Prospective Action Planning: Autistic Children Go Their Own Way

Poster Presentation
Friday, May 11, 2018: 5:30 PM-7:00 PM
Hall Grote Zaal (de Doelen ICC Rotterdam)
C. Becchio1, A. Cavallo2, L. Romeo3, C. Ansuini4, J. Podda1, E. Veneselli5, F. Battaglia1 and M. Pontil1, (1)Italian Institute of Technology, Genova, Italy, (2)University of Turin, TORINO, Italy, (3)Università Politecnica delle Marche, Ancona, Italy, (4)C'MON Unit, Istituto Italiano di Tecnologia, Genoa, Italy, (5)Istituto G. Gaslini, Genova, Italy

The simple act of picking up a water glass involves predictive control mechanism. People alter their manipulative behavior as a function of what they plan to do with the objects. Failure to develop this form of prospective motor control has been proposed to contribute to faults in higher mind functions of individuals with autism spectrum disorders (ASDs; Trevarthen and Delafield-Butt, 2013), however, no study to date has quantified the extent to movement parameters specify future action plans in ASD.


The present study combined kinematic analysis with machine learning techniques to determine whether and how movement parameters specify self-action and other-action plans in ASD and TD. The specification of action plans was analyzed at three levels: i) individual level, to investigate whether individuals with ASD show consistent motor patterns specifying self- and other-action plans; ii) within-group level: to determine whether specification of self- and other-action plans is similar across individuals with ASD; iii) between-group level: to determine whether specification of self- and other-action plans is similar across ASD and TD individuals.


We used a sequential object manipulation task to test prospective motor control in children with ASD (n = 20) and IQ-matched typically developing (TD) children (n = 20). To quantify changes in behavior as a function of self-action plans, we first trained a support vector machine to classify grasping movements as a function of onward self-actions (place, pour, and pass). We next attempted to verify whether ASD and TD children altered their initial grasp in anticipation of co-actor’s forthcoming action by training the same classifier to distinguish grasps as a function of onward other-actions (pass-to place and pass-to pour).


Self-action plans. Specification of self-action plans detected at the individual level was similar across individuals with ASD. As shown by cross-classification, ASD and TD children activated similar motor plans in anticipation of forthcoming self-demands.

Other-action plans. Both ASD and TD children altered their grasp in anticipation of other-demands, showing consistent within-group patterns. However, a cross-group analysis in which a classifier was trained on one group and tested on the other, failed to achieve greater than chance performance, suggesting that specification of other-action plans varied across ASD and TD children.


While motor parameters specify future action plans in both ASD and TD individuals, children with autism appear to have their own way to accommodate other-action plans. This may contribute to bilateral difficulties in reciprocal social interaction, reinforcing the view that the difficulties that autistic individuals experience in social cognition are rooted in atypical motor patterns.