The pretheoretical notions of logical consequence and of a logical expression are linked in vague and complex ways to modal and pragmatic intuitions. I offer an introduction to the difficulties that these intuitions create when one attempts to give precise characterizations of those notions. Special attention is given to Tarski’s theories of logical consequence and logical constancy. I note that the Tarskian theory of logical consequence has fared better in the face of the difficulties than the Tarskian theory of logical constancy. Other theories of these notions are explained and criticized.

Recently it has become almost the received wisdom in certain quarters that Kripke models are appropriate only for something like metaphysical modalities, and not for logical modalities. Here the line of thought leading to Kripke models, and reasons why they are no less appropriate for logical than for other modalities, are explained. It is also indicated where the fallacy in the argument leading to the contrary conclusion lies. The lessons learned are then applied to the question of the status of the formula.

Gómez-Torrente’s papers have made important contributions to vindicate Tarski’s model-theoretic account of the logical properties in the face of Etchemendy’s criticisms. However, at some points his vindication depends on interpreting the Tarskian account as purportedly modally deflationary, i.e., as not intended to capture the intuitive modal element in the logical properties, that logical consequence is (epistemic or alethic) necessary truth-preservation. Here it is argued that the views expressed in Tarski’s seminal work do not support this modally deflationary interpretation, even if Tarski himself was sceptical about modalities.

The time-honored view that logic is a non-empirical enterprise is still widely accepted, but it is not always recognized that there are (at least) two distinct ways in which this view can be made precise. One way focuses on the knowledge we can have of logical matters, the other on the nature of the logical consequence relation itself. More specifically; the first way embodies the claim that knowledge of whether the logical consequence relation holds in a particular case is knowledge that can be had a priori (if at all). The second way presupposes a distinction between structural and non-structural properties and relations, and it holds that logical consequence is to be defined exdusively in terms of the former. It is shown that the two ways are not coextensive by giving an example of a logic that is non-empirical in the second way but not in the first.

Tarski’s definition of logical consequence can take different forms when implemented in second order languages, depending on what counts as a model. In the canonical, or standard, version, a model is just an ordinary structure and the (monadic) second-order variables are meant to range over all subsets of its domain. We discuss the dependence of canonical second-order consequence on set theory and raise doubts on the assumption that canonical consequence is a definite relation.

The paper argues that a philosophically informative and mathematically precise characterization is possible by (i) describing a particular proposal for such a characterization, (ii) showing that certain criticisms of this proposal are incorrect, and (iii) discussing the general issue of what a characterization of logical constants aims at achieving.

I offer some brief remarks in reply to comments and criticisms of my earlier work on logical consequence and logical constants. I concentrate on criticisms, especially García-Carpintero’s charge that myviews make no room for modal intuitions about logical consequence, and Sher’s attempted rebuttal of my critique of her theory of logical constants. I show that García-Carpintero’s charge is based on misunderstandings, and that Sher’s attempted rebuttal actually reveals new problems for her theory.

One of the most disturbing problems of social decision making and indeed quite difficult to resolve is the need to reconcile individual rationality with that of society. While individual rationalizing rules indicateways to maximize benefits without any restriction, the collective point of view tilts toward the limitation of individual maximization. This is the very core of the so-called Prisoner’s Dilemma which is but a formalway of saying that there is good reason for every individual to defect on a bargain; the ‘good reason’, that generally takes the gentler form of a self-exception, is that if a player defects and his opponent does not, then the former profits. The purpose of this study is to show the highly erosive character of this self-exceptions and also to make evident its perverse normative power.