Putnam construed the aim of Carnap’s program of inductive logic as the specification of a “universal learning machine,” and presented a diagonal proof against the very possibility of such a thing. Yet the ideas of Solomonoff and Levin lead to a mathematical foundation of precisely those aspects of Carnap’s program that Putnam took issue with, and in particular, resurrect the notion of a universal mechanical rule for induction. In this paper, I take up the question whether the Solomonoff–Levin proposal is (...) successful in this respect. I expose the general strategy to evade Putnam’s argument, leading to a broader discussion of the outer limits of mechanized induction. I argue that this strategy ultimately still succumbs to diagonalization, reinforcing Putnam’s impossibility claim. (shrink)

Carnap's Inductive Logic, like most philosophical discussions of induction, is designed for the case of independent trials. To take account of periodicities, and more generally of order, the account must be extended. From both a physical and a probabilistic point of view, the first and fundamental step is to extend Carnap's inductive logic to the case of finite Markov chains. Kuipers (1988) and Martin (1967) suggest a natural way in which this can be done. The probabilistic character of Carnapian inductive (...) logic(s) for Markov chains and their relationship to Carnap's inductive logic(s) is discussed at various levels of Bayesian analysis. (shrink)

It is well known that Rudolf Carnap’s original system of inductive logic failed to provide an adequate account of analogical reasoning. Since this problem was identified, there has been no shortage of proposals for how to incorporate analogy into inductive inference. Most alternatives to Carnap’s system, unlike his original one, have not been derived from first principles; this makes it to some extent unclear what the epistemic situations are to which they apply. This paper derives a new analogical inductive logic (...) from a set of axioms which extend Carnap’s postulates in a natural way. The key insights come from Bruno de Finetti’s ideas about analogy. The axioms of the new system capture epistemic conditions that call for a strong kind of analogical reasoning. The new system has a number of merits, but is also subject to limitations. I shall discuss both, together with some possible ways to generalize the approach taken in this paper. (shrink)

The purpose of this article is to invigorate debate concerning the nature of analogy, and to broaden the scope of current conceptions of analogy. We argue that analogizing is not a single or even a fundamental cognitive process. The argument relies on an analysis of the history of the concept of analogy, case studies on the use of analogy in scientific problem solving, cognitive research on analogy comprehension and problem solving, and a survey of computational mechanisms of analogy comprehension. Analogizing (...) is regarded as a macrocognitive phenomenon having a number of supporting processes. These include the apperception of resemblances and distinctions, metaphor, and the balancing of semantic flexibility and inference constraint. Psychological theories and computational models have generally relied on (a) a sparse set of ontological concepts (a property called “similarity” and a structuralist categorization of types of semantic relations), (b) a single form category (i.e., the classic four-term analogy), and (c) a single set of morphological distinctions (e.g., verbal vs. pictorial analogies). This article presents a classification based on a “naturalistic” exploration of the variety of uses of analogical reasoning in pragmatically distinct contexts. The resultant taxonomy distinguishes pre-hoc, ad-hoc, post-hoc, pro-hoc, contra-hoc, and trans-hoc analogy. Each will require its own macrocognitive modeling, and each presents an opportunity for research on phenomena of reasoning that have been neglected. (shrink)

The argument from analogy is examined from the point of view of Carnap's confirmation theory. It is argued that if inductive arguments are to be applicable to the real world, they must contain elementary analogical inferences. Carnap's system as originally developed (theλ -system) is not strong enough to take account of analogical arguments, but it is shown that the new system, which he has announced but not published in detail (theη -system), is capable of satisfying the conditions of inductive analogy. (...) Finally it is shown that an elementary analysis of analogical inference yields postulates of the η -system with a minimum of arbitrary assumptions. (shrink)

