Paradox of Confirmation

Think of confirmation in the context of the Ravens Paradox this way. The likelihood ratio measure of incremental confirmation gives us, for an observed Black Raven and for an observed non-Black non-Raven, respectively, the following “full” likelihood ratios.

Freitag (2015) and Schramm (2014) have proposed different, although converging, solutions of Goodman’s New Riddle of Induction. Answering their proposals, Dorst (2016 and 2018) has used the fictitious character of a ‘grue-speaker’ as his principal device for criticizing counterfactual-based treatments of the Riddle. In this paper, I argue that Dorst’s arguments fail: On the observation of no other than green emeralds, the ‘grue-speaker’ cannot use the symmetry between the ‘green’- and ‘grue’-languages for claiming ‘grue’- instead of ‘green’-evidence, and the counterfactuals (...) involved (explicitly by Schramm and implicitly by Freitag) remain unaffected by Dorst’s proposal for how to evaluate them. (shrink)

The article presents the empirical confirmation to the black hole and white hole juxtapose theory. The author based the experiment on the multi- mission multi-spectral space telescope data conducted remotely with the NASA Data Challenge and Harvard- Smithsonian Micro-Observatory. Since the loss of the original manuscript, the author reformulated the mathematics during the research. The observation developed a resonance observation technique that observed the white hole to the moon’s direction with the sun. The data reduction of the white hole and (...) other observations provides an explanation for the undetected gravitons and some empirical physical features of the white hole. (shrink)

Simple random sampling resolutions of the raven paradox relevantly diverge from scientific practice. We develop a stratified random sampling model, yielding a better fit and apparently rehabilitating simple random sampling as a legitimate idealization. However, neither accommodates a second concern, the objection from potential bias. We develop a third model that crucially invokes causal considerations, yielding a novel resolution that handles both concerns. This approach resembles Inference to the Best Explanation (IBE) and relates the generalization’s confirmation to confirmation of an (...) associated law. We give it an objective Bayesian formalization and discuss the compatibility of Bayesianism and IBE. (shrink)

After long arguments between positivism and falsificationism, the verification of universal hypotheses was replaced with the confirmation of uncertain major premises. Unfortunately, Hemple proposed the Raven Paradox. Then, Carnap used the increment of logical probability as the confirmation measure. So far, many confirmation measures have been proposed. Measure F proposed by Kemeny and Oppenheim among them possesses symmetries and asymmetries proposed by Elles and Fitelson, monotonicity proposed by Greco et al., and normalizing property suggested by many researchers. Based on the (...) semantic information theory, a measure b* similar to F is derived from the medical test. Like the likelihood ratio, measures b* and F can only indicate the quality of channels or the testing means instead of the quality of probability predictions. Furthermore, it is still not easy to use b*, F, or another measure to clarify the Raven Paradox. For this reason, measure c* similar to the correct rate is derived. Measure c* supports the Nicod Criterion and undermines the Equivalence Condition, and hence, can be used to eliminate the Raven Paradox. An example indicates that measures F and b* are helpful for diagnosing the infection of Novel Coronavirus, whereas most popular confirmation measures are not. Another example reveals that all popular confirmation measures cannot be used to explain that a black raven can confirm “Ravens are black” more strongly than a piece of chalk. Measures F, b*, and c* indicate that the existence of fewer counterexamples is more important than more positive examples’ existence, and hence, are compatible with Popper’s falsification thought. (shrink)

Many researchers want to unify probability and logic by defining logical probability or probabilistic logic reasonably. This paper tries to unify statistics and logic so that we can use both statistical probability and logical probability at the same time. For this purpose, this paper proposes the P–T probability framework, which is assembled with Shannon’s statistical probability framework for communication, Kolmogorov’s probability axioms for logical probability, and Zadeh’s membership functions used as truth functions. Two kinds of probabilities are connected by an (...) extended Bayes’ theorem, with which we can convert a likelihood function and a truth function from one to another. Hence, we can train truth functions by sampling distributions. This probability framework was developed in the author’s long-term studies on semantic information, statistical learning, and color vision. This paper first proposes the P–T probability framework and explains different probabilities in it by its applications to semantic information theory. Then, this framework and the semantic information methods are applied to statistical learning, statistical mechanics, hypothesis evaluation, confirmation, and Bayesian reasoning. Theoretical applications illustrate the reasonability and practicability of this framework. This framework is helpful for interpretable AI. To interpret neural networks, we need further study. (shrink)

