The development of the sunk cost bias

Abstract

The sunk cost bias is when people overvalue objects or projects because they have already invested time, money, or effort into them. Most sunk cost research over the past 50 years investigated the bias in adults, exploring the conditions in which people expect themselves and others to be biased by sunk costs. However, very little work has examined the developmental origins of the bias, despite much evidence that young children reason about costs for a host of predictions and inferences about others. Across three papers, this dissertation examines children’s (N = 990) and adults’ (N = 934) sunk cost predictions. Chapter Two first explored whether children predict others’ actions are biased by sunk costs. After seeing agents collect two identical objects but being able to keep only one, adults expected agents will be biased by sunk costs and choose high-cost objects. However, 5- to 6-year-olds chose between high- and low-cost objects equally. Across four experiments, children consistently failed to anticipate that sunk costs biased others’ choices, their own hypothetical choices, or their choices in interpersonal contexts where costs are sunk by others. Children were not insensitive to costs, though, as children predicted agents would collect low-cost objects in the future. Together, the findings from this chapter show that children do not anticipate sunk cost bias across various scenarios. Chapter 3 tested between two accounts for why children overlook sunk costs when predicting actions. On one account, children do not see sunk costs as causing future outcomes, while on another, they can recognize this causal link but do not see actions as avoiding losses. In three experiments, 5-7-year-olds again did not expect sunk costs to bias others’ actions, as they responded at chance when predicting which objects agents would keep. However, children reasoned about sunk costs to predict emotion, anticipating that agents would feel sadder about high-cost objects. Together, the findings of this chapter support the view that children see sunk costs as causally relevant but do not expect actions to compensate for losses. Chapter Four examined whether children can be prompted to anticipate the sunk cost bias. Before predicting which objects agents would keep, children were asked about effort, waste, or negative emotion. In three experiments, children around age 6 predicted the sunk cost bias when prompted with effort and around age 7 when prompted with waste. Prompting children with waste did not always lead to sunk cost predictions, though, and children only showed some sensitivity to predicting the bias with negative emotion. Overall, this dissertation shows that children do not spontaneously predict the sunk cost bias. Yet, children are not entirely unable to reason about sunk costs, as they can recognize how sunk costs relate to waste, effort, and negative emotion, and predict the bias when prompted. This work deepens our understanding of children’s cost-based reasoning and the developmental trajectory of the sunk cost bias. This work also contributes to theories of the bias and raises questions about the role of experience and theory of mind in the emergence of sunk cost predictions.