Alan Hajek

We introduce a St. Petersburg-like game, which we call the ‘Pasadena game’, in which we toss a coin until it lands heads for the first time. Your pay-offs grow without bound, and alternate in sign (rewards alternate with penalties). The expectation of the game is a conditionally convergent series. As such, its terms can be rearranged to yield any sum whatsoever, including positive infinity and negative infinity. Thus, we can apparently make the game seem as desirable or undesirable as we want, simply by reordering the pay-off table, yet the game remains unchanged throughout. Formally speaking, the expectation does not exist; but we contend that this presents a serious problem for decision theory, since it goes silent when we want it to speak. We argue that the Pasadena game is more paradoxical than the St. Petersburg game in several respects. We give a brief review of the relevant mathematics of infinite series. We then consider and rebut a number of replies to our paradox: that there is a privileged ordering to the expectation series; that decision theory should be restricted to finite state spaces; and that it should be restricted to bounded utility functions. We conclude that the paradox remains live.

David Lewis claims that a simple sort of anti-Humeanism-that the rational agent desires something to the extent he believes it to be good-can be given a decision-theoretic formulation, which Lewis calls 'Desire as Belief' (DAB). Given the (widely held) assumption that Jeffrey conditionalising is a rationally permissible way to change one's mind in the face of new evidence, Lewis proves that DAB leads to absurdity. Thus, according to Lewis, the simple form of anti-Humeanism stands refuted. In this paper we investigate whether Lewis's case against DAB can be strengthened by examining how it fares under rival versions of decision theory, including other conceptions of rational ways to change one's mind. We argue that the anti-Humean may escape Lewis's argument either by adopting a version of causal decision theory, or by claiming that the refutation only applies to hyper-idealised rational agents, or by denying that the decision-theoretic framework has the expressive capacity to formulate anti-Humeanism.

This is the sequel to my “Fifteen Arguments Against Finite Frequentism” ( Erkenntnis 1997), the second half of a long paper that attacks the two main forms of frequentism about probability. Hypothetical frequentism asserts: The probability of an attribute A in a reference class B is p iff the limit of the relative frequency of A ’s among the B ’s would be p if there were an infinite sequence of B ’s. I offer fifteen arguments against this analysis. I consider various frequentist responses, which I argue ultimately fail. I end with a positive proposal of my own, ‘hyper-hypothetical frequentism’, which I argue avoids several of the problems with hypothetical frequentism. It identifies probability with relative frequency in a hyperfinite sequence of trials. However, I argue that this account also fails, and that the prospects for frequentism are dim.

Bayesianism is our leading theory of uncertainty. Epistemology is defined as the theory of knowledge. So “Bayesian Epistemology” may sound like an oxymoron. Bayesianism, after all, studies the properties and dynamics of degrees of belief, understood to be probabilities. Traditional epistemology, on the other hand, places the singularly non-probabilistic notion of knowledge at centre stage, and to the extent that it traffics in belief, that notion does not come in degrees. So how can there be a Bayesian epistemology?

A decade ago, Harris Nover and I introduced the Pasadena game, which we argued gives rise to a new paradox in decision theory even more troubling than the St Petersburg paradox. Gwiazda's and Smith's articles in this volume both offer revisionist solutions. I critically engage with both articles. They invite reflections on a number of deep issues in the foundations of decision theory, which I hope to bring out. These issues include: some ways in which orthodox decision theory might be supplemented; the role of simulations of such infinite games; the role of small probabilities, and of idealization, in decision theory; tolerance about practical norms; and alternative ways of understanding decision theory's job description.

Probabilities figure centrally in much of the literature on the semantics of conditionals. I find this surprising: it accords a special status to conditionals that other parts of language apparently do not share. I critically discuss two notable ‘probabilities first’ accounts of counterfactuals, due to Edgington and Leitgeb. According to Edgington, counterfactuals lack truth values but have probabilities. I argue that this combination gives rise to a number of problems. According to Leitgeb, counterfactuals have truth conditions-roughly, a counterfactual is true when the corresponding conditional chance is sufficiently high. I argue that problems arise from the disparity between truth and high chance, between approximate truth and high chance, and from counterfactuals for which the corresponding conditional chances are undefined. However, Edgington, Leitgeb and I can unite in opposition to Stalnaker and Lewis-style ‘similarity’ accounts of counterfactuals.

