Clinical and Demographic Variables
The AUT group (M = 34.65) had higher AQ scores than the N-AUT group (M = 16.46) (t(57) = -8.208, p < .001, d = 2.21) (See Table 1). The DASS-21 total scores and its subscales of depression, anxiety, and stress were calculated for each participant. The AUT group had significantly higher DASS-21 total scores compared to the N-AUT group, t(57) = -3.121, p = .004, d = 0.92. When each subscale was compared, the autistic students scored significantly higher on anxiety, t(57) = -3.058, p = .005, d = 0.90, and stress, t(57) = -4.035, p < .001, d = 1.04, but not on the depression subscale, t(57) = -1.603, p = .120, d = 0.47. Similarly, the AUT group reported significantly higher social anxiety scores measured with LSAS compared to N-AUT, t(56) = -6.254, p < .001, d = 1.90. Specifically, the LSAS subscales of fear, t(56) = -6.572, p < .001, d = 1.97, and avoidance, t(56) = -5.506, p < .001, d = 1.67, were significantly higher for autistic participants compared to their non-autistic peers.
Table 1
Clinical Scores of autistic traits, social anxiety, and depression in the AUT (N = 20 for DASS-21 and N = 19 for LSAS measures) and N-AUT group (N = 39)
| AUT Mean (SD) Range | N-AUT Mean (SD) Range | p value (group differences) |
AQ Total | | 34.65 (8.74) 16–48 | 16.46 (7.69) 4–34 | < .001 |
DASS-21 Total | | 42.20 (25.43) 10–108 | 22.95 (16.08) 2–82 | .004 |
| Anxiety subscale | 11.70 (8.93) 0–36 | 5.08 (5.25) 0–18 | .005 |
| Depression subscale | 11.60 (10.03) 0–32 | 7.59 (6.93) 0–36 | .12 |
| Stress subscale | 18.90 (9.85) 6–42 | 10.05 (6.84) 0–28 | < .001 |
LSAS Total | | 76.74 (26.05) 36–137 | 36.69 (14.36) 7–73 | < .001 |
| Fear subscale | 40.26 (12.77) 15–69 | 19.23 (8.04) 5–38 | < .001 |
| Avoidance subscale | 36.48 (13.98) 20–68 | 17.46 (7.99) 2–35 | < .001 |
AQ: Autism Quotient. DASS-21: Depression Anxiety and Stress Scales- 21 Items. LSAS: Liebowitz Social Anxiety Scale.
Behavioural Results
Social Judgement Task
One-sample t-tests to check whether the expectation scores differed significantly from chance (50%) found no positive expectation bias in either group (see Table 2; AUT; t(19) = -1.879, p = 0.90, d = 0.40, N-AUT; t(38) = .754, p = .456, d = 0.12). However, on average, the AUT group had lower positive expectation scores (lower tendency to predict being liked; group mean 44.25%) compared to the N-AUT group (group mean 51.45%) and this group difference was significant, t(57) = 2.038, p = .046, d = 0.54. The exact analysis with the responses from the AJT can be found in the Supplementary information (S1).
Table 2
Behavioural results of Social and Age Judgement Tasks
Task | Responses (SD) | AUT | N-AUT |
AJT | Positive expectation score (% of ‘Yes’ judgements) | 51.70 (9.32) | 54.97 (10.36) |
SJT | Positive expectation score (% of ‘Yes’ judgements) | 44.25 (14.38) | 51.45 (12.02) |
Psychophysiology Results
Social Judgement Task
The main goal of the current study was to investigate whether autistic participants responded differently to unexpected social rejection (Yes/No trials) compared to non-autistic participants by examining their cardiac responses. Figure 1 illustrates IBI responses associated with feedback processing in the SJT in the N-AUT (Fig. 1a) and the AUT groups (Fig. 1b). As expected, cardiac responses were lengthened following feedback at IBI0 (slower heartbeats), followed by a recovery as the IBI responses got shorter again (faster heartbeats). More importantly, cardiac responses at IBI3 and IBI4 during the ‘Yes/No’ feedback trials in the N-AUT group were longer compared to all other feedback trials, indicating more cardiac slowing to unexpected social rejection (see Fig. 1a). However, cardiac responses of IBI3 and IBI4 during both ‘Yes/No’ and ‘No/No’ feedback trials were longer in the AUT group compared to ‘Yes/Yes’ and ‘No/Yes’ feedback trials (see Fig. 1b).
A mixed-model RM ANOVA with within-subject factor Feedback type (4; No/No, Yes/No, No/Yes, Yes/Yes) and between-subject factor Group (2; AUT, N-AUT) was conducted to explore effects of social rejection trials on cardiac responses (mean of IBI3 and IBI4). The main effect of Group was not significant, F(1, 57) = .028, p = .867, ηp2 < .001. However, there was a main effect of Feedback type, F(3, 171) = 13.361, p < .001, ηp2 = .190, and a significant interaction effect of Feedback type and Group, F(3, 171) = 3.224, p = .024, ηp2 = .054, although this effect was small. The interaction effect of Feedback type and Group was followed up by carrying out ANOVAs with each feedback type per group. In both groups, there was a main effect of Feedback type (AUT; F(3, 57) = 8.701, p < .001, ηp2 = .314, N-AUT; F(3,114) = 10.852, p < .001, ηp2 = .222). Bonferroni corrected post hoc comparisons revealed that the cardiac responses for the 'Yes/No' condition (unexpected social rejection) in the N-AUT group were significantly longer compared to all other feedback conditions (all ps < .004) and there were not any significant differences between any other feedback conditions (all ps > .004). In the AUT group, cardiac responses for the ‘Yes/No’ condition was significantly longer than ‘No/Yes’ (p = .002) and ‘Yes/Yes’ conditions (p = .002). However, there was not a significant difference in cardiac responses between the ‘Yes/No’ condition and ‘No/No’ condition (p > .999). These results suggested that cardiac slowing was observed specifically during unexpected social rejection in the N-AUT group, while both unexpected and expected social rejection led to a cardiac slowing in the AUT group.
