Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Mar 1.
Published in final edited form as: J Autism Dev Disord. 2024 Feb 3;55(3):832–842. doi: 10.1007/s10803-024-06268-8

Valence and Intensity of Emotional Expression in Autistic and Non-Autistic Toddlers

Jessie B Northrup 1, Carla A Mazefsky 2, Taylor N Day 3
PMCID: PMC11297193  NIHMSID: NIHMS1964584  PMID: 38315319

Abstract

Purpose:

Differences in emotional experience and expression have long been recognized as common in the presentation of autism, yet research examining emotional expression in early childhood is limited, with mixed findings. Understanding emotional reactivity and expression in autism in early life is an essential step towards uncovering the mechanisms of these risks and identifying targets for intervention.

Methods:

The present study examined emotional expression in autistic (N = 17) and non-autistic (N = 20) toddlers (mean age = 25.27; SD = 1.88) during emotion elicitation tasks aimed at eliciting joy, frustration, and unease. Video recorded tasks were coded in ten second intervals for emotional valence and intensity, and the following variables were computed: proportion of time in positive, neutral, and negative affect; maximum intensity of positive and negative affect; and range of affect (i.e., most negative to most positive intensity).

Results:

Autistic toddlers spent more time in neutral facial expressions, less time displaying positive affect, and had somewhat less intense positive emotional expression than non-autistic peers. Small differences were apparent in intensity of negative affect expression, while no differences emerged in duration of time spent in negative affect.

Conclusion:

Findings emphasize that differences may be more apparent in duration, rather than intensity of emotional expression, and that it may be particularly important to examine periods of “neutral” affect in young autistic children. Future research should consider the best ways to understand emotional reactivity in this population considering their unique interests, challenges, and communication styles.

Keywords: autism, toddlers, emotion expression, emotional reactivity, early childhood

Introduction

Autism is a neurodevelopmental condition characterized by differences and challenges in the areas of social interaction and communication as well as preferences for repetition and routine, intense and nonconventional interests, and sensory challenges (American Psychiatric Association, 2013). Differences in emotional experience and expression have also long been recognized as a common part of the presentation of autism. For example, reduced facial expressions and social smiling are included in many diagnostic evaluations (e.g., the ADOS-2; Lord et al., 2012) and have been reported in the context of unstructured play interactions with parents (Snow, Hertzig, & Shapiro, 1987) and semi-structured interactions with experimenters (Kasari, Sigman, Mundy, & Yirmiya, 1990; Yirmiya, Kasari, Sigman, & Mundy, 1989). Moreover, intense negative emotional reactions, challenges with emotion regulation, and risk for mental health disorders and suicidality are increasingly recognized as prevalent challenges in autism (Cibralic, Kohlhoff, Wallace, McMahon, & Eapen, 2019; Conner et al., 2021; O’Halloran, Coey, & Wilson, 2022). Understanding emotional reactivity and expression in autism in early life is an essential step towards uncovering the mechanisms of these risks and identifying targets for intervention.

Studies that have compared parent-reported temperamental features in young autistic children compared to non-autistic peers have reported reduced positive affect and increased negative affect in autistic children starting in toddlerhood (Garon et al., 2009, 2016; Macari, Koller, Campbell, & Chawarska, 2017). Only a handful of observational studies have examined emotional reactivity and expression in young autistic children, and results have been mixed. Jahromi, Meek, & Ober-Reynolds (2012) did not find differences in facial/bodily negativity intensity or duration between a group of autistic children and non-autistic peers (matched on verbal ability and mental age; mean chronological age autism group: 59 months, non-autism group: 50 months) engaging in frustrating tasks. However, they did find greater intensity and duration of resignation in the autistic children and a marginally significant difference in negative vocalizations in the autism group. Hirschler-Guttenberg, Golan, Ostfeld-Etzion, & Feldman (2015) examined positive and negative emotionality in preschoolers (mean age: 58.47 months) during tasks designed to elicit fear (i.e., the masks task from the Laboratory Temperament Assessment Battery; Lab-TAB) and joy (i.e., puppets task from the lab-TAB). Autistic children displayed more negative emotionality during the masks task, but only when they were with their fathers (i.e., not with their mothers) and less positive emotionality overall. Macari and colleagues (2018) examined the intensity, as measured by peak emotional expression, of expressions of fear, joy, and anger in response to tasks design to elicit these emotions in autistic and non-autistic toddlers (mean age around 20 months). Autistic toddlers displayed less intense fear than developmentally matched neurotypical peers and more intense frustration/anger than developmentally delayed peers (but not than neurotypical peers). No differences were found in intensity of joyful expressions. Similarly, Day, Mazefsky, and Wetherby (2022) found that overall affect was more negative in response to frustration tasks in autistic toddlers (mean age 25 months) compared to non-autistic peers, but did not find differences in affect during fear eliciting tasks.

Some of the variability in findings across these studies may be the result of differences in the way emotional expression was quantified. For example, Hirschler-Guttenberg et al. (2015) examined amount of time displaying negative and positive emotionality, Macari et al. (2018) reported on peak emotional intensity, and Day et al. (2022) reported on mean affect (from negative to positive). Each of these variables captures only part of the larger picture of emotional expression, potentially missing other important features. For example, examining only peak emotion or average emotion may miss differences in amount of time spent in negative affect, while examining only duration of affect might miss important differences in intensity. Notably, none of these studies have specifically examined time spent in neutral affect or overall range of affective expression (from negative to positive), despite evidence that reduced facial expression is common in autistic individuals (Trevisan, Hoskyn, & Birmingham, 2018).

