From adolescent cliques to entire nations, whenever humans get together in groups, they develop ways of doing things that are particular to their group. Imitation is a key mechanism through which children learn the socially acceptable behaviours of such social groups. I investigate what kinds of imitation differences exist in autism and what they might mean for autistic people’s social interactions and bonding with others.

Recently, my collaborators and I developed a method called CAMI (short for Computerised Assessment of Motor Imitation). CAMI uses motion data from two people to evaluate how well one person imitated the other. CAMI does this by automatically assessing the relative importance of each joint and exporting a single value ranging from 0 to 1, which indicates the person’s imitation performance.


We have shown that imitation ability as measured by CAMI alone could identify which children had an autism diagnosis with over 85% accuracy. At present, we are working on optimising CAMI for off-the-shelf 2D cameras and on testing its diagnostic and therapeutic utility in young children.


Inge-Marie Eigsti (University of Connecticut)

Adam Eggebrecht (Washington University)

Daniel Lidstone (Kennedy Krieger Institute)
Stewart Mostofsky (Kennedy Krieger Institute)
Carolina Pacheco (Johns Hopkins University)

René Vidal (Johns Hopkins University)

Key Publications