NeuroAI and Neural Development

Autism spectrum disorders

Autism spectrum disorders (ASDs) are neurodevelopmental in origin. We are trying to understand more about ASDs through several parallel approaches.

• Zebrafish autism mutants. We are examining how the development of behavior and neural circuits are altered in zebrafish mutant for genes that in humans are associated with ASDs. In this recent paper in The Journal of Neuroscience we showed that zebrafish fmr1 mutants, a gene which when mutated in humans causes Fragile X Syndrome, show sensory defensiveness, altered hunting and social behavior, and altered patterns of neural activity.
• Zebrafish sleep. We recently used Hidden Markov Models to show that zebrafish sleep consists of distinct light and deep sleep substates (paper). We are now investigating how mutations in autism-associated genes change sleep architecture in zebrafish.
• Machine learning analysis of motor behavior in human infants. In collaboration with Natasha Marrus we are using recently-developed techniques for automated human pose extraction from video data, combined with advanced machine learning techniques, to understand how behavioral phenotypes are altered in human infants subseqently diagnosed with autism.
• Computational modeling of neural circuits. By incorporating the changes in excitatory-inhibitory balance that are believed to underly autism into recurrent neural network (RNN) models, we are trying to understand the changes that autism induces at the level of circuit computations. We recently reviewed RNN models of neural circuit development here.