Automatic conversational scene analysis in children with Asperger syndrome/high-functioning autism and typically...

Area of Science:

• Neuroscience
• Developmental Psychology
• Computational Linguistics

Background:

• Individuals with Asperger syndrome/High-Functioning Autism (AS/HFA) experience mindblindness, a difficulty attributing mental states.
• This deficit is hypothesized to impact real-time conversational dynamics.

Purpose of the Study:

• To investigate if mindblindness influences conversational interaction patterns in AS/HFA.
• To identify quantifiable speech/silence dynamics that differentiate AS/HFA from typically developing individuals.

Main Methods:

• Conversations were modeled as interacting Markov processes using Gaussian mixture and observed influence models.
• Novel speech/silence patterns, termed Steady Conversational Periods (SCPs), were analyzed.
• Discriminative classifiers analyzed model parameters to distinguish participant groups.

Main Results:

• Steady Conversational Periods (SCPs) captured conversational dynamics in both AS/HFA and control groups.
• Analysis successfully distinguished the conversational dialogs of children and adolescents with AS/HFA from age-matched controls.
• Meaning-free speech/silence sequences were found to encode typical and atypical conversational dynamics.

Conclusions:

• Conversational synchronization, reflected in speech/silence patterns, is altered in individuals with AS/HFA.
• These findings suggest a direct link between theory of mind abilities and basic speech initiative behaviors.
• The study provides a quantitative method for analyzing conversational dynamics in neurodevelopmental conditions.