To identify the contributions, and from Autism Speaks. factors that underpin expression recognition impairments at an individual level will require multi-modal studies Availability of data and materials and/or designs that assess a range of cognitive functions The data set generated and analysed during this study is available from the within each individual. We have included an abridged corresponding author upon request. Loth et al. Molecular Autism (2018) 9:7 Page 10 of 11 Authors’ contributions 12. Capps L, Yirmiya N, Sigman M. Understanding of simple and complex EL designed the study and wrote the first and final draft. LG designed the emotions in non-retarded children with autism. J Child Psychol Psychiatry. study materials and contributed to the writing of the manuscript. JA carried 1992;33:1169–82. out the test-retest reliability study, contributed to data collection, prepared 13. Clark TF, Winkielman P, McIntosh DN. Autism and the extraction of emotion all figures, and contributed to the writing of the manuscript. EW contributed from briefly presented facial expressions: stumbling at the first step of to data collection and commented on the manuscript. ACD contributed to empathy. Emotion. 2008;8:803–9. the data collection and commented on the manuscript. BD designed the 14. Wingenbach TSH, Ashwin C. Brosnan: diminished sensitivity and specificity study and contributed to the writing of the manuscript. All authors read and at recognising facial emotion expressions of varying intensity underlie approved the final manuscript. emotion-specific recognition deficits in autism spectrum disorders. Res Autism Spectr Disord. 2017;34:52–61. 15. Kennedy DP, Adolphs R. Perception of emotions from facial expressions in Ethics approval and consent to participate high-functioning adults with autism. Neuropsychologia. 2012;50:3313–9. The study was approved by the South London and Maudsley NHS Trust 16. Wang S, Adolphs R. Reduced specificity in emotion judgment in people ethics committee and the UCL Research Ethics Committee. All with autism spectrum disorder. Neuropsychologia. 2017;99:286–95. participants—and in the case of minors—a parent gave informed written 17. Sucksmith E, Allison C, Baron-Cohen S, Chakrabarti B, Hoekstra RA. Empathy consent. and emotion recognition in people with autism, first-degree relatives, and controls. Neuropsychologia. 2013;51:98–105. Consent for publication 18. Pelphrey KA, Sasson NJ, Reznick JS, Paul G, Goldman BD, Piven J. Visual Not applicable scanning of faces in autism. J Autism Dev Disord. 2002;32:249–61. 19. Dawson G, Webb SJ, Carver L, Panagiotides H, McPartland J. Young children Competing interests with autism show atypical brain responses to fearful versus neutral facial The authors declare that they have no competing interests. expressions of emotion. Dev Sci. 2004;7:340–59. 20. Ashwin C, Baron-Cohen S, Wheelwright S, O'Riordan M, Bullmore ET. Differential activation of the amygdala and the ‘social brain’ during fearful Publisher’sNote face-processing in Asperger syndrome. Neuropsychologia. 2007;45:2–14. Springer Nature remains neutral with regard to jurisdictional claims in 21. Wang AT, Dapretto M, Hariri AR, Sigman M, Bookheimer SY. Neural published maps and institutional affiliations. correlates of facial affect processing in children and adolescents with autism spectrum disorder. J Am Acad Child Adolesc Psychiatry. 2004;43:481–90. Author details 22. Macdonald H, Rutter M, Howlin P, Rios P, Le Conteur A, Evered C, Folstein S. Sackler Institute for Translational Neurodevelopment, Institute of Psychiatry, Recognition and expression of emotional cues by autistic and normal Psychology and Neuroscience, King’s College London, De Crespigny Park, adults. J Child Psychol Psychiatry. 