How is knowledge about the world organised in our minds? Studies with people whose memory has been affected by brain damage, have pointed to a hierarchy of factual, ‘semantic’ knowledge, in which more general information must be accessed first on the way to more specific information.
Presented with a photo of a dog, patients with semantic dementia will often only be able to identify it as an ‘animal’. Presented with a desk, such patients will only recognise it as ‘furniture’. That is, they seem to have lost their object-specific knowledge but retained their knowledge of higher-order, ‘superordinate’ categories. This has led psychologists to propose that superordinate information is somehow less vulnerable to brain damage.
But now Glyn Humphreys and Emer Forde at Birmingham University have reported on a patient, FK, who appears to exhibit the opposite pattern. When he was a 22-year-old student, FK suffered frontal, temporal and occipital brain damage from carbon monoxide poisoning. Now FK is good at naming things at their ‘base-level’ (e.g. dog, desk, hammer), but is severely impaired at matching them with their superordinate categories (e.g. animal, furniture, tool).
This pattern “runs contrary to almost all of the neuropsychological literature on patients with semantic dementia”, the authors said. “The data are clearly inconsistent with strictly hierarchical accounts of semantic memory, in which access to superordinate knowledge is a prerequisite for accessing other forms of knowledge”.
The authors said their findings also challenge the idea that we have different memory stores for each of our senses. Patient FK’s ability to name different items was consistent regardless of whether they were presented to him by sight, sound, or touch. This “fits better with the idea that we have one semantic system…rather than a semantic system differentiated by modality”, they said.
Humphreys, G.W. & Forde, M.E. (2005). Naming a giraffe but not an animal: base-level but not superordinate naming in a patient with impaired semantics. Cognitive Neuropsychology, 22, 539-558.