Papers

Only Causal Decision Theory Defers to Experts
Journal of Philosophy, forthcoming
Synopsis: I show that causal decision theory is the only suppositional decision theory that says it's always good to defer to a better informed expert (ike your future self or the objective chances) when you're making a decision.
Desire-as-Belief in Context
Mind, forthcoming
Synopsis: I give a context-sensitive version of the Desire-as-Belief Thesis and prove its tenability.
A Behavioural and Representational Evaluation of Goal-Directedness in Language Model Agents
Co-authored with Raghu Arghal, Fade Chen, Niall Dalton, Evgenii Kortukov, Angelos Nalmpantis, Moksh Nirvaan, Gabriele Sarti, and Mario Giulianelli
Proceedings of the 43rd International Conference on Machine Learning (ICML), 2026
Synopsis: We develop a novel framework for interpreting what goals an AI system is pursuing.
Dutch Books, Indicative Conditionals, and Rational Updating
Co-authored with Snow Zhang
Proceedings of the 24th Annual Amsterdam Colloquium, 2025
Synopsis: We discuss how you should update your credence in an indicative conditional, when you learn a piece of factual information.
Causal Decision Theory, Context, and Determinism
Philosophy and Phenomenological Research, 2024
Synopsis: I respond to "deterministic" counterexamples to causal decision theory (CDT), by drawing attention to the fact that counterfactuals are context-sensitive.
The Punctuated Equilibrium of Scientific Change: a Bayesian Network Model
Co-authored with Patrick Grim, Frank Seidl, Isabell N. Astor, and Caroline Diaso
Synthese, 2022
Synopsis: We use Bayesian networks to understand how evidence flows through scientific communities.
Scientific Theories as Bayesian Nets: Structure and Evidence Sensitivity
Co-authored with Patrick Grim, Frank Seidl, Hinton E. Rago, Isabell N. Astor, Caroline Diaso, and Peter Ryner
Philosophy of Science, 2022
Synopsis: We show how scientific theories can be modeled as Bayesian networks.

Work in Progress

A paper on decision theory and utilitarianism
R&R (email me for a draft)
Synopsis: I prove a version of Harsanyi's impartial observer theorem, in the suppositional decision theory setting.
A paper on indicative conditionals and Dutch books
Co-authored with Snow Zhang
under review (email me for a draft)
Synopsis: We discuss how you should update your credence in an indicative conditional, when you learn a piece of factual information. (This is the full version of our paper in the Amsterdam proceedings.)
A paper on the Off-switch Game
under review (email me for a draft)
Synopsis: I prove a generalization of Hadfield-Menell et al.'s (2017) Off-switch Game result about AI alignment.
Emergent Alignment and the Projectability of Ethical Personas
Co-authored with Guillermo Del Pinal, Youngchan Lee, and Alejandro Pérez Carballo
under review (arXiv preprint)
Synopsis: We give empirical results demonstrating "emergent alignment" in Large Language Models.
A paper on self-recommending decision theories
under review (email me for a draft)
Synopsis: I prove that no suppositional decision theory is self-recommending.
A paper on updating on conditionals
under review (email me for a draft)
Synopsis: I draw out a relationship between van Fraassen's Judy Benjamin Problem and Stalnaker's Thesis.
Another paper on utilitarianism
in progress (draft available soon!)
Synopsis: I discuss the classic objection that utilitarianism is "self-defeating". I prove that, on certain precisifcations of this term, utilitarianism is not self-defeating---although a number of other moral theories are.
A paper on accuracy and epistemic modality
Co-authored with Mikayla Kelley, Richard Roth, and Snow Zhang (a.k.a. the accu-vengers)
in progress (draft available soon!)
Synopsis: We investigate what happens to epistemic utility theory when we introduce epistemic modals and indicative conditionals.
A paper on goal-representations in Large Language Models
Co-authored with Raghu Arghal, Niall Dalton, Guillermo Del Pinal, and Mario Giulianelli
in progress (draft avialble soon!)
Synopsis: We use methods from mechanistic interpretability to locate desire-like representations in language models, show that these are distinct from belief-like representations, and demonstrate how the two combine to produce certain courses of action taken by a model.
A paper on counterfactuals
Co-authored with Paolo Santorio
in progress (draft available soon!)
Synopsis: We give a new account of counterfactual similarity between worlds, based on causal models.