Skip to main content
NCBI home page
Social Cognitive and Affective Neuroscience logoLink to Social Cognitive and Affective Neuroscience
. 2016 Feb 9;11(6):907–914. doi: 10.1093/scan/nsw015

Anxiety and social deficits have distinct relationships with amygdala function in autism spectrum disorder

John D Herrington 1,2,✉, Judith S Miller 1,2, Juhi Pandey 1, Robert T Schultz 1,2,3
PMCID: PMC4884314  PMID: 26865425

Abstract

Current neural models of autism spectrum disorder (ASD) and anxiety disorders suggest hyperactivation of amygdala in anxiety, but hypoactivation of amygdala in ASD. The objectives of this study were to (i) test the hypothesis that amygdala activity measured by functional magnetic resonance imaging (fMRI) represents a hybrid signal of opposing social functions and anxiety symptoms, and (ii) determine whether longstanding findings of decreased amygdala activation in ASD apply only to those individuals with ASD and low levels of anxiety. During fMRI scanning, 81 youth with ASD and 67 non-ASD control participants completed a face recognition paradigm that elicits robust amygdala activation. Only individuals with ASD and low anxiety levels (a subsample of 28 participants) showed decreased amygdala activation relative to controls. In the ASD group, anxiety symptoms were positively correlated with amygdala activity across the full ASD group, whereas core ASD symptoms (including social deficits) were negatively correlated. Results indicate that hypoactivation of amygdala in ASD, a suggestive finding first reported nearly 20 years ago, can be masked by comorbid anxiety—thus bringing enhanced clarity to this line of work. Amygdala activity represents a hybrid signal of emotion and social processes that cannot be reduced to either alone.

Keywords: autism spectrum disorder, anxiety, functional MRI, amygdala, social information processing

Introduction

Abnormal amygdala function is perhaps the most agreed upon single neurobiological marker of autism spectrum disorder (ASD) (Baron-Cohen et al., 2000; Dalton et al., 2005; Schultz, 2005; Kleinhans et al., 2009; Kleinhans et al., 2010; Pelphrey, et al., 2011). The majority of the literature on this topic attributes deficits in social perception and affiliation to decreased amygdala activity—consistent with the role of this structure in social functions more generally (Amaral, 2002; Adolphs, 2010; Pelphrey et al., 2011). However, this literature has yet to be fully reconciled with the equally robust research associating amygdala function with anxiety (Phelps and LeDoux, 2005; Wilensky et al., 2006). This is problematic, as these literatures lead to opposing predictions about the nature of amygdala activity. Diminished amygdala function has been repeatedly associated with deficits in social intelligence, motivation and perception, particularly among individuals with ASD (Baron-Cohen, 2000; Schultz, 2005; Pelphrey et al., 2011; Chevallier et al., 2012). Neurobiological models of anxiety disorders also focus on amygdala deficits, but for a different reason—two decades of neuroimaging research strongly associates anxiety with ‘increased’ amygdala activity during the perception and experience of emotion, and during fear learning (Phelps and LeDoux, 2005; Wilensky et al., 2006). Despite the advanced theoretical work done to understand amygdala function (i.e. Amaral, 2002; Whalen, 2007; Adolphs, 2010; Pessoa, 2010), it remains unclear how to reconcile this opposing pattern of social and affective amygdala functions when considering that anxiety disorders (i.e. increased amygdala activity) co-occur frequently with poor social functioning (i.e. decreased amygdala activity) in those with ASD. The integration of these research trajectories is critical to our understanding the precise role of amygdala in ASD.

Because there is now a substantial literature associating amygdala deficits with core symptoms of ASD (particularly social motivation and perception—see Baron-Cohen et al., 1999; Chevallier et al., 2012), this study focuses primarily on the largely unexplored influence of anxiety on these findings. Awareness is growing that individuals with ASD experience anxiety at rates in excess of not only the general population but also populations with other neurodevelopmental disorders (Evans et al., 2005; Green and Ben-Sasson, 2010; Rodgers et al., 2012). Prevalence estimates of anxiety disorders in ASD vary widely (Simonoff, 2005), with more recent studies indicating rates in excess of 40% (Leyfer et al., 2006; Sukhodolsky et al., 2008). It appears likely that most if not all anxiety disorders occur at an elevated rate in ASD (Sukhodolsky et al., 2008; Hallett et al., 2013).

This study examines whether discrepancies in amygdala findings can be understood in terms of the relative balance of co-occurring anxiety and overall ASD severity. Of the four studies reporting increased rather than decreased amygdala activation in ASD (Dalton et al., 2005; Kleinhans et al., 2010; Monk et al., 2010; Weng et al., 2011), only two included measures of anxiety. Weng et al. (2011) reported increased amygdala activation relative to controls, and did not find a significant relationship between amygdala activation and scores on the Spence Child Anxiety Scale. Kleinhans et al. (2010), on the other hand, found a significant relationship between amygdala activation and social anxiety symptoms in ASD (measured via the Social Anxiety and Distress Scale). Each of the studies used faces as stimuli. The present study examines amygdala activation in a large sample of individuals with ASD (N = 81) during a face perception task that provides highly reliable amygdala activation (Herrington et al., 2011). The primary hypothesis of this study is that core ASD and anxiety symptoms have unique and opposite effects on amygdala signal [as measured by functional magnetic resonance imaging (fMRI)], such that longstanding findings of decreased amygdala activity in ASD are only likely to manifest in non-anxious samples.