Present Study

The present study aimed to characterize emotional valence (negative, neutral, and positive), intensity, and range (i.e., distance between most intense negative and most intense positive emotion) in response to emotionally evocative tasks in autistic and non-autistic toddlers. Children participated in nine tasks from the Lab-TAB aimed at eliciting joy, frustration, and unease1. The study reports on the same sample and data described in Day et al. (2022; described briefly above), in which authors reported on mean valance of facial, bodily, and vocal affect during each task. Results indicated that autistic toddlers differed from non-autistic children in mean valence of their facial, bodily, and vocal expression on a few tasks. Specifically, autistic children displayed more negatively valanced facial expressions during one frustration task and one joy task and more negatively valanced vocal expressions and bodily expressions during one frustration task. No differences were found in response to the unease tasks.

As mentioned above, findings regarding mean valence may miss important aspects of the emotional experience of children. For example, a mean facial expression valence of zero might indicate that a child was neutral the entire time or may indicate equal amounts of intense positive and negative affect. Thus, knowing about the proportion of time spent in different emotional valence states (positive, negative, and neutral) and about the peak intensity of positive and negative affect would provide a more complete picture of emotional reactivity. The present study aimed to fill this gap. Specifically, we calculated the proportion of time spent in positive, neutral, and negative affect, peak positive and negative affect, and the range of affect expressed within each type of emotion elicitation task: joy, frustration, and unease. The present study focused solely on facial expressions, since correlations in the initial study between facial, bodily, and vocal expressions were high (ranging from .69 to .73).

Analyses were aimed at asking three primary questions: 1) Does emotional intensity and valence differ depending on the type of emotion being elicited (main effect of task type)?; 2) Do autistic children differ from non-autistic children in emotional valence and intensity (main effect of diagnosis)?; 3) Do autistic children differ from non-autistic children in their response to particular emotion elicitations (interaction between task type and diagnosis)?

Methods

Participants

Children in this study were recruited from the FIRST WORDS project, an ongoing, prospective longitudinal study focused on screening and early identification of autism. As part of the FIRST WORDS project, children were screened for communication delays and autism via the Early Screening for Autism and Communication Disorders (ESAC; Wetherby et al., 2021, 2015). Parents were then invited for further evaluation following the ESAC if: (1) his/her child received a positive screen for a communication delay or autism based on parent-report; (2) he/she expressed concern about his/her child’s development; or (3) his/her child was randomly selected after receiving a negative screen. Invited children completed a diagnostic evaluation between 18–26 months, which included the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) Toddler Module (Luyster et al., 2009), Mullen Scales of Early Learning (MSEL; Mullen, 1995), Vineland Adaptive Behavior Scales, Third Edition (VABS-3; Sparrow, Cicchetti, & Saulnier, 2016), Systematic Observation of Red Flags of ASD (SORF; Dow, Day, Kutta, Nottke, & Wetherby, 2020) based on a home observation, and parent-report questionnaires. Evaluations were conducted by master’s or PhD-level clinicians who had extensive experience in early child development and autism, including research reliability on the ADOS-2. Toddlers received a diagnostic determination based on the clinician’s clinical impression and assessment scores.

Children from the FIRST WORDS project were recruited to participate in the present study, focused on emotion reactivity and regulation, if they were between the ages of 22 and 28 months between September 2018 and June 2019 and were able to travel to one of the four evaluation sites (Tallahassee, Panama City, Tampa, Ft. Myers/Naples). Children were excluded from the present study if a diagnosis of autism was deferred during the diagnostic evaluation (i.e., autism could not be confirmed or ruled out by the clinician). In total, 66 families were invited to participate. Twenty-nine families (43.9%; 18 children with an autism diagnosis; 11 children without an autism diagnosis) declined participation. The remaining 37 families (17 children with an autism diagnosis, 20 without an autism diagnosis) participated in the study.

The proportions for participants vs. non-participants were comparable with regards to autism diagnostic status, χ2 = 1.18, p = .277, and reported ethnicity (n = 43), χ2 = 1.52, p = .22. Families who declined participation were somewhat more likely to be male, χ2 = 3.02, p = .08, and were less likely to identify as White, based on information from parents who disclosed race (n = 47), χ2 = 10.44, p = .001.

Procedures

The parent and child attended a lab visit lasting approximately 1–1.5 hours which included a series of tasks selected from the well-studied Laboratory Temperament Assessment Battery (Lab-TAB; Goldsmith, Reilly, Lemery, Longley, & Prescott, 1999; Goldsmith & Rothbart, 1999; Planalp, Van Hulle, Gagne, & Goldsmith, 2017), a standardized assessment of early temperament. The Lab-TAB is made up of short activities, or probes, that are designed to mimic everyday situations that elicit emotional reactions. For the present study, nine tasks designed to elicit positive affect, frustration, and unease were selected (see Table 1 for descriptions of all tasks). Minor modifications were made to tasks to ensure developmental appropriateness for the current sample (e.g., parent remaining in room for duration of visit, but occupied with completing questionnaires and instructed only to respond to their child’s initiations). The accompanying guardian (89.2% female caregiver, 10.8% male caregiver) was in the room for the duration of the visit. Task order was standardized to balance positive tasks with frustration and unease tasks (see Table 1), with a brief activity (short video) to allow the child to return to baseline following each task and longer structured breaks after every third task. All frustration and unease tasks incorporated a positive resolution phase (e.g., after having a toy taken away, the toy was returned to the child for play). The last author (TND) was the examiner for all tasks and could not be naïve to diagnostic status due to clinical training and specialization in the early diagnosis of autism.

Table 1:

Description of Laboratory Temperament Assessment Battery (Lab-TAB) tasks. Tasks are listed in order they were presented during the lab visit.

