Science Watch

Psychologist Phyllis Koenig, PhD, explains her research on semantic memory with a simple metaphor: If you want to build a tower out of blocks and you can't, it may be that you don't have the blocks to do it--a problem with supplies. Alternatively, it may be that you don't know how to stack them--a problem with process.

It's similar for using semantic memory: the part of memory that allows us to name and categorize everything we sense. Similar to building the tower, if you can't remember what an object is, you might not have the information stored in your brain--a lack of content. Alternatively, you might not be able to put your pieces of knowledge together--a breakdown in the process.

For at least 20 years, researchers have debated which part of this equation breaks down to cause semantic memory problems, particularly in patients with Alzheimer's disease, whose frequent difficulty with semantic memory can interfere significantly with their quality of life. By most accounts, those who believe that a loss of content is the primary culprit have held most of the cards.

But new research by Koenig and her colleagues at the University of Pennsylvania's School of Medicine provides new evidence for the idea that the processes involved in semantic memory may be equally important.

In a study published in the March issue of Neuropsychology (Vol. 21, No. 2), they find that people with Alzheimer's disease retain the ability to form new semantic memories when they use a simple, automatic process to place objects into categories.

They struggle only when asked to use a more complex rule-based process to access semantic memory, which requires higher order processing.

This research not only says something about what goes wrong in Alzheimer's disease, but also sheds light on how semantic memory operates in the brain. Specifically, it shows how the two aspects of memory are intertwined, says University of California San Diego neuropsychologist David Salmon, PhD.

"While the study doesn't rule out the possibility that there's semantic degradation in Alzheimer's disease, it does allow us to rule in impaired access to semantic knowledge," he notes.

What's more, the finding could lend insight into the biological underpinnings of various aspects of Alzheimer's disease, says psychologist Alex Martin, PhD, senior investigator in the Cognitive Neuropsychology Section of the Laboratory of Brain and Cognition at the National Institute of Mental Health.

"It would really be interesting to try to link these behavioral deficits with different patterns of neuropathology," he says.

Categorizing 'crutters'

As a first step toward that goal, Koenig and her colleagues took on the challenge of showing that the brain uses more than one process--requiring very different brain circuits--to form and use semantic memory.

"There's a relatively rapid association process that allows us to recognize a typical chair, for example," explains the head of Koenig's lab, neurologist Murray Grossman, MD, who co-authored the paper. "But if you see something unusual, you can't just rely on this rapid process. You need to piece together knowledge represented in semantic memory using a more complex cognitive process, which includes executive functions such as attending to the specific features of an object that are important for its meaning."

It's one or both of these processes that Koenig and Grossman believe may break down for some people suffering from semantic memory deficits.

To test this idea, they, along with Columbia University's Ed Smith, PhD, examined the ability to form and manipulate new semantic memories in people with Alzheimer's disease and among people without cognitive impairment.

Koenig created 64 drawings of novel, but biologically plausible, animals, using all variations of six features: snout, legs, color, neck, teeth and tail. A series of tests determined that people primarily used snout, legs, color and neck to sort the animals into categories while they generally ignored teeth and tail.

The researchers then randomly selected one drawing to act as a prototypical example of a fictitious category of animals called "crutters." The 20 animals that matched the prototype on at least three of the most salient features were considered crutters. The rest of the animals--having two, one or no salient features in common with the prototype--varied in how much they looked like crutters.

Once they developed the crutter categorization system, the researchers trained study participants on the task and then asked them to decide which of the 64 animals were crutters. Participants had to use either direct comparison to make their assessments--judging which animals were crutters based on their overall resemblance to the picture of the prototype--or a stated rule that members of the category matched the prototype on three of the four target features.

The findings were more robust than the researchers anticipated. On average, people with Alzheimer's disease were as good as age-matched controls at using comparison-based judgments about the novel animals. Their performance suffered only when they had to use rule-based judgments.

The researchers got even stronger results when they limited their analysis to those patients who had demonstrated deficits in semantic memory. Those without semantic memory deficits--but who were equally impaired on a general scale of dementia--were able to make rule-based judgments almost as well as people without cognitive deficits.

"The implication of this finding is that the rule-based loss is not trivial, and may be the core cause of the [semantic memory deficits] in Alzheimer's disease," says McGill University neurologist Howard Chertkow, MD, who studies semantic memory deficits in people with the disease. "This research proves that an approach to semantic memory deterioration that focuses on impairment in process, rather than on storage of long-term information, can yield fascinating and informative results."

Connecting behavior to brain

Another finding from this study may also help Koenig and Grossman connect their behavioral findings to specific types of brain pathology. Specifically, those Alzheimer's patients who did poorly on the rule-based judgments also scored poorly on tests that measured higher order executive function, which controls functions such as planning and rule use. By connecting semantic memory deficits to problems with executive function, these findings suggest that the area of the brain that controls executive function may be critical to rule-based semantic memory. It also leaves open the idea that other types of semantic memory deficits are associated with damage to different parts of the brain.

This research strongly supports the idea that semantic memory involves multiple processes supported by

different brain circuits, says Salmon. As part of a larger program of investigation at Penn that uses brain imaging to link areas of degeneration to specific deficits, Koenig and her colleagues have homed in on those circuits in healthy young adults. In particular, they have found that rule-based categorization involves specific portions of frontal cortex that appear to contribute to executive functioning.

"That's exactly what you'd want to do," says NIMH's Martin. "Piece out what part of the patients' impairment reflects an actual loss of knowledge, which would be expected to be tied to pathology in the left temporal lobe, and what part reflects a processing deficit--for example, impaired retrieval--perhaps tied to frontal lobe pathology."

Beth Azar is a freelance science writer in Portland, Ore.