Materials and methods

Participants

In total, 148 children (81 ASD, 67 typically developing controls, or TDCs) completed a neuropsychological and psychodiagnostic battery, and a 5-min fMRI task involving face identity recognition. (See Table 1 for participant characteristics.) Participants with an ASD were diagnosed via consensus clinical best estimate procedure among experienced autism clinicians, guided by results from the Autism Diagnostic Observation Schedule (Lord et al., 2002) and Autism Diagnostic Interview—Revised (Le Couteur et al., 2003) (clinicians were trained to research reliability on both instruments). The ASD and TDC groups were age-matched (mean age = 12.5 for both groups).

Table 1.

Participant characteristics

Group N (F/M) Age IQ (GCA)* ADOS Comm ADOS social ADOS stereo ADI social ADI Comm ADI Rep. ADI onset SCQ total* SCARED total*
ASD 81 (18/63) 12.5/2.6 101/23.0 3.3/1.8 7.7/2.9 2.1/1.9 19.1/6.0 15.0/4.6 5.3/2.4 2.9/1.5 20.9/6.0 20.4/12.5
TDC 67 (21/46) 12.5/2.7 115/16.1 N/A N/A N/A N/A N/A N/A N/A 1.3/1.8 6.0/5.4
Open in a new tab

Notes. Except for the N column (sample size, broken down by Female/Male), all data are presented as mean/s.d. IQ is captured by GCA (General Conceptual Ability, measured via the Differential Abilities Scale Second Edition (Elliot, 2007), a standardized score, mean/s.d. = 100/15). ADOS, Autism Diagnostic Observation Schedule—Generic (Lord et al., 2002) (administered to the ASD group only), with data for the Communication, Social and Stereotypical Behavior subscales; ADI, Autism Diagnostic Interview—Revised (Le Couteur et al., 2003) (administered to the ASD group only). All ASD participants met criteria for a diagnosis of ASD using a combination of the ADOS scores, the Autism Diagnostic Interview—Revised (to obtain participant history of ASD symptoms), and expert clinical judgment. SCQ, Chandler et al. (2007). SCARED, Screen for Child Anxiety Related Emotional Disorders, parent report (Birmaher et al., 1997). Headings with an asterisk indicate significant differences between groups at P < 0.05 (obtained via t-test).

The validity and reliability of standard anxiety inventories among children with ASD are under active investigation (Wood and Gadow, 2010; Van Steensel et al., 2012; Hallett et al., 2013). The parent report version of the Screen for Child Anxiety Related Emotional Disorders (SCARED; Birmaher et al., 1997) was selected for two reasons. First, a recent study by Van Steensel et al. (2012) found that the psychometric properties of the SCARED are robust and comparable to those found in non-ASD samples. Second, the specific anxiety diagnoses covered by the SCARED (separation anxiety, generalized anxiety, social anxiety and panic/somatic symptoms) have well-established relationships with amygdala function (i.e. Thomas et al., 2001; Guyer et al., 2008; Blackford and Pine, 2012). Social deficits in ASD were measured via the Social Communication Questionnaire (SCQ) (Chandler et al., 2007), a widely used and validated measure of core ASD deficits beginning early in development (including those related to social functions).

The ASD group demonstrated elevated scores and a fairly broad score range across the two questionnaires of interest in fMRI analyses. The average SCQ score for the ASD group was 20.9 (s.d. = 6.0, range = 9–36). The ASD group presented with significant anxiety spanning a very large range, with an average total score of 20.4 (s.d. = 12.5, range = 1–60), where 25 is the clinical cutoff for the instrument [though see Kerns et al. (2015) for an argument in favor of lowering this cutoff in ASD]. The TDC group, on the other hand, presented with minimal ASD (SCQ mean = 1.3, s.d. = 1.8, range = 0–9) or anxiety symptoms (mean SCARED total = 6.0, s.d. = 5.4, range 0–25). (See the Supplementary Data for more details about group characteristics and inclusion/exclusion criteria, as well as additional information on the experimental paradigm, scan parameters and fMRI analyses.)

Experimental design

Participants completed a 5-min facial identity discrimination task [facial stimuli are generally regarded as yielding the most robust amygdala activation of any stimulus category in both clinical and non-clinical samples (Sergerie et al., 2008)]. Side-by-side pairs of faces were presented simultaneously for 3500 ms (followed by a blank screen for 500 ms) in blocks of six trials. Participants were required to indicate with a button press if the people depicted were the same person or not. As a control task, side-by-side pairs of houses were used (for more details, see Herrington et al., 2011).

MRI data acquisition, reduction and analysis

Imaging data were collected on a Siemens Verio 3T MRI scanner (Siemens, Malvern, PA) using a 12-channel head coil. During the facial identity task, 132 gradient-echo echo-planar images were collected. (See the Supplementary Data for more details on scan parameters.)

Preprocessing

MRI data were analyzed using the FMRB Software Library (FSL). All participants had less the one voxel of maximum volume-to-volume displacement. The final groups of ASD and TDC participants did not differ significantly from one another in the absolute value of average volume-to-volume displacement for any rotation or translation parameters or root mean square displacement (average root mean square (RMS) displacement was 0.31 mm for ASD and 0.24 mm for TDC). fMRI data were spatially (Full Width at Half Maximum (FWHM) = 5 mm) and temporally (1/48 Hz high-pass) filtered.