Task Emotion Type Description
Stranger Approach Unease An unfamiliar female adult (i.e., research assistant the child had not yet met) wearing a baseball hat and large sunglasses slowly approached and attempted to interact with the child.
Make That Car Go Joy The child and examiner raced cars down ramps for 2 minutes.
End of the Line Frustration The child was told to stop playing with a desirable toy and it was removed from his/her reach for 30 seconds.
Attractive Toy in a Transparent Box Frustration Two-minute task in which a desired toy is placed in a clear locked box. The child was provided the box with no means of unlocking it.
Popping Bubbles Joy The child was encouraged to use his/her hands and feet to pop bubbles.
Remote Controlled Spider Unease An unfamiliar animal moved towards and away from the child several times.
Halloween Mask Unease The research assistant wore a Halloween mask and kneeled in front of the child for 30 seconds.
Modified Peek-a-Boo Joy The examiner played peek-a-boo with the child using the playhouse as the prop to hide.
Gentle Arm Restraint by Parent Frustration The parent physically restrained the child so that he/she could not reach an interesting toy for 30 seconds.
Open in a new tab

Informed consent for the parent’s and child’s participation was collected from the parent prior to the start of the study visit. Parents also consented to share archival data collected as part of their child’s participation in the FIRST WORDS project. Procedures were approved by the Florida State University’s Institutional Review Board. Overall, parental acceptability of the procedures was high. Parents consented to all the tasks, with the exception of one parent who declined the child’s participation in the Mask task only. No families ended the study before all Lab-TAB measures were presented. One child did not complete the Remote-Control Spider task because of technical difficulties.

Measures

Assessment Data.

Children underwent a full diagnostic assessment prior to being recruited for the current study. Those data were made available to determine eligibility for the present study. The following measures are included in descriptive data or analyses: ADOS–2 Toddler Module calibrated severity scores (CSS; range: 1–10) to estimate autism symptom severity (Esler et al., 2015), MSEL Verbal and Non-verbal T-scores (Mullen, 1995) to assess developmental level, and demographics collected via parent-report. Assessments occurred on average 4.46 months (SD = 2.27; Range: 2.27–9.41) before the Lab-TAB visit and time between assessment and Lab-TAB visit did not differ between autistic and non-autistic toddlers, t (314) = 0.15, p = .87.

Parents also completed the Child Behavior Checklist (CBCL/1.5–5; Achenbach & Rescorla, 2000) around the time of the Lab-TAB assessment. The CBCL/1.5–5 is a 99-item parent report measure of children’s psychiatric and behavioral functioning. The CBCL provides composite scale t-scores for Internalizing and Externalizing Problems as well as a Total Problems score encompassing all 99 items. Higher scores indicate more problems in these areas.

Coding Protocol.

Trained undergraduate research assistants coded specified behaviors from video recordings of the Lab-TAB tasks using a detailed coding system tapping emotional reactivity and regulation. Videos were coded in 10-second intervals for each of the nine tasks; only the time window for the intended emotion was coded and periods of task setup and recovery were not included. The coding protocol was modified from several different coding schemes, including those from the Lab-TAB manuals (Goldsmith et al., 1999; Goldsmith & Rothbart, 1999), the Autism & Developmental Disorders Inpatient Research Collaborative (ADDRIC) coding manual (Northrup et al., 2020), and Nuske and colleagues’ work (2017). Emotion regulation strategies, and valence of emotional reactivity as expressed in facial expression, vocalizations, and body movements/positioning were coded. The full coding system and descriptive results are described in Day et al. (2022). The present study focuses only on the coding of facial expression. Toddlers’ emotional facial expressions were characterized based on valence and intensity of affect, on a −3 (highly negative) to +3 (highly positive) scale. A code of 0 indicated no change in expression from the child’s baseline state (i.e., neutral or flat affect); a code of −1 or +1 indicated subtle, fleeting, or ambiguous negative/positive affect; a code of −2 or +2 indicated an obvious expression of negative/positive affect (e.g., a clear frown or smile); and a code of −3 and +3 indicated a high-intensity expression of negative/positive affect typically involving multiple indicators of emotion (e.g., big smile and raised eyebrows for intense joy).

A gestalt level of engagement for each task was also coded, in which the child had to show awareness of the task (e.g., briefly look, approach, or make some communicative bid to present adults or to relevant materials). If the child was disengaged in the task, codes were not applied because the emotional state was likely unrelated to the task itself. A total of five children did not engage in one task each; they engaged in all other tasks. Four of these children (two autism, two non-autism) did not shift their attention in any capacity during the Scary Mask task. In addition, one autistic child did not actively participate in the Make That Car Go task.

Coders were trained by the last author (TND) on coding scheme definitions and then calibrated with TND and each other before proceeding to independent coding. Eight of the 37 videos (21.6%) were coded by two research assistants to assess inter-rater reliability. The kappa value was 0.69 (0.41 – 0.83) for facial affect. The full coding manual is available upon request. All research assistants were naïve to diagnostic status and to the hypotheses of the study.

Data Analysis

The following six variables were calculated from the coded data: Proportion of intervals spent in positive, negative, and neutral affect, maximum affect (peak positive affect), minimum affect (peak negative affect), and range of affect. Proportion of intervals spent in each affective state was calculated by dividing the number of intervals rated as positive (i.e., greater than zero), negative (i.e., less than 0), and neutral (i.e., zero) by the total number of intervals for that task. Peak positive and negative affect were, respectively, the highest and lowest coded values for affect within each task. Finally, range of affect was calculated as the distance between the minimum and maximum affect ratings.

We ran linear mixed-effect models using the lme4 package (Bates, Mächler, Bolker, & Walker, 2015) in R (R Core Team, 2023) with each of the six variables as dependent variables in six separate models. Lab-TAB task type (positive, frustration, unease), diagnosis (autism, non-autism), and the interaction between task and diagnosis were entered as fixed-effects. A random intercept for participant was included in all models. For each model, we then successively tested whether including a random intercept for task (capturing variability among the nine individual tasks) and a random slope for task type for each participant (capturing individual differences in the effect of task type) would improve model fit (using the anova command in R to conduct a chi-square analysis of deviance). For each model, random effects that improved model fit were included (see supplemental materials for all final models). Significance was calculated using the lmerTest package (Kuznetsova, Brockhoff, & Christensen, 2017), which applies Satterthwaite’s method to estimate degrees of freedom and generate p-values for mixed models. Significant main effects and interactions were followed up with post-hoc comparisons with Sidak adjustment using the emmeans package in R with effect sizes calculated using the eff_size function (Lenth, 2023).

