Accessing Peer Social Interaction: Using Authorable Virtual Peer Technology as a Component of a Group Social Skills Intervention Program
Andrea Tartaro, Justine Cassell, Courtney Ratz, Jennifer Lira, Valeria Nanclares-Nogués · 2014 · ACM Transactions on Accessible Computing · doi:10.1145/2700434
Summary
This paper evaluates the Authorable Virtual Peer (AVP), a life-sized animated character technology designed to support social skills development in children with autism spectrum disorders. The AVP differs from other virtual agent approaches by offering three modes of interaction: face-to-face conversation with the virtual peer, authoring the virtual peer's behaviors, and operating the virtual peer for another child. This design is grounded in constructionist learning theory—the idea that children learn best when actively building artifacts rather than passively receiving instruction. The researchers integrated the AVP into an 11-week social skills group intervention program for seven children aged 7-11 with autism diagnoses. The program followed Baker's Structured Learning curriculum, covering skills like introducing oneself, asking questions, giving feedback, and handling mistakes. Each weekly session included group instruction, AVP interaction (for half the group), and peer role-play practice. The study used a within-subjects counterbalanced design where children alternated between AVP and no-AVP conditions across weeks, allowing comparison of how each child performed after interacting with the technology versus without it. The research specifically targets reciprocity—the back-and-forth exchange that makes conversation feel natural and connected. Children with autism often struggle with reciprocal interaction, offering responses that seem off-topic or failing to maintain conversational flow. The AVP provides a safe space to practice these skills before attempting them with human peers.
Key findings
The study found statistically significant effects of AVP interaction on three key reciprocity components: giving feedback (R²=0.42, p<0.0001), responding to questions (R²=0.31, p<0.0001), and sharing information (R²=0.51, p<0.0001). Children demonstrated more appropriate use of these reciprocal behaviors during peer role-play sessions when they had first practiced with the AVP compared to sessions without prior AVP interaction. On the Social Responsiveness Scale (SRS), a standardized measure completed by parents, there was a significant improvement in the social communication subscale (p<0.05) from pre- to post-intervention. This subscale specifically measures expressive social communication—difficulty making friends, trouble getting ideas across in conversation, and answering questions indirectly. Other SRS subscales did not show significant changes, which the authors note is common in short-term intervention studies. Qualitative examples illustrated the differences vividly: one child began the program with awkward, off-topic utterances during introductions but by week 6 (after AVP exposure) engaged in natural reciprocal exchanges about shared interests. The combination of authoring behaviors and then performing them appeared to give children agency over their learning while scaffolding appropriate social responses.
Relevance
This research demonstrates how technology can support, rather than replace, human social interaction for children with autism. The AVP serves as a bridge—a lower-stakes environment where children can practice reciprocal conversation skills before applying them with peers. This has practical implications for social skills programs that may want to incorporate virtual peer technology as a preparation tool. For accessibility practitioners, the study highlights the importance of constructionist design principles: letting users actively build and control technology rather than simply consuming it. The three-mode interaction (face-to-face, authoring, operating) gives children multiple ways to engage based on their comfort level and learning style. The small sample size (n=7) limits generalizability, and the authors acknowledge this as a "proof of concept" study rather than a definitive clinical trial. Future research would need larger samples and longer intervention periods. However, the significant effects on social communication and the specific reciprocity components offer promising evidence that AVP technology merits further development and evaluation in therapeutic settings.
Tags: autism · social skills · virtual peer · social communication · assistive technology · constructionism · reciprocity · children