Lack of Embodiment of Action Words in the Autistic Brain

Background:  

Atypical organisation of the semantic system in autism spectrum conditions (ASC) is implied by previous research (Harris et al, 2006) but remains understudied. In typical individuals, word meaning is embodied in sensorimotor systems related to experiencing that concept in the world. Action-related words (e.g. “kick”) activate frontocentral “mirror” motor systems in a somatotopic fashion reflecting their semantic relationship with the effectors, and visual object-related words (e.g.  “cat”) activate occipitotemporal visual-processing areas.  In ASC, a different learning style and atypical neural connections might give rise to substantially different neural organisation of action- and object-related words.  We predicted there would be atypical representation of action words in ASC due to atypical connections in the pathway connecting temporal to frontocentral cortices and, most importantly, atypical mirror neuron functioning in the networks of multimodal, sensorimotor circuits typically involved in representing action words (Pulvermuller & Fadiga, 2010).

Objectives:  

To investigate the functional organisation of conceptual knowledge in ASC, focusing on action- and object-related words.

Methods:  

Event-related fMRI was employed to investigate activity evoked by passive reading of well-matched action and object words in 18 subjects with an ASC (all previously diagnosed using gold-standard instruments) and 18 controls matched for age and IQ, both groups equally balanced for gender. In order to investigate the behavioural correlates of activity, participants completed a speeded semantic decision task several weeks after scanning.

Results:  

Alongside a general reduction in word-related inferior frontal and precentral activity in ASC, striking differences in the representation of action words appeared: action words evoked robust motor activity in controls but not in ASC subjects. A regions of interest analysis confirmed this group difference in the motor system (f (1, 34) = 5.280, p < .03), which was specific to action words; activity to object words overlapped substantially between groups in both frontal and temporal cortices. Semantic judgements revealed an interaction of word category and group, indicating that autistic subjects were significantly slower in the processing of action words (F (1, 30) = 4.291, p < .05). A significant correlation showed that motor system activation was associated with faster reaction times to action words (r =.509, p <.01).

Conclusions:  

The normal embodiment of action word meaning in the motor system was absent in ASC, though other words did not show this difference. A correlation between activity in motor cortex and the speed of processing showed that as activation decreased, people with autism were slower to process action words, indicating the functional importance of the motor system in processing action words. Lack of activity in the frontocentral motor systems and slowness of semantic decisions in ASC suggests deficits in embodied motor cognition at the action-semantic level. Our results indicate that conceptual-semantic knowledge differs substantially between ASC and typical controls as predicted by the mirror neuron theory of ASC, which links the sociolinguistic and mindreading deficits of autism to the robustly-reported inactivity of mirror systems replicated in the present study.