Semantic Memory: Neurobiological Perspective
Consider the following questions: Do acorns have stems? Do you tune a guitar by tightening and loosening the strings? Do deer have white noses? If you were able to answer these questions correctly (yes, yes, no), then you were using semantic memory. Much of semantic memory relates to the brain systems that enable us to store and retrieve semantic knowledge. This article explores the neurobiology of this process.
Multiple Long-Term Memory Systems
When you remember that acorns have stems, you are using your semantic memory. When you remember the time you gathered acorns with your father in fifth grade, you are using your episodic memory. In both cases you are retrieving information from long-term memory. Are there, in fact, two separate memory systems for these two types of memories?
In 1972, the psychologist Endel Tulving first argued that there are two separate memory systems. Tulving distinguished between memories that have an autobiographical reference (i.e., episodic memories) and memories that do not (i.e., semantic memories). Cognitive psychologists have disagreed about whether episodic and semantic memories reflect two distinct systems or whether they are part of a common long-term memory system. However, neuropsychologists have provided compelling evidence that there are two distinct systems: one for our knowledge of events and one for our knowledge of facts. This evidence comes from patients who have selective impairments of semantic memory.
Impairments if semantic memory correlate with several etiologies of brain damage. The most common cause of semantic-memory impairments is Alzheimer's disease, a progressive dementia that causes widespread cortical degeneration. Semantic memory impairments also attend acute illnesses such as stroke or infection. These conditions, however, commonly involve both episodic and semantic memory impairments. Yet there is one condition, semantic dementia, in which patients exhibit a progressive loss of semantic memory (Hodges, Patterson, Oxbury, and Funnell, 1992). By definition, patients with semantic dementia have an impairment in semantic memory but normal episodic memory.
One of the earliest symptoms of semantic dementia is anomia, an impairment in the ability to find words when speaking or writing. Patients with this disorder also have problems with semantic memory tasks that do not require language at all. The pyramids and palm trees test is a common test of nonverbal knowledge that requires the patient to identify semantic relationships between pictures (Howard and Patterson, 1992). For example, the patient might be asked to indicate which picture belongs with a pyramid: a palm tree or a fir tree. This task does not require language comprehension or production (i.e., the patient is simply pointing at pairs of pictures), but it does require semantic memory. Patients with semantic dementia are unable to answer simple questions of this type. The selective loss of semantic memory in semantic dementia may indicate idea that semantic memory and episodic memory are two distinct long term memory systems.
In all of the etiologies of brain damage associated with semantic memory loss, there is widespread and often bilateral damage to the temporal lobes. Semantic dementia may result from a variant of Alzheimer's disease in which degeneration is initially restricted to the temporal lobes. Semantic dementia may also result from Pick's disease, another degenerative disease that is associated with temporal lobe atrophy. Pick's disease does not affect the medial temporal lobes, including the hippocampus, which is a part of the brain associated with episodic memory.
Separate or Unitary Semantic Stores
When Tulving first defined semantic memory, he described it as a "mental thesaurus" that supported language use. Since that time, many investigators of semantic memory have broadened their definitions of semantic memory to include both verbal and nonverbal knowledge. Clearly, semantic memory can be accessed from both verbal labels and visual depictions. Many psychologists, however, have questioned whether there is a unitary semantic memory store or whether there are separate stores for verbal and non-verbal memory.
We have already seen one piece of evidence for a unitary semantic memory store: As mentioned above, patients with semantic dementia have impairments of both verbal and nonverbal tests of semantic knowledge. Curiously, a few patients have deficits that seem restricted to a single test modality (e.g., Beauvois, 1982). But it is unclear whether this pattern reflects a true loss of semantic memory or merely an impairment in accessing semantic memory from one kind of input.