1989;30:865–77. Denmark Hill, London SE5 8AF, UK. Department of Forensic and 23. Boraston Z, Blakemore SJ, Chilvers R, Skuse D. Impaired sadness Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and recognition is linked to social interaction deficit in autism. Neuroscience, King’s College London, London, UK. Division of Psychology, Neuropsychologia. 2007;45:1501–10. Department of Life Sciences, Brunel University, London, UK. Department of 24. Buitelaar JK, van der Wees M, Swaab-Barneveld H, van der Gaag RJ. Theory of Psychological and Brain Sciences, Dartmouth College, Hanover, USA. mind and emotion-recognition functioning in autistic spectrum disorders and in psychiatric control and normal children. Dev Psychopathol. 1999;11:39–58. Received: 26 August 2017 Accepted: 10 January 2018 25. Garrido L, Furl N, Draganski B, Weiskopf N, Stevens J, Tan GC, Driver J, Dolan RJ, Duchaine B. Voxel-based morphometry reveals reduced grey matter volume in the temporal cortex of developmental prosopagnosics. Brain. References 2009;132:3443–55. 1. American Psychiatric Association. Diagnostic and statistical manual of 26. Wechsler D. Wechsler Abbreviated Scale of Intelligence - Second Edition mental disorders. Fifth ed. Arlington: American Psychiatric Publishing; 2013. (WASI-II). London: Pearson; 2011. 2. Charman T. Variability in neurodevelopmental disorders: evidence from 27. Wechsler D. Wechsler Adult Intelligence Scale. Third ed. London: Pearson; 1997. autism. In: Riby D, Van Herwegen J, editors. Applied research and key 28. Baron-Cohen S, Wheelwright S, Skinner R, Martin J, Clubley E. The autism- issues in neurodevelopmental disorders. Hove: Routledge Psychology spectrum quotient (AQ): evidence from Asperger syndrome/high- Press; 2015. p. 177–40. functioning autism, males and females, scientists and mathematicians. 3. Loth E, Spooren W, Ham LM, Isaac MB, Auriche-Benichou C, Banaschewski T, J Autism Dev Disord. 2001;31:5–17. Baron-Cohen S, Broich K, Bolte S, Bourgeron T, et al. Identification and 29. Lai MC, Lombardo MV, Auyeung B, Chakrabarti B, Baron-Cohen S. Sex/ validation of biomarkers for autism spectrum disorders. Nat Rev Drug gender differences and autism: setting the scene for future research. J Am Discov. 2016;15:70–3. Acad Child Adolesc Psychiatry. 2015;54:11–24. 4. Biomarkers Definition Working Group. Biomarkers and surrogate endpoints: 30. George D, Mallery P. SPSS for Windows step by step: a simple guide and preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001; reference. Boston: Allyn & Bacon; 2003. 69:89–95. 31. Kaplan R, Saccuzzo D. Psychological testing: principles, applications and 5. Hobson RP. The autistic child's appraisal of expressions of emotion. J Child issues (5th ed). Belmont: Wadsworth; 2001. Psychol Psychiatry. 1986;27:321–42. 32. Baron-Cohen S. Autism: the empathizing-systemizing (E-S) theory. Ann N Y 6. Uljarevic M, Hamilton A. Recognition of emotions in autism: a formal meta- Acad Sci. 2009;1156:68–80. analysis. J Autism Dev Disord. 2012;43(7):1517-26. 33. Lombardo MV, Lai M-C, Auyeung B, Holt RJ, Allison C, Smith P, Chakrabarti 7. Harms MB, Martin A, Wallace GL. Facial emotion recognition in autism B, Ruigrok ANV, Suckling J, Bullmore E, et al. Unsupervised data-driven spectrum disorders: a review of behavioral and neuroimaging studies. stratification of mentalizing heterogeneity in autism. Sci Rep Oct. 2016;18:18. Neuropsychol Rev. 2010;20:290–322. 34. Oakley BF, Brewer R, Bird G, Catmur C. Theory of mind is not theory of 8. Celani G, Battacchi MW, Arcidiacono L. The understanding of the emotional emotion: a cautionary note on the reading the mind in the eyes test. meaning of facial expressions in people with autism. J Autism Dev Disord. J Abnorm Psychol. 2016;125:818–23. 1999;29:57–66. 