Evidentialists say that a necessary condition of sound epistemic reasoning is that our beliefs reflect only our evidence. This thesis arguably conflicts with standard Bayesianism, due to the importance of prior probabilities in the latter. Some evidentialists have responded by modelling belief-states using imprecise probabilities (Joyce 2005). However, Roger White (2010) and Aron Vallinder (2018) argue that this Imprecise Bayesianism is incompatible with evidentialism due to “inertia”, where Imprecise Bayesian agents become stuck in a state of ambivalence towards hypotheses. Additionally, (...) escapes from inertia apparently only create further conflicts with evidentialism. This dilemma gives a reason for evidentialist imprecise probabilists to look for alternatives without inertia. I shall argue that Henry E. Kyburg’s approach offers an evidentialist-friendly imprecise probability theory without inertia, and that its relevant anti-inertia features are independently justified. I also connect the traditional epistemological debates concerning the “ethics of belief” more systematically with formal epistemology than has been hitherto done. (shrink)

Supra-Bayesianism is the Bayesian response to learning the opinions of others. Probability pooling constitutes an alternative response. One natural question is whether there are cases where probability pooling gives the supra-Bayesian result. This has been called the problem of Bayes-compatibility for pooling functions. It is known that in a common prior setting, under standard assumptions, linear pooling cannot be nontrivially Bayes-compatible. We show by contrast that geometric pooling can be nontrivially Bayes-compatible. Indeed, we show that, under certain assumptions, geometric and (...) Bayes-compatible pooling are equivalent. Granting supra-Bayesianism its usual normative status, one upshot of our study is thus that, in a certain class of epistemic contexts, geometric pooling enjoys a normative advantage over linear pooling as a social learning mechanism. We discuss the philosophical ramifications of this advantage, which we show to be robust to variations in our statement of the Bayes-compatibility problem. (shrink)

Epistemic injustice is injustice to a person qua knower. In one form of this phenomenon a speaker’s testimony is denied credence in a way that wrongs them. I argue that the received definition of this testimonial injustice relies too heavily on epistemic criteria that cannot explain why the moral concept of injustice should be invoked. I give an account of the nature of the wrong of epistemic injustice that has it depend not on the accuracy of judgments that are used (...) or made in the process of deciding whether to listen to or trust a speaker, but on whether the basis of the decision about a speaker is their reliability or their identity, and the account explains why the latter is a moral wrong. A key difference between the two accounts is how they classify the use of true statistical generalizations connecting identity and reliability. The received view implies that this cannot be an injustice, while the view proposed here implies that it can. As such the new view appears to imply a conflict between moral and epistemic obligations: it is morally wrong to use true statistical generalizations in certain contexts, yet they are part of our evidence, and we are epistemically obligated to take all of our evidence into account. I reconcile these two thoughts without adopting the currently popular view that a belief’s being morally wrong makes it epistemically unjustified, and I argue that following the principle of total evidence encourages epistemic justice rather than thwarting it. (shrink)

Many who think that naked statistical evidence alone is inadequate for a trial verdict think that use of probability is the problem, and something other than probability – knowledge, full belief, causal relations – is the solution. I argue that the issue of whether naked statistical evidence is weak can be formulated within the probabilistic idiom, as the question whether likelihoods or only posterior probabilities should be taken into account in our judgment of a case. This question also identifies a (...) major difference between the Process Reliabilist and Probabilistic Tracking views of knowledge and other concepts. Though both are externalist, and probabilistic, epistemic theories, Tracking does and Process Reliabilism does not put conditions on likelihoods. So Tracking implies that a naked statistic is not adequate evidence about an individual, and does not yield knowledge, whereas the Reliabilist thinks it gives justified belief and knowledge. Not only does the Tracking view imply that naked statistical evidence is insufficient for a verdict, but it gives us resources to explain why, in terms of risk and the special conditions of a trial. (shrink)

This chapter presents a typology of the different kinds of inductive inferences we can draw from our evidence, based on the explanatory relationship between evidence and conclusion. Drawing on the literature on graphical models of explanation, I divide inductive inferences into (a) downwards inferences, which proceed from cause to effect, (b) upwards inferences, which proceed from effect to cause, and (c) sideways inferences, which proceed first from effect to cause and then from that cause to an additional effect. I further (...) distinguish between direct and indirect forms of downwards and upwards inferences. I then show how we can subsume canonical forms of inductive inference mentioned in the literature, such as inference to the best explanation, enumerative induction, and analogical inference, under this typology. Along the way, I explore connections with probability and confirmation, epistemic defeat, the relation between abduction and enumerative induction, the compatibility of IBE and Bayesianism, and theories of epistemic justification. (shrink)

