The Intersection of Working Memory and Emotion Recognition in Autism Spectrum Disorders

Background:  The executive functioning (EF) theory of autism spectrum disorders (ASD) grounds key impairments within the cognitive realm, whereas social-cognitive theories view social functioning impairments as primary. The current study examines aspects of both EF and social-cognitive functioning in ASD. Specifically, the focus is narrowed within EF to working memory (WM), operationalized with the n-back task. Many studies report deficits in verbal WM in ASD, whereas visual/spatial WM may be intact. EF based on emotion and motivation has been referred to as “hot” EF. Studies have examined WM using n-back for facial identity in ASD, but not for emotional expression; whereas, n-back for emotional expression has been investigated in controls, but not in ASD.

Objectives:  The current study investigates “hot” executive function in ASD via, the n-back using emotionally expressive faces, as a model of social interaction. Hypotheses included less accuracy in ASD for 2- and 3-back conditions relative to controls in the context of similar 0 and 1 back performance. Reaction time for correct trials was analyzed for exploratory purposes.

Methods:  The sample included young adults 17-30 years old with ASD (N=16, N_Males=11, M_age=21.63, SD_age=3.74 and a typically developing sample matched on age (N=29, N_Males=2, M_age=20.34, SD_age=3.13). Stimuli consisted of cropped facial expressions (NimStim). Six basic emotions were balanced across all conditions. The 0-back served as baseline measure of short-term memory for emotional expression, followed by 1-, 2-, and 3-back tasks, 2 blocks each in pseudorandom order, counterbalanced within groups. 

Results:  Accuracy and reaction time were analyzed with 2 x 3 mixed ANOVA and post-hoc t-tests. Results demonstrated that the ASD group had significantly lower accuracy in 0-back, t (19.562) = 3.564, p=.002, d=1, as well as in each WM load condition (ps<.001 to .007). There was no significant interaction between group and n-back level for accuracy. There was no main effect of group on reaction time. The effect of working memory load on reaction time was significant, F (1.703, 3.56), p=.04, η²_partial = 0.076, with 1-back (M=76.30, SD= 10.66) significantly faster than 2-back (M=70.68, SD= 12.28) for the whole sample. The interaction term was significant and post-hoc within group analyses of changes in reaction time as working memory load increased revealed that the control group’s reaction time significantly increased from 1- to 2-back, F (1,28)=13.455, p=.001, η²_partial=.325, whereas the ASD group’s RT was not significantly different, F (1, 15)= .784, p=.390, η²_partial =.05. Both ASD and control participants performed near chance on the 3-back (ASD accuracy 45-60%, control accuracy range 50-59%). 

Conclusions:  

The results suggest that the n-back with emotional expressions is too challenging at 3-back for both control and ASD populations. With limited exposure time, short-term memory for emotional expression is compromised in ASD, which is in contrast to previous findings of excellent short memory for facial expressions in controls. When visual stimuli include emotionally relevant information, working memory is deficient in ASD even at basic vigilance (0-back), despite evidence for intact visual/spatial working memory for non-social/non-emotional n-back stimuli.