Abstract

The ability to extract regularities from the environment is arguably an adaptive characteristic of intelligent systems. In the context of speech, statistical word-learning is thought to be an important mechanism for language acquisition. By taking into account individual differences in speech auditory-motor synchronization, an independent component analysis of fMRI data reveals that the neural substrates of this cognitive ability are not shared across individuals. While a network of auditory and superior pre/motor regions is universally activated to produce learning, a fronto-parietal network is instead additionally and selectively engaged by some individuals, boosting their performance. Interfering with the use of this network via articulatory suppression (producing irrelevant speech during learning) normalizes performance across the entire sample. Crucially, the engagement of this network predicts speech auditory-motor synchrony, directly relating this cognitive skill with language abilities.