The concepts of coherence and confirmation are closely intertwined: according to a prominent proposal coherence is nothing but mutual confirmation. Accordingly, it should come as no surprise that both are confronted with similar problems. As regards Bayesian confirmation measures these are illustrated by the problem of tacking by conjunction. On the other hand, Bayesian coherence measures face the problem of belief individuation. In this paper we want to outline the benefit of an approach to coherence and confirmation based on content (...) elements. It will be shown that the resulting concepts, called genuine coherence and genuine confirmation, can be used in order to solve the two mentioned problems. In a final section we present some results on degrees of genuine coherence and genuine confirmation. (shrink)

"All ravens are black" is logically but not confirmationally equivalent with "all non-black things are non-ravens." But this is impossible, given that logical equivalence guarantees confirmational equivalence. In this paper, this paradox is solved.

Objectivism about evidential support is the thesis that facts about the degree to which a body of evidence supports a hypothesis are objective rather than depending on subjective factors like one’s own language or epistemic values. Objectivism about evidential support is key to defending a synchronic, time-slice-centric conception of epistemic rationality, on which what you ought to believe at a time depends only on what evidence you have at that time, and not on how you were at previous times. Here, (...) I defend a version of objectivism about evidential support on which facts about evidential support are grounded in facts about explanatoriness. (shrink)

Confirmation and falsification are different strategies for testing theories and characterizing the outcomes of those tests. Roughly speaking, confirmation is the act of using evidence or reason to verify or certify that a statement is true, definite, or approximately true, whereas falsification is the act of classifying a statement as false in the light of observation reports. After expounding the intellectual history behind confirmation and falsificationism, reaching back to Plato and Aristotle, I survey some of the main controversial issues and (...) arguments that pertain to the choice between these strategies: the Raven Paradox, the Duhem/Quine problem and the Grue Paradox. Finally, I outline an evolutionary criticism of inductive Bayesian approaches based on my assumption of doxastic involuntarism. (shrink)

Historically, Nelson Goodman’s paradox involving the predicates ‘grue’ and ‘bleen’ has been taken to furnish a serious blow to Carl Hempel’s theory of confirmation in particular and to purely formal theories of confirmation in general. In this paper, I argue that Goodman’s paradox is no more serious of a threat to Hempel’s theory of confirmation than is Hempel’s own paradox of the ravens. I proceed by developing a suggestion from R. D. Rosenkrantz into an argument for the conclusion that these (...) paradoxes are, in fact, equivalent. My argument, if successful, is of both historical and philosophical interest. Goodman himself maintained that Hempel’s theory of confirmation was capable of handling the paradox of the ravens. And Hempel eventually conceded that Goodman’s paradox showed that there could be no adequate, purely syntactical theory of confirmation. The conclusion of my argument entails, by contrast, that Hempel’s theory of confirmation is incapable of handling Goodman’s paradox if and only if it is incapable of handling the paradox of the ravens. It also entails that for any adequate solution to one of these paradoxes, there is a corresponding and equally adequate solution to the other. (shrink)

The canonical Bayesian solution to the ravens paradox faces a problem: it entails that black non-ravens disconfirm the hypothesis that all ravens are black. I provide a new solution that avoids this problem. On my solution, black ravens confirm that all ravens are black, while non-black non-ravens and black non-ravens are neutral. My approach is grounded in certain relations of epistemic dependence, which, in turn, are grounded in the fact that the kind raven is more natural than the kind black. (...) The solution applies to any generalization “All F’s are G” in which F is more natural than G. (shrink)

Fitelson (1999) demonstrates that the validity of various arguments within Bayesian confirmation theory depends on which confirmation measure is adopted. The present paper adds to the results set out in Fitelson (1999), expanding on them in two principal respects. First, it considers more confirmation measures. Second, it shows that there are important arguments within Bayesian confirmation theory and that there is no confirmation measure that renders them all valid. Finally, the paper reviews the ramifications that this "strengthened problem of measure (...) sensitivity" has for Bayesian confirmation theory and discusses whether it points at pluralism about notions of confirmation. (shrink)

Bayesians regard their solution to the paradox of confirmation as grounds for preferring their theory of confirmation to Hempel’s. They point out that, unlike Hempel, they can at least say that a black raven confirms “All ravens are black” more than a white shoe. However, I argue that this alleged advantage is cancelled out by the fact that Bayesians are equally committed to the view that a white shoe confirms “All non-black things are non-ravens” less than a black raven. In (...) light of this, I reexamine the dialectic between Hempel and the Bayesians. (shrink)