Subject-level fMRI analyses

Face-specific brain activation was estimated for each participant by calculating a general linear model (GLM) on each voxel. The GLM analyses calculated per-voxel statistical maps containing parameter estimates (betas) corresponding to each condition, and a statistical map containing face vs house contrast parameters. These latter maps comprised the dependent variable (DV) used in group-level fMRI analyses described below.

Group-level fMRI analyses

Group-level analyses follow two parallel approaches: per-voxel t-tests and region-of-interest (ROI) analysis (both using face > house beta maps as the DV). We implemented this parallel approach not to provide separate inferences about a priori hypotheses (this would represent circularity of analysis), but to illustrate different aspects of the same empirical relationship between amygdala, ASD symptoms and anxiety. While per-voxel tests better allow for localization of amygdala signals (and is by far the most widely used approach in the ASD amygdala literature—for example, see Baron-Cohen et al., 1999; Dalton et al., 2005; Kleinhans et al., 2009, 2010; Monk et al., 2010; Weng et al., 2011), the ROI analysis allowed for a more precise characterization of the specific anxiety symptoms contributing to variance in amygdala activation.

Per-voxel analyses

Significant clusters from per-voxel analyses reported here survived family-wise error correction, implemented via the simultaneous application of a per-voxel threshold (P < 0.001) and the application of a cluster size threshold derived from Monte Carlo simulations (generated via the program AlphaSim; see Supplementary Data). Per-voxel analyses (i.e. statistical mapping) were implemented to test the hypothesis that only ASD individuals with low anxiety levels show decreased amygdala activation relative to TDCs. Face-selective brain activation was examined separately for the ASD and TDC groups via per-voxel t-tests comparing beta maps to zero, and compared between groups via independent t-tests.

A set of post hoc per-voxel tests were conducted to confirm that the null group effect identified from the above analysis (see Results) was not attributable to ancillary group differences. Although matched for age, the TDC group scored significantly higher than the ASD group in general intelligence, as measured by the Differential Abilities Scale General Conceptual Ability score (DAS-II [Elliot, 2007] GCA, analogous with IQ; data were missing for two participants in the ASD group). Groups also differed in behavioral performance during the fMRI task (see Results; performance data were not collected for one TDC and two ASD participants). The possibility that GCA and/or accuracy might have diminished amygdala differences was addressed by replicating the above per-voxel analyses using matched subgroups of participants, selected by eliminating ASD participants with the lowest GCA and accuracy scores until there was no significant difference between groups (P > 0.10).

A final per-voxel analysis examined whether diminished amygdala activation in ASD could be observed when comparing ASD and TDC participants who were matched for anxiety level. Statistical maps were generated by contrasting a group of ASD participants with minimal anxiety symptoms (low scores on the SCARED) with the TDC group. This low-anxiety ASD subgroup was selected using the same procedure as described above for GCA and task accuracy.

ROI analyses

For ROI analysis, the peak z-value within amygdala was extracted from each individual’s face > house contrast map, and used as the DV. The search spaces for isolating the peak ROIs were derived from the left and right amygdalae from the Harvard-Oxford subcortical atlas (Flitney et al., 2007). Left and right amygdalae were included as independent variables (IVs) in multivariate linear regression (canonical correlation). The primary IVs of interest included the total anxiety score from the SCARED, and the total score from the SCQ (lifetime). Because of inclusion criteria, SCARED anxiety ratings for the TDC group were at floor; the TDC group was therefore excluded from ROI analyses involving the SCARED. In order to probe the relative contributions of core ASD and anxiety symptoms within specific diagnostic domains (i.e. panic, separation anxiety, social anxiety and generalized anxiety), results from canonical correlation analysis were further decomposed via univariate multiple regression.

Results

Per-voxel analysis

Within- and between-group effects: TDC and ASD

When analyzed separately, the ASD and TDC groups showed robust activation of bilateral amygdala (Figure 1A for TDC, 1B for ASD). However, despite previous findings of abnormal amygdala function in ASD, the two groups did not show significant differences in either left or right amygdala activity using per-voxel t-tests.

Fig. 1.

Fig. 1.

Open in a new tab

Decreased amygdala activation among individuals with ASD and low anxiety. Amygdala activation is circled in green (face > house contrast). Statistical maps are presented using a per-voxel threshold of P < 0.001. All amygdala clusters shown survived multiple comparisons correction (cluster thresholding via Monte Carlo Simulation; see Supplementary Data). (A) TDC group only. (B) ASD group only. (C) TDC vs a low-anxiety ASD subgroup (N = 28), as measured via the SCARED total score. (D) TDC vs a low-anxiety ASD subgroup, as measured via the SCARED Separation Anxiety Score. This last panel is included because ROI analyses (described under Results) indicated that the Separation Anxiety (and Panic/Somatic) subscales of the SCARED had the strongest relationship with amygdala function.

Post hoc tests were conducted to confirm that this null result was not influenced by differences in intellectual level (GCA). The TDC group had significantly higher GCA scores [101 for ASD and 115 for TDC, t(144) = 4.043, P < 0.001]. To account for this, an ASD subgroup was identified (N = 62) that did not significantly differ from TDCs in their GCA scores, and the group comparison was re-run as before. As with the full sample, there was no significant amygdala difference between groups.

