Risk in Decision Theory

A decision theory is self-recommending if, when you ask it which decision theory you should use, it considers itself to be among the permissible options. I show that many alternatives to expected utility theory are not self-recommending, and I argue that this tells against them.

Orthodox causal decision theory is unstable. Its advice changes as you make up your mind about what you will do. Several have objected to this kind of instability and explored stable alternatives. Here, I'll show that explorers in search of stability must part with a vestige of their homeland. There is no plausible stable decision theory which satisfies Savage's Sure Thing Principle. So those in search of stability must learn to live without it.

A novel theory of comparative risk is developed and defended. Extant theories are criticized for failing the tests of extensional and formal adequacy. A unified diagnosis is proposed: extant theories consider risk to be a univariable function, but risk is a multivariate function. According to the theory proposed, which we call the unified theory of risk, the riskiness of a proposition is a function of both the proportion and the modal closeness of the possible worlds at which the proposition holds. (...) It is argued the theory passes the tests of extensional and formal adequacy. (shrink)

There are infinitely many ways the world might be, and there may well be infinitely many people in it. These facts raise moral paradoxes. We explore a conflict between two highly attractive principles: a Pareto principle that says that what is better for everyone is better overall, and a statewise dominance principle that says that what is sure to turn out better is better on balance. We refine and generalize this paradox, showing that the problem is faced by many theories (...) of interpersonal aggregation besides utilitarianism, and by many decision theories besides expected value theory. Considering the range of consistent responses, we find all of them to be quite radical. (shrink)

I present a new puzzle about choice under uncertainty for agents whose preferences are sensitive to multiple dimensions of outcomes in such a way as to be incomplete. In response, I develop a new theory of choice under uncertainty for incomplete preferences. I connect the puzzle to central questions in epistemology about the nature of rational requirements, and ask whether it shows that preferences are rationally required to be complete.

A moral requirement R1 is said to be lexically prior to a moral requirement R2 just in case we are morally obliged to uphold R1 at the expense of R2 – no matter how many times R2 must be violated thereby. While lexical priority is a feature of many ethical theories, and arguably a part of common sense morality, attempts to model it within the framework of decision theory have led to a series of problems – a fact which is (...) sometimes spun as a ‘decision theoretic critique’ of lexical priority. In this paper, I develop an enriched decision theoretic framework that is capable of modelling lexical priority while avoiding all extant problems. This will involve introducing several new ingredients into the standard decision theoretic framework, including multidimensional utilities, de minimis risks, and the means to represent two different conceptions of risk. (shrink)

In a recent paper, Björn Lundgren and H. Orri Stefánsson (forthcoming) present three objections to normic de minimis expected utility theory (NDEUT) – a decision theoretic framework defended in Smith (2024). In this paper, I respond to these objections and outline some possible ways in which NDEUT might be modified or further developed. Like any de minimis framework, NDEUT employs a risk threshold to sort possibilities into those that must be considered when making a decision, and those that can be (...) legitimately ignored – the so-called ‘de minimis risks’. While this threshold would usually take the form of a probability value, in NDEUT the de minimis risks are identified instead on the basis of their abnormality. In the first section I will set up some of the formalism used in NDEUT and consider Lundgren and Stefánsson’s first objection – which is essentially formal in nature. In the second section I will delve further into the interpretation of NDEUT and consider the second and third objections, which turn on particular examples. (shrink)

A social planner who evaluates risky public policies in light of the other risks with which their society will be faced should judge favourably some such policies even though they would deem them too risky when considered in isolation. I suggest that a longtermist would—or at least should—evaluate risky polices in light of their prediction about future risks; hence, longtermism supports social risk-taking. I consider two formal versions of this argument, discuss the conditions needed for the argument to be valid, (...) and briefly compare these conditions to some risky policy options with which actual public decision-makers are faced. (shrink)

Should you be willing to forego any sure good for a tiny probability of a vastly greater good? Fanatics say you should, anti-fanatics say you should not. Anti-fanaticism has great intuitive appeal. But, I argue, these intuitions are untenable, because satisfying them in their full generality is incompatible with three very plausible principles: acyclicity, a minimal dominance principle, and the principle that any outcome can be made better or worse. This argument against anti-fanaticism can be turned into a positive argument (...) for a weak version of fanaticism, but only from significantly more contentious premises. In combination, these facts suggest that those who find fanaticism counterintuitive should favor not anti-fanaticism, but an intermediate position that permits agents to have incomplete preferences that are neither fanatical nor anti-fanatical. (shrink)

Various decision theories share a troubling implication. They imply that, for any finite amount of value, it would be better to wager it all for a vanishingly small probability of some greater value. Counterintuitive as it might be, this _fanaticism_ has seemingly compelling independent arguments in its favour. In this paper, I consider perhaps the most _prima facie_ compelling such argument: an _Egyptology argument_ (an analogue of the Egyptology argument from population ethics). I show that, despite recent objections from Russell (...) (Noûs, 2023) and Goodsell (Analysis 81(3):420–426, 2021), the argument’s premises can be justified and defended, and the argument itself remains compelling. (shrink)

Recent claims that artificial intelligence (AI) systems demonstrate “emergent abilities” have fueled excitement but also fear grounded in the prospect that such systems may enable a wider range of parties to make unprecedented advances in areas that include the development of chemical or biological weapons. Ambiguity surrounding the term “emergent abilities” has added avoidable uncertainty to a topic that has the potential to destabilize the strategic landscape, including the perception of key parties about the viability of nonproliferation efforts. To avert (...) these problems in the future, scientists, developers, policymakers, and other parties should take credible steps to strengthen the health of the scientific ecosystem around AI. (shrink)

