Rational for What?

What I Would Ask David Lewis About Counterfactuals If I Could

There are many philosophical questions that I wish I could ask David Lewis. Several of them concern counterfactuals. They include the following: You came very close to the view that most counterfactuals are false at points in your writings. Why didn't you take the extra step and endorse the view? You were famously a contextualist about counterfactuals, yet that seems in tension with various things that you say. How much of a contextualist were you? You were not a contextualist at all about indicative conditionals. Why did you give such different analyses of counterfactuals and indicative conditionals? I also have some questions about the logic of counterfactuals, concerning their antecedents. What do you make of objections to your treatment of: antecedent strengthening; simplification of disjunctive antecedents; counterfactuals with true antecedents; counterfactuals with impossible antecedents? I did get to ask David why he liked Australia so much, and I will share his answer.

The Price Is Wrong: The Two-Boxer's Dismal Prospects

We'll look at a fascinating problem for two boxers, and why recent attempts to solve it (including those belonging to at least two conference attendees) are unsuccessful.

Truthmakers and Mental Models

I explore the possibility that recent work in 'truthmaker semantics' by Kit Fine, Stephen Yablo, and other philosophers, logicians, and linguists, together with the decades-long program of 'mental model theory' carried out by Philip Johnson-Laird, Ruth Byrne, and their collaborators, are converging on a shared system of semantic representation---one based on exact verifiers, or mental models of such verifiers, standing in mereological part-whole relations to one another---that both captures the semantic competence of ideal speakers and meshes with the typical performance of real cognitively bounded subjects in natural language deduction tasks. At the interface of these domains, I propose a new psycholinguistic theory of comprehension and deduction, according to which we possess a basic, purely mereological and hyperintensional competence to track 'containment' or 'parthood' relations between premises and conclusions, while it is only through further, cognitively costly modal reasoning that we may achieve the ideal of classical validity.

Externalist Decision Theory for Evidentialists

Many epistemologists accept evidence externalism: the view that the content of your total evidence might not entail that it is your total evidence. This talk explores the consequences of evidence externalism for decision theory. Traditional decision theory is built on internalist assumptions. I argue that a form of evidence externalism---what I call "evidentialist externalism"---can be imported into decision theory while preserving many of the subjective, evidence-sensitive, and predictive roles that have motivated internalism in decision theory. This is so even though evidentialist externalism has some nontraditional consequences: that rational agents aren't always in a position to know what their decision-theoretic options are, or which options are rational, or even what option they are taking.

Do the Chances Update by Conditionalization?

This talk focuses on Caie (2015)'s temporal chance deference norm—roughly, the idea that vindicated credence functions defer to the current actual chances. I show how a non-conditionalizing update rule, based on Goldstein (2020)'s conditionalization-with-normalization, can maximize expected accuracy over a language capable of expressing this norm. But if the chance itself does not update by conditionalization, what hope is there for us?

Epistemic Actualism

According to veritism, your epistemic goal is to believe what's true. According to actualist veritism, your epistemic goal is to believe what's actually true. At first glance, it seems that the pursuit of truth and the pursuit of actual truth will lead to the same destination. This will follow for evidential decision theorists; but if we are causal decision theorists (as I am), then pursuing truth and pursuing actual truth will lead us to different destinations. In this paper, I assume causal decision theory and argue for actualist veritism over veritism. My argument is that veritism does not respect the direction of fit of belief, since it incentivizes changing the world to match your beliefs, rather than changing your beliefs to match the world as it is. In contrast, actualist veritism honors the direction of fit of belief, and only incentivizes changing your beliefs to match the world as it is.

Against Proxy Optimization

I discuss conditions under which maximizing a proxy utility function is harmful and suggest that this poses a problem for applying decision theory.

Deference Done with Certainty

Many experts are modest—they are uncertain whether their opinions are deference-worthy. We can't defer to such experts by adopting their probabilities as our own, on pain of inconsistency. So how should we defer to them? This paper proposes a new deference principle for modest experts, which I call Total Certainty. Roughly, it says: if you are certain that the expert prefers option X to option Y given your total evidence, then you should prefer X to Y. I give a value-theoretic characterization of this principle and compare it with other deference principles that have been proposed in the literature.