Semantic memory Flashcards
Models of encoding structure
Collins and Quillian 1969 - Hierarchical, a tree of knowledge you travel down to retrieve some info, with different nodes from superordinate (animal) to basic (bird) to subordinate (ostrich) to associated semantic features (‘runs fast’).
McClelland et al 1995 - memories are first stored in the hippocampus, which then supports reinstatement of existing memories in the neocortex (aka consolidation). Synaptic changes in the neocortex means the existing memory is altered a little on each reinstatement, and remote memory is based on these neocortical connectivity changes.
Conditions affecting semantic memory
Stroke - large lesions often cause aphasia, which can be associated with semantic impairment
Herpes simplex encephalitis - can impair semantic, but also often involves episodic and various other functions
Alzheimer’s - starts with medial temporal lobe atrophy and episodic impairment, but as it progresses can involve lateral TL and semantic memory and other functions
Semantic dementia - first described by Warrington 1975. In its early stage, there is atrophy in anterolateral temporal lobes and a specific impairment in semantic memory (see Grahams et al 2000, 2001). Often affects left hemisphere more than right.
Modality-specific effects - for
Warrington and McCarthy found a stroke patient TOB who was semantically impaired when presented with words but not with pictures, and later found two patients impaired at word naming but not when given verbal descriptions. They suggested there were different semantic stores.
Modality-specific effects - against
Hodges et al 1995 - A longitudinal study of semantic dementia patient JL showed knowledge loss that reflects a ‘pruning back’ of Collins and Quillian’s semantic memory tree, consistent across all modalities of input and output.
Vandenberghe et al 1996 - PET scanning of patients in a semantic knowledge task involving three levels (the first baseline, the latter two using more complex semantic knowledge), showed activity in the same area when the task involved using pictures as words. Areas included parieto-temporal junction, fusiform-inferior occipital junction, middle temporal lobe, inferior frontal lobe. So no verbal/picture dissociation. Maybe stroke patients have damage to one of the access routes to a single unitary semantic knowledge store?
Semantic hub model
Patterson et al 2007 - Review of many imaging studies showed consistent temporal pole activation regardless of type of stimulus used.
Perhaps the temporal pole is a ‘hub’ used for semantic retrieval, and integration of different bits of knowledge, with the different categories represented around the neocortex.
This would account for semantic deficits both in selective temporal pole lesions, and widespread cortical stroke lesions.
Category-specific effects - for, and then modifications
A double dissociation was found between living and non-living things in stroke patients, even when controlling for familiarity, age of acquisition of stimuli. Warrington and Shallice proposed a sensory-functional distinction, where living things were put in a ‘sensory’ store and objects were categorised based on their use.
But one of the patients they reported as impaired on living things only was later found to be fine on body parts, and impaired on precious stones and musical instruments.
Functional neuroimaging studies suggest multiple levels of category-specific processing, with variability in activity levels when naming people, objects, tools. Martin and Chao 2001 reviewed imaging studies and suggested that category-specific knowledge is distributed in networks that parallel the sensory, motor and language processing areas in brain, e.g. tools represented in motion perception and motor planning areas, characteristic object colours in colour perception area.
