Experimental research on choice in shared resource problems focuses on contextual and individual variation. Experimenters are interested in how factors including incentive structures, social dynamics, and cultural differences influence choices. By contrast, theoretical research focuses on the development of formal models that are robust to these sources of variation. The aim of this paper is to provide a model of social decision-making that overcomes this divergence between experiment and theory. This model is based on the idea that individual and contextual variation in the way people invest in shared resources is the product of variation in their social motivation. The paper builds on two well-developed theories of social choice behavior – the Experience Weighted Attractions model based on selfish reinforcement learning, and the Conditional Cooperation model based on social beliefs and preferences – with a view to understanding choice in terms of selection between the social motives represented by the two models. I show that this approach can accurately predict the social motives used to generate choice in artificial simulations, and that the model can outperform the EWA and CC models alone. I then apply the model to three prominent experimental cases. I show that increasing the return on public good investment decreases cooperative social motives; that allowing peer punishment in a public goods game increases cooperative motives; and that culture mediates the effects of punishment on social motive selection. The paper concludes with discussion of some of the theoretical and experimental avenues opened up by the approach offered.