Decision Theory

Research on the credit economy in academic science has typically assumed that academics are expected credit maximizers without argument. How might this assumption be justified? And how might the measure-theoretic foundations of credit be secured? Two approaches are considered. One in which credit is equated with citations and one, based on von Neumann's expected utility theorem, in which credit is constructed from academics' preferences over lotteries among research records. The latter is shown to be the weakest possible defense of expected credit maximization in a formally precise sense, so those assuming expected credit maximization are committed to it.

Despite efforts to teach agents about their privilege by minimizing cost of information, Kinney & Bright argue risksensitive frameworks like Buchak's allow privileged agents to rationally shield themselves from this costless and relevant information. In response, I show that uncertainty about information's relevance may block one from rationally upholding ignorance. I explore the implications and interpretations of the agent's uncertainty; these educational initiatives may not be as doomed as suggested, and agents may feel better having learned something but rationally decline to learn it now. This has upshots for the viability of risksensitive expected utility theory in explaining elitegroup ignorance.

In a recent paper Winsberg (2021) argues in favor of research into geoengineering by relying on Good's theorem, which states that conducting research maximizes one's expected utility. However, Good's theorem sometimes fails for risk-avoidant agents (Buchak 2010). Since risk-avoidance captures some of the 'precautionary' intuitions that critics of geoengineering share, it is important to see if research into geoengineering would maximize one's utility if risk-avoidance is taken into account. I show that there are further considerations to be taken into account if one wants to conclude that conducting research into geoengineering maximizes utility based on Good's results.

If you can’t or don’t want to ascribe probabilities to the consequences of your actions, classic causal decision theory won’t let you reap the undeniable benefits of causal reasoning for decision making. The following theory fixes this problem. I explain why it’s good to have a causal decision theory that applies to non-deterministic yet non-probabilistic decision problems. I then introduce the underdeterministic framework and subsequently use it to formulate underdeterministic decision theory. The theory applies to decisions with infinitely many possible consequences and to agents who can’t decide on a single causal model representing the decision problem.