Bayesians have a seemingly attractive account of rational credal states in terms of coherence. An agent's set of credences are synchronically coherent just in case they conform to the probability calculus. Some Bayesians impose a further putative coherence constraint called regularity: roughly, if X is possible, then it is assigned positive probability. I look at two versions of regularity – logical and metaphysical – and I canvass various defences of it as a rationality norm. Combining regularity with synchronic coherence, we have a set of constraints known as strict coherence.I argue that strict coherence is untenable. In particular, I attack regularity as a rationality norm. First, I rebut each of the various defences of regularity. Then I argue directly against regularity: it conflicts with the Bayesian decision‐theoretic treatment of rational action. Thus, seemingly plausible theoretical and pragmatic norms turn out to be inconsistent.[Barbossa is about to kill Will,but Jack Sparrow shows up:]Barbossa: It's not possible!Jack: Not probable.To be uncertain is to be uncomfortable,but to be certain is to be ridiculous

Four important arguments for probabilism—the Dutch Book, representation theorem, calibration, and gradational accuracy arguments—have a strikingly similar structure. Each begins with a mathematical theorem, a conditional with an existentially quantified consequent, of the general form: if your credences are not probabilities, then there is a way in which your rationality is impugned. Each argument concludes that rationality requires your credences to be probabilities. I contend that each argument is invalid as formulated. In each case there is a mirror-image theorem and a corresponding argument of exactly equal strength that concludes that rationality requires your credences not to be probabilities. Some further consideration is needed to break this symmetry in favour of probabilism.

Frank Ramsey (1931) wrote: If two people are arguing 'if p will q?' and both are in doubt as to p, they are adding p hypothetically to their stock of knowledge and arguing on that basis about q. We can say that they are fixing their degrees of belief in q given p. Let us take the first sentence the way it is often taken, as proposing the following test for the acceptability of an indicative conditional: ‘If p then q’ is acceptable to a subject S iff, were S to accept p and consider q, S would accept q. Now consider an indicative conditional of the form (1) If p, then I believe p. Suppose that you accept p and consider ‘I believe p’. To accept p while rejecting ‘I believe p’ is tantamount to accepting the Moore-paradoxical sentence ‘p and I do not believe p’, and so is irrational. To accept p while suspending judgment about ‘I believe p’ is irrational for similar reasons. So rationality requires that if you accept p and consider ‘I believe p’, you accept ‘I believe p’.

The so-called ‘Adams’ Thesis’ is often understood as the claim that the assertibility of an indicative conditional equals the corresponding conditional probability—schematically: $${({\rm AT})}\qquad\qquad\quad As(A\rightarrow B)=P({B|A}),{\rm provided}\quad P(A)\neq 0.$$ The Thesis is taken by many to be a touchstone of any theorizing about indicative conditionals. Yet it is unclear exactly what the Thesis is . I suggest some precise statements of it. I then rebut a number of arguments that have been given in its favor. Finally, I offer a new argument against it. I appeal to an old triviality result of mine against ‘Stalnaker’s Thesis’ that the probability of a conditional equals the corresponding conditional probability. I showed that for all finite-ranged probability functions, there are strictly more distinct values of conditional probabilities than there are distinct values of probabilities of conditionals, so they cannot all be paired up as Stalnaker’s Thesis promises. Conditional probabilities are too fine-grained to coincide with probabilities of conditionals across the board. If the assertibilities of conditionals are to coincide with conditional probabilities across the board, then assertibilities must be finer-grained than probabilities. I contend that this is implausible—it is surely the other way round. I generalize this argument to other interpretations of ‘ As ’, including ‘acceptability’ and ‘assentability’. I find it hard to see how any such figure of merit for conditionals can systematically align with the corresponding conditional probabilities