A similar post hoc analysis was carried out for task accuracy. As predicted, the ASD group was less accurate for both face and house trials [main effect of Group, F(1,143) = 18.94, P < 0.001], and especially for faces [Group × Stimulus interaction, F(1,143) = 9.14, P = 0.003; mean accuracy for faces/houses was 70/84% in ASD and 82/91% in TDC]. Again, per-voxel analysis of subgroups matched for accuracy (N = 47 per group) showed no significant amygdala differences.

Between-group effects: low-anxiety ASD subgroup vs TDC

In order to test whether decreased amygdala activation is observed in ASD when accounting for anxiety, a subset of ASD participants (N = 28) were selected who did not significantly differ from TDCs in their overall anxiety level [SCARED total score; t(93) = 1.60, P = 0.11]. This group did not significant differ from the rest of the ASD group in terms of face accuracy, t(77) = 0.21, P = 0.83, so differences in comparisons between the low anxiety ASD group and the full ASD group could not be accounted for by differences in behavioral performance. Mean SCARED scores were very low for both the ASD subset and TDC groups, indicating minimal anxiety symptoms (7.79 for ASD and 6.0 for TDC, on a scale where > 25 indicates clinically significant anxiety). Although the comparison of the entire ASD group to TDC (reported above) yielded no significant per-voxel differences in amygdala activation, the comparison of this low-anxiety ASD group showed a significant decrease in amygdala activation bilaterally (Figure 1C).

Even with low overall scores, it was possible that endorsed items might have loaded onto individual subdomains across participants, and therefore represented mild manifestations of anxiety (the SCARED total score is a straight sum of all subdomain scores, which in turn are sums of individual items). Therefore, as a final analysis, we implemented a more conservative grouping procedure. ROI analyses (described below) indicated that Separation Anxiety and Panic/Somatic symptoms drove the correlations between SCARED total score and amygdala; Figure 1D shows increased amygdala activation in the TDC group compared with those ASD participants who fell in the bottom quartile (N = 19) on the Separation Anxiety scale. This same procedure was run for the Panic/Somatic Scale, with highly overlapping statistical maps. The overall pattern indicates that the decrease in amygdala activity associated with ASD can be superimposed on an increase associated with co-occurring anxiety, leading to an absence of an overall group difference relative to TDCs.

ROI analysis

ROI analyses were conducted in order to more precisely decompose the contributions of core ASD and anxiety symptoms to amygdala activation. Initial analyses compared peak amygdala signal between ASD and TDC groups, and were moderately consistent with the results from the per-voxel tests above. Peak values from the ROI analysis above showed significantly decreased activation in the right amygdala of the ASD group relative to TDCs, t(146) = 2.15, P = 0.03 (the same analysis for the left amygdala peak was not significant). However, this effect was no longer significant when covarying accuracy [F(1,142) = 1.19, P = 0.28] or GCA [F(1,143) = 1.48, P = 0.23].

Canonical correlation analysis was implemented to identify the relative contributions of anxiety and core ASD symptoms to amygdala activation. Because the TDC group was selected to be free of significant clinical symptoms, their scores on the SCARED were at floor; canonical correlation analysis therefore focused on the ASD group only. The analysis predicted amygdala activation (left and right hemispheres simultaneously) from IVs including the total SCARED and SCQ scores in the full sample. A total of two latent IV and DV factors could be calculated from the present model, resulting in two possible canonical correlations. The first of these was statistically significant, Wilks Lambda = 0.86, F(4,154) = 3.14, P = 0.016 (Table 2; the second model was not significant and is not considered further). Individual DV correlations indicated that the latent DV loaded primarily on activation within left amygdala (r = 0.90; right amygdala r = 0.28). As predicted, the latent IV loaded heavily on both the SCARED (0.71) and SCQ (−0.60), in opposite directions—despite the fact that the SCARED and SCQ were themselves positively correlated (r = 0.58, P < 0.01).

Table 2.

Anxiety, ASD symptoms and amygdala activation

(A) Canonical correlation: Wilks Lambda = 0.86, F(4,154) = 3.14, P = 0.016
DV correlations with latent factor
 IV correlations with latent factor
Variable R Variable R
Left amygdala 0.90 SCARED total 0.71
Right amygdala 0.28 SCQ total −0.59
(B) Stepwise linear regression predicting left amygdala: overall F(2,78) = 5.14, P < 0.01
IV T P-value Partial correlation ®
SCARED total 2.56 0.01 0.28
SCQ total −2.28 0.03 −0.25
(C) Correlations between left amygdala and SCARED subscales (partialing SCQ)
Total Panic Separation Generalized Social
0.28 (P = 0.01) 0.29 (P = 0.01) 0.29 (P < 0.01) 0.19 (P = 0.09) 0.11 (P = 0.32)
Open in a new tab

Notes. All analyses on data from the ASD group only. Panel A: canonical correlation predicting left and right amygdala activation from SCARED total score and the SCQ (Chandler et al., 2007) total score. Data are presented from the first canonical correlation model only (the second model was not statistically significant—see text). As the observed effect was driven primary by left amygdala, further analyses were carried out on this structure only. Panel B: stepwise linear regression deconstructing the relationship between left amygdala, SCARED and SCQ. Panel C: bivariate correlations (Pearson r) between left amygdala and SCARED subscales, conditioned on SCQ. Panic: panic/somatic symptoms; Separation: separation anxiety; Generalized: generalized anxiety; Social: social anxiety.

