1. Introduction. It is generally admitted that a large part of man's knowledge is based on inductive arguments. Hence any philosophical theory concerning the nature of inductive arguments constitutes an epistemological theory. Any such philosophical theory of induction must, if it is to be satisfactory, take adequate account of Hume's criticism of inductive arguments. One way of treating his criticism is to say that the validity of inductive arguments is in an important sense relative to some broad factual assumptions about the general nature of the universe and that these general factual assumptions are presupposed in a certain way by the users of inductive arguments (8, 10). Let us call any theory of this general type a postulate theory of induction.

This article shows that a slight variation of the argument in Milne 1996 yields the log-likelihood ratio l rather than the log-ratio measure r as “the one true measure of confirmation.”

Prominent defenders of the extended cognition thesis have looked to evolutionary theory for support. Roughly, the idea is that natural selection leads one to expect that cognitive strategies should exploit the environment, and exploitation of the right sort results in a cognitive system that extends beyond the head of the organism. I argue that proper appreciation of evolutionary theory should create no such expectation. This leaves open whether cognitive systems might in fact bear a relationship to the environment that leads to their extension.

The place of induction in the framing and test of scientific hypotheses is investigated. The meaning of ‘induction’ is first equated with generalization on the basis of case examination. Two kinds of induction are then distinguished: the inference of generals from particulars (first degree induction), and the generalization of generalizations (second degree induction). Induction is claimed to play a role in the framing of modest empirical generalizations and in the extension of every sort of generalizations—not however in the invention of high-level hypotheses containing theoretical predicates. It is maintained, on the other hand, that induction by enumeration is essential in the empirical test of the lowest-level consequences of scientific theories, since it occurs in the drawing of “conclusions” from the examination of empirical evidence. But it is also held that the empirical test is insufficient, and must be supplemented with theorification, or the expansion of isolated hypotheses into theories. Refutation is not viewed as a substitute for confirmation but as its complement, since the very notion of unfavorable case is meaningful only in connection with the concept of positive instance. Although the existence of an inductive method is disclaimed, it is maintained that the various patterns of plausible reasoning (inductive inference included) are worth being investigated. It is concluded that scientific research follows neither the advice of inductivism nor the injunction of deductivism, but takes a middle course in which induction is instrumental both heuristically and methodologically, although the over-all pattern of research is hypothetico-deductive.

Michael Weisberg and Kenneth Reisman argue that the Volterra Principle can be derived from multiple predator-prey models and that, therefore, the Volterra Principle is a prime example for robustness analysis. In the current article, I give new results regarding the Volterra Principle, extending Weisberg’s and Reisman’s work, and I discuss the consequences of these results for robustness analysis. I argue that we do not end up with multiple, independent models but rather with one general model. I identify the kind of situation in which this generalization approach may occur, and I analyze the generalized Volterra Principle from an explanatory perspective.

Evolutionary convergence is often appealed to in support of claims about multiple realization. The idea is that convergence shows that the same function can be realized by different kinds of structures. I argue here that the nature of convergence is more complicated than it might appear at first look. Broad claims about convergence are made by biologists during general discussions of the mechanisms of evolution. In their specialized work, though, biologists are often more limited in the claims they make. I will examine a standard example to show how claims about convergence can be oversimplified.

The failure of the attempt by Michael Polanyi to capture the social organization of science by comparing it to the operation of a market bears salutary lessons for modern philosophers of science in their rush to appropriate market models and metaphors. In this case, an initially plausible invisible hand argument ended up as crude propaganda for the uniquely privileged social support of science.

This paper involves one crucial assumption; namely, that the statistical predictions of quantum mechanics for Bell's variant of the EPR experiment will continue to be verified as detector efficiencies are improved and the need for coincidence counters is eliminated. This assumption entails that any hidden-variables theory for quantum mechanics must violate Bell's inequality—the inequality derived in Bell (1964). It is shown here that four locality conditions are involved in the derivation of Bell's inequality; and that a violation of any of the four locality conditions will either entail the existence of superluminal influences or the existence of superluminal signals (superluminal influences that can be used to transmit information), if conspiratorial theories can be ruled out. The attempts so far to rule out conspiratorial theories are all found to be rather dubious, but there are other considerations developed here that rule them out convincingly. Finally, it is demonstrated that violations of each of the four locality conditions can be used to transmit information superluminally, if certain auxiliary conditions are satisfied. This is of particular interest because one of these conditions corresponds to a condition dubbed “completeness” by Jon Jarrett. Jarrett and others have suggested that violations of completeness cannot be used to send information superluminally. Demonstrating otherwise is, perhaps, the most significant result obtained in this paper.

