Decision Making in Social Networks

We investigate how the network topology of social networks impacts decision making. First we look at sequential models of decision making with feedforward network toplogies. We see how rational agents incorporate knowledge of the network topology in order to make an optimal estimates of an unknown parameter. We give a condition for making this optimal estimate in terms of the row space of a matrix which encodes the network topology. We then show what this condition means for infinitely large networks. Then we extend a model of evidence accumulation for a two-alternative task to general networks where agents are allowed to communicate decisions. We detail the process rational agents must undergo and give detailed computations of the process for basic network structures.