The relationship between anxiety, core ASD symptoms and left amygdala activation was examined further via stepwise linear regression. As suggested by the canonical correlation results, the full model, including SCARED and SCQ total scores, significantly predicted left amygdala activity, F(2,78) = 5.14, P < 0.01. When entered last in the regression equation, the SCARED total score captured the most variance (partial r = 0.28, P = 0.01), with the SCQ coming next (partial r = −0.25, P = 0.03). A subsequent set of exploratory partial correlations were conducted examining which SCARED subdomain (Panic/Somatic, Generalized Anxiety, Separation Anxiety and Social Anxiety) best accounted for amygdala activity. When conditioning the correlations on the SCQ, the Separation Anxiety (r = 0.29, P < 0.01) and Panic (r = 0.30, P < 0.01) subdomains were significantly correlated with left amygdala activation, the Generalized Anxiety (r = 0.19, P = 0.09) subdomain showed a statistical trend, and the Social Anxiety subdomain was not significant (r = 0.11, ns).

A series of post hoc analyses were conducted to examine whether the portions of amygdala associated with anxiety and core ASD symptoms could be spatially distinguished from one another. These analyses roughly paralleled the ROI analyses above. First, we implemented independent t-tests on x, y and z coordinates (separately) between the full ASD and TDC groups (all coordinates in this study are reported in Montreal Neurological Institute [MNI] stereotactic space). None of the coordinates for the left amygdala differed between groups (all P’s > 0.2). The x coordinate for right amygdala differed at P = 0.04 (though this is not likely to survive multiple comparisons correction), whereas the y and z coordinate P-values were both above 0.6. Next, a series of four multiple regression analyses were carried out predicting SCARED and SCQ total scores (in separate analyses) from the x, y and z coordinates of left and right amygdala (in separate analyses) for the ASD group only. All of these models had P-values > 0.2. Furthermore, none of the incremental contributions of any of the three dimensions reached P < 0.05 (the most significant was again the x coordinate for right amygdala, P = 0.07). Finally, coordinates were compared between the anxiety quartile in the ASD group, and the TDC group (paralleling the per-voxel analysis above). As with the overall ASD vs TDC group comparison, the x coordinate of right amygdala was significant at P = 0.03; the next highest P-value was 0.38. The overall pattern indicates that the effects of anxiety and core ASD symptoms could not be localized to separate subregions within the amygdala that showed the strongest relationship with these symptoms (i.e. left amygdala). As for right amygdala, there appeared to be a modest relationship between ASD and TDC groups in the x coordinate of peak signal. However, even this result needs to be interpreted in light of the fact that most of right amygdala showed decreased activation for the low-quartile ASD group vs TDC groups (Figure 1D).

A final set of post hoc analyses examined the relationships between behavioral performance and ROI activity in the ASD group. The SCARED total score correlated with face accuracy at trend level, r = −0.15, P = 0.07, but not correlated with face RT or with house accuracy or RT. The SCQ correlated with face accuracy (r = −0.39, P < 0.01) and house accuracy (r = −0.24, P < 0.01), but not with RT for either faces or houses. When entering face accuracy as a covariate in the above multiple regression analysis, SCQ and SCARED total scores still predicted left amygdala activation in the same directions (Ps of 0.05 and 0.02, respectively).

Discussion

Findings from this study suggest that individual differences in social behavior and in anxiety have independent, opposite relationships with amygdala activity among individuals with ASD. This has significant theoretical implications for our understanding of amygdala function in general and ASD in particular. Specifically, the present data add an additional level of complexity to longstanding models of amygdala’s role in social and emotional information processing, and how these relate to amygdala function among individuals with ASD. In the context of social function, amygdala is generally construed as supporting social perception and affiliation. However, amygdala is also widely viewed as supporting negative affect. Data from this study suggest that amygdala activity during face perception reflects both of these processes simultaneously.

This finding complicates the received wisdom from the cognitive neuroscience of ASD that social impairment is related to diminished amygdala function. Data from this study, and many others, provide support for this hypothesis; but data from this study may actually be construed as supporting a ‘secondary’ role for social impairment vis-a-vis anxiety (insofar as correlations between amygdala and core ASD symptoms were smaller than those for anxiety). Importantly, almost no studies to date have been appropriately designed or sufficiently powered to observe whether anxiety symptoms also affect amygdala activity in ASD (the vast majority of studies on this topic includes samples of <30 per group). Awareness is increasing that studies from neurobiology are generally underpowered to detect anything but very large effects, and that limited sample sizes greatly increase the chance of Type 1 error and diminish the reliability of results (Button et al., 2013).

Present and future theories of amygdala function need to better accommodate the countervailing functions supported by amygdala, where a relative increase in activity within the structure can, in some contexts, indicate a propensity for social engagement, and in other contexts, indicate the presence of anxiety. Although our examination of peak amygdala coordinates converged on the same portion of amygdala driving both anxiety and core ASD effects, it is still possible that this hybrid signal reflects two distinct signals within amygdala (i.e. different amygdala nuclei). If this proves to be the case, future amygdala theories of ASD need instead to focus on subregions, or on the network behavior of these regions (for example, see Bickart et al., 2012, 2014). There is a clear need for more research using high-resolution scanning sequences optimized for amygdala signal, so that we may be in a better position to robustly address this question.

