FaceSay - Social Skills Games That Work

Background: A big challenge for children and adults on the spectrum, as well as educators and providers, is generalizing Social Skills to everyday life.  FaceSay is one of the first, and as far as we know, the only technology intervention to show generalization to the playground in a randomized controlled study (N=49, at the University of Alabama, Birmingham, publication in preparation).  Because FaceSay is scaleable - relatively low cost, little or no training, accessible as a download, can be used independently by higher functioning students, complementary to other interventions - it has the potential to provide a large benefit.

Objectives: The overal goal was to determine if a computer-based social skills intervention for children with autism or Asperger Syndrome is effective in improving specific social skills. The specific aims of the study were to investigate the effects of an avatar assistant on the emotional cognition of children with an ASD’s emotional. The study also examined the social skills effects of the intervention.
Methods: Forty-nine children with autism or Asperger Syndrome participated in the project.The children had previously received a diagnosis of autism (n=25) or Asperger Syndrome (n=24) according to the criteria specified by the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1994). The children were recruited from several sources, including the Autism Society of Alabama, Glenwood Inc, Mitchell’s Place, and local elementary schools.
The children were randomly assigned to the training group (FaceSay) or the control group (Tux Paint Software). All children were asked to attend two 2 sessions per week (10-25 minutes each) for a total of 12 sessions. Children who attended greater than 83% of the sessions were included in the analyses 
Results:

FaceSay  Intervention Groups -->>	N=11, Cars=36, IQ=55	N=13, Cars=34, IQ=92	Comment
Emotion Recognition	F= 6.40       p < .05	F= 23.04       p < .001	Higher KBIT related to higher Emotion scores
Benton Face Recognition	F= .069       p > .05	F= 8.29         p < .01
SSRS Parent Self Report	F= 10.36     p < .05	F= 4.36         p = .05
Blinded Grad Student Observations	F= 5.05       p < .05	F= 13.61       p < .001	Generalization to the Playground

Conclusions: While all of the results are encouraging, the generalization is the most encouraging and the biggest surprise.  Based on results from other strong projects, the researchers expected that no generalization would be found.  Additional research is underway to try to replicate and better measure the generalization.  One of the biggest challenges in measuring generalization is how to make a controlled measure of an uncontrolled interaction?   We are exploring the use of LifeShirts, as well as new equipment from SeeingMachines and software from Michael Spezio to automatically measure dwell time, eye vs mouth in naturalistic interactions between teacher and student.   In addition, a study with neurotypical preschool children using FaceSay has shown promise for using FaceSay inclusively and "prophylactically".  One recent development is the exploration by Sally Roger's team that attention to faces may play a role not just in emotion recognition, but in imitation.