Results

Table 2 presents descriptive statistics for each of the dependent variables of interest for the autism and non-autism groups.

Table 2:

Descriptive Statistics of Demographic and Developmental Characteristics

non-autism (n = 20) autism (n = 17) Statistical comparison
Age at Visit [months, M (SD)] 24.60 (1.79) 26.06 (1.71) t=−2.52, p=.02
Sex [female, n (%)] 7 (35%) 12 (71%) χ2 =3.34, p=.07
Race [white, n (%)] 18 (90%) 8 (67%) χ2 =1.37, p=.24
Ethnicity [non-Hispanic, n (%)] 19 (95%) 8 (73%) χ2 =1.46, p=.23
Maternal education [4-year degree or higher, n (%)] 14 (70%) 7 (54%) χ2 =.33, p=.57
MSEL Verbal T-score [M (SD)] 52.55 (10.68) 28.32 (7.69) t=7.79, p<.001
MSEL Non-Verbal T-score [M (SD)] 53.17 (6.81) 40.35 (10.33) t=4.52, p<.001
ADOS-2 Toddler Module CSS [M (SD)] 2.40 (.94) 7.59 (1.87) t=−10.90, p<.001
CBCL Internalizing T-score [M (SD)] 45.61 (9.65) 58.94 (10.1) t=−3.92, p<.001
CBCL Externalizing T-score [M (SD)] 50.11 (8.82) 57.50 (9.82) t=−2.30, p=.02
CBCL Total Problems T-score [M (SD)] 48.44 (9.31) 60.13 (9.46) t=−3.62, p=.001
Open in a new tab

Note. Some parents of autistic children declined to report race (n=5), ethnicity (n=6), and maternal education (n=4); ADOS-2 Toddler Module CSS = Autism Diagnostic Observation Schedule, Second Edition Toddler Module Calibrated Severity Score (range: 1–10); MSEL = Mullen Scales of Early Learning; CBCL = Child Behavior Checklist; CBCL scores missing for three participants (2 non-autism; 1 autism); T-values are reported for continuous data and χ2 are reported for categorical data.

Preliminary Analyses

Demographic and developmental characteristics for the 37 toddlers who participated in the present study are summarized in Table 2, including comparisons between the autistic and non-autistic toddlers. Significant differences between the autism and non-autism groups were found in age at first visit, MSEL verbal and non-verbal T-scores, and CBCL internalizing, externalizing, and total problem t-scores (see Table 2). Autistic children were on average older, had lower verbal and non-verbal abilities, and had higher internalizing, externalizing, and total problems scores on the CBCL. Although not statistically significant, autistic children were also somewhat more likely to be male, p = .07.

Given differences between autism and non-autism groups in age, sex, MSEL scores, and CBCL scores, we first examined whether the dependent variables of interest were related to these variables. T-tests indicated that female children spent a higher proportion of time in positive affect, t(34.3) = 2.37, p = 0.02, and less time in negative affect, t(34.34) = −2.04, p = 0.048, than male children. No other variables differed by child sex. None of the variables of interest were significantly correlated with child age in months. Proportion of time spent in positive affect was positively correlated with MSEL verbal T-score, r = .37, p = .025. No other variable was correlated with MSEL verbal or non-verbal scores. Finally, none of the variables of interest were significantly correlated with CBCL internalizing, externalizing, or total problems t-scores. Below, we controlled for variables where appropriate (e.g., control for child sex and verbal ability in analysis of proportion of time spent in positive affect).

Proportion of Time Spent in Positive, Negative, and Neutral Affect

Table 4 contains the results of models predicting proportion of time spent in positive, negative, and neutral affect and Figure 1 displays the means for autism and non-autism children for proportion of time spent in each affective state during each of the three task types.

Table 4.

Linear mixed-effects model ANOVA results for proportion time spent in positive, negative, and neutral affect.

Proportion time spent in positive affect Proportion time spent in negative affect Proportion time spent in neutral affect
Task type F (2, 285.34) = 79.23 p < .001 F (2, 9.01) = 6.79 p = 0.016 F (2, 6.72) = 1.22 p = 0.353
autism diagnosis F (1, 33.26) = 4.23 p = 0.048 F (1, 37.08) = 0.55 p = 0.464 F (1, 34.94) = 5.1 p = 0.03
Task type * autism diagnosis F (2, 285.56) = 1.09 p = 0.336 F (2, 38.23) = 2.37 p = 0.107 F (2, 279.44) = 4.41 p = 0.013
Sex of Child F (1, 33.19) = 1.55 p = 0.223 F (1, 62.73) = 1.52 p = 0.222
Mullen verbal t-score F (1, 33.12) = 0.3 p = 0.59
Peak Positive Affect Peak Negative Affect Range affect
Task type F (2, 11.22) = 11.36 p = 0.002 F (2, 10.27) = 11.62 p = 0.002 F (2, 7.31) = 0.45 p = 0.656
Autism diagnosis F (1, 41.44) = 3.69 p = 0.062 F (1, 35.03) = 3.99 p = 0.054 F (1, 35.73) = 0.21 p = 0.648
Task type * autism diagnosis F (2, 45.49) = 0.06 p = 0.944 F (2, 43.48) = 1.46 p = 0.244 F (2, 280.75) = 1.29 p = 0.277
Open in a new tab

Figure 1.

Figure 1.

Open in a new tab

Mean proportion of time spent in positive, neutral and negative affect across emotion elicitation tasks (joy, frustration, unease) for autistic and non-autistic toddlers.