I vindicate Hume’s argument against belief in miracle reports against a prevalent objection. Hume has us balance the probability of a miracle’s occurrence against the probability of its being falsely attested to, and argues that the latter must inevitably be the greater; thus, reason requires us to reject any miracle report. The "flaw" in this reasoning, according to Butler and many others, is that it proves too much--it counsels us to never believe historians, newspaper reports of lottery results, and so on; and this is clearly absurd. I show that this objection is misguided: far from providing counterexamples to Hume’s "balancing principle", as I call it, these cases actually confirm it, as some simple calculations of probabilities show

The Cable Guy is coming. You have to be home in order for him to install your new cable service, but to your chagrin he cannot tell you exactly when he will come. He will definitely come between 8.a.m. and 4 p.m. tomorrow, but you have no more information than that. I offer to keep you company while you wait. To make things more interesting, we decide now to bet on the Cable Guy’s arrival time. We subdivide the relevant part of the day into two 4-hour long intervals, ‘morning’: (8, 12], and ‘afternoon’: (12, 4). You nominate an interval on which you will bet. If he arrives during your interval, you win and I will pay you $10; otherwise, I win and you will pay me $10. Notice that we stipulate that if he arrives exactly on the stroke of noon, then (8, 12] is the winning interval, since it is closed on the right; but we agree that this event has probability 0 (we have a very precise clock!). At first you think: obviously there is no reason to favour one interval over the other. Your probability distribution of his arrival time is uniform over the 8 a.m. – 4 p.m. period, and thus assigns probability 1/2 to each of the two 4-hour periods at issue. Whichever period you nominate, then, your expected utility is the same. The two choices are equally rational. But then you reason as follows. Suppose that you choose the morning interval. Then there will certainly be a period during which you will regard the other interval as..

“Pascal's Wager” is the name given to an argument due to Blaise Pascal for believing, or for at least taking steps to believe, in God. The name is somewhat misleading, for in a single paragraph of his Pensées, Pascal apparently presents at least three such arguments, each of which might be called a ‘wager’ — it is only the final of these that is traditionally referred to as “Pascal's Wager”. We find in it the extraordinary confluence of several important strands of thought: the justification of theism; probability theory and decision theory, used here for almost the first time in history; pragmatism; voluntarism (the thesis that belief is a matter of the will); and the use of the concept of infinity

Counterfactuals are a species of conditionals. They are propositions or sentences, expressed by or equivalent to subjunctive conditionals of the form 'if it were the case that A, then it would be the case that B', or 'if it had been the case that A, then it would have been the case that B'; A is called the antecedent, and B the consequent. Counterfactual reasoning typically involves the entertaining of hypothetical states of affairs: the antecedent is believed or presumed to be false, or contrary-to-the-fact, but its truth is imagined or supposed. Counterfactual reasoning is thus a form of modal reasoning, kindred to reasoning about necessity or possibility, and in contrast to reasoning about the way things actually are. The philosophical study of conditionals goes back at least as far as the Stoics of ancient Greece, although their systems of logic apparently did not accord the counterfactual any emphasis. The rise in interest in counterfactuals has been a rather recent phenomenon, as it started to become clear to philosophers that counterfactuals are implicated in a host of other important concepts—laws of nature, confirmation, causation, scientific explanation, knowledge, perception, dispositions, free action, etc. The significance of counterfactuals has also become increasingly appreciated in the..

This paper is a response to Paul Bartha’s ‘Making Do Without Expectations’. We provide an assessment of the strengths and limitations of two notable extensions of standard decision theory: relative expectation theory and Paul Bartha’s relative utility theory. These extensions are designed to provide intuitive answers to some well-known problems in decision theory involving gaps in expectations. We argue that both RET and RUT go some way towards providing solutions to the problems in question but neither extension solves all the relevant problems.