The principle of indifference has fallen from grace in contemporary philosophy, yet some papers have recently sought to vindicate its plausibility. This paper follows suit. In it, I articulate a version of the principle and provide what appears to be a novel argument in favour of it. The argument relies on a thought experiment where, intuitively, an agent’s confidence in any particular outcome being true should decrease with the addition of outcomes to the relevant space of possible outcomes. Put simply: (...) the greater the number of outcomes, the weaker your confidence should be in any one of those outcomes. The argument holds that this intuition favours the principle of indifference. I point out that, in contrast, the intuition is also incompatible with a major alternative to the principle which prescribes imprecise credences: the so-called wide interval view. Consequently, the argument may also be seen as an argument against the wide interval view. (shrink)

In the age of big data and a machine epistemology that can anticipate, predict, and intervene on events in our lives, the problem once again is that a few individuals possess the knowledge of how to regulate these activities. But the question we face now is not how to share such knowledge more widely, but rather of how to enjoy the public benefits bestowed by this knowledge without freely sharing it. It is not merely personal privacy that is at stake (...) but a range of unsung benefits that come from ignorance and forgetting, traits that are inherently human and integral to the functioning of our society. (shrink)

Why are conditional degrees of belief in an observation E, given a statistical hypothesis H, aligned with the objective probabilities expressed by H? After showing that standard replies are not satisfactory, I develop a suppositional analysis of conditional degree of belief, transferring Ramsey’s classical proposal to statistical inference. The analysis saves the alignment, explains the role of chance-credence coordination, and rebuts the charge of arbitrary assessment of evidence in Bayesian inference. Finally, I explore the implications of this analysis for Bayesian (...) reasoning with idealized models in science. (shrink)

Roger White objects to interval-valued credence theories because they produce a counterintuitive “dilation” effect in a story he calls the Coin Game. We respond that results in the Coin Game were bound to be counterintuitive anyway, because the story involves an agent who learns a biconditional. Biconditional updates produce surprising results whether the credences involved are ranged or precise, so White's story is no counterexample to ranged credence theories.

Consider someone whom you know to be an expert about some issue. She knows at least as much as you do and reasons impeccably. The issue is a straightforward case of statistical inference that raises no deep problems of epistemology. You happen to know the expert’s opinion on this issue. Should you defer to her by adopting her opinion as your own? An affirmative answer may appear mandatory. But this paper argues that a crucial factor in answering this question is (...) the description under which you identify the expert. Experts may be identified under various descriptions (e.g., “the expert appearing on Channel 7 at 5 pm”). Section 2 provides cases that illustrate how deference to a known expert may be appropriate under some descriptions, but not others. Section 3 proposes that, in these cases, the differentiating factor is self-identification: deference to an expert under a given description is appropriate when the expert self-identifies under that description. Section 4 presents a formal framework that demonstrates the sufficiency of self-identification for appropriate deference within a significant class of cases. Section 5 notes a way in which this phenomenon allows for a form of “agreeing to disagree” in Robert Aumann’s sense. Section 6 illustrates the practical application of these results to the case of testimonials. (shrink)

The article begins by describing two longstanding problems associated with direct inference. One problem concerns the role of uninformative frequency statements in inferring probabilities by direct inference. A second problem concerns the role of frequency statements with gerrymandered reference classes. I show that past approaches to the problem associated with uninformative frequency statements yield the wrong conclusions in some cases. I propose a modification of Kyburg’s approach to the problem that yields the right conclusions. Past theories of direct inference have (...) postponed treatment of the problem associated with gerrymandered reference classes by appealing to an unexplicated notion of projectability . I address the lacuna in past theories by introducing criteria for being a relevant statistic . The prescription that only relevant statistics play a role in direct inference corresponds to the sort of projectability constraints envisioned by past theories. (shrink)

Additionally, the text shows how to develop computationally feasible methods to mesh with this framework.

A general form is proposed for epistemological theories, the relevant factors being: the family of epistemic judgments, the epistemic state, the epistemic commitment, and the family of possible epistemic inputs. First a simple theory is examined in which the states are probability functions, and the subject of probability kinematics introduced by Richard Jeffrey is explored. Then a second theory is examined in which the state has as constituents a body of information and a recipe that determines the accepted epistemic judgments (...) on the basis of this corpus. Through an examination of several approaches to the statistical syllogism, a relation is again established with Jeffrey's generalized conditionalization. (shrink)

Examples growing out of the Newcomb problem have convinced many people that decision theory should proceed in terms of some kind of causal probability. I endorse this view and define and investigate a variety of causal probability. My definition is related to Skyrms' definition, but proceeds in terms of objective probabilities rather than subjective probabilities and avoids taking causal dependence as a primitive concept.