In the model predicting proportion of time spent in positive affect, the main effect of task type and main effect of diagnosis were both significant. Autistic children had significantly less positive affect then non-autistic children, regardless of task type (p = .047; effect size [ES] = .68). Children (regardless of diagnosis) displayed more positive affect in the joy tasks than in the unease (p < .001; ES = 1.73) or frustration tasks (p < .001; ES = 1.90). The difference between the frustration tasks and the unease tasks was not significant (p = .55, ES = .16).

Turning to proportion of time spent in negative affect, there was a significant main effect of task type. The main effect of diagnosis and interaction between diagnosis and task type were not significant. Post-hoc analyses indicated that the children displayed more negative affect in the frustration tasks than the joy tasks (p = .017; ES = 1.73). The difference between the amount of negative affect displayed in the joy tasks verses the unease tasks was large, but non-significant (p = .13; ES = 1.09). The difference between the frustration tasks and the unease tasks was also non-significant (p = .47, ES = .65)

Finally, the model predicting proportion of time spent in neutral affect revealed a significant main effect of diagnosis and significant interaction between task type and diagnosis. Post-hoc analyses revealed that the autism group had more neutral affect than the non-autism group in the joy (p = .03; ES = .58) and unease tasks (p = .003, ES = .84), but not in the frustration task (p = .85; ES = .05). For the autism group, the frustration and joy tasks had a similar amount of neutral affect, while the unease tasks had nearly twice as much. For the non-autism group, the unease and frustration tasks had a similar amount of neutral affect, while the Joy tasks had only about half as much.

Peak Positive and Negative Affect

Results of models predicting peak positive and negative affect (maximum and minimum affect) are presented in Table 4. In the model predicting peak positive affect, we found a main effect of task type. Peak positive affect was higher in the Joy tasks than in the unease (p = .03, ES = 1.18) or frustration tasks (p = .002; ES = 1.76) but did not differ between the unease and frustration tasks (p = .32; ES = .58). Non-autistic children had somewhat higher peak positive affect than autistic children, although the main effect of diagnosis did not quite reach significance (p = .06; ES = .35). The interaction between task type and diagnosis was not significant.

With regards to peak negative affect, we again found a significant main effect of task type. Children displayed more intense negative affect in the frustration (p = .009; ES = 1.45) and unease (p = .03; ES = 1.12) than in the joy tasks. The frustration and unease tasks did not differ from one another (p = .71; ES = .34). In addition, autistic children had more intense negative affect than non-autistic children, although again this difference did not quite reach significance (p = .053; ES = .424). The interaction between task type and diagnosis was not significant.

Range of Affect

Finally, we examined the total range of affect (i.e., distance between the minimum and maximum affect ratings) displayed in each task. Range of affect did not differ by task type or diagnostic group.

Discussion

This study adds to the literature on emotional reactivity and expression in autism by reporting on valence, intensity, and range of emotional expression across three different emotion elicitation tasks in toddlers with and without an autism diagnosis. Autistic toddlers spent more time in neutral facial expressions, less time displaying positive affect, and had somewhat less intense positive emotional expression than non-autistic peers. With regards to negative emotion, small differences were apparent in intensity of negative affect expression, while no differences emerged in duration of time spent in negative affect.

In a previously published study with the same data, Day et al. (2022) reported that autistic toddler’s mean affect was less positive/more negative than non-autistic toddlers, particularly in frustration and joy tasks. The present study adds to our understanding of these original results, indicating that these mean differences were driven largely by increased time in neutral affect and decreased time in positive affect in the autism group (see Figure 1). Autistic toddlers did not spend significantly more time in negative affect than their non-autistic peers, and while they did display somewhat less intense positive affect and more intense negative affect, these were small effects that ultimately did not reach significance. Notably, contrary to our expectation, we also did not find differences in the range of emotional expression between autistic and non-autistic toddlers, further emphasizing that differences are more apparent in the duration of time spent in affective states, rather than in the intensity of emotional expression when it occurs.

The present paper adds to the mixed literature on emotional expression in response to emotion elicitation tasks in young autistic children. We replicate several other studies that have failed to show overall differences (or found only small/marginal differences) in negative facial affect between autistic and non-autistic children. Macari et al., (2018) found only a marginal effect for peak intensity of anger expressions in response to anger tasks and reduced fear expression in response to unease tasks. Jahromi et al. (2012) found no differences in facial negativity in response to frustration tasks, and Hirschler-Guttenberg et al. (2015) did not find differences in duration of negativity during an unease task when the child was with their mother (although the autistic children did display more negativity when with their father). Taken together with the present study, these results suggest that evidence from observational studies for increased intensity and duration of negative emotion expression in autistic children is relatively weak.

Our findings of reduced duration of positive emotion in autistic children are consistent with findings from Hirschler-Guttenberg et al. (2015) that showed less time spent in positive emotionality in autistic children during Lab-TAB tasks, and with parent reports of reduced positive affect in young autistic children (Macari, 2017, Garon, 2009, 2016). Our findings also suggest that there may be reduced intensity of positive expression, although these effects were small and did not reach significance. Macari et al. (2018) found no significant difference in intensity of positive affect in response to joy-eliciting tasks. Further research with larger samples will be necessary to clarify these results.

Notably, the present study is the first, to our knowledge to specifically examine duration of neutral facial expression during emotion elicitation tasks. This is surprising given that reduced emotional expression is commonly described in autism, is a part of many diagnostic evaluations, and has been reported in a number of studies examining facial expression in autism across the lifespan (see Trevisan, Hoskyn, & Birmingham, 2018 for a review). Results suggest more time spent in neutral expression among autistic toddlers than non-autistic toddlers, particularly during tasks designed to elicit joy and unease in this age group.