A de minimis risk is defined as a risk that is so small that it may be legitimately ignored when making a decision. While ignoring small risks is common in our day-to-day decision making, attempts to introduce the notion of a de minimis risk into the framework of decision theory have run up against a series of well-known difficulties. In this paper, I will develop an enriched decision theoretic framework that is capable of overcoming two major obstacles to the modelling (...) of de minimis risk. The key move is to introduce, into decision theory, a non-probabilistic conception of risk known as normic risk. (shrink)

Bartha and DesRoches (Synthese 199(3–4):8701–8740, 2021) and Steel and Bartha (Risk Analysis 43(2):260–268, 2023) argue that we should understand the precautionary principle as the injunction to maximise lexical utilities. They show that the lexical utility model has important pragmatic advantages. Moreover, the model has the theoretical advantage of satisfying all axioms of expected utility theory except continuity. In this paper I raise a trilemma for any attempt at modelling the precautionary principle with lexical utilities: it permits choice cycles or leads (...) to paralysis or implies that the smallest value difference that is possible in a context has extreme axiological implications. (shrink)

Suppose an agent is choosing between rescuing more people with a lower probability of success, and rescuing fewer with a higher probability of success. How should they choose? Our moral judgments about such cases are not well-studied, unlike the closely analogous non-moral preferences over monetary gambles. In this paper, I present an empirical study which aims to elicit the moral analogues of our risk preferences, and to assess whether one kind of evidence—concerning how they depend on outcome probabilities—can debunk them. (...) I find significant heterogeneity in our moral risk preferences—in particular, moral risk-seeking and risk-neutrality are surprisingly popular. I also find that subjects’ judgments aren't probability-dependent, thus providing an empirical defense against debunking arguments from probability dependence. (shrink)

According to a common judgement, a social planner should often use a lottery to decide which of two people should receive a good. This judgement undermines one of the best-known arguments for utilitarianism, due to John C. Harsanyi, and more generally undermines axiomatic arguments for utilitarianism and similar views. In this paper we ask which combinations of views about (a) the social planner’s attitude to risk and inequality, and (b) the subjects’ attitudes to risk are consistent with the aforementioned judgement. (...) We find that the class of combinations of views that can plausibly accommodate this judgement is quite limited. But one theory does better than others: the theory of chance-sensitive utility. (shrink)

When you are indifferent between two options, it’s rationally permissible to take either. One way to decide between two such options is to flip a fair coin, taking one option if it lands heads and the other if it lands tails. Is it rationally permissible to employ such a tie-breaking procedure? Intuitively, yes. However, if you are genuinely risk-averse—in particular, if you adhere to risk-weighted expected utility theory (Buchak in Risk and rationality, Oxford University Press, 2013) and have a strictly (...) convex risk-function—the answer will often be no: the REU of deciding by coin-flip will be lower than the REU of choosing one of the options outright (so long as at least one of the options is a nondegenerate gamble). This turns out to be a significant worry for risk-weighted expected utility theory. I argue that it adds real bite to established worries about diachronic consistency afflicting views, like risk-weighted expected utility theory, that violate Independence. And that, while these worries might be surmountable, surmounting them comes at a price. (shrink)

All ordinary decisions involve some risk. If I go outside for a walk, I may trip and injure myself. But if I don’t go for a walk, I slightly increase my chances of cardiovascular disease. Typically, we disregard most small risks. When, for practical purposes, is it appropriate for one to ignore risk? This issue looms large because many activities performed by those in wealthy societies, such as driving a car, in some way risk contributing to climate harms. Are these (...) activities morally appropriate? -/- In this paper, I first summarize and respond to some arguments that purport to show that it is appropriate to ignore or discount very small risks. I argue that because our rationality is bounded, it is impossible for us to include every small risk in our decision-making process, and so we may reasonably use heuristics to guide many decisions. However, contrary to some thinkers, I argue that this does not violate the spirit of expected value theory; it merely shows that we should adopt a so-called "two-level" view. Our use of heuristics allows for the reasonable ignoring of some risks, and this perhaps explains why one might be inclined to think that individual climate-related risks are negligible. However, virtually all greenhouse-gas emitting activities in fact have some climate risk on the negative side of the ledger, and the use of heuristics does not permit the general ignoring of climate-change-related risk by individuals on grounds of expediency of judgment and decision-making. (shrink)

This paper aims to scrutinize how the notion of risk should be understood and applied to possibly catastrophic cases. I begin with clarifying the standard usage of the notion of risk, particularly emphasizing the conceptual relation between risk and probability. Then, I investigate how to make decisions in the case of seemingly catastrophic disasters by contrasting the precautionary principle with the preventive (prevention) principle. Finally, I examine what kind of causal thinking tends to be actually adopted when we make decisions (...) on emergent cases. My arguments are mainly based on Japan's 2011 massive earthquake and nuclear power station accident. Masaki Ichinose. (shrink)

How should you evaluate your choices when you’re unsure what their outcomes will be? One popular answer is to rank your options in terms of their expected utilities. But what should you do when you think that the value of their respective outcomes might be incommensurable? In the face of incommensurable values, it no longer makes sense to speak of ranking your options according to expected utility. Are there any general principles to guide us when facing decisions of this kind? (...) If only! This chapter develops an impossibility result: it holds that there are a handful of independently plausible constraints that no such decision theory can jointly satisfy. The result, while depressing, can be used to helpfully classify extant approaches based on which of the constraints they violate. (shrink)