Hypothetico-deductive (H-D) confirmation builds on the idea that confirming evidence consists of successful predictions that deductively follow from the hypothesis under test. This article reviews scope, history and recent development of the venerable H-D account: First, we motivate the approach and clarify its relationship to Bayesian confirmation theory. Second, we explain and discuss the tacking paradoxes which exploit the fact that H-D confirmation gives no account of evidential relevance. Third, we review several recent proposals that aim at a sounder and (...) more comprehensive formulation of H-D confirmation. Finally, we conclude that the reputation of hypothetico-deductive confirmation as outdated and hopeless is undeserved: not only can the technical problems be addressed satisfactorily, the hypothetico-deductive method is also highly relevant for scientific practice. (shrink)

I argue that the standard Bayesian solution to the ravens paradox— generally accepted as the most successful solution to the paradox—is insufficiently general. I give an instance of the paradox which is not solved by the standard Bayesian solution. I defend a new, more general solution, which is compatible with the Bayesian account of confirmation. As a solution to the paradox, I argue that the ravens hypothesis ought not to be held equivalent to its contrapositive; more interestingly, I argue that (...) how we formally represent hypotheses ought to vary with the context of inquiry. This explains why the paradox is compelling, while dealing with standard objections to holding hypotheses inequivalent to their contrapositives. (shrink)

To clarify and illuminate the place of probability in science Ellery Eells and James H. Fetzer have brought together some of the most distinguished philosophers ...

The (recent, Bayesian) cognitive science literature on the Wason Task (WT) has been modeled largely after the (not-so-recent, Bayesian) philosophy of science literature on the Paradox of Confirmation (POC). In this paper, we apply some insights from more recent Bayesian approaches to the (POC) to analogous models of (WT). This involves, first, retracing the history of the (POC), and, then, re-examining the (WT) with these historico-philosophical insights in mind.

The Paradox of the Ravens (a.k.a,, The Paradox of Confirmation) is indeed an old chestnut. A great many things have been written and said about this paradox and its implications for the logic of evidential support. The first part of this paper will provide a brief survey of the early history of the paradox. This will include the original formulation of the paradox and the early responses of Hempel, Goodman, and Quine. The second part of the paper will describe attempts (...) to resolve the paradox within a Bayesian framework, and show how to improve upon them. This part begins with a discussion of how probabilistic methods can help to clarify the statement of the paradox itself. And it describes some of the early responses to probabilistic explications. We then inspect the assumptions employed by traditional (canonical) Bayesian approaches to the paradox. These assumptions may appear to be overly strong. So, drawing on weaker assumptions, we formulate a new-and-improved Bayesian confirmation-theoretic resolution of the Paradox of the Ravens. (shrink)

The (recent, Bayesian) cognitive science literature on The Wason Task (WT) has been modeled largely after the (not-so-recent, Bayesian) philosophy of science literature on The Paradox of Confirmation (POC). In this paper, we apply some insights from more recent Bayesian approaches to the (POC) to analogous models of (WT). This involves, first, retracing the history of the (POC), and, then, reexamining the (WT) with these historico-philosophical insights in mind.

This paper is devoted to analysis of co-called paradox of confirmation formulated by C. G. Hempel in the 1930s. In particular, the author proposes a solution of this puzzle. The proposal consists in refining the concept of confirmation by adding a clause that if A confirms a hypothesis h, the former must be a logical consequence of a latter, eventually derived with the help of additional assumptions. This leads to an additional constraint requiring that confirmations act relatively to sets of (...) reference. Finally, if h and h’ are logically equivalent, a sentence A confirms both to the same degree if and only if related sets of reference are the same. (shrink)

Exhumation and study of the 1945 paradox of confirmation brings out the defect of its formulation. In the context of quantifier conditional-probability logic it is shown that a repair can be accomplished if the truth-functional conditional used in the statement of the paradox is replaced with a connective that is appropriate to the probabilistic context. Description of the quantifier probability logic involved in the resolution of the paradox is presented in stages. Careful distinction is maintained between a formal logic language (...) and its semantics, as the same language may be outfitted with different semantics. An acquaintance with sections 1? 5 of Hailperin (2006) covering the sentential aspects of probability logic is assumed as background information for quantifier probability logic. (shrink)