There are several possible explanations for why we might see more neutral facial expressions during emotion elicitation tasks in young autistic children. Young autistic children may, in fact be experiencing fewer emotions during these tasks, either because the tasks are not as evocative for them as for neurotypical children or because they do not experience these emotions as readily in general. Alternatively, young autistic children may simply not communicate their emotions via facial affect to the same degree as peers, despite experiencing the emotions internally. For example, autistic toddlers may only communicate positive and negative emotions when they are experienced at a higher level of intensity, which would explain why we found larger differences in duration of emotional expressions, but small or non-existent differences in intensity or range of emotion. Moreover, the positive correlation we found between child verbal ability and positive affect may support the idea that differences in emotional expression may be at least partially explained by differences in communication of emotion.

Work incorporating physiological measures of arousal may help to clarify these results. For example, Zantinge, van Rijn, Stockmann, & Swaab (2018) reported on the concordance between heart rate arousal and behavioral expressions of fear/unease during an elicitation task. They found a positive correlation between physiological arousal and behavioral expression in the neurotypical group, but no correlation between these measures in the autism group, suggesting that the behavioral expressions of emotional arousal may not be matching physiological expressions for young autistic children. More research utilizing both physiological and behavioral measures is needed to better understand differences in emotional expression in autism.

Finally, the Lab-TAB tasks were designed to elicit emotions in neurotypical children, and therefore differences in emotional expression may be due to differences in how effective these tasks are at eliciting these emotions in autistic children. For example, Jacques, Courchesne, Mineau, Dawson, and Mottron (2022) examined frequency and duration of time spent in positive, negative, and neutral emotions during the Montreal Stimulating Play Situation, a paradigm meant to elicit positive emotions that was developed specifically to incorporate potential autistic interests and did not find differences between autistic and non-autistic children in the frequency or duration of positive, negative, or neutral emotions. The use of frustration and unease tasks designed for neurotypical children may also explain why we (and others using Lab-TAB tasks) did not find convincing results for differences in duration or intensity of negative affect, despite reports of increased negative reactivity from parents of young autistic children (Samson et al., 2014; Samson, Wells, Phillips, Hardan, & Gross, 2015). In line with this notion, the lack of correlation between negative reactivity to the Lab-TAB tasks and CBCL scores in the present study suggests that the Lab-TAB is not eliciting the types of emotional and behavioral reactions that parents are reporting in everyday life. Autistic children may have unique triggers (e.g., sensory stimuli, unexpected changes to routine) compared to neurotypical children, and future research should consider developing emotion elicitation tasks or paradigms that are more individually tailored to individual children’s unique interests and challenges.

Limitations and Conclusions

While the present study has a number of strengths, including a sample of very young children (i.e., under 2.5 years of age), detailed observational coding of emotional expression, the use of numerous and varied emotion elicitation tasks, and reporting of several variables to describe emotional expression, it also has important limitations. First and foremost, our study sample was relatively small and therefore may have been under powered to detect small effects. Null results should be interpreted with caution. Our sample was also homogenous with regards to race, and the two diagnostic groups (autism and non-autism) differed on age and sex. Further research with larger and more diverse samples is needed to help characterize early emotional expression and reactivity. Finally, observational coding of affect from video recordings is a difficult and time-consuming task. It should be noted that interrater reliability was mostly moderate and therefore some caution is warranted in interpreting results.

Despite these limitations, the present study offers valuable insight to our knowledge of emotional reactivity and expression in young autistic children. Findings emphasize that differences may be more apparent in duration, rather than intensity of expression, and that it may be particularly important to examine periods of “neutral” affect in this population. From a clinical perspective, these results raise important questions about the distinction between the experience of emotion and communication of that emotion. Future research should consider the best ways to understand emotional reactivity in young autistic children considering their unique interests, challenges, and communication styles. In particular, we suggest that collection of physiological data along with observational data, and the use of tasks specifically designed with autistic children in mind would help to further advance our understanding of this important topic.

Supplementary Material

Supplemental Materials

Figure 2.

Figure 2.

Open in a new tab

Means and 95% confidence intervals for peak intensity positive and negative affect across emotion elicitation tasks (joy, frustration, unease) for autistic and non-autistic toddlers.

Table 3.

Descriptive Statistics for all Primary Variables of Interest

Non-ASD (N = 20) ASD (N = 17)
M (SD) Range M (SD) Range
Positive Tasks
 Proportion Positive Affect 0.85 (0.2) 0.2 – 1 0.61 (0.35) 0 – 1
 Proportion Negative Affect 0.03 (0.08) 0 – 0.4 0.13 (0.25) 0 – 1
 Proportion Neutral Affect 0.13 (0.17) 0 – 0.6 0.25 (0.29) 0 – 0.92
 Peak Negative Affect 0.48 (1.11) −2 – 2 −0.28 (1.53) −3 – 3
 Peak Positive Affect 2.55 (0.59) 1 – 3 2.1 (1.18) −3 – 3
 Range Affect 2.07 (1.01) 0 – 4 2.38 (1.38) 0 – 6
Frustrating Tasks
 Proportion Positive Affect 0.29 (0.35) 0 – 1 0.16 (0.25) 0 – 1
 Proportion Negative Affect 0.44 (0.39) 0 – 1 0.56 (0.37) 0 – 1
 Proportion Neutral Affect 0.27 (0.28) 0 – 0.86 0.28 (0.28) 0 – 1
 Peak Negative Affect −1.23 (1.36) −3 – 3 −1.75 (1.32) −3 – 2
 Peak Positive Affect 0.55 (1.72) −3 – 3 0.14 (1.65) −3 – 3
 Range Affect 1.78 (1.37) 0 – 6 1.88 (1.35) 0 – 5
Unease Tasks
 Proportion Positive Affect 0.34 (0.35) 0 – 1 0.19 (0.24) 0 – 0.8
 Proportion Negative Affect 0.36 (0.37) 0 – 1 0.32 (0.36) 0 – 1
 Proportion Neutral Affect 0.3 (0.31) 0 – 1 0.48 (0.31) 0 – 1
 Peak Negative Affect −1.07 (1.42) −3 – 2 −1.17 (1.2) −3 – 0
 Peak Positive Affect 1.17 (1.45) −3 – 3 0.87 (1.12) −1 – 3
 Range Affect 2.24 (1.43) 0 – 6 2.04 (1.08) 0 – 4
Open in a new tab

Note. M = Mean; SD = Standard Deviation

Acknowledgements:

This research was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development grant R01HD078410 (PI: Wetherby). Dr. Jessie Northrup was supported by NIMH K23 MH127420 during the preparation of this manuscript. These data were collected as part of Dr. Taylor Day’s dissertation study. A special thank you to Dr. Amy Wetherby for her mentorship and guidance as well as the FIRST WORDS Project staff for their support and to the participating families.