The “Cosmological Constant Problem” (CCP) is widely considered a crisis in contemporary theoretical physics. Unfortunately, the search for its resolution is hampered by open disagreement about what is, strictly, the problem. This disagreement stems from the observation that the CCP is not a problem within any of our current theories, and nearly all of the details of those future theories for which the CCP could be made a problem are up for grabs. Given this state of affairs, I discuss how one ought to make sense of the role of the CCP in physics and generalize some lessons from it.

Most attempts to define culture as used in the cultural evolution literature treat culture as a single phenomenon that can be given a single nondisjunctive definition. In this article I argue that, really, cultural evolutionists employ a variety of distinct but closely related concepts of culture. I show how the main prominent attempts to define a culture concept fail to properly capture all the uses of “culture” employed in cultural evolutionary work. I offer a description of some of the most important culture concepts used by cultural evolutionists.

This article explores some ways in which philosophical theories of causation and empirical investigations into causal learning and judgment can mutually inform one another.

Is the quantum-logic interpretation dead? Its near total absence from current discussions about the interpretation of quantum theory suggests so. While mathematical work on quantum logic continues largely unabated, interest in the quantum-logic interpretation seems to be almost nil, at least in Anglo-American philosophy of physics.

This paper has the immodest purpose of changing that fact. I shall argue that while the quantum-logic interpretation faces challenges, it remains a live option. The usual objections either miss the mark, or admit a reasonable answer, or fail to decide the issue conclusively.

This paper accounts for broad definitions of memory, which extend to paradigmatic memory phenomena, like episodic memory in humans, and phenomena in worms and sea snails. These definitions may seem too broad, suggesting that they extend to phenomena that don’t count as memory or illustrate that memory is not a natural kind. However, these responses fail to consider a definition as a hypothesis. As opposed to construing definitions as expressing memory’s properties, a definition as a hypothesis is the basis to test inferences about phenomena. A definition as a hypothesis is valuable when the “kinding” of phenomena is ongoing.

Mechanisms are usually viewed as hierarchical, with lower levels of a mechanism influencing, and decomposing, its higher-level behavior. To draw quantitative predictions from a model of a mechanism, the model must capture this hierarchical aspect. Recursive Bayesian networks (RBNs) were put forward by Lorenzo Casini et al. as a means to model mechanistic hierarchies by decomposing variables into their constituting causal networks. The proposal was criticized by Alexander Gebharter. He proposes an alternative formalism, which instead decomposes arrows. Here, I defend RBNs from the criticism and argue that they offer a better representation of mechanistic hierarchies than the rival account.

Almost all computational models of the mind and brain ignore details about neurotransmitters, hormones, and other molecules. The neglect of neurochemistry in cognitive science would be appropriate if the computational properties of brains relevant to explaining mental functioning were in fact electrical rather than chemical. But there is considerable evidence that chemical complexity really does matter to brain computation, including the role of proteins in intracellular computation, the operations of synapses and neurotransmitters, and the effects of neuromodulators such as hormones. Neurochemical computation has implications for understanding emotions, cognition, and artificial intelligence.

We show that as a chain of confirmation becomes longer, confirmation dwindles under screening-off. For example, if E confirms H1, H1 confirms H2, and H1 screens off E from H2, then the degree to which E confirms H2 is less than the degree to which E confirms H1. Although there are many measures of confirmation, our result holds on any measure that satisfies the Weak Law of Likelihood. We apply our result to testimony cases, relate it to the Data-Processing Inequality in information theory, and extend it in two respects so that it covers a broader range of cases.

This paper counters an objection raised against one of Bertrand Russell's lesser-known epistemological views, viz. “structural realism” (SR). In short, SR holds that at most we have knowledge of the structure of the external (i.e., physical) world. M. H. A. Newman's allegedly fatal objection is that SR is either trivial or false. I argue that the accusation of triviality is itself empty since it fails to establish that SR knowledge claims are uninformative. Moreover, appealing to Quine's notion of ontological relativity, I suggest that far from being false, SR knowledge claims seem to be the most that we can hope for.

In this paper, I investigate the nature of a priori biological laws in connection with the idea that laws must be empirical. I argue that the epistemic functions of a priori biological laws in biology are the same as those of empirical laws in physics. Thus, the requirement that laws be empirical is idle in connection with how laws operate in science. This result presents a choice between sticking with an unmotivated philosophical requirement and taking the functional equivalence of laws seriously and modifying our philosophical account. I favor the latter.

Multiarm bandit problems have been used to model the selection of competing scientific theories by boundedly rational agents. In this article, I define a variable-arm bandit problem, which allows the set of scientific theories to vary over time. I show that Roth-Erev reinforcement learning, which solves multiarm bandit problems in the limit, cannot solve this problem in a reasonable time. However, social learning via preferential attachment combined with individual reinforcement learning, which discounts the past, does.