Changing Social Network Interactions and Dementia in Rural Honduras Villages Abstract Social connections may reduce dementia risk. But determining the extent and direction of any such causal effect is difficult. The relationship between social networks and dementia in developing world settings is also very incompletely understood. Here, we will exploit rare data about 2,011 people in 12 villages in Honduras who are part of an ongoing longitudinal study in order to evaluate how social network interactions are associated with dementia in this rural elderly population, using cross- sectional, longitudinal, and coincidentally experimental approaches. We assess dementia using a validated dementia assessment tool suitable for use in this rural population (PhotoTest). In Aim 1, we will first quantify the association between social network position and dementia. A primary hypothesis is that older adults with dementia will be socially marginalized within village-level social networks, even after accounting for pertinent confounders. Next, in Aim 2, we will assess how changes in the social networks of older villagers across a two-year interval are associated with their subsequent dementia status, exploiting the longitudinality of our data. A primary hypothesis is that a decline in social network degree and centrality will be associated with higher subsequent dementia risk. Finally, in Aim 3, we will assess how exogenously induced changes (induced by an underlying randomized trial that offered public health information to randomly chosen members of each village with monthly visits over a two-year period) in the networks of older villagers are associated with their subsequent dementia status. Here, since the network connections of the elderly change, under experimental pressure, over the two-year period from Wave 1 to Wave 3, we hypothesize that measures of dementia ascertained subsequently will be adversely affected. A primary hypothesis is that exogenously caused social marginalization of older adults will increase their dementia risk, compared to the elderly not experiencing such shocks to their social network.