Another important agenda for future research on this topic is to more closely examine the role of attentional mechanisms in mediating or moderating the amygdala/anxiety relationship in ASD. In particular, numerous studies indicate that amygdala activity increases when focusing on another person’s eyes (Kawashima et al., 1999; Gamer and Buchel, 2009; Sauer et al., 2014), and there is some evidence to suggest that this increase might be exaggerated in ASD (Dalton et al., 2005). It is possible that individuals with ASD and co-occurring anxiety attend differently to the eyes than individuals with ASD who do not have anxiety—which in turn leads to abnormal amygdala activity (though the causal pathway of this effect might just as easily move in the opposite direction—amygdala-mediated fear responses leading to altered gaze patterns). Future research should consider whether eye gaze patterns have explanatory power in understanding why individual with ASD and anxiety show increased amygdala activation.

This study leaves unanswered some important questions about the nature of anxiety in ASD, and how specific manifestations of anxiety may or may not be associated with abnormal activation in amygdala. Although there are reliability and criterion validity data on the use of the parent-report SCARED in ASD samples (Van Steensel et al., 2012), it is unclear whether this is the best instrument to inform our understanding of amygdala function, or how well it captures the presence of anxiety disorder diagnoses in ASD (for considerations of anxiety measurement and construct validity in ASD, see Wood and Gadow, 2010). Furthermore, a more complete picture of panic/somatic and separation anxiety symptoms—the two anxiety domains most closely coupled with amygdala activation in the present ASD sample—is clearly required. Formal panic disorder is thought to be relatively infrequent in childhood (though rates increase in adolescence), and the observed relationship between panic symptoms and amygdala function may or may not be related to sensory sensitivities that are common in ASD. It is possible that panic symptoms and sensory sensitivities are both related to hyperarousal in ASD, which in turn may be related to increased amygdala function (Mazefsky et al., 2013). As for the separation anxiety/amygdala correlation—it is very difficult to interpret the presence of separation anxiety among children with ASD without more detailed diagnostic information. A critical agenda for future research on this topic is identifying the dimension of anxiety that best accounts for amygdala signal.

There are additional considerations for future research that relate to study design. The present task design was not optimized to test the intriguing hypothesis that anxiety in ASD is associated with abnormal patterns of amygdala habitation. Abnormal amygdala habituation has been associated with anxiety in typically developing adults and children (i.e. Rauch et al., 2006; Hare et al., 2008), and has more recently been documented in ASD (Kleinhans et al., 2009; Swartz et al., 2013; Wiggins et al., 2014); it is possible that these patterns are moderated by anxiety symptoms in ASD samples. Relatedly, there is pressing need in this area of research to include a control group of typically developing children with significant anxiety symptoms (i.e. a formal anxiety disorder). In particular, the inclusion of such a group will improve our ability to tease apart neural signatures of anxiety that are or are not ASD-specific (Kerns et al., 2014).

Data from this study argue strongly in favor of dimensional approaches to understanding the relationship between brain function and manifestations of psychopathology. Because the diagnosis of ASD does not contain specific content related to anxiety, it is not surprising that the presence of increased anxiety in ASD samples is almost universally overlooked in studies that take categorical approaches (i.e. group comparisons of ASD vs control). However, we are growing increasingly aware that the functions of brain areas involved in psychiatric disorders like ASD are not likely to respect diagnostic boundaries (Insel et al., 2010; Sanislow et al., 2010). The relationship between anxiety and amygdala function in ASD is a prime example of what we stand to learn from examining psychological disorders as emergent manifestations of brain systems with diverse functions—rather than making strict inferences about the functions of brain systems in terms of diagnostic classification alone.

Supplementary Material

Supplementary Data
supp_11_6_907__index.html (911B, html)

Acknowledgements

We are very grateful to the many families who participated in this research.

Funding

The design and conduct of the study, collection, management and analysis were supported by grants from the Pennsylvania Department of Health (SAP #4100042728 to R.T.S.), the National Institute of Child Health and Development (P30 HD026979, to Marc Yudkoff) and National Institute of Mental Health (RC1MH08879 and R01 MH073084-01 to R.T.S.). The analysis, preparation, review and approval of the manuscript were also supported by funds from Shire Pharmaceuticals. Additionally, R.T.S. reported receiving lecture fees and research funds from Pfizer. J.D.H., J.S.M. and R.T.S. reported having received lecture fees and/or research funds from Shire Pharmaceuticals. J.P. reported no biomedical financial interests or potential conflicts of interest. Portions of this manuscript were presented at the 12th Annual International Meeting for Autism Research, San Sebastian, Spain (2013).

Supplementary data

Supplementary data are available at SCAN online.

Conflict of interest. None declared.