The argument from analogy is examined from the standpoint of Carnap's confirmation theory. Carnap's own discussion of analogy in relation to his c*— function is restricted to cases where the analogues are known to be similar, but not known to be different in any respect. It has been argued by the author in a previous work,, and by P. Achinstein, that typical analogy arguments involve known differences between the analogues as well as similarities. Achinstein shows that for such arguments none (...) of Carnap's Δ— system of conflrmation functions gives satisfactory values, and it is further shown in the present paper that for these arguments the confirmation never rises above its initial value, irrespective of evidence drown from an analogue. it is argued that even if inductive arguments are to be applicable to the real world, they must in principle be capable of taking into account known differences between the instances of an inductive generalization. Hence Carnap's Δ— system is inadequate as an explication of both induction and analogy.Three conditions are stated as necessary for any confirmation theory which gives a satisfactory explication of analogical inference:I. If two individuals a and b are known to agree in certain properties and differ in others, and if in addition a has a further property, then the confirmation of the hypothesis that b also has this property is greater than its initial confirmation, at least if the weight of the similarities is sufficiently great compared with the weight of the differences.II. The confirmation‐value increases or decreases with the weight of the similarities between the analogues compared with the weight of their differences.III. The confirmation that b has a certain property is greater if a has that property than if a does not have it.The Δ— system sastisfles none of these conditions. Carnap and Stegmüller have now, however, presented a new anxiom system for confirmation functions. the — system, Which is introduced primarily to deal with languages whose primitive predicates fall into families such that each individual can only be qualified by one predicate of each family. They construct a c‐ function which, as well as dealing with a language having two families of such predicates, also depends, unlike the functions in the Δ— system, on similarities between otherwise different individuals. The system is not developed far enough in this Appendix to enable the authors to discuss the confirmation in analogy arguments satisfying conditions I — III above, but it is shown in the present paper that, with a simple generalization of the measure‐function defined by Carnap and Stegmüller, and in the simplest non‐trivial case, condition I — III are all satisfied, indeed a stronger form of condition I is satisfied, in which no known similarity of the analogues is postulated.Finally it is remarked that Carnap and Stegmüller's method of introducing their Δ— system is extremely arbitrary and ad hoc, and an alternative method of introducing is suggested. Bearing in mind the fundamental nature of the argument from analogy in all application of induction to the real world, it is suggesetd that the fundamental inductive inference needing explication is not the inference of‘P1b, from ‘P,1a, but the analogical infeence of‘P2b, from‘P1P2a.P1b,. It is posinitial confirmation of‘P2b, and the confirmation of‘P2b, given‘P2a” and that the later two confirmation‐values are equivalent to those in the Δ— system. It is then shown that Carnap and Stegmüller's new measure‐function follows from their axiom system together with these postulates. Thus some of the arbitrariness of the — system is removed by general considérations regarding inductive and analogical inference. (shrink)

The object of this paper is to look at the extent and nature of the uses of analogy during the ªrst century following the so-called scientiªc revolution. Using the research tool provided by JSTOR we systematically analyze the uses of “analog” and its cognates (analogies, analogous, etc.) in the Philosophical Transactions of the Royal Society of London for the period 1665–1780. In addition to giving the possibility of evaluating quantitatively the proportion of papers explicitly using analogies, this approach makes it (...) possible to go beyond the maybe idiosyncratic cases of Descartes, Kepler, Galileo, and other much studied giants of the so-called Scientiªc Revolution. As a result a classiªcation of types of uses is proposed. Relations between types of analogies and research ªelds are also established. In this paper we are less interested in discussing the “real nature” or “essence” or even the cognitive limitations of analogical thinking than in describing its various uses and different meanings as they changed over the course of a century. (shrink)

Work on analogy has been done from a number of disciplinary perspectives throughout the history of Western thought. This work is a multidisciplinary guide to theorizing about analogy. It contains 1,406 references, primarily to journal articles and monographs, and primarily to English language material. classical through to contemporary sources are included. The work is classified into eight different sections (with a number of subsections). A brief introduction to each section is provided. Keywords and key expressions of importance to research on (...) analogy are discussed in the introductory material. Electronic resources for conducting research on analogy are listed as well. (shrink)

According to a standard account of evidence, one piece of information is stronger evidence for an hypothesis than is another iff the probability of the hypothesis on the one is greater than it is on the other. This condition, I argue, is neither necessary nor sufficient because various factors can strengthen the evidence for an hypothesis without increasing (and even decreasing) its probability. Contrary to what probabilists claim, I show that this obtains even if a probability function can take these (...) evidential factors into account in ways they suggest and yield a unique probability value. Nor will the problem be solved by appealing to second-order probabilities. (shrink)

An a priori thesis about evidence, defended by many, states that the only empirical fact that can affect the truth of an objective evidence claim of the form ‘e is evidence for h’ (or ‘e confirms h to degree r’) is the truth of e; all other considerations are a priori. By examining cases involving evidential flaws, I challange this claim and defend an empirical concept of evidence. In accordance with such a concept, whether, and the extent to which, e, (...) if true, confirms h is an empirical, not a priori, fact. (shrink)