This paper argues that if a person's beliefs are idealized as a set of sentences (theoretical, observational, and mixed) then the device of Ramsey sentences provides a treatment, of the mind-brain problem, that has at least four noteworthy characteristics. First, sentences asserting correlations between one's own brain state and one's own “private” experiences are, on such treatment, reconstrued as neither causal, coreferential, nor as meaning postulates, but as clauses in an overall hypothesis (Ramsey sentence) whose only nonlogical constants have “private” meanings. Second, sentences asserting psycho-physical correlations in general, or in other individuals, remain theoretical and susceptible to scientific reduction, though not prejudged to be so. Third, communication between persons having mutually exclusive, “private,” observational vocabularies can be made intelligible. Fourth, it becomes possible in principle (though not necessary) that the world could ultimately be given a total description in a scientific language, Lφ, without mental primitives. However, such a language would be susceptible to interpretation by a given individual using a metalanguage whose primitives (not among those of Lφ) were privately observational to that individual. His procedure in accomplishing such an interpretation would be to construct a Tarskian truth definition in a metalanguage MLφ and then to incorporate it, by Ramsifying its descriptive terms, in his own overall Ramsey sentence. The physical language, Lφ, while not containing a person's primitives, would, of course, be able to define their physical correlates, thus providing a certain sort of mutual mirroring, and a certain sort of relativity.

It is argued that the main problem with “the problem of the direction of time” is to figure out what the problem is or is supposed to be. Towards this end, an attempt is made to disentangle and to classify some of the many issues which have been discussed under the label of ‘the direction of time’. Secondly, some technical apparatus is introduced in the hope of producing a sharper formulation of the issues than they have received in the philosophical literature. Finally, some tentative suggestions about the central issues are offered. In particuar, it is suggested that entropy and irreversibility are much less crucial to the central issues than most philosophers would have us believe. This suggestion is not made because of any firm conviction of its correctness but rather because it helps to focus the discussion on some basic but long neglected assumptions which underlie traditional approaches.

A brief discussion is offered of what it means to say that a set of statements provides D-N explanation with special emphasis given to approximative D-N explanation. An idealized theory is seen to provide approximative D-N explanation. An ideal case provides explanation only if postulates are offered which connect the ideal antecedent condition with actual conditions. Such postulates will help in accounting for deviations between what the consequent of the ideal case entails and what actually occurs. Three ways are presented for deciding on the acceptability of idealizations: (i) as actual conditions more closely approximate the ideal antecedent conditions specified in an ideal case, the more closely actual cases ought to approximate the consequent of the ideal case; (ii) the additional postulates connecting ideal antecedent conditions with actual conditions ought to account for deviations between what is entailed by an idealized theory and what actually occurs ; and (iii) an idealized theory is explained by a more general theory when certain state variables in the more general theory are given extreme values. It is argued that (iii) is not a necessary condition for the acceptability of idealizations.

This paper takes the first steps in the construction of a theory of event identity as that theory applies to historical sentences. The theory is extensional throughout. Following statements of criteria of adequacy for the construction, Davidson's method of regimenting sentences is adopted in order to allow for variables ranging over events. Events in this theory are only partially construed, that is, to the extent of treating them as concrete individuals rather than as classes or repeatable universals. The paper concludes with a statement of several theorems and definitions and an example of how the theory works.

An inductive method Cλ in the λ-system of Carnap [1] is immodest, on evidence e, iff its estimate, on e, of its own accuracy is higher than its estimate, on e, of the accuracy of any rival method Cλ′. Immodesty seems to be a condition of stable trust: if you trusted a modest Cλ, you should start by trusting its advice to replace it by a rival that it estimates to be more accurate. One might guess that any Cλ would be immodest on any evidence. But in [2] I proved that, under a certain accuracy measure taken from Carnap [1], §§ 20–21, there would be exactly one immodest Cλ. Unfortunately, that one sometimes turns out to be C0 (the straight rule); and since nobody in his right mind would trust C0 we are then left with no acceptable Cλ. Stephen Spielman [3] has proposed a remedy: an estimate of accuracy, on evidence e, should disregard accuracy under circumstances that are ruled out by e. Spielman proves that if this change is made, any Cλ is immodest on any evidence.

A large number of definitions of the concept “model” have been given by various authors in recent years. Thirty-seven definitions are listed by A. I. Uyemov in a recent monograph (see [9]). This list is somewhat one-sided since it contains a disproportionate number of references to the work of Soviet authors. However, most of the important definitions given by Western writers are included. I shall give three definitions, all of great generality, so that various types of models, replicas, maps, theories and, most importantly, cybernetic models, are included. The advantage of this high degree of generality, or all-inclusiveness, is that the task of the investigator of uses of models in science, is reduced to classifying the various types of models, their construction, function, and structure, thus avoiding indecisive arguments (historically important, of course) such as the Duhem–Campbell controversy, so perceptively discussed by Mary Hesse ([2], [3]).