Social networks and Alzheimer's disease

This seems interesting (press release via Science Blog):

"Many elderly people who have the tangles and plaques associated with Alzheimer's disease don't clinically experience cognitive impairment or dementia," said Bennett. "Our findings suggest that social networks are related to something that offers a 'protective reserve' capacity that spares them the clinical manifestations of Alzheimer's disease."

We already know that Alzheimer's progression is slower or less severe in people with higher education levels, higher verbal expression early in adulthood, and continued mental exertion later in life. But what seems interesting here is the autopsy data, which show that the Alzheimer's-related plaques are still developing in people with less severe symptoms, so that it appears the social networks are having a protective effect in the context of people who would otherwise have had worse symptoms:

The relationship between the amount of Alzheimer's disease pathology and cognitive performance changed with the size of the social network. As the size of the social network increased, the same amount of pathology had less effect on cognitive test scores. In other words, for persons without much pathology, social network size had little effect on cognition. However, as the amount of pathology increased, the apparent protective effect on cognition also increased. Thus, social network size appears to have offered a protective reserve capacity despite the fact that their brains had the tangles and plaques indicative of Alzheimer's disease.

One hypothesis would be that people have differing reserve capacities to make up for plaque-related declines in function, and that these depend on social interactions. But in that case, the mechanism would be of great interest with respect to the evolution of social interactions -- for example, is it neurotrasmitter-mediated, or hormonally mediated, or does it require the use of particular anatomical structures?

Depending on the mechanism, we might expect social networks to have strong effects on cognitive performance in non-Alzheimer's older people -- or maybe even in younger people also.