I argue that Bayesians need two distinct notions of probability. We need the usual degree-of-belief notion that is central to the Bayesian account of rational decision. But Bayesians also need a separate notion of probability that represents the degree to which evidence supports hypotheses. Although degree-of-belief is well suited to the theory of rational decision, Bayesians have tried to apply it to the realm of hypothesis confirmation as well. This double duty leads to the problem of old evidence, a problem (...) that, we will see, is much more extensive than usually recognized. I will argue that degree-of-support is distinct from degree-of-belief, that it is not just a kind of counterfactual degree-of-belief, and that it supplements degree-of-belief in a way that resolves the problems of old evidence and provides a richer account of the logic of scientific inference and belief. (shrink)

The analyses of explanation and causal beliefs are heavily dependent on using probability functions as models of epistemic states. There are, however, several aspects of beliefs that are not captured by such a representation and which affect the outcome of the analyses. One dimension that has been neglected in this article is the temporal aspect of the beliefs. The description of a single event naturally involves the time it occurred. Some analyses of causation postulate that the cause must not occur (...) later than the effect. If we want this kind of causality it is easy to add the appropriate clause to ( CAUS ). An alternative is not to rule out backwards causation or causal loops a priori , but expect that ( CAUS ), via the properties of the contraction P C - , will result in the desired temporal relation between C and E . One way of ensuring this is to postulate that when the probability function P is contracted to P C - , the probabilities of all events that occurred before C remain the same in P C - as in P . This means that all beliefs about the history of events up to C are left unaltered in the construction of the hypothetical state of belief P C - . In conclusion, I hope to have shown that, in spite of these limitations, ( EXP ) and ( CAUS ) provide viable analyses of explanation and causality between single events for the case when epistemic states can be described by probability functions. I have also shown that the two analyses can be used to explicate the close connections between the two notions. These analyses reduce the problems of explanation and causality, hopefully in a non-circular way, to the problem of identifying contractions of states of belief. (shrink)

We examine the notion of conditionals and the role of conditionals in inductive logics and arguments. We identify three mistakes commonly made in the study of, or motivation for, non-classical logics. A nonmonotonic consequence relation based on evidential probability is formulated. With respect to this acceptance relation some rules of inference of System P are unsound, and we propose refinements that hold in our framework.

The object of this paper is to show why recent research in the psychology of deductive and probabilistic reasoning does not have.

R.A.Fisher introduced the fiducial argument as a means for obtaining something from nothing. He thought that on some occasions it was legitimate to obtain a posterior probability distribution over a range of simple statistical hypotheses without commitment to a prior distribution [4].H.Jeffreys thought he could tame Fisher by casting his argument in a Bayesian mold through a derivation of the fiducial posterior from a suitably constructed ignorance prior via Bayes’ theorem and conditionalization on the data of experimentation. According to Jeffreys, (...) Fisher was using something to obtain something after all ([7],pp.381-383).D.V.Lindley furthered the process of taming by specifying allegedly necessary and sufficient conditions for the consistency of fiducial reasoning in contexts of one parameter estimation with Bayesian requirements [14]. I. Hacking exploited Lindley’s result to supply his own ingenious method for taming Fisher([5],ch. 9).T.Seidenfeld has recently demonstrated that Lindley’s conditions are insufficient for consistency and that reconstructions like Jeffreys’ or Hacking’s lead to contradiction ([15], PP. 721-727). (shrink)

In this paper we dispel the supposed ``admissibility troubles'' for Bayesian accounts of direct inference proposed by Wallmann and Hawthorne, which concern the existence of surprising, unintuitive defeaters even for mundane cases of direct inference. We show that if one follows the majority of authors in the field in using classical probability spaces unimbued with any additional structure, one should expect similar phenomena to arise and should consider them unproblematic in themselves: defeaters abound! We then show that the framework of (...) Higher Probability Spaces allows the natural modelling of the discussed cases which produces no troubles of this kind. (shrink)

We consider Geanakoplos and Polemarchakis’s generalization of Aumman’s famous result on “agreeing to disagree", in the context of imprecise probability. The main purpose is to reveal a connection between the possibility of agreeing to disagree and the interesting and anomalous phenomenon known as dilation. We show that for two agents who share the same set of priors and update by conditioning on every prior, it is impossible to agree to disagree on the lower or upper probability of a hypothesis unless (...) a certain dilation occurs. With some common topological assumptions, the result entails that it is impossible to agree not to have the same set of posterior probabilities unless dilation is present. This result may be used to generate sufficient conditions for guaranteed full agreement in the generalized Aumman-setting for some important models of imprecise priors, and we illustrate the potential with an agreement result involving the density ratio classes. We also provide a formulation of our results in terms of “dilation-averse” agents who ignore information about the value of a dilating partition but otherwise update by full Bayesian conditioning. (shrink)

Systems of logico-probabilistic reasoning characterize inference from conditional assertions that express high conditional probabilities. In this paper we investigate four prominent LP systems, the systems _O, P_, _Z_, and _QC_. These systems differ in the number of inferences they licence _. LP systems that license more inferences enjoy the possible reward of deriving more true and informative conclusions, but with this possible reward comes the risk of drawing more false or uninformative conclusions. In the first part of the paper, we (...) present the four systems and extend each of them by theorems that allow one to compute almost-tight lower-probability-bounds for the conclusion of an inference, given lower-probability-bounds for its premises. In the second part of the paper, we investigate by means of computer simulations which of the four systems provides the best balance of reward versus risk. Our results suggest that system _Z_ offers the best balance. (shrink)

The question addressed in this paper is how the degree of justification of a belief is determined. A conclusion may be supported by several different arguments, the arguments typically being defeasible, and there may also be arguments of varying strengths for defeaters for some of the supporting arguments. What is sought is a way of computing the “on sum” degree of justification of a conclusion in terms of the degrees of justification of all relevant premises and the strengths of all (...) relevant reasons. (shrink)

In this paper I attempt to tie together a longstanding dispute between Henry Kyburg and Isaac Levi concerning statistical inferences. The debate, which centers around the example of Petersen the Swede, concerns Kyburg's and Levi's accounts of randomness and choosing reference classes. I argue that both Kyburg and Levi have missed the real significance of their dispute, that Levi's claim that Kyburg violates Confirmational Conditionalization is insufficient, and that Kyburg has failed to show that Levi's criteria for choosing reference class (...) are problematic. Rather, the significance of the Petersen case is to show that other aspects of their respective systems are defective: for Levi his account of credal judgments other than direct inference, and for Kyburg his explanation of how indexes are associated with a body of knowledge. (shrink)

Dilation occurs when an interval probability estimate of some event E is properly included in the interval probability estimate of E conditional on every event F of some partition, which means that one’s initial estimate of E becomes less precise no matter how an experiment turns out. Critics maintain that dilation is a pathological feature of imprecise probability models, while others have thought the problem is with Bayesian updating. However, two points are often overlooked: (1) knowing that E is stochastically (...) independent of F (for all F in a partition of the underlying state space) is sufficient to avoid dilation, but (2) stochastic independence is not the only independence concept at play within imprecise probability models. In this paper we give a simple characterization of dilation formulated in terms of deviation from stochastic independence, propose a measure of dilation, and distinguish between proper and improper dilation. Through this we revisit the most sensational examples of dilation, which play up independence between dilator and dilatee, and find the sensationalism undermined by either fallacious reasoning with imprecise probabilities or improperly constructed imprecise probability models. (shrink)

Examples growing out of the Newcomb problem have convinced many people that decision theory should proceed in terms of some kind of causal probability. I endorse this view and define and investigate a variety of causal probability. My definition is related to Skyrms' definition, but proceeds in terms of objective probabilities rather than subjective probabilities and avoids taking causal dependence as a primitive concept.

According to Williamson , the difference between empirical subjective Bayesians and objective Bayesians is that, while both hold reasonable credence to be calibrated to evidence, the objectivist also takes such credence to be as equivocal as such calibration allows. However, Williamson’s prescription for equivocation generates constraints on reasonable credence that are objectionable. Herein Williamson’s calibration norm is explicated in a novel way that permits an alternative equivocation norm. On this alternative account, evidence calibrated probability functions are recognised as implications of (...) evidence calibrated density functions defined over chance hypotheses. The objective Bayesian equivocates between these calibrated density functions rather than between the calibrated probability functions themselves. The result is an objective Bayesianism that avoids the main problem afflicting Williamson’s original proposal. (shrink)