Refractoriness within the semantic system: investigations on the access and the content of semantic memory

The starting purpose of this project was to investigate some issues related to the mechanisms underlying the efficient access to concepts within the semantic memory systems. These issues were mainly related to the role of refractoriness in explaining the comprehension deficits underlying semantic access. The insights derived from this first approach were then used to formulate and test hypotheses about the organization of the contents of the semantic system itself. The first part of the thesis presents an investigation of the semantic abilities of an unselected case-series of patients affected by tumours to either the left or right temporal lobes in order to detect possible semantic access difficulties. Semantic access deficits are typically attributed to the semantic system becoming temporarily refractory to repeated activation. Previous investigations on the topic were mainly based on single case reports, mainly on stroke patients. The rare examples of group studies suggested moreover the possibility that the syndrome might not be functionally unitary. The tasks used in the study were two word-to-picture matching tasks aimed to control for the typical variables held to be able to distinguish semantic access from degradation syndromes (consistency of access, semantic relatedness, word frequency, presentation rate and serial position). In the group of tumour patients tested access deficits were consistently found in patients with high grade tumours in the left posterior superior temporal lobe. However, the patients were overall only weakly affected by the typical temporal factors (presentation rate and serial position) characterizing an access syndrome as refractory. The pattern of deficit, together with the localization data, suggested that the deficit described is qualitatively different from typical semantic access syndromes and possibly caused by the disconnection of posterior temporal lexical input areas from the semantic system. In the second study we tried to answer the question whether semantic access deficits are caused by the co-occurrence of two causes (refractoriness and a lexicalsemantic disconnection) or whether the presence of refractoriness in itself is sufficient to induce all the behavioural effects described in access syndromes. A second aim of the study was moreover to investigate the precise locus of refractory behaviour, since refractory effects have also been reported in naming tasks in which the possibility exists that the interference might be located at a post-semantic lexical stage of processing. To address these issues a series of three behavioural experiments on healthy subjects was conducted. The tasks used were speeded versions of the same word-to picture matching tasks used in the previous study. A speeded paradigm was adopted in order to induce a mild refractory state also in healthy participants. The results showed that it was possible to induce, in the group of subjects tested, a performance similar to that of refractory semantic access patients. Since no post-semantic stage of processing is assumed to be necessary to perform these tasks it was argued that refractoriness arises due to interference occurring between representations within the semantic system itself. In the second part of the project, the finding that refractoriness arises due to interference involving semantic representations themselves, was used to investigate issues related to the organization of the content within the semantic memory. In particular, a second series of behavioural experiments was performed to investigate whether the way an object is manipulated is indeed a feature that defines manipulable objects at a semantic level. The tasks used were speeded word-to-picture matching tasks similar to those previously described. A significantly greater interference was found in the recognition of objects sharing similar manipulation than in the recognition of objects sharing only visual similarity. Moreover the repeated presentation of objects with similar manipulation created a ‘negative’ serial position effect (with error increasing over presentations), while the repeated presentation of objects sharing only visual similarity created an opposite ‘positive’ serial position effect (learning). The role of manipulability in the semantic representation of manipulable objects was further investigated in the last study of this work. In a second unselected group of brain tumour patients the ability to name living things and artifacts was investigated. Artifacts were manipulable objects, varying in the degree of their manipulability. Results from both behavioural and Voxel-based Lesion Symptom Mapping (VLSM) analyses showed that the only patients showing a selective deficit in naming artifacts (particularly highly manipulable objects) were patients with lesions in the posterior middle and superior portions of the left temporal lobe, an area lying within the basin of those regions involved in processing object-directed actions and previously linked to the processing of manipulable objects in a wide range of studies. The results of these last two studies support ‘property-based networks’ accounts of semantic knowledge rather than ‘undifferentiated network’ accounts. Overall this series of studies represents an attempt to better understand the mechanisms that underlie the access to semantic representations and, indirectly, the structure of representations stored within semantic networks. The insights obtained about the mechanisms of access to stored semantic representations were used as a tool to investigate the structures of the same semantic representations. A combination of different approaches was used (from behavioural speeded interference paradigms on healthy subjects, to neuropsychological case series investigations, as well as Voxel-based Lesion Symptom Mapping technique), to ‘cross-validate’ the results obtained at any level of analysis.