Another method for investigating questions about semantic memory is to noninvasively measure regional changes in metabolism or blood flow in the brain of a healthy subject who is performing a cognitive task (e.g., functional magnetic-resonance imaging). Several neuroimaging studies have tested the existence of a unitary semantic store. In one such study (Vandenberghe et al., 1996), brain imaging was conducted during either a verbal semantic task (word matching) or a visual semantic task (similar to the pyramids and palm trees test). The activation of many common regions of the brain in these tests is consistent with the theory that there is a unitary semantic memory store for both verbal and nonverbal retrieval.
Categories of Semantic Memory
For decades cognitive psychologists have been debating the two issues discussed above: whether semantic memory is distinct from episodic memory and whether there is a unitary semantic store. The evidence reviewed above simply fuels the controversy with evidence from neurobiology. There is, however, another question about semantic memory that originated with neuropsychologists: Are there separate stores of semantic memory for different taxonomic categories, such as animals and tools?
This might seem like a strange hypothesis, but the evidence for it is very compelling. Imagine administering a simple picture-naming task to a brain-damaged patient. You show the patient pictures of all sorts of objects, such as a trellis, a compass, and an abacus, and the patient can name them with little difficulty. However, when you show the patient a picture of something as simple as a duck or a bee, the patient struggles and fails to come up with the name. This very pattern of behavior has been reported in dozens of patients, most of them afflicted with herpes simplex encephalitis (a brain infection), who have semantic memory impairments restricted to categories of natural items (Warrington and Shallice, 1984).
One hypothesis for this peculiar pattern of impairments is that natural items (e.g., plants and animals) are more difficult to identify because they are less familiar or more visually complex (Funnell and Sheridan, 1992). Carefully controlled studies have largely ruled out this supposition, however (Farah, McMullen, and Meyer, 1991). There have been a few cases of the reverse pattern: patients with impaired knowledge of man-made objects but with relatively well-preserved knowledge of natural objects (Warrington and McCarthy, 1987). Although rare, this disorder provides stronger evidence for the hypothesis that there are separate memory stores for natural objects and human-made objects. Some researchers (Caramazza and Shelton, 1998) have argued that separate stores have developed for categories for which there is some evolutionary significance (e.g., plants, animals, and tools).
Domains of Semantic Memory
In the preceding section, the evidence for separate semantic memory stores for different categories seemed clear-cut. However, many investigators believe that there is another, more plausible, interpretation of these category-specific deficits. The following two questions illustrate an important difference between our knowledge of natural objects and our knowledge of human-made objects: How would you define tiger so that someone could distinguish it from similar concepts (e.g., leopards, lions, and bobcats)? You would probably mention something about the color and size of tigers. In contrast, how would you define a clock (as distinct from a radio or a lamp)? In this case, you are probably less likely to talk about color or size or even shape, because clocks come in all varieties. Instead, you would probably talk about the function of clocks. Hence the defining attributes of natural items are more likely to be visual than those of human-made items.
This distinction leads to another hypothesis about a distinction within semantic memory: there may be different memory stores for different kinds, or domains, of semantic knowledge. Knowledge of any concept may be distributed across a variety of distinct memory stores that are tied to sensorimotor systems (Allport, 1985). According to this hypothesis, patients who appear to have a deficit pertaining to natural objects in fact have a visual-knowledge deficit that is most evident when tested on items defined by visual attributes (i.e., natural objects). The difficulty of distinguishing absolutely between object category and knowledge domain makes it is hard to experimentally distinguish between these two hypotheses. Some investigators have pointed to interesting exceptions to the natural/man-made distinction that are more consistent with a visual/functional dichotomy. For example, patients described as having a deficit pertaining to natural objects may also have impairments with musical instruments, which are largely defined by visual attributes (Warrington and Shallice, 1984).
