Sven Ove Hansson

A number of seminal papers on the logic of belief change by Alchourrön, Gärden-fors, and Makinson have given rise to what is now known as the AGM paradigm. The present discussion note is a response to Neil Tennant's [1994], which aims at a critical appraisal of the AGM approach and the introduction of an alternative approach. We show that important parts of Tennants's critical remarks are based on misunderstandings or on lack of information. In the course of doing this, we attend to some central philosophical issues in the theory of belief change, such as the choice of a representation for belief states and the meaning of an idealized rational agent.

An operator of contraction for a belief set (a theory) can be obtained by assigning to it a belief base and an operator of partial meet contraction for that base. It is argued that closure of the base under disjunction is an intuitively reasonable condition. Axiomatic characterizations are given of the contractions of belief sets that can be generated by (various types of) partial meet contraction on disjunctively closed bases. The corresponding revision operators are also characterized. Finally, some results are reported on operations on bases that are closed under material implication.

Five examples are given of major philosophical discussions in which technology needs to be taken into account. In the philosophy of science, the notion of mechanism has a central role. It has a technological origin, and its interpretation has links to technology. In the philosophy of mind, a series of technological analogues have had a deep influence on our understanding of human cognition: automata and watches, telegraphy and telephony, and most recently computers. The discussion on free will largely concerns, in Locke’s words, whether we can “put morality and mechanism together.” Notions of computation and automata that have been abstracted from the behavior of technological devices are key concepts both in logic and in the philosophy of mathematics. Finally, bioethics is largely concerned with the ethical issues that new technologies give rise to in healthcare. As these examples show, there is no lack of technology-related subject matter in philosophy, but there is a lack of sustained attention to it.

A definition of pseudoscience is proposed, according to which a statement is pseudoscientific if and only if it (1) pertains to an issue within the domains of science, (2) is not epistemically warranted, and (3) is part of a doctrine whose major proponents try to create the impression that it is epistemically warranted. This approach has the advantage of separating the definition of pseudoscience from the justification of the claim that science represents the most epistemically warranted statements. The definition is used to explain why proponents of widely divergent criteria for the demarcation between science and pseudoscience tend to be in almost complete agreement on the particular demarcations that should presumably be based on these general criteria

A basic framework for studies of changes in preference is introduced, and four types of changes in preference are identified.Revision by a sentence such as ‘A is better thanB’ means that a preference forA overB is acquired. The result ofcontraction by ‘A is better thanB’ is that the subject no longer holdsA to be better thanB. Inaddition andsubtraction, an alternative is added to, or subtracted from, respectively, the set of alternatives that are under consideration. Formal models of these four types of change are introduced, and they are shown to satisfy plausible postulates for rational changes in preferences.

A descriptor is a set of sentences that are truth-functional combinations of expressions of the form \ , where \ is a metalinguistic belief predicate and p a sentence in the object language in which beliefs are expressed. Descriptor revision ) is an operation of belief change that takes us from a belief set K to a new belief set \ where \ is a descriptor representing the success condition. Previously studied operations of belief change are special cases of descriptor revision, hence sentential revision can be represented as \ , contraction as \ , multiple contraction as \ , replacement as \ , etc. General models of descriptor revision are constructed and axiomatically characterized. The common selection mechanisms of AGM style belief change cannot be used, but they can be replaced by choice functions operating directly on the set of potential outcomes . The restrictions of this construction to sentential revision ) and sentential contraction give rise to operations with plausible properties that are also studied in some some detail

Research is subject to more stringent ethical requirements than most other human activities, and a procedure that is otherwise allowed may be forbidden in research. Hence, risk-taking is more restricted in scientific research than in most non-research contexts, and privacy is better protected in scientific questionnaires than in marketing surveys. Potential arguments for this difference are scrutinized. The case in its favour appears to be weak. A stronger case can be made in favour of a difference in the opposite direction: If perilous or otherwise problematic activities have to be performed it is usually better to perform them in a research context where they are properly evaluated so that guidance is obtained for the future. However, retreating from current ethical demands on research is not a desirable direction to go. Instead, research ethics can serve to inspire the introduction of more stringent ethical principles in other social sectors

Rationality postulates for preferences are developed from two basic decision theoretic principles, namely: (1) the logic of preference is determined by paradigmatic cases in which preferences are choice-guiding, and (2) excessive comparison costs should be avoided. It is shown how the logical requirements on preferences depend on the structure of comparison costs. The preference postulates necessary for choice guidance in a single decision problem are much weaker than completeness and transitivity. Stronger postulates, such as completeness and transitivity, can be derived under the further assumption that the original preference relation should also be capable of guiding choice after any restriction of the original set of alternatives.

A general theory of coherence is proposed, in which systemic and relational coherence are shown to be interdefinable. When this theory is applied to sets of sentences, it turns out that logical closure obscures the distinctions that are needed for a meaningful analysis of coherence. It is concluded that references to “all beliefs” in coherentist phrases such as “all beliefs support each other” have to be modified so that merely derived beliefs are excluded. Therefore, in order to avoid absurd conclusions, coherentists have to accept a weak version of epistemic priority, that sorts out merely derived beliefs. Furthermore, it is shown that in belief revision theory, coherence cannot be adequately represented by logical closure, but has to be represented separately.