Footnotes

Statements and Declarations:

Conflict of interest: The authors have no conflicts of interest to declare.

1

These tasks are typically called “fear” tasks. They are designed to elicit emotions such as unease, worry or nervousness to novel or social stimuli. We feel the word “fear” implies a level of negative emotion that is unwarranted by the tasks themselves, which all mimic experiences preschoolers encounter in regular life, and therefore have chosen a term we feel more accurately describes these tasks: ‘unease’.

Contributor Information

Jessie B. Northrup, University of Pittsburgh School of Medicine, Department of Psychiatry, Pittsburgh, PA USA

Carla A. Mazefsky, University of Pittsburgh School of Medicine, Department of Psychiatry, Pittsburgh, PA USA

Taylor N. Day, University of Pittsburgh School of Medicine, Department of Psychiatry, Pittsburgh, PA USA

References

  1. Achenbach TM, & Rescorla LA (2000). Manual for the ASEBA preschool forms and profiles (Vol. 30). Burlington, VT: Research Center for Children, Youth, & Families, University of Vermont. [Google Scholar]
  2. Association AP (2013). Diagnostic and statistical manual of mental disorders (DSM-5®). American Psychiatric Pub. [DOI] [PubMed] [Google Scholar]
  3. Cibralic S, Kohlhoff J, Wallace N, McMahon C, & Eapen V. (2019). A systematic review of emotion regulation in children with Autism Spectrum Disorder. Research in Autism Spectrum Disorders, 68, 101422. 10.1016/j.rasd.2019.101422 [DOI] [Google Scholar]
  4. Conner CM, Golt J, Shaffer R, Righi G, Siegel M, & Mazefsky CA (2021). Emotion Dysregulation is Substantially Elevated in Autism Compared to the General Population: Impact on Psychiatric Services. Autism Research, 14(1), 169–181. 10.1002/aur.2450 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Day TN, Mazefsky CA, & Wetherby AM (2022). Characterizing difficulties with emotion regulation in toddlers with autism spectrum disorder. Research in Autism Spectrum Disorders, 96, 101992. 10.1016/j.rasd.2022.101992 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Dow D, Day TN, Kutta TJ, Nottke C, & Wetherby AM (2020). Screening for autism spectrum disorder in a naturalistic home setting using the systematic observation of red flags (SORF) at 18–24 months. Autism Research, 13(1), 122–133. 10.1002/aur.2226 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Esler AN, Hus V, Guthrie W, Wetherby AM, Ellis S, & Catherine W. (2015). The Autism Diagnostic Observation Schedule, Toddler Module: Standardized severity scores. Journal of Autism and Developmental Disorders, 45, 2704–2720. 10.1007/s10803-015-2432-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Garon N, Bryson SE, Zwaigenbaum L, Smith IM, Brian J, Roberts W, & Szatmari P. (2009). Temperament and its relationship to autistic symptoms in a high-risk infant sib cohort. Journal of Abnormal Child Psychology, 37(1), 59–78. 10.1007/s10802-008-9258-0 [DOI] [PubMed] [Google Scholar]
  9. Garon N, Zwaigenbaum L, Bryson S, Smith IM, Brian J, Roncadin C, … Roberts W. (2016). Temperament and its association with autism symptoms in a high-risk population. Journal of Abnormal Child Psychology, 44(4), 757–769. 10.1007/s10802-015-0064-1 [DOI] [PubMed] [Google Scholar]
  10. Goldsmith HH, Reilly J, Lemery KS, Longley S, & Prescott A. (1999). The Laboratory Temperament Assessment Battery Manual (Lab-TAB): Preschool Version.
  11. Goldsmith HH, & Rothbart MK (1999). Laboratory Temperament Assessment Battery (Lab-TAB): Locomotor Version. [DOI] [PMC free article] [PubMed]
  12. Hirschler-Guttenberg Y, Golan O, Ostfeld-Etzion S, & Feldman R. (2015). Mothering, fathering, and the regulation of negative and positive emotions in high-functioning preschoolers with autism spectrum disorder. Journal of Child Psychology and Psychiatry and Allied Disciplines, 56(5), 530–539. 10.1111/jcpp.12311 [DOI] [PubMed] [Google Scholar]
  13. Jacques C, Courchesne V, Mineau S, Dawson M, & Mottron L. (2022). Positive, negative, neutral—or unknown? The perceived valence of emotions expressed by young autistic children in a novel context suited to autism. Autism, 26(7), 1833–1848. 10.1177/13623613211068221 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Jahromi LB, Meek SE, & Ober-Reynolds S. (2012). Emotion regulation in the context of frustration in children with high functioning autism and their typical peers. Journal of Child Psychology and Psychiatry and Allied Disciplines, 53(12), 1250–1258. 10.1111/j.1469-7610.2012.02560.x [DOI] [PubMed] [Google Scholar]
  15. Kasari C, Sigman M, Mundy P, & Yirmiya N. (1990). Affective sharing in the context of joint attention interactions of Normal, Autistic, and Mentally Retarded Children. Journal of Autism & Developmental Disorders, 20(1), 87–100. 10.1007/BF02206859 [DOI] [PubMed] [Google Scholar]
  16. Kuznetsova A, Brockhoff PB, & Christensen RHB (2017). {lmerTest} Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82(13), 1–26. 10.18637/jss.v082.i13 [DOI] [Google Scholar]
  17. Lenth RV (2023). emmeans: Estimated Marginal Means, aka Least-Squares Means. Retrieved from https://cran.r-project.org/package=emmeans
