Simulating individual differences in language ability and genetic differences in FOXP2 using a neural network model of the SRT task

Recent work has shown that individual differences in language development are related to differences in procedural learning, as measured by the serial reaction time (SRT) task. Performance on this task has also been shown to be associated with common genetic variants in FOXP2. To investigate what these differences can tell us about the functional properties of language processing, we present a computational model of the SRT task. We varied parameters in the model to observe their effects on performance in the task. We found that the combined effect of several model parameters produced changes in the learning trajectory that were similar to those observed behaviorally.