Epistemic Utility

I investigate two limitations of common approaches to the alethic aspect of epistemic rationality. These approaches are either normative (what a reasoner ought to/may
believe?) or evaluative (how rational is a reasoner?), where evaluative approaches are often comparative (one reasoner is evaluated compared to another). The fact is that most of them fail in describing a standard of rationality satisfiable (i) on the face of blindspots and (ii) by finite reasoners (reasoners with limited cognitive resources). I propose a non-comparative evaluative function α, which accepts a set of beliefs as input and returns a numerical evaluation, and use this function to define a standard of rationality satisfiable on the face of blindspots and by finite reasoners. Function α also generates norms and evaluation methods (comparative and non-comparative) that can be used to investigate the rationality of finite reasoners and do not exhibit problems with blindspots. It is a result of this investigation that the project of providing alethic norms for finite reasoners is defective. In addition, the use of function α sheds light on important epistemological issues, such as the paradoxes of lottery and preface and the principles of clutter avoidance and reflection.

We use a theorem from Schervish (1989) to explain the relationship between accuracy and practical success. If an agent is pragmatically rational, she'll quantify the expected loss of her credence with a strictly proper scoring rule. Which scoring rule is right for her will depend on the sorts of decisions she expects to face. We relate this pragmatic conception of inaccuracy to the purely epistemic one popular among epistemic utility theorists.

According to epistemic utility theory, epistemic rationality is teleological: epistemic norms are instrumental norms that have the aim of acquiring accuracy. What’s definitive of these norms is that they can be expected to lead to the acquisition of accuracy when followed. While there’s much to be said in favor of this approach, it turns out that it faces a couple of worrisome extensional problems involving the future. The first problem involves credences about the future, and the second problem involves future credences. Examining prominent solutions to a different extensional problem for this approach reinforces the severity of the two problems involving the future. Reflecting on these problems reveals the source: the teleological assumption that epistemic rationality aims at acquiring accuracy.

This is a paper for a symposium on Richard Pettigrew's "Accuracy and the Laws of Credence." Pettigrew offers new axiomatic constraints on legitimate measures of inaccuracy. His most interesting axiom, which he calls Decomposition, stipulates that legitimate measures of inaccuracy evaluate a credence in part based on its level of calibration at a world. I argue that if calibration is valuable, as Pettigrew claims, then this fact is an explanandum for accuracy-first epistemologists, not an explanans for three reasons. First, the intuitive case for the importance of calibration
isn’t as strong as Pettigrew believes. Second, calibration is a perniciously global property that both contravenes Pettigrew’s own views about the nature of credence functions themselves and undercuts the achievements and ambitions of accuracy-first epistemology. Finally, Decomposition introduces a new kind of value compatible with but separate from accuracy-proper in violation of Pettigrew’s alethic monism.

Bayesian epistemologists support the norms of probabilism and conditionalization using Dutch book and accuracy arguments. These arguments assume that rationality requires agents to maximize practical or epistemic value in every doxastic state, which is evaluated from a subjective point of view (e.g., the agent’s expectancy of value). The accuracy arguments also presuppose that agents are opinionated. The goal of this paper is to discuss the assumptions of these arguments, including the measure of epistemic value. I have designed AI agents based on the Bayesian model and a nonmonotonic framework and tested how they achieve practical and epistemic value in conditions in which an alternative set of assumptions holds. In one of the tested conditions, the nonmonotonic agent, which is not opinionated and fulfills neither probabilism nor conditionalization, outperforms the Bayesian in the measure of epistemic value that I argue for in the paper (α-value). I discuss the consequences of these results for the epistemology of rationality.

According to epistemic utility theory, epistemic rationality is teleological: epistemic norms are instrumental norms that have the aim of acquiring accuracy. What’s definitive of these norms is that they can be expected to lead to the acquisition of accuracy when followed. While there’s much to be said in favor of this approach, it turns out that it faces a couple of worrisome extensional problems involving the future. The first problem involves credences about the future, and the second problem involves future credences. Examining prominent solutions to a different extensional problem for this approach reinforces the severity of the two problems involving the future. Reflecting on these problems reveals the source: the teleological assumption that epistemic rationality aims at acquiring accuracy.

Umas das principais dificuldades associadas à interpretação Everettiana da mecânica quântica é a de fazer sentido da noção de probabilidade. Não está claro como poderia haver probabilidades num contexto em que não haveria incerteza ou alternativas não realizadas. Nas últimas décadas, foram propostas soluções para o problema. Numa dessas soluções, a regra de Born é derivada a partir de postulados da teoria da decisão. Argumento que a introdução de questões sobre racionalidade prática na discussão sobre existência de probabilidades na interpretação Everettiana da mecânica quântica está fora de lugar. Proponho que a derivação da regra de Born deveria partir de postulados da teoria da utilidade epistêmica (racionalidade teórica). Também proponho a exploração de outras relações entre racionalidade teórica e o formalismo probabilístico da mecânica quântica.