Abstract
Research suggests that semantic memory deficits can occur in at least three ways. Patients can (1) show amodal degradation of concepts within the semantic store itself, such as in semantic dementia (SD), (2) have an impairment of semantic control, leading to difficulty accessing appropriate knowledge in line with current goals or context, as in semantic aphasia (SA), and (3) experience a semantic deficit in only one modality following degraded input from sensory cortex. Patients with SA show damage to prefrontal cortex which extends posteriorly (PF+), or damage restricted to temporoparietal regions (TP-only), and have deficits of semantic control and ‘access’ across word and picture tasks, consistent with the view that their problems arise from impaired multimodal control processes. This thesis aims to explore the nature of these deficits, in four themes. (1) “Refractory effects” in SA patients are explored across modalities – i.e., these patients are shown to experience declining accuracy in cyclical matching tasks when semantically-related sets are presented rapidly and repeatedly. (2) We studied one case study with ‘verbal-only’ refractory effects, to investigate an apparent anomaly in the literature – the existence of patients who have ‘access’ deficits which are restricted to a single modality. These patients challenge the notion that semantic control processes are modality-general. We assessed the hypothesis that multimodal semantic control/ access impairments can follow a modality-specific pattern if paired with an input deficit of a single modality. (3) We explore the effect of lesion location on behavioural performance of semantic aphasia (SA) patients, who have PF+ or TP-only lesions by bringing together data published previously in different papers, together with some new SA cases. Past research suggests SA patients with these two lesions may show similar deficits of semantic control, yet the functional neuroimaging literature proposes a unique role for the prefrontal cortex. PF+ patients were less fluent, showed more associative picture naming errors, and overall somewhat stronger SA characteristics (e.g., they were more inconsistent, and less affected by frequency). (4) Semantic control recruits a wide cortical network, in both the left hemisphere (LH) and right hemisphere (RH). Semantic representations in the RH are partially distinct from the LH, including specialised knowledge of faces and metaphors. Our aim was to test whether damage to RH control regions would negatively affect performance on semantic control tasks which use items stored in the RH, in a similar way to our SA patients in the LH. Overall, the results suggest that semantic control operates in an amodal fashion, with deficits found across modalities. There was evidence to suggest a wide network involved in semantic control beyond the prefrontal cortex – including left posterior cortex and right hemisphere regions. However, these regions are subtly distinct in their role in semantic control.