  18. Lord C, Rutter M, DiLavore P, Risi S, Gotham K, & Bishop S. (2012). Autism diagnostic observation schedule–2nd edition (ADOS-2). In Los Angeles, CA: Western Psychological Corporation. [Google Scholar]
  19. Luyster R, Gotham K, Guthrie W, Coffing M, Petrak R, Pierce K, … Lord C. (2009). The Autism Diagnostic Observation Schedule — Toddler Module : A New Module of a Standardized Diagnostic Measure for Autism Spectrum Disorders. Journal of Autism and Developmental Disorders, 39, 1305–1320. 10.1007/s10803-009-0746-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Macari SL, DiNicola L, Kane-Grade F, Prince E, Vernetti A, Powell K, … Chawarska K. (2018). Emotional Expressivity in Toddlers With Autism Spectrum Disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 57(11), 828–836.e2. 10.1016/j.jaac.2018.07.872 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Macari SL, Koller J, Campbell DJ, & Chawarska K. (2017). Temperamental markers in toddlers with autism spectrum disorder. Journal of Child Psychology and Psychiatry and Allied Disciplines, 58(7), 819–828. 10.1111/jcpp.12710 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mullen EM (1995). Mullen scales of early learning. Pearson San Antonio, TX. [Google Scholar]
  23. Northrup JB, Goodwin M, Montrenes J, Vezzoli J, Golt J, Peura CB, … Mazefsky C. (2020). Observed emotional reactivity in response to frustration tasks in psychiatrically hospitalized youth with autism spectrum disorder. Autism, 24(4), 968–982. 10.1177/1362361320908108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Nuske HJ, Hedley D, Woollacott A, Thomson P, Macari S, & Dissanayake C. (2017). Developmental delays in emotion regulation strategies in preschoolers with autism. Autism Research, 10(11), 1808–1822. 10.1002/aur.1827 [DOI] [PubMed] [Google Scholar]
  25. O’Halloran L, Coey P, & Wilson C. (2022). Suicidality in autistic youth: A systematic review and meta-analysis. Clinical Psychology Review, 93, 102144. 10.1016/j.cpr.2022.102144 [DOI] [PubMed] [Google Scholar]
  26. Planalp EM, Van Hulle C, Gagne JR, & Goldsmith HH (2017). The Infant Version of the Laboratory Temperament Assessment Battery (Lab-TAB): Measurement Properties and Implications for Concepts of Temperament . Frontiers in Psychology, Vol. 8. Retrieved from https://www.frontiersin.org/articles/10.3389/fpsyg.2017.00846 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. R Core Team. (2023). R: A Language and Environment for Statistical Computing. Retrieved from https://www.r-project.org
  28. Samson AC, Phillips JM, Parker KJ, Shah S, Gross JJ, & Hardan AY (2014). Emotion dysregulation and the core features of autism spectrum disorder. Journal of Autism and Developmental Disorders, 44(7), 1766–1772. 10.1007/s10803-013-2022-5 [DOI] [PubMed] [Google Scholar]
  29. Samson AC, Wells WM, Phillips JM, Hardan AY, & Gross JJ (2015). Emotion regulation in autism spectrum disorder: evidence from parent interviews and children’s daily diaries. Journal of Child Psychology and Psychiatry and Allied Disciplines, 56(8), 903–913. 10.1111/jcpp.12370 [DOI] [PubMed] [Google Scholar]
  30. Snow ME, Hertzig ME, & Shapiro T. (1987). Expression of Emotion in Young Autistic Children. Journal of the American Academy of Child and Adolescent Psychiatry, 26(6), 836–838. 10.1097/00004583-198726060-00006 [DOI] [PubMed] [Google Scholar]
  31. Sparrow SS, Cicchetti DV, & Saulnier CA (2016). Vineland Adaptive Behavior Scales, Third Edition. San Antonio, TX: Psychological Corporation. [Google Scholar]
  32. Trevisan DA, Hoskyn M, & Birmingham E. (2018). Facial Expression Production in Autism: A Meta-Analysis. Autism Research, 11(12), 1586–1601. 10.1002/AUR.2037 [DOI] [PubMed] [Google Scholar]
  33. Wetherby AM, Guthrie W, Hooker JL, Delehanty A, Day TN, Woods J, … Lord C. (2021). The Early Screening for Autism and Communication Disorders: Field-testing an autism-specific screening tool for children 12 to 36 months of age. Autism, 25(7), 2112–2123. 10.1177/13623613211012526 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Wetherby AM, Guthrie W, Petkova E, Woods JJ, Lord C, Voccola D, … Rozenbilt L. (2015). Broadband and autism-specific screening using the Early Screening for Autism and Communication Disorders (ESAC): Moving from paper to the Smart ESAC for children 12 to 36 months of age. Oral presentation at the International Meeting for Autism Research, Salt Lake City, UT. [Google Scholar]
  35. Yirmiya N, Kasari C, Sigman M, & Mundy P. (1989). Facial Expressions of Affect in Autistic, Mentally Retarded and Normal Children. Journal of Child Psychology and Psychiatry, 30(5), 725–735. 10.1111/j.1469-7610.1989.tb00785.x [DOI] [PubMed] [Google Scholar]
  36. Zantinge G, van Rijn S, Stockmann L, & Swaab H. (2018). Concordance between physiological arousal and emotion expression during fear in young children with autism spectrum disorders. Autism, 23(3), 629–638. 10.1177/1362361318766439 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Materials
NIHMS1964584-supplement-Supplemental_Materials.docx (13.8KB, docx)

RESOURCES