Losses and gains in social dilemmas

Abstract

People constantly make decisions that affect them and other people. The outcomes of such decisions are positive, negative, or 0, but we don’t know whether, and if so, how, the sign of such outcomes affects people’s decisions. This thesis tries to answer this question.

Chapter 1 outlines social dilemmas and some approaches for understanding them, particularly Prospect Theory. I highlight the specific question of how gains and losses influence decision-making: first by discussing loss aversion from non-social studies, then by reviewing the evidence from studies on gains and losses in social dilemmas. I conclude that there is no good answer to the original question, hence the need for further experiments.

Chapter 2 presents the results from 7 behavioural studies in which we directly investigated the effect of losses and gains in social dilemmas (2✕2 games). The results suggest that people try to avoid losses (loss avoidance), independent of whether this increases or decreases cooperation. Especially for short interactions, these effects can be quite large and have a strong influence on people’s social decisions. There was no consistent evidence for loss aversion.

Chapter 3 presents a theoretical development of linearly-additive decomposed 2✕2 games with which the framing of a 2✕2 game can be changed without changing the game itself, especially in the context of losses and gains. Decomposed 2✕2 games have existed for decades, but only in the context of symmetric Prisoner’s Dilemmas. I generalise decomposed games to all 2✕2 games (including asymmetric games and games with ties) and to asymmetric decompositions. I show which 2✕2 games can be decomposed in a linearly-additive way, how to decompose a given game, and provide some conceptual differences between standard matrix-form games and decomposed games. This chapter includes no empirical results, but shows how decomposed games can be used to manipulate the sign of the actions and outcomes independently of each other. I outline how this feature of decomposed games can be used to empirically study more nuanced aspects of gains and losses in social dilemmas.

The 4th chapter takes everything we have learnt in the preceding chapters and suggests a new, modified value function for Prospect Theory. This new value function consists of a 3-part separation into losses, zero, and gains by including separate intercepts for gains and losses. I also outline specific novel predictions that this model makes and how future studies could compare this new model to existing value functions.

The 5th (and final) chapter summarises the entire thesis and provides pointers for future research.