PROJECT SUMMARY The overall goal of this project is to develop and test a detailed neural and computational account of the brain mechanisms underlying speech motor sequence planning and motor program initiation and their breakdown in stuttering. Persistent developmental stuttering affects more than three million people in the United States, and it can have profound adverse effects on quality of life. Despite its prevalence and negative impact, stuttering has resisted explanation and effective treatment, due in large part to a poor understanding of the neural processing impairments underlying the disorder. This project aims to remove this critical barrier to progress through an integrated combination of behavioral, neurostimulation, and neuroimaging experiments and associated neurocomputational modeling. The studies in Aim 1 will characterize the neural mechanisms underlying sub- syllabic sequencing in neurotypical individuals. In Study 1.1, we will use functional magnetic resonance imaging (fMRI) to test the hypotheses that (1) phonological working memory in left posterior inferior frontal sulcus utilizes an onset-nucleus-coda representation rather than representing an entire syllable as a single item, whereas (2) ventral premotor cortex uses a syllable-based representation in which a fully learned syllable is represented by a single motor program. Study 1.2 uses non-invasive neurostimulation to directly test hypotheses concerning the neural substrates of improved performance accuracy (hypothesized to involve the cerebellum in concert with motor cortical areas) versus speed (hypothesized to involve left posterior inferior frontal sulcus) when learning novel syllables. In Aim 2 we investigate sequencing at the multi-syllabic level, including the effects of word learning in adults with and without stuttering (Study 2.1) and children with and without stuttering (Study 2.2). These studies will test the hypotheses that (i) novel nonword repetition performance is impaired in both children and adults who stutter compared to neurotypical speakers, (ii) learning a multi-syllabic word reduces working memory load compared to producing the same syllables prior to learning them as a word, and (iii) this reduced working memory load will reduce error rate differences between individuals who do and do not stutter and will increase fluency in those who stutter. In addition, Study 2.1 uses fMRI to probe the neural mechanisms involved in word learning, thereby testing the model-based hypotheses that (i) anterior inferior frontal sulcus is the site of a word buffer whose load is decreased by learning a multi-syllabic word compared to producing syllables in a novel nonword combination, and (ii) impaired initiation of motor programs, rather than impaired working memory per se, is the central contributor to stuttering. Together these studies will result in an improved neurocomputational account of the brain mechanisms underlying the sequencing and initiation of speech ...