Neuroimaging studies have also been used to test this hypothesis (e.g., Thompson-Schill, Aguirre, D'Esposito, and Farah, 1999). For example, when subjects are asked to name natural objects (e.g., animals), brain activation is observed in areas associated with vision; in contrast, when subjects are asked to name human-made objects (e.g., tools), brain activation is observed in areas concerned with motor control (Martin et al., 1995). Results of this sort have been taken as evidence that the fundamental distinction within semantic memory is one of domain of knowledge and not taxonomic category.
There are other sources of evidence that semantic memory is organized according to domain. Recall that patients with semantic dementia have global deficits in semantic memory when tested either verbally or visually. However, there is one domain in which these patients show preserved semantic knowledge: Patients with semantic dementia often continue to demonstrate normal use of objects long after they fail to identify these objects correctly on other tests of semantic memory. This pattern of behavior stands in stark contrast to another disorder, ideational apraxia, an impairment in object use in patients who test normally for traditional semantic memory (Ochipa, Rothi, and Heilman, 1989). Thus, there may be a semantic memory store for object use that is distinct from other domains of semantic memory.
Neurobiological studies of semantic memory have addressed some fundamental questions of semantic memory and have raised a few new questions as well. Disorders that selectively impair semantic memory provide evidence that semantic memory is a memory system that is distinct from episodic memory. Neuroimaging studies have indicated that verbal and nonverbal semantic retrieval depends on a common memory store. Within this semantic store, however, there appear to be distinctions based either on taxonomic category or type of knowledge. Neuroimaging and neuropsychological studies support the theory that semantic memory is distributed across different sensorimotor systems, such as object appearance and object use.
Allport, D. A. (1985). Distributed memory, modular subsystems and dysphasia. In S. K. Newman and R. Epstein, eds., Current perspectives in dysphasia. Edinburgh: Churchill Livingstone.
Beauvois, M. F. (1982). Optic aphasia: A process of interaction between vision and language. Philosophical Transactions of the Royal Society of London, 298, 35-47.
Caramazza, A., and Shelton, J. R. (1998). Domain-specific knowledge systems in the brain the animate-inanimate distinction. Journal of Cognitive Neuroscience 10, 1-34.
Farah, M. J., McMullen, P. A., and Meyer, M. M. (1991). Can recognition of living things be selectively impaired? Neuropsychologia, 29, 185-193.
Funnell, E., and Sheridan, J. (1992). Categories of knowledge? Unfamiliar aspects of living and nonliving things. Cognitive Neuropsychology 9, 135-153.
Hodges, J. R., Patterson, K., Oxbury, S., and Funnell, E. (1992). Semantic dementia. Progressive fluent aphasia with temporal lobe atrophy. Brain 115, 1,783-1,806.
Howard, D., and Patterson, K. (1992). Pyraminds and palm trees: A test of semantic access from pictures and words. Bury St. Edmunds: Thames Valley Test Company.
Martin, A., Haxby, J. V., Lalonde, F. M., Wiggs, C. L., and Ungerleider, L. G. (1995). Discrete cortical regions associated with knowledge of color and knowledge of action. Science 270, 102-105.
Ochipa, C., Rothi, L. J., and Heilman, K. M. (1989). Ideational apraxia: A deficit in tool selection and use. Annals of Neurology 25, 190-193.
Thompson-Schill, S. L., Aguirre, G. K., D'Esposito, M., and Farah, M. J. (1999). A neural basis for category and modality specificity of semantic knowledge. Neuropsychologia 37, 671-676.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving and W. Donaldson, eds., Organization of memory. New York: Academic Press.
Vandenberghe, R., Price, C., Wise, R., Josephs, O., and Frackowiak, R. S. (1996). Functional anatomy of a common semantic system for words and pictures. Nature 383, 254-256.
Warrington, E. K., and McCarthy, R. A. (1987). Categories of knowledge. Further fractionations and an attempted integration. Brain 110, 1,273-1,296.
Warrington, E. K., and Shallice, T. (1984). Category specific semantic impairments. Brain 107, 829-854.
Revised bySharon L.Thompson-Schill