Age Judgement Task
To test whether the cardiac responses to SJT were specific to social rejection and not to general negative cognitive feedback, the same analysis was repeated using the data from the AJT. Figure 2 illustrates the IBI responses associated with feedback processing in the AJT in the N-AUT (Fig. 2a) and the AUT group (Fig. 2b). As expected, the cardiac responses were lengthened following feedback (IBI0) and recovered again when they were shortened towards the end of the trial.
The same mixed-model RM ANOVA results showed that the main effect of Group was not significant F(1, 57) = .946, p = .335, ηp2 = .016; nor was the main effect of Feedback type, F(3, 171) = 3.015, p = .088, ηp2 = .050, and neither was the interaction effect for Feedback type and Group, F(3, 171) = .443, p = .722, ηp2 = .008. These results suggested that the feedback conditions in the AJT were not associated with different patterns of cardiac response in either group.
Comparing cardiac responses between SJT and AJT
Subsequent analyses were conducted to test whether the cardiac responses to unexpected social rejection (e.g. social judgement) were more pronounced than the responses to cognitive negative feedback processing (e.g. age judgement). To test this prediction, cardiac responses to unexpected social rejection (‘Yes/No’ feedback condition) in the SJT were compared with the cardiac responses to negative feedback in the AJT; ‘Yes/No’ (incongruent older) and ‘No/Yes’ (incongruent younger) for each group separately. Cardiac responses of N-AUT participants to unexpected social rejection were significantly longer than the cardiac responses to ‘Yes/No’ condition, F(1, 38) = 9.281, p = .004, ηp2 = .196, and ‘No/Yes’ condition in the AJT, F(1, 38) = 9.817, p = .003, ηp2 = .206. Therefore, the additional cardiac slowing in N-AUT participants was specific for unexpected social rejection. However, these comparisons did not differ from each other in the AUT group; comparing the SJT ‘Yes/No’ condition to the AJT Yes/No condition, F(1, 19) = .091, p = .766, ηp2 = .005, and ‘No/Yes’ condition, F(1, 19) = .148, p = .705, ηp2 = .008.
Individual differences in experiences of social rejection
Given the individual differences within both groups, especially the AUT group, understanding the relationship between social rejection sensitivity and mental health has clinical implications regardless of diagnosis. One-tailed Pearson correlations were performed to test whether behavioural and psychophysiological responses to social rejection were associated with autistic traits, social anxiety, and depression scores across participants. As can be seen in Table 3, negative expectation scores indicated by percentage of ‘No’ responses in the SJT were significantly correlated with autistic traits, r = .356, p = .006 (moderate positive correlation), social anxiety, r = .507, p < .001 (strong positive correlation), and depression scores, r = .374, p = .004 (moderate positive correlation). Participants with a higher negative prediction score (higher tendency to predict being rejected by others) were more likely to report higher autistic traits, higher social anxiety, and more depression symptoms. On the other hand, cardiac responses to unexpected social rejection were not significantly correlated with any of the self-report measures.
Table 3
Correlations between experiences of social rejection and autistic traits, social anxiety, and depression symptoms across participants (N = 58)
Variables | 1 | 2 | 3 | 4 | 5 |
1. Negative expectation score | 1 | − .037 | .356** | .507** | .374** |
2. Cardiac responses to unexpected social rejection | | 1 | .019 | − .175 | − .170 |
3. AQ total score | | | 1 | .760** | .379** |
4. LSAS total score | | | | 1 | .471** |
5. DASS-21 depression | | | | | 1 |
AQ: Autism Quotient. LSAS: Liebowitz Social Anxiety Scale. DASS-21: Depression Anxiety and Stress Scales- 21 Items. ** Correlation is significant at the 0.01 level (1-tailed).
Hierarchical regression with AQ, LSAS, DASS-21 depression scores and cardiac responses to unexpected social rejection as predictors of negative expectation scores found only social anxiety as a significant predictor, β = .507, t(57) = 4.402, p < .001, accounting for 24.4% of the variance. Thus, higher self-reported social anxiety significantly predicted higher tendency to expect being rejected by others. Adding depression symptoms, autistic traits, and cardiac responses to unexpected social rejection did not improve the model as they were not significant predictors of negative prediction scores (DASS-21 depression subscale; β = .231, t(57) = 1.806, p = .076, AQ; β = − .013, t(57) = − .075, p = .941, cardiac responses to unexpected social rejection; β = .070, t(57) = .582, p = .563).