The control of language production and its neural substrates in Parkinson's disease

Abstract

Individuals with Parkinson’s disease (PD) exhibit deficits across a range of language measures. The underlying source of these impairments has yet to be discerned, and this reflects an overarching lack of clarity regarding the participation of the basal ganglia in language production. Current accounts suggest that these nuclei play a secondary role in language processing as a product of involvement in cognitive control. In the language realm, cognitive control encompasses the processes of verbal selection and suppression. This account is consistent with the anatomical connectivity of the basal ganglia, which exchange feedback with regions of the prefrontal cortex subserving cognitive control. What remains to be discerned is the manner in which these networks participate in spoken language production, and the nature of their disruption in PD. This thesis therefore aimed to determine whether the language impairments observed in PD arise as a result of disrupted verbal selection and verbal suppression processes, and to elucidate the associated changes in frontostriatal activity.A measure that has been widely utilised to study these processes is the Hayling Sentence Completion Task (HSCT). Critical appraisal of this task and consideration of its design in the context of other commonly employed paradigms scaffolded the conceptualisation of this thesis. A number of task limitations were identified: ability to isolate components of cognitive control, variable temporal parameters, minimal consideration of input modality, variation in response requirement, and minimal use of imaging.Four studies were designed to address the limitations outlined above. The first utilised an object-based negative priming paradigm in which participants were required to name a picture that had previously served as a distractor. It was found that individuals with PD were unimpaired in their ability to suppress an irrelevant representation, recording a negative priming effect equivalent to healthy controls. The second study employed a hybridisation of the HSCT and a competitor priming paradigm to allow for observation of the time-course of verbal suppression. This involved integration of sentence completion trials (in which the prepotent response had to be suppressed in favour of a word unrelated to the sentence) and subsequent naming of a picture representing the suppressed response or its semantic relation. Again, the PD group performed commensurately with the control group, with the exception of error processing. The PD group were found to make significantly more errors in picture naming when the sentence completion trial immediately prior had been executed incorrectly. This effect was postulated to reflect the intermittent failure of frontal systems responsible for modulating cognitive control, however data analysis was based on a limited number of valid trials and conclusions were thus highly speculative.A third study was based on the observation that individuals with PD present with impairments in the suppression of a strongly prepotent response, as measured in the HSCT. However, it was suggested that this task does not consider the influence of strategy generation as separate from suppression ability. A novel fMRI variation on the HSCT was therefore utilised to address this concern. Behavioural results again demonstrated no group differences, however the PD participants presented with altered patterns of activity in task-relevant frontostriatal circuits, including hyperactivation of the striatum and dorsolateral prefrontal cortex. This hyperactivation was interpreted as evidence of compensatory mechanisms, recruited to bolster disease-driven signal loss in these circuits.The final study considered verbal selection. This process recruits cognitive control when a number of alternative linguistic units may appropriately fulfil a task requirement, and are thus in competition. Deficits in verbal selection have been identified in PD cohorts, however this evidence is largely drawn from studies of single-word processing. It was further noted that the HSCT only included sentence stems with few competing alternatives. The study therefore utilised a variation on the HSCT, combined with fMRI, which required participants to complete sentence stems with systematically varied selection demands. Again, behavioural results revealed no main effect of group (although a group-by-condition interaction indicated that control participants recorded a significant variation in accuracy between low and medium constraint, whereas PD did not). However, imaging data revealed a significant decrease in activity in the task-relevant ventrolateral prefrontal cortex and striatum in the PD group during conditions of increased selection demand. As this group were considered to be in a mild-moderate stage of disease severity and were taking dopaminergic medication, it was speculated that results may reflect an overmedication effect in frontostriatal pathways that were as yet unaffected by disease pathology.The overall conclusion of this thesis was that verbal selection and suppression processes were largely intact in this cohort of individuals with mild-moderate PD, and this appeared to be supported by compensatory neural mechanisms, acting to bolster the output of frontostriatal circuitry. These findings help explain the heterogeneity of cognitive-linguistic deficits observed in PD, and could have future applications in the development of treatment protocols that capitalise on these compensatory mechanisms.