The thesis that probabilities of conditionals are conditional probabilities has putatively been refuted many times by so-called ‘triviality results’, although it has also enjoyed a number of resurrections. In this paper I assault it yet again with a new such result. I begin by motivating the thesis and discussing some of the philosophical ramifications of its fluctuating fortunes. I will canvas various reasons, old and new, why the thesis seems plausible, and why we should care about its fate. I will look at some objections to Lewis’s famous triviality results, and thus some reasons for the pursuit of further triviality results. I will generalize Lewis’s results in ways that meet the objections. I will conclude with some reflections on the demise of the thesis—or otherwise

This paper revisits the Pasadena game (Nover and Háyek 2004), a St Petersburg-like game whose expectation is undefined. We discuss serveral respects in which the Pasadena game is even more troublesome for decision theory than the St Petersburg game. Colyvan (2006) argues that the decision problem of whether or not to play the Pasadena game is ‘ill-posed’. He goes on to advocate a ‘pluralism’ regarding decision rules, which embraces dominance reasoning as well as maximizing expected utility. We rebut Colyvan’s argument, offering several considerations in favour of the Pasadena decision problem being well posed. To be sure, current decision theory, which is underpinned by various preference axioms, leaves indeterminate how one should value the Pasadena game. But we suggest that determinacy might be achieved by adding further preference axioms. We conclude by opening the door to a far greater plurality of decision rules. We suggest how the goal of unifying these rules might guide future research.

I came to philosophy as a refugee from mathematics and statistics. I was impressed by their power at codifying and precisifying antecedently understood but rather nebulous concepts, and at clarifying and exploring their interrelations. I enjoyed learning many of the great theorems of probability theory—equations rich in ‘P’s of this and of that. But I wondered what is this ‘P’? What do statements of probability mean? When I asked one of my professors, he looked at me like I needed medication. That medication was provided by philosophy, and I found it first during my Masters at the University of Western Ontario, working with Bill Harper, and then during my Ph.D. at Princeton, working with Bas van Fraassen, David Lewis, and Richard Jeffrey—all deft practitioners of formal methods. I found that philosophers had been asking my question about ‘P’ since about 1650, but they were still struggling to find definitive answers. I was also introduced to a host of other philosophical problems, and it became clear to me within nanoseconds of arriving at U.W.O. that I wanted to spend my life pursuing some of them. But I kept being drawn back to the formal methods of mathematics, and in particular of probability theory. It may be worthwhile to pause for a moment and to ask “What are formal methods?” Of course, it’s easy to come up with examples: the use of various logical systems, computational algorithms, causal graphs, information theory, probability theory and mathematics more generally. What do they have in common? They are all abstract representational systems. Sometimes the systems are studied in their own..

Much is asked of the concept of chance. It has been thought to play various roles, some in tension with or even incompatible with others. Chance has been characterized negatively, as the absence of causation; yet also positively—the ancient Greek τυχη´ reifies it—as a cause of events that are not governed by laws of nature, or as a feature of the laws themselves. Chance events have been understood epistemically as those whose causes are unknown; yet also objectively as a distinct ontological kind, sometimes called ‘pure’ chance events. Chance gives rise to individual unpredictability and disorder; yet it yields collective predictability and order—stable long-run statistics, and in the limit, aggregate behavior susceptible to precise mathematical theorems. Some authors believe that to posit chances is to abjure explanation; yet others think that chances are themselves explanatory. During the Enlightenment, talk of ‘chance’ was regarded as unscientific, unphilosophical, the stuff of superstition or ignorance; yet today it is often taken to be a fundamental notion of our most successful scientific theory, quantum mechanics, and a central concept of contemporary metaphysics. Chance has both negative and positive associations in daily life. The old word in English for it, hazard, which derives from French and originally from Arabic, still has unwelcome connotations of risk; ‘chance’ evokes uncertainty, uncontrollability, and chaos. Yet chance is also allied with luck, fortune, freedom from constraint, and diversity. And it apparently has various practical uses and benefits. It forms the basis of randomized trials in statistics, and of mixed strategies in decision theory and game theory; it is appealed to in order to resolve problems of fair division and other ethical..