35. Castelli F, Frith C, Happe F, Frith U. Autism, Asperger syndrome and brain 9. Baron-Cohen S, Jolliffe T, Mortimore C, Robertson M. Another advanced test mechanisms for the attribution of mental states to animated shapes. Brain. of theory of mind: evidence from very high functioning adults with autism 2002;125:1839–49. or Asperger syndrome. J Child Psychol Psychiatry. 1997;38:813–22. 10. Castelli F. Understanding emotions from standardized facial expressions in 36. Rinehart N, Tonge B, Brereton A, Bradshaw J. Attentional blink in young autism and normal development. Autism. 2005;9:428–49. people with high-functioning autism and Asperger’s disorder. Autism. 2010; 11. Rutherford MD, Towns AM. Scan path differences and similarities during 14:47–66. emotion perception in those with and without autism spectrum disorders. 37. Wright B, Clarke N, Jordan J, Young AW, Clarke P, Miles J, Nation K, Clarke L, J Autism Dev Disord. 2008;38:1371–81. Williams C. Emotion recognition in faces and the use of visual context in Loth et al. Molecular Autism (2018) 9:7 Page 11 of 11 young people with high-functioning autism spectrum disorders. Autism. 2008;12:607–26. 38. Happe F, Frith U. The weak coherence account: detail-focused cognitive style in autism spectrum disorders. J Autism Dev Disord. 2006;36:5–25. 39. Loth E, Gómez JC, Happé F. When seeing depends on knowing: adults with autism spectrum conditions show diminished top-down processes in the visual perception of degraded faces but not degraded objects. Neuropsychologia. 2010;48:1227–36. 40. Verhallen RJ, Bosten JM, Goodbourn PT, Lawrance-Owen AJ, Bargary G, Mollon JD. General and specific factors in the processing of faces. Vis Res. 2017;141: 217-27. 41. Bird G, Cook R. Mixed emotions: the contribution of alexithymia to the emotional symptoms of autism. Transl Psychiatry. 2013;3:e285. 42. Cook R, Brewer R, Shah P, Bird G. Alexithymia, not autism, predicts poor recognition of emotional facial expressions. Psychol Sci. 2013;24:723–32. 43. Berthoz S, Haviland MG, Riggs ML, Perdereau F, Bungener C. Assessing alexithymia in French-speaking samples: psychometric properties of the Observer Alexithymia Scale-French translation. Eur Psychiatry. 2005;20:497–502. 44. Lombardo MV, Barnes JL, Wheelwright SJ, Baron-Cohen S. Self-referential cognition and empathy in autism. PLoS One. 2007;2:e883. 45. Nemiah JC, Freyberger H, Sifneos PE. Alexithymia: a view of the psychosomatic process. In: Hill OW, editor. Modern trends in psychosomatic medicine. London: Butterworths; 1976. 46. Klin A, Jones W, Schultz R, Volkmar F, Cohen D. Visual fixation patterns during viewing of naturalistic social situations as predictors of social competence in individuals with autism. Arch Gen Psychiatry. 2002;59:809–16. 47. Loth E, Charman T, Mason L, Tillmann J, Jones EJH, Wooldridge C, Ahmad J, Auyeung B, Brogna C, Ambrosino S, et al. The EU-AIMS Longitudinal European Autism Project (LEAP): design and methodologies to identify and validate stratification biomarkers for autism spectrum disorders. Mol Autism. 2017;8:24. https://doi.org/10. 1186/s13229-017-0146-8. 48. Charman T, Loth E, Tillmann J, Crawley D, Wooldridge C, Goyard D, Ahmad J, Auyeung B, Ambrosino S, Banaschewski T, et al. The EU-AIMS Longitudinal European Autism Project (LEAP): clinical characterisation. Mol Autism. 2017; 8:27. https://doi.org/10.1186/s13229-017-0145-9. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries � Our selector tool helps you to find the most relevant journal � We provide round the clock customer support � Convenient online submission � Thorough peer review � Inclusion in PubMed and all major indexing services � Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit

Journal

Molecular Autism – Springer Journals

Published: Dec 1, 2018

Keywords: neurology; neurosciences; neuropsychology; psychiatry; pediatrics; human genetics

There are no references for this article.