References

  1. Adolphs R. (2010). What does the amygdala contribute to social cognition? Annals of the New York Academy of Sciences, 1191, 42–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Amaral D.G. (2002). The primate amygdala and the neurobiology of social behavior: implications for understanding social anxiety. Biological Psychiatry, 51, 11–7. [DOI] [PubMed] [Google Scholar]
  3. Baron-Cohen S., Ring H.A., Wheelwright S.J., et al. (1999). Social intelligence in the normal and autistic brain: an fMRI study. European Journal of Neuroscience, 11, 1891–8. [DOI] [PubMed] [Google Scholar]
  4. Baron-Cohen S., Ring H.A., Bullmore E.T., Wheelwright S., Ashwin C., Williams S.C.R. (2000). The amygdala theory of autism. Neuroscience and Biobehavioral Reviews, 24, 355–64. [DOI] [PubMed] [Google Scholar]
  5. Bickart K.C., Dickerson B.C., Barrett L.F. (2014). The amygdala as a hub in brain networks that support social life. Neuropsychologia, 63, 235–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bickart K.C., Hollenbeck M.C., Barrett L.F., Dickerson B.C. (2012). Intrinsic amygdala-cortical functional connectivity predicts social network size in humans. Journal of Neuroscience, 32, 14729–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Birmaher B., Khetarpal S., Brent D., et al. (1997). The Screen for Child Anxiety Related Emotional Disorders (SCARED): scale construction and psychometric characteristics. Journalof the American Academy of Child and Adolescent Psychiatry, 36, 545–53. [DOI] [PubMed] [Google Scholar]
  8. Blackford J.U., Pine D.S. (2012). Neural substrates of childhood anxiety disorders: a review of neuroimaging findings. Child Adolescent Psychiatry Clinics of North America, 21, 501–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Button K.S., Ioannidis J.P.A., Mokrysz C., et al. (2013). Power failure: why small sample size undermines the reliability of neuroscience. Nature Reviews Neuroscience, 14, 365–76. [DOI] [PubMed] [Google Scholar]
  10. Chandler S., Charman T., Baird G., et al. (2007). Validation of the social communication questionnaire in a population cohort of children with autism spectrum disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 46, 1324–32. [DOI] [PubMed] [Google Scholar]
  11. Chevallier C., Kohls G., Troiani V., Brodkin E.S., Schultz R.T. (2012). The social motivation theory of autism. Trends in Cognitive Sciences, 16, 231–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dalton K.M., Nacewicz B.M., Johnstone T., et al. (2005). Gaze fixation and the neural circuitry of face processing in autism. Nature Neuroscience, 8, 519–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Elliot C. (2007). The Differential Abilities Scale, 2nd edn Harcourt Assessments, Inc, San Antonio, Texas. [Google Scholar]
  14. Evans D.W., Canavera K., Kleinpeter F.L., Maccubbin E., Taga K. (2005). The fears, phobias and anxieties of children with autism spectrum disorders and down syndrome: comparisons with developmentally and chronologically age matched children. Child Psychiatry and Human Development, 36, 3–26. [DOI] [PubMed] [Google Scholar]
  15. Flitney D., Webster M., Patenaude, et al. (2007). Anatomical brain atlases and their application in the FSLView visualisation tool. Presented at the Thirteenth Annual Meeting of the Organization for Human Brain Mapping, Chicago, IL. [Google Scholar]
  16. Gamer M., Buchel C. (2009). Amygdala activation predicts gaze toward fearful eyes. Journal of Neuroscience, 29, 9123–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Green S.A., Ben-Sasson A. (2010). Anxiety disorders and sensory over-responsivity in children with autism spectrum disorders: is there a causal relationship? Journal of Autism and Developmental Disorders, 40, 1495–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Guyer A.E., Lau J.Y.F., McClure-Tone E.B., et al. (2008). Amygdala and ventrolateral prefrontal cortex function during anticipated peer evaluation in pediatric social anxiety. Archives of General Psychiatry, 65, 1303–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hallett V., Lecavalier L., Sukhodolsky D. G., et al. (2013). Exploring the manifestations of anxiety in children with autism spectrum disorders. Journal of Autism Developmental Disorders, 43, 2341–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hare T.A., Tottenham N., Galvan A., et al. (2008). Biological substrates of emotional reactivity and regulation in adolescence during an emotional go-nogo task. Biological Psychiatry, 63, 927–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Herrington J.D., Taylor J.M., Grupe D.W., Curby K.M., Schultz R.T. (2011). Bidirectional communication between amygdala and fusiform gyrus during facial recognition. NeuroImage, 56, 2348–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Insel T., Cuthbert B., Garvey M., et al. (2010). Research domain criteria (RDoC): toward a new classification framework for research on mental disorders. American Journal of Psychiatry, 167, 748–51. [DOI] [PubMed] [Google Scholar]
  23. Kawashima R., Sugiura M., Kato T., et al. (1999). The human amygdala plays an important role in gaze monitoring. A PET study. Brain, 122, 779–83. [DOI] [PubMed] [Google Scholar]
  24. Kerns C.M., Maddox B.B., Kendall P.C., et al. (2014). Traditional and atypical presentations of anxiety in youth with autism spectrum disorder. Journal of Autism and Developmental Disorders, 44, 2851–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kerns C.M., Maddox B.B., Kendall P.C., et al. (2015). Brief measures of anxiety in non-treatment-seeking youth with autism spectrum disorder. Autism, , 19(8), 969–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kleinhans N.M., Johnson L.C., Richards T., et al. (2009). Reduced neural habituation in the amygdala and social impairments in autism spectrum disorders. American Journal of Psychiatry, 166, 467–75. [DOI] [PubMed] [Google Scholar]
