green numbers give full details.     |    back to list of philosophers     |     unexpand these ideas

Ideas of Wesley Salmon, by Text

[American, 1925 - 2001, Professor at the University of Pittsburgh.]

1970 Statistical Explanation
p.2 An explanation is a table of statistical information
     Full Idea: On Salmon's statistical relevance view, an explanation is a table of statistical information.
     From: report of Wesley Salmon (Statistical Explanation [1970]) by Michael Strevens - No Understanding without Explanation 1
     A reaction: [He cites W.Salmon 1970] When put like that the view sounds incredibly implausible, but maybe a reading of Salmon would improve the case for it.
1980 Causality: Production and Propagation
§2 p.155 Instead of localised events, I take enduring and extended processes as basic to causation
     Full Idea: I propose to approach causality by taking processes rather than events as basic entities. Events are relatively localised in space and time, while processes have much greater temporal duration, and, in many cases, much greater spatial extent.
     From: Wesley Salmon (Causality: Production and Propagation [1980], §2)
     A reaction: This strikes me as an incredibly promising proposal, not just in our understanding of causation, but for our general metaphysics and understanding of nature. See Idea 4931, for example. Vague events and processes blend into one another.
§4 p.164 A causal interaction is when two processes intersect, and correlated modifications persist afterwards
     Full Idea: When two processes intersect, and they undergo correlated modifications which persist after the intersection, I shall say that the intersection is a causal interaction. I take this as a fundamental causal concept.
     From: Wesley Salmon (Causality: Production and Propagation [1980], §4)
     A reaction: There may be a problem individuating processes, just as there is for events. I like this approach to causation, which is ontologically sparse, and fits in with the scientific worldview. Change of properties sounds precise, but isn't. Stick to processes.
§8 p.170 Cause must come first in propagations of causal interactions, but interactions are simultaneous
     Full Idea: In a typical cause-effect situation (a 'propagation') cause must precede effect, for propagation over a finite time interval is an essential feature. In an 'interaction', an intersection of processes resulting in change, we have simultaneity.
     From: Wesley Salmon (Causality: Production and Propagation [1980], §8)
     A reaction: This takes the direction of time as axiomatic, and quite right too. Salmon isn't addressing the real difficulty, though, which is that the resultant laws are usually held to be time-reversible, which is a bit of a puzzle.
1980 Probabilistic Causality
p.137 p.137 Probabilistic causal concepts are widely used in everyday life and in science
     Full Idea: Probabilistic causal concepts are used in innumerable contexts of everyday life and science. ...In causes of cancer, road accidents, or food poisoning, for example.
     From: Wesley Salmon (Probabilistic Causality [1980], p.137)
     A reaction: [Second half compresses his examples] This strikes me as rather a weak point. No one ever thought that a particular road accident was actually caused by the high probability of it at a particular location. Causes are in the mechanisms.
1984 Causal Connections
p.111 Salmon says processes rather than events should be basic in a theory of physical causation
     Full Idea: Salmon argues that processes rather than events should be the basic entities in a theory of physical causation.
     From: report of Wesley Salmon (Causal Connections [1984]) by Stathis Psillos - Causation and Explanation §4.2
     A reaction: It increasingly strikes me that the concept of a 'process' ought to be ontologically basic. Edelman says the mind is a process. An 'event' is too loose, and a 'fact' too vague, and heaven knows what Hume meant by an 'object'.
1984 Scientific Explanation and the Causal Structure of the World
p.211 Causation produces productive mechanisms; to understand the world, understand these mechanisms
     Full Idea