When do individuals with ASD mimic? Passive observation versus task engagement

Several studies examined facial mimicry in high-functioning individuals with autism spectrum disorder (ASD). Some found deficits in spontaneous mimicry. That is, in an unprompted setting, ASD individuals do not smile to a smile or frown to a frown. In one experiment, participants' task was to “just watch” large pictures of happy and angry expressions presented for several seconds (McIntosh et al. Reference McIntosh, Reichmann-Decker, Winkielman and Wilbarger2006). Facial electromyography revealed congruent facial reactions only in control, but not ASD, participants. Importantly, ASD participants reacted typically when mimicry was explicitly encouraged, thus controlling for nonspecific motor or perceptual deficits. In another experiment, participants' task was to “just watch” a video in which a student described his adventures in an amusement park, displaying happy expressions (Stel et al. Reference Stel, van den and Smeets2008). Analysis of experimenter-coded facial expressions revealed less spontaneous mimicry in ASD than in control participants. This effect was observed despite no differences in the amount of time spent looking at the screen and no differences in voluntary mimicry. In yet another study, the task was a passive observation of 2-second presentations of several expressions (Dapretto et al. 2006). fMRI analysis showed that ASD individuals had less activity in several areas, including the premotor cortex (inferior frontal gyrus), amygdala, insula, and the striatum. In short, “just watching faces” does not trigger embodied simulation in ASD individuals. Interestingly, similar ASD impairments were found in studies that explored spontaneous mimicry of non-emotional actions (e.g., hand movements) with EEG measures of motor cortex activation, such as mu-wave suppression. Thus, ASD participants show a typical level of mu-wave suppression when performing an action, but less mu-wave suppression (i.e., less motor cortex activation) to observation of an action, as compared to typical participants (Oberman et al. Reference Oberman, Hubbard, McCleery, Ramachandran and Pineda2005).

But the story is more complex. Studies of non-emotional mimicry suggest that ASD participants occasionally show “spontaneous mirroring.” For example, reduction in action imitation in ASD is found only for actions displayed from an egocentric perspective (away from the observer), but not from an allocentric perspective (toward the observer) (Theoret et al. Reference Theoret, Halligan, Kobayashi, Fregni, Tager-Flusberg and Pascual-Leone2005). In another example, ASD individuals show little EEG mu-wave suppression when watching an action performed by a stranger, but normal mu-wave suppression when watching an action by a family member (Oberman et al. Reference Oberman, Ramachandran and Pineda2008). In the domain of emotional mimicry, one study found no impairments of spontaneous mimicry, measured with EMG, when ASD individuals were given the task of integrating visual and auditory cues to emotion (Magnee et al. Reference Magnee, de Gelder, van Engeland and Kemner2007). A recent study showed that giving ASD participants a task of recognizing expressions (rather than just “watching them”) led to a comparable level of mimicry as the typical group, even though ASD mimicry was slower (Oberman et al. Reference Oberman, Winkielman and Ramachandran2009). In short, it appears that “task engagement” is necessary to trigger mimicry in ASD individuals.

But what is “task engagement”? What factors turn on and off embodied simulation? This is not simply the question of attention and general motivation. After all, all studies ensured that individuals stay “on task.” One idea is that factors like “allocentric versus egocentric perspective,” “family versus stranger,” and “emotion identification versus observation” work similarly to “eye gaze,” discussed in the target article. Specifically, all these factors switch the processing strategy from the disembodied, “pattern-matching” strategy, to one that presumably requires access to the underlying states of the individuals modeling the expressions or actions.

Of course, this idea needs to be tested directly. Here is one possibility. It is known that in perceiving expressions, ASD individuals rely more heavily on a “cold” rule-based strategy (Rutherford & McIntosh Reference Rutherford and McIntosh2007). As a result, they are more likely than controls to accept as realistic the exaggerated images of expressions (smiles that are stronger than real smiles), presumably because those expressions represent “best fits” to the rule. However, this “rule-based” processing should diminish when they are dealing with faces that smile and look “at them” and are produced by family members, rather than total strangers. Importantly, the task needs to be realistically challenging – it cannot be easily solvable by a simple pattern-matching strategy (e.g., recognizing a distinct expression containing pure happiness, or a simple emoticon, like the Walmart's smiley). Under such conditions, the embodiment (mimicry) should contribute to emotion recognition. This is critical, because as Niedenthal et al. note, mimicry contributes to performance of typical participants only when the perceptual task is hard, the participant cares, and the subjective emotional response is a useful signal for the recognition task (Oberman et al. Reference Oberman, Winkielman and Ramachandran2007).

I hope that this exchange inspires fruitful research that will benefit not only theoretical understanding of emotion processing, but also better research on atypical social functioning (see Winkielman et al. [Reference Winkielman, McIntosh and Oberman2009] for further discussion).