Date of Award

Spring 1-2012

Document Type


Degree Name

Doctor of Philosophy (PhD)


Psychology & Neuroscience

First Advisor

Yuko Munakata

Second Advisor

Marie T. Banich

Third Advisor

Akira Miyake

Fourth Advisor

Mark A. Whisman

Fifth Advisor

Christopher Lowry


When we speak, we must constantly retrieve and select words in the face of multiple competing alternatives. Previous research has left many questions unanswered about how we achieve these fundamental cognitive control processes. This dissertation contributes to answering these questions at three levels. First, using well-controlled tasks and measures, we ask what specific aspects of language production drive cognitive control demands, as indexed by slower RTs to produce a verbal response. Second, we apply these unconfounded measures to fMRI experiments, to ask what neural substrates support cognitive control during language production. Third, we ask how these brain areas support cognitive control processes, by first simulating possible mechanisms in a neural network model and then empirically testing model predictions using pharmacological and clinical methods. In sum, the dissertation research suggests that cognitive control is needed during language production when responses compete with alternative task-relevant response options (underdetermined selection), compete with prepotent responses (prepotent selection), or are difficult to retrieve from semantic memory (controlled retrieval), and these demands interact both behaviorally and neurally. Shared neural substrates in left ventrolateral prefrontal cortex (VLPFC) support both underdetermined selection and controlled retrieval, while left VLPFC is not activated by prepotent selection demands. In contrast, an area of left dorsolateral prefrontal cortex (DLPFC) is sensitive to both underdetermined and prepotent competition. Neural network simulations suggest that competitive lateral inhibition in VLPFC is key for underdetermined selection, while other mechanisms subserved by VLPFC support controlled retrieval, and top-down biasing from DLPFC is critical for prepotent selection. As predicted by the model, increased inhibition under the GABA agonist midazolam improved selection, while anxiety (linked to reduced GABAergic function) was associated with impaired selection and reduced engagement of left VLPFC during selection. These findings enable a synthesis and reinterpretation of prior evidence, and suggest that language production is affected by both selection and retrieval mechanisms subserved by left VLPFC and DLPFC, and these processes interact in meaningful ways. Better understanding these fundamental aspects of language production may ultimately have implications for better understanding and treating impairments associated with prefrontal damage, as well as anxiety and depression.