  27. Kleinhans N.M., Richards T., Weaver K., et al. (2010). Association between amygdala response to emotional faces and social anxiety in autism spectrum disorders. Neuropsychologia, 48, 3665–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Le Couteur A., Lord C., Rutter M. (2003). The Autism Diagnostic Interview-Revised (ADI-R). Western Psychological Services, Los Angeles, CA. [Google Scholar]
  29. Leyfer O.T., Folstein S.E., Bacalman S., et al. (2006). Comorbid psychiatric disorders in children with autism: interview development and rates of disorders. Journal of Autism and Developmental Disorders, 36, 849–61. [DOI] [PubMed] [Google Scholar]
  30. Lord C., Rutter M., DiLavore P.C., Risi S. (2002). Autism Diagnostic Observation Schedule. Western Psychological Services, Los Angeles, CA. [Google Scholar]
  31. Mazefsky C.A., Herrington J., Siegel M., et al. (2013). The role of emotion regulation in Autism spectrum disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 52, 679–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Monk C.S., Weng S.-J., Wiggins J.L., et al. (2010). Neural circuitry of emotional face processing in autism spectrum disorders. Journal of Psychiatry and Neuroscience, 35, 105–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Pelphrey K.A., Shultz S., Hudac C.M., Vander Wyk B.C. (2011). Research Review: constraining heterogeneity: the social brain and its development in autism spectrum disorder. Journal of Child Psychology and Psychiatry, 52, 631–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Pessoa L. (2010). Emotion and cognition and the amygdala: from ‘what is it?’ to ‘what’s to be done?’ Neuropsychologia, 48, 3416–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Phelps E.A., LeDoux J.E. (2005). Contributions of the amygdala to emotion processing: from animal models to human behavior. Neuron, 48, 175–87. [DOI] [PubMed] [Google Scholar]
  36. Rauch S.L., Shin L.M., Wright C.I. (2006). Neuroimaging studies of amygdala function in anxiety disorders. Annals of the New York Academy of Sciences, 985, 389–410. [DOI] [PubMed] [Google Scholar]
  37. Rodgers J., Riby D.M., Janes E., Connolly B., McConachie H. (2012). Anxiety and repetitive behaviours in autism spectrum disorders and Williams syndrome: a cross-syndrome comparison. Journal of Autism and Developmental Disorders, 42, 175–80. [DOI] [PubMed] [Google Scholar]
  38. Sanislow C.A., Pine D.S., Quinn K.J., et al. (2010). Developing constructs for psychopathology research: research domain criteria. Journal of Abnormal Psychology, 119, 631–9. [DOI] [PubMed] [Google Scholar]
  39. Sauer A., Mothes-Lasch M., Miltner W.H.R., Straube T. (2014). Effects of gaze direction, head orientation and valence of facial expression on amygdala activity. Social Cognitive and Affective Neuroscience, 9, 1246–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Schultz R.T. (2005). Developmental deficits in social perception in autism: the role of the amygdala and fusiform face area. International Journal of Developmental Neuroscience, 23, 125–41. [DOI] [PubMed] [Google Scholar]
  41. Sergerie K., Chochol C., Armony J.L. (2008). The role of the amygdala in emotional processing: a quantitative meta-analysis of functional neuroimaging studies. Neuroscience and Biobehavioral Reviews, 32, 811–30. [DOI] [PubMed] [Google Scholar]
  42. Simonoff E. (2005). Children with psychiatric disorders and learning disabilities. BMJ, 330, 742–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. van Steensel F.J.A., Deutschman A.A., Bögels S.M. (2012). Examining the Screen for Child Anxiety-Related Emotional Disorder-71 as an assessment tool for anxiety in children with high-functioning autism spectrum disorders. Autism, 17, 681–92. [DOI] [PubMed] [Google Scholar]
  44. Sukhodolsky D.G., Scahill L., Gadow K., et al. (2008). Parent-rated anxiety symptoms in children with pervasive developmental disorders: frequency and association with core autism symptoms and cognitive functioning. Journal of Abnormal Child Psychology, 36, 117–28. [DOI] [PubMed] [Google Scholar]
  45. Swartz J.R., Wiggins J.L., Carrasco M., Lord C., Monk C.S. (2013). Amygdala habituation and prefrontal functional connectivity in youth with autism spectrum disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 52, 84–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Thomas K., Drevets W., Dahl R., et al. (2001). Amygdala response to fearful faces in anxious and depressed children. Archives of General Psychiatry, 58, 1057–63. [DOI] [PubMed] [Google Scholar]
  47. Weng S.-J., Carrasco M., Swartz J.R., et al. (2011). Neural activation to emotional faces in adolescents with autism spectrum disorders. Journal of Child Psychology and Psychiatry, 52, 296–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Whalen P. (2007). The uncertainty of it all. Trends in Cognitive Sciences, 11, 499–500. [DOI] [PubMed] [Google Scholar]
  49. Wiggins J.L., Swartz J.R., Martin D.M., Lord C., Monk C.S. (2014). Serotonin transporter genotype impacts amygdala habituation in youth with autism spectrum disorders. Social Cognitive and Affective Neuroscience, 9, 832–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wilensky A.E., Schafe G.E., Kristensen M.P., LeDoux J.E. (2006). Rethinking the fear circuit: the central nucleus of the amygdala is required for the acquisition, consolidation, and expression of Pavlovian fear conditioning. Journal of Neuroscience, 26, 12387–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wood J.J., Gadow K.D. (2010). Exploring the nature and function of anxiety in youth with autism spectrum disorders. Clinical Psychology: Science and Practice, 17, 281–92. [Google Scholar]

Associated Data

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

Supplementary Materials

Supplementary Data
supp_11_6_907__index.html (911B, html)
supp_nsw015_scan-15-377-File003.pdf (63.6KB, pdf)

Articles from Social Cognitive and Affective Neuroscience are provided here courtesy of Oxford University Press

RESOURCES