Shultz et al report in a letter to Nature (requires subscription) on a phylogenetic analysis of the evolution of social stability in primates. This is a pretty complicated phylogenetic analysis involving Bayesian estimates of evolutionary events based on the distribution of social characteristics across the phylogenetic tree.
Their main findings are consistent with a picture of primate social evolution proceeding directly from solitary individuals to large multi-male/multi-female groups sometime around 50 million years ago (mya). Other types of social groups such as pair living and single-male/multi-female groups appear to have come about significantly later, around 16 mya. This "reversible--jump model" seems pretty non-intuitive, but is the best model for their data set. It will be interesting to see how this is received by primatologists.
Their results also support a "step-wise" acquisition of stable social groupings which starts out with loose aggregations of individuals banding together for predation defense in association with becoming diurnal. More stable social associations are a secondary occurrence that only come about later in time.