This paper starts by indicating the analysis of Hempel's conditions of adequacy for any relation of confirmation (Hempel, 1945) as presented in Huber (submitted). There I argue contra Carnap (1962, Section 87) that Hempel felt the need for two concepts of confirmation: one aiming at plausible theories and another aiming at informative theories. However, he also realized that these two concepts are conflicting, and he gave up the concept of confirmation aiming at informative theories. The main part of the paper (...) consists in working out the claim that one can have Hempel's cake and eat it too - in the sense that there is a logic of theory assessment that takes into account both of the two conflicting aspects of plausibility and informativeness. According to the semantics of this logic, a is an acceptable theory for evidence β if and only if a is both sufficiently plausible given β and sufficiently informative about β. This is spelt out in terms of ranking functions (Spohn, 1988) and shown to represent the syntactically specified notion of an assessment relation. The paper then compares these acceptability relations to explanatory and confirmatory consequence relations (Flach, 2000) as well as to nonmonotonic consequence relations (Kraus et al., 1990). It concludes by relating the plausibility-informativeness approach to Carnap's positive relevance account, thereby shedding new light on Carnap's analysis as well as solving another problem of confirmation theory. (shrink)

Astronomers study the behaviour of the stars; philosophers of science study the behaviour of the astronomers. Philosophers of science, alongside historians and sociologists of science, are in the business of accounting for how science works and what it achieves. There is more to the philosophy of science than principled descriptions of scientific activity, since there are also all the normative questions of justification and warrant, but the descriptive task is an important part of the discipline and the primary focus of (...) the present essay. (shrink)

Brief consideration of (1) Peirce’s ‘logic of vagueness’, (2) his categories, and (3) the concepts of overdetermination and underdetermination, vagueness and generality, and inconsistency and incompleteness, along with (4) the abrogation of classical Aristotelian principles of logic, bear out the complexity of all relatively rich sign systems. Given this complexity, there is semiotic indeterminacy, which suggests sign limitations, and at the same time it promises semiotic freedom, giving rise to sign proliferation the yield of which is pluralistic, inter-relational semiosis. This (...) proliferation of signs owes its perpetual flowing change in time to the inapplicability of classical logical principles, namely Non-Contradiction and Excluded-Middle, with respect to elements of vagueness and generality in all signs. Hempel’s ‘Inductivity Paradox’ and Goodman’s ‘New Riddle of Induction’ bear out the limitation and freedom of sign making and sign taking. A concrete cultural example, the Spaniards’ world including the Virgin of Guadalupe and the Aztecs world including their Goddess, Tonantzín, are given a Hempel-Goodman interpretation to reveal the ambiguous, vague, and complex nature of intercultural sign systems, further suggesting pluralism. In fact, when taking the ‘limitativetheorems’ of Gödel, Turing, and Chaitin into account, pluralism becomes undeniable, in view of the inconsistency-incompleteness of complex systems. A model for embracing and coping with pluralism suggests itself in the form of contextualized novelty seeking relativism. This form of pluralism takes overdetermination, largely characteristic of Peirce’s Firstness, and underdetermination largely characteristic of Peirce’s Thirdness, into its embrace to reveal a global context capable of elucidating local contexts the collection of which is considerably less than that global view. The entirety of this global context is impossible to encompass, given our inevitable finitude and fallibilism. Yet, we usually manage to cope with processual pluralism, within the play of semiosis. (shrink)

To solve the highly counterintuitive paradox of confirmation represented by the statement, “A pair of red shoes confirms that all ravens are black,” Hempel employed a strategy that retained the equivalence condition but abandoned Nicod’s irrelevance condition. However, his use of the equivalence condition is fairly ad hoc, raising doubts about its applicability to this problem. Furthermore, applying the irrelevance condition from Nicod’s criterion does not necessarily lead to paradoxes, nor does discarding it prevent the emergence of paradoxes. Hempel’s approach (...) fails to adequately resolve the paradox. (shrink)

Hempel first introduced the paradox of confirmation in (Hempel 1937). Since then, a very extensive literature on the paradox has evolved (Vranas 2004). Much of this literature can be seen as responding to Hempel’s subsequent discussions and analyses of the paradox in (Hempel 1945). Recently, it was noted that Hempel’s intuitive (and plausible) resolution of the paradox was inconsistent with his official theory of confirmation (Fitelson & Hawthorne 2006). In this article, we will try to explain how this inconsistency affects (...) the historical dialectic about the paradox and how it illuminates the nature of confirmation. In the end, we will argue that Hempel’s intuitions about the paradox of confirmation were (basically) correct, and that it is his theory that should be rejected, in favor of a (broadly) Bayesian account of confirmation. (shrink)

Hypothetico-deductivists have struggled to develop qualitative confirmation theories not raising the so-called tacking by disjunction paradox. In this paper, I analyze the difficulties yielded by the paradox and I argue that the hypothetico-deductivist solutions given by Gemes (1998) and Kuipers (2000) are questionable because they do not fit such analysis. I then show that the paradox yields no difficulty for the Bayesian who appeals to the Total Evidence Condition. I finally argue that the same strategy is unavailable to the hypothetico-deductivist.

'IRS' is our term for the logical empiricist idea that the best way to understand the epistemic bearing of observational evidence on scientific theories is to model it in terms of Inferential Relations among Sentences representing the evidence, and sentences representing hypotheses the evidence is used to evaluate. Developing ideas from our earlier work, including 'Saving the Phenomena'(Phil Review 97, 1988, p.303-52 )we argue that the bearing of observational evidence on theory depends upon causal connections and error characteristics of the (...) processes by which data is produced and used to detect features of phenomena. Neither of these depends upon, or is greatly illuminated by a consideration of, formal relations among observation and theoretical sentences or propositions. By taking causal structures and error characteristics, you too can evade the IRS. In doing so, you can gain insight into Hempel’s raven paradox, theory loading, and other issues from the standard philosophical literature on confirmation theory. (shrink)

Naive deductivist accounts of confirmation have the undesirable consequence that if E confirms H, then E also confirms the conjunction H·X, for any X—even if X is completely irrelevant to E and H. Bayesian accounts of confirmation may appear to have the same problem. In a recent article in this journal Fitelson (2002) argued that existing Bayesian attempts to resolve of this problem are inadequate in several important respects. Fitelson then proposes a new‐and‐improved Bayesian account that overcomes the problem of (...) irrelevant conjunction, and does so in a more general setting than past attempts. We will show how to simplify and improve upon Fitelson's solution. (shrink)

In this paper, I focus on the so-called "tacking by disjunction problem". Namely, the problem to the effect that, if a hypothesis H is confirmed by a statement E, H is confirmed by the disjunction E v F, for whatever statement F. I show that the attempt to settle this difficulty made by Grimes 1990, in a paper apparently forgotten by today methodologists, is irremediably faulty.

According to Hempel's paradox, evidence (E) that an object is a nonblack nonraven confirms the hypothesis (H) that every raven is black. According to the standard Bayesian solution, E does confirm H but only to a minute degree. This solution relies on the almost never explicitly defended assumption that the probability of H should not be affected by evidence that an object is nonblack. I argue that this assumption is implausible, and I propose a way out for Bayesians. Introduction Hempel's (...) paradox, the standard Bayesian solution, and the disputed assumption Attempts to defend the disputed assumption Attempts to refute the disputed assumption A way out for Bayesians Conclusion. (shrink)

Three confirmation principles discussed by Hempel are the Converse Consequence Condition, the Special Consequence Condition and the Entailment Condition. Le Morvan (1999) has argued that, when the choice among confirmation principles is just about them, it is the Converse Consequence Condition that must be rejected. In this paper, I make this argument definitive. In doing that, I will provide an indisputable proof that the simple conjunction of the Converse Consequence Condition and the Entailment Condition yields a disastrous consequence.

Naive deductive accounts of confirmation have the undesirable consequence that if E confirms H, then E also confirms the conjunction H & X, for any X—even if X is utterly irrelevant to H (and E). Bayesian accounts of confirmation also have this property (in the case of deductive evidence). Several Bayesians have attempted to soften the impact of this fact by arguing that—according to Bayesian accounts of confirmation— E will confirm the conjunction H & X less strongly than E confirms (...) H (again, in the case of deductive evidence). I argue that existing Bayesian “resolutions” of this problem are inadequate in several important respects. In the end, I suggest a new‐and‐improved Bayesian account (and understanding) of the problem of irrelevant conjunction. (shrink)

According to Bayesian confirmation theory, evidence E (incrementally) confirms (or supports) a hypothesis H (roughly) just in case E and H are positively probabilistically correlated (under an appropriate probability function Pr). There are many logically equivalent ways of saying that E and H are correlated under Pr. Surprisingly, this leads to a plethora of non-equivalent quantitative measures of the degree to which E confirms H (under Pr). In fact, many non-equivalent Bayesian measures of the degree to which E confirms (or (...) supports) H have been proposed and defended in the literature on inductive logic. I provide a thorough historical survey of the various proposals, and a detailed discussion of the philosophical ramifications of the differences between them. I argue that the set of candidate measures can be narrowed drastically by just a few intuitive and simple desiderata. In the end, I provide some novel and compelling reasons to think that the correct measure of degree of evidential support (within a Bayesian framework) is the (log) likelihood ratio. The central analyses of this research have had some useful and interesting byproducts, including: (i ) a new Bayesian account of (confirmationally) independent evidence, which has applications to several important problems in con- firmation theory, including the problem of the (confirmational) value of evidential diversity, and (ii ) novel resolutions of several problems in Bayesian confirmation theory, motivated by the use of the (log) likelihood ratio measure, including a reply to the Popper-Miller critique of probabilistic induction, and a new analysis and resolution of the problem of irrelevant conjunction (a.k.a., the tacking problem). (shrink)

Contemporary Bayesian confirmation theorists measure degree of (incremental) confirmation using a variety of non-equivalent relevance measures. As a result, a great many of the arguments surrounding quantitative Bayesian confirmation theory are implicitly sensitive to choice of measure of confirmation. Such arguments are enthymematic, since they tacitly presuppose that certain relevance measures should be used (for various purposes) rather than other relevance measures that have been proposed and defended in the philosophical literature. I present a survey of this pervasive class of (...) Bayesian confirmation-theoretic enthymemes, and a brief analysis of some recent attempts to resolve the problem of measure sensitivity. (shrink)

The adverse reaction to Hempel's 'ravens paradox' embodied in giving it that description is compared with the usual reaction of experimental subjects to the Wason selection task.

The expression of a physical law has usually been understood as a description. Among the entities that different philosophers have maintained that nomic expressions describe are regularities among events or states of affairs, natural necessities, and relations among universals. I maintain that nomic expressions are better understood not as serving a descriptive function but as in many respects playing a prescriptive role; the expression of a physical law, I argue, says that one ought to regard certain claims as able to (...) justify certain other claims. On the basis of this proposal, I advance a response to Hempel's "problem of provisos" and a conception of inductive inference as the reasoned adoption of claims prescribing that certain inferences be considered good. I argue that on this account of laws, the contraposition of a nomic expression need not express a law; by properly distinguishing the confirmation of a claim's nomologicality from the confirmation of a claim's truth, this account of inductive inference can avoid Hempel's raven paradox. I hold that this approach to confirmation leads to a novel conception of the relation between empirical and non-empirical criteria of theory-choice, as well as to proposals concerning the grue problem and the role of theoretical terms in the confirmation of a claim's nomologicality. I further argue that in order to offer any reasons at all for treating persons as qualified to make a certain kind of observation-report in virtue of their behavior under certain circumstances, one must be prepared to adopt nomic expressions that endorse what Hume considered to be risky inferences. I contend that this result constitutes a promising basis for accounts of scientific progress and of the rationality of inductive inference. By deliberating about which expressions to adopt as nomic, i.e., by reasoning about what claims we ought to count as good reasons for what other claims, we can get an argumentative grip on our justificatory practices; as a progressively greater portion of our language-use comes into accord with the prescriptions that we have adopted as nomic expressions, the proprieties that govern our language-use become increasingly of our own design. (shrink)

I present a puzzle which seems simple, but is found to have interesting implications for confirmation. Its dissolution also helps us to throw light on the relationship between first- and second-order probabilities construed as rational degrees of belief.

The concepts of supportive evidence and of relevant evidence seem very closely related to each other. Supportive evidence is clearly always relevant as well. But must relevant evidence be defined as evidence which is either supportive or weakeking? In an explicit or implicit manner, this is indeed the position of many philosophers. The paradox of ideal evidence, however, shows us that this is to restrictive. Besides increasing or decreasing the probability attached to some hypothesis, evidence can alter or interact with (...) the background assumptions underlying the hypothesis.In most circumstances, the (post hoc) relevance of evidence can indeed be judged by its effect on the confidence one attaches to hypotheses. Occasionally, as in the circumstances described by Example I, and more generally called “the Paradox of Ideal Evidence”, the relevance of evidence to an hypothesis can only be understood by appeal to a broader sense. (shrink)