About 30 percent of Parkinson's patients will become demented as the disease progresses, but the cognitive impairment is distinct from that in Alzheimer's.

Trevor Robbins, of Cambridge University in England, described experiments in which PD patients temporarily discontinued their drug regimen and took various learning tests. The researchers imaged, with PET, the brains of these patients while they conducted the tests. They found that short-term memory and spatial working memory, which we use to plan moves in a chess game, for example, improve with L-dopa therapy. However, L-dopa impairs another kind of learning called reversal learning, which involves cognitive flexibility such as being quiet while in class but speaking up in discussion. Yet other forms of learning were unaffected by L-dopa.

This points to a current problem with L-dopa therapy. It floods the entire brain with dopamine but not all brain areas are equally dopamine-depleted in Parkinson's. Those aspects of cognitive function that depend on relatively more dopamine-poor brain regions are boosted by L-dopa treatment while other aspects might get worse because the brain areas underlying them are becoming overstimulated with therapy, Robbins said.

Mark Gluck reported on the use of computational modeling to understand cognitive deficits in Parkinson's. His research asked the question of how neurodegeneration of the substantia nigra-striatal projection-affects other brain areas that connect to this projection. This work is also geared toward understanding the role the basal ganglia-which house both the substantia nigra and the striatum-play in cognition as compared to the memory functions of the hippocampus and frontal cortex. Gluck's computer models build simulations that predict the effect of drugs and can then be tested in human experiments.

For example, Gluck et al., subjected people with Parkinson's to a test that requires incremental categorization and building predictions based on disparate, partially predictive clues. Healthy people initially use a simple strategy to solve the task but then progress to a more complex one as they develop a "gut feeling" for the test. People with PD remain stuck with the simple strategy, Gluck reports. The computer model largely predicted this difference, Gluck said, and his group is currently studying which brain areas are responsible for it.

The learning deficits in Parkinson's are distinct from those in Alzheimer's, Gluck noted, possibly reflecting the fact that PD damages the striatum while early AD involves the medial temporal lobe (which includes the hippocampus) but largely spares the basal ganglia. In support of this notion, Gluck and coworkers found that people with PD were slow to learn associations between cues and outcomes but were good at applying newly learned associations to other contexts. People with early AD, however, had less trouble making initial associations in the same test but were unable to transfer them to other situations. This reflects decreasing ability, in AD, to distinguish relevant from irrelevant information, Gluck said. Intriguingly, the slow learning of PD patients disappeared once the patients came off L-dopa, said Gluck, corroborating Robbins' finding that drug therapy actually induces some cognitive deficits that are commonly attributed to the disease.

Barbara Knowlton of University of California, Los Angeles, reported that people with Parkinson's show marked deficits in habit learning, which happens subconsciously as a person keeps repeating a certain task, for example driving to work seemingly on autopilot. In a small imaging study with seven PD patients, Knowlton found that habit learning depends heavily on the caudate nucleus, a part of the striatum. The Parkinson's patients tended to compensate for poor habit-learning by calling on their hippocampus, which normally is more involved in declarative, conscious learning of new facts and events. This may explain why people with PD feel that many tasks they once performed with ease now seem to require a conscious effort, Knowlton writes.

In a symposium on this topic on Sunday, Emmeline Edwards of the National Institute of Neurologic Diseases and Stroke noted that NINDS is expanding its funding of neurobiological mechanisms underlying cognitive problems in Parkinson's.-Gabrielle Strobel.

Reference:Gluck MA, Robbins T, Knowlton B, Edwards E. Society for Neuroscience 2001.