Sample characteristics
Six participants including children with ASD- (n = 2), ADHD (n = 1), and ASD+ (n = 3) were excluded as they could not accomplish the behavioral tasks. Furthermore, an initial screening based on omission rate of mask-task was conducted to yield 2 outlier subjects (Z score beyond ± 3 of the sample as a whole). As a result, a total of 98 subjects (24 ASD-, 23 ADHD, 23 ASD+, 28 NTC) were available for final analysis.
Table 1 showed the sociodemographic, clinical and performance information of all the participants. There were no significant differences in age (H (3,94) = 0.412, p = 0.745) and sex distribution (χ2 = 2.048, p = 0.562) among the four groups. There were group differences in FIQ (F (3,94) = 5.395, p = 0.002) and PSI (H (3,94) = 5.625, p = 0.001), and the post hoc test showed that ASD + group presented lower FIQ than NTC (p = 0.001), and all clinical groups scored lower PSI than NTC (all pairwise comparisons, p ≤ 0.005).
Table 1
Demographic characteristics of the participants(n = 98)
| ASD- | ADHD | ASD+ | NTC | Statistic | P | Post hoc (Bonferroni corrected) |
N(female) | 24(3) | 23(2) | 23(3) | 28(6) | 2.048 | 0.562 | - |
Age | 7.83 ± 0.36* | 8.34 ± 0.29 | 7.97 ± 0.38* | 8.18 ± 0.33* | 0.412 | 0.745 | - |
FIQ# | 97.75 ± 3.61 | 98.61 ± 3.29 | 92.09 ± 3.60 | 108.71 ± 1.80 | 5.395 | 0.002 | ASD + < NTC |
VCI# | 97.17 ± 3.33 | 103.65 ± 2.81 | 93.09 ± 3.81 | 111.25 ± 1.98 | 7.429 | < 0.001 | ASD-, ASD + < NTC |
PRI# | 104.13 ± 3.73 | 104.00 ± 3.66 | 98.39 ± 3.64 | 110.18 ± 2.06 | 2.276 | 0.085 | - |
WMI | 100.79 ± 3.81 | 95.39 ± 2.95 | 94.17 ± 3.17 | 101.75 ± 2.63 | 1.469 | 0.228 | - |
PSI | 87.50 ± 2.77* | 88.57 ± 2.95 | 86.78 ± 4.01* | 100.71 ± 1.85 | 5.625 | 0.001 | ASD-, ADHD, ASD + < NTC |
CARS# | 31.44 ± 0.45 | 20.00 ± 0.37 | 31.37 ± 0.55 | 16.71 ± 0.29* | 80.265 | < 0.001 | ASD-, ASD + > ADHD > NTC |
SNAP-IV | | | | | | | |
total score | 18.29 ± 1.05 | 28.70 ± 1.38 | 35.78 ± 1.37 | 15.29 ± 1.16 | 35.030 | < 0.001 | ADHD, ASD + > ASD-, NTC |
Inattention | 11.38 ± 0.54 | 17.00 ± 0.89 | 17.13 ± 0.45 | 10.07 ± 0.61 | 31.226 | < 0.001 | ADHD, ASD + > ASD-, NTC |
hyperactivity | 6.67 ± 0.76 | 11.70 ± 0.94 | 12.65 ± 1.23 | 5.21 ± 0.70 | 16.232 | < 0.001 | ADHD, ASD + > ASD-, NTC |
SRS | | | | | | | |
total score | 69.38 ± 4.52 | 62.52 ± 3.79 | 96.78 ± 3.89 | 45.57 ± 3.29 | 30.821 | < 0.001 | ASD + > ASD-, ADHD > NTC |
awareness | 9.54 ± 0.59 | 9.78 ± 0.51 | 12.35 ± 0.46 | 8.00 ± 0.37 | 14.148 | < 0.001 | ASD + > ASD-, ADHD, NTC |
cognition | 13.13 ± 0.99 | 12.83 ± 0.88 | 19.57 ± 0.85 | 8.75 ± 0.78 | 26.515 | < 0.001 | ASD + > ASD-, ADHD > NTC |
communication | 23.92 ± 1.77 | 19.87 ± 1.45 | 31.52 ± 1.61 | 13.36 ± 1.38* | 24.698 | < 0.001 | ASD + > ASD-, ADHD > NTC |
motivation | 10.38 ± 0.70 | 9.96 ± 0.72* | 13.52 ± 4.51* | 9.43 ± 0.81* | 5.171 | 0.002 | ASD + > ADHD, NTC |
autistic behavior# | 12.42 ± 1.28 | 10.09 ± 1.21 | 19.83 ± 5.47 | 6.04 ± 0.61 | 46.217 | < 0.001 | ASD + > ASD-, ADHD, NTC; ASD- > NTC |
Note: # variance inequality; * distributed non-normally; ASD-, autism spectrum disorder (without ADHD symptoms); ADHD, attention-deficit/hyperactivity disorder; ASD+, ASD children with ADHD symptoms; NTC, neurotypical controls; FIQ, full intellectual quotient; VCI, verbal comprehensive index; PRI, perceptual reasoning index; WMI, working memory index; PSI, processing speed index. |
As for clinical characteristics, Kruskal-Wallis test and ANOVA showed between-group differences from the score of CARS (H (3,94) = 80.265, p < 0.001), SRS (F (3,94) = 30.821, p < 0.001) and SNAP-IV (F (3,94) = 35.030, p < 0.001). Multiple comparisons indicated that in CARS, ASD- and ASD + scored higher than ADHD and NTC (all pairwise comparisons, p < 0.001), and ADHD scored higher than NTC (p = 0.04). In SRS total and subdomain scores, children with ASD + were higher than other three groups (all pairwise comparisons, p < 0.001), and both ASD- and ADHD groups were higher than NTC (ASD- vs NTC, p < 0.001; ADHD vs NTC, p = 0.013). In SNAP-IV, post hoc test demonstrated ADHD and ASD + scored higher than ASD- and NTC group (all pairwise comparisons, p < 0.001).
The ability of face encoding and retrieving
The ANOVA results showed the group differences on face encoding and retrieving performance (see Table S1). In the FET, RT of mask (H (3,94) = 18.185, p < 0.001), RT of encoding (F (3,94) = 3.868, p = 0.012) and ACC (F (3,94) = 7.027, p < 0.001) were statistically different among the four groups. Multiple comparisons illustrated that the RT-mask of ASD- and ASD + were longer than NTC (ASD- vs NTC, p = 0.005; ASD + vs NTC, p = 0.001), and the result was similar in the RT-encoding (ASD- vs NTC, p = 0.027; ASD + vs NTC, p = 0.045). While children with ASD- and ASD + presented smaller ACC-encoding than NTC (ASD- vs NTC, p < 0.001; ASD + vs NTC, p = 0.014), and children with ASD- had a smaller ACC-encoding than ADHD (p = 0.024). In the facet of FRT, we found there were group differences in RT of mask (H (3,94) = 8.809, p = 0.032) and ACC (F (3,94) = 12.118, p < 0.001). Multiple comparisons indicated that children with ASD- had a longer RT-mask than NTC (p = 0.029), and children with ASD- and ASD + had smaller ACC-retrieving than ADHD and NTC (all pairwise comparisons, p < 0.005).
In order to control the influence of participants’ age span and the basic response speed, we further conducted analysis of covariance (ANCOVA) after controlling age and PSI [47]. The results revealed that there were no significant interactions between group and age (F (3,94) = 1.131, p = 0.341), neither between group and PSI (F (3,94) = 1.215, p = 0.309). Moreover, only group differences of ACC-encoding (F (3,94) = 5.145, p = 0.002) and ACC-retrieving (F (3,94) = 9.641, p < 0.001) remained significant. In FET, children with ASD- showed a smaller ACC than children with ADHD (p = 0.026) and NTC (p = 0.003). In FRT, both children with ASD- and those with ASD + presented a lower ACC than ADHD and NTC (ASD- vs ADHD, p < 0.001; ASD- vs NTC, p = 0.001; ASD + vs ADHD, p = 0.003; ASD + vs NTC, p = 0.023) (see Fig. 2).
The correlation between VSWM performance and symptoms of ASD and ADHD
Table 2 exhibited the results of Pearson correlation analysis (or Spearman analysis for variables that couldn’t meet the prerequisite of normal-distribution and homogeneity of variance) between VSWM performance and the score of SRS and SNAP-IV. In ASD- group, the RT-encoding was negatively correlated with scores of SRS including total score (r=-0.411, p = 0.046), social awareness (r=-0.462, p = 0.023), social cognition (r=-0.455, p = 0.025) and autistic behaviors (r=-0.418, p = 0.042); while the ACC-retrieving (r=-0.418, p = 0.042) and RT-retrieving (r=-0.406, p = 0.049) was associated with social cognition. By contrast, in ADHD group, only ACC-retrieving was negatively correlated with social motivation (r=-0.428, p = 0.041). As for children with ASD+, the ACC-encoding was associated with score of SRS including total score (r=-0.433, p = 0.039) and social motivation (r=-0.521, p = 0.011); the RT-encoding was correlated with total score (r=-0.497, p = 0.016) and hyperactivity (r=-0.550, p = 0.007) of SNAP-IV; and RT-retrieving was associated with social awareness (r=-0.593, p = 0.003). Moreover, NTC group didn’t exhibit significant correlations of VSWM with the score of SRS and SNAP-IV.
Table 2
The correlation between VSWM performance and core symptoms of ASD and ADHD.
| ASD- (n = 24) | | ADHD (n = 23) | | ASD+ (n = 23) |
| Encoding | Retrieving | | Encoding | Retrieving | | Encoding | Retrieving |
| ACC | RT | ACC | RT | | ACC | RT | ACC | RT | | ACC | RT | ACC | RT |
SRS | | | | | | | | | | | | | | |
total score | -0.271 | -0.411* | -0.219 | -0.366 | | -0.124 | 0.012 | -0.278 | 0.231 | | -0.433* | -0.108 | -0.403 | -0.313 |
awareness | -0.154 | -0.462* | 0.045 | -0.245 | | 0.036 | -0.273 | -0.135 | 0.196 | | -0.129 | -0.314 | 0.151 | -0.593** |
cognition | -0.339 | -0.455* | -0.418* | -0.406* | | -0.118 | -0.032 | -0.061 | 0.101 | | -0.372 | 0.000 | -0.317 | -0.243 |
communication | -0.194 | -0.294 | -0.227 | -0.324 | | -0.072 | 0.000 | -0.263 | 0.174 | | -0.286 | -0.183 | -0.395 | -0.210 |
motivation | -0.202 | -0.107 | -0.033 | -0.199 | | 0.100 | 0.132 | -0.428* | 0.412 | | -0.521* | 0.210 | -0.382 | -0.230 |
autistic behavior | -0.244 | -0.418* | -0.138 | -0.307 | | 0.290 | 0.097 | -0.200 | 0.113 | | -0.337 | -0.157 | -0.331 | -0.164 |
SNAP | | | | | | | | | | | | | | |
total score | -0.266 | -0.199 | -0.254 | -0.260 | | -0.345 | 0.129 | -0.077 | 0.423 | | 0.167 | -0.497* | 0.160 | -0.389 |
inattention | -0.089 | -0.131 | -0.317 | -0.295 | | -0.280 | 0.016 | 0.014 | 0.298 | | 0.317 | -0.002 | -0.011 | -0.197 |
hyperactivity | -0.307 | -0.183 | -0.122 | -0.147 | | -0.239 | 0.174 | -0.127 | 0.337 | | 0.071 | -0.550** | 0.181 | -0.360 |
Note: * P < 0.05; **P < 0.01; VSWM, visuospatial working memory; ASD-, autism spectrum disorder (without ADHD symptoms); ADHD, attention-deficit/hyperactivity disorder; ASD+, ASD children with ADHD symptoms; ACC, accuracy; RT, response time. |
The relationship between EF and VSWM
In the comparison of EF, all clinical groups had lower score of CC than NTC (H (3,94) = 18.781, p < 0.001; all pairwise comparison, p < 0.05). Participants with ADHD and those with ASD + scored higher than NTC in the index of RE (H (3,94) = 16.886, p = 0.001; ADHD vs NTC, p = 0.001; ASD + vs NTC, p = 0.006). There existed no group differences in RPE and FMS (see Table S2).
The Spearman correlation analysis (see Table S3) illustrated that ACC-retrieving was positively associated with CC in ASD- group (r = 0.444, p = 0.030). In addition, RT-retrieving was correlated with CC (r=-0.501, p = 0.015), RE (r = 0.425, p = 0.043) and RPE (r=-0.583, p = 0.018) in ADHD group. And for children with ASD+, RT-encoding showed positive relationship with CC (r = 0.546, p = 0.007) and FMS (r = 0.463, p = 0.026). No significant relationship between EF and VSWM performance was found in NTC.
To explore how different indices of EF predicted the performance of face encoding and retrieving, we employed multi-linear regression analysis (stepwise) adjusting for age and FIQ. The results showed in Table 3 illustrated that CC predicted 22.5% individual variation of ACC-retrieving in ASD- group (F (1,22) = 7.660, p = 0.011, R2adjusted = 0.225), and RPE could predict 23.3% individual difference of RT-retrieving in ADHD group (F (1,21) = 7.672, p = 0.011, R2adjusted = 0.233). While in ASD + group, RPE and FMS predicted 27.2% performance of RT-encoding (F (1,20) = 5.104, p = 0.016, R2adjusted = 0.272), and RPE predicted 32.2% performance of RT-retrieving (F(1,21) = 11.438, P = 0.003, R2adjusted = 0.322). None indices of EF predicted the performance of face encoding and retrieving in NTC group.
Table 3
Multiple regression analysis between EF and VSWM in different groups
Group | Dependent | Independent | B(SE) | b’ | t | P | R2 | R2changed |
ASD- | ACC-Retrieving | CC | 0.043 (0.016) | 0.508 | 2.768 | 0.011 | 0.285 | 0.258 |
ADHD | RT-Retrieving | RPE | -40.781(14.724) | -0.517 | -2.770 | 0.011 | 0.268 | 0.268 |
ASD+ | RT-Encoding | FMS | 163.741 (72.155) | 0.414 | 2.269 | 0.034 | 0.338 | 0.145 |
| | RPE | 15.541 (7.435) | 0.381 | 2.090 | 0.050 | | |
| RT-Retrieving | RPE | 23.284 (6.855) | 0.594 | 3.382 | 0.003 | 0.353 | 0.353 |
Abbreviation: ASD-, autism spectrum disorder (without ADHD symptoms); ADHD, attention-deficit/hyperactivity disorder; ASD+, ASD children with ADHD symptoms; CC, categories completed; RE, errors responses; RPE, perseverative responses errors; FMS, failure to maintain set; ACC, accuracy; RT, response time. |