Semantic memory is organized into a series of hierarchical networks, with major concepts represented as nodes and properties/features associated with each concept
Warrington (1975) describes cases of progressive semantic degeneration (Semantic Dementia) which follows similar principles to this hierarchical model (lower properties lost first)
Semantic memory is organized by semantic relatedness/distance, with activation spreading from one concept to related concepts, decreasing as it gets further away
Some SD patients display "reverse concreteness effects", performing better on abstract than concrete words, but this is not a consistent finding across tasks and patients
Argued for subdivisions of the two properties: Sensory property divided into visual, auditory, taste, and tactile sensations; Functional property divided into entity behaviors and functional information
Hoffman & Lambon-Ralph (2013) noted that 'verbal listing' used to capture features of object categories (e.g. Cree & McRae, 2003) may not always easily capture sensory knowledge
Hoffman & Lambon-Ralph (2013) found that their ratings system generated a novel pattern - a clear distinction between different types of artifact, with mechanical devices (e.g. vehicles, appliances, instruments) distinct from other non-living objects
Current imaging evidence suggests a complex pattern of neuroanatomical specialisation in object representation, which cannot be accounted for by a simple sensory-functional dichotomy
Some patients present with a focal impairment of a particular semantic category, which might suggest semantic memory has such explicit categories as part of its functional architecture
Other researchers have argued that semantic representations may be differentiated into different 'kinds' of information (sensory/functional) or different 'modalities' of information (colour, smell, sound etc.), with features being critical rather than category labels per se
Semantic dementia cases have focal atrophy to the anterior temporal lobes (ATL), leading to the suggestion this region is a 'semantic hub' with modality specific semantic information held in 'spoke' regions linked to it
Early fMRI studies did not show ATL activation, likely due to signal drop-out in these regions, but later studies using distortion-corrected fMRI have confirmed ATL activation in semantic tasks
rTMS studies inhibiting left or right ATL increased picture naming RTs and affected semantic judgement tasks, with all types of concepts (living/nonliving) affected, suggesting a domain general system
Pobric et al. (2010) found that applying rTMS to inferior temporal lobes (praxis region) induced a nonliving worse than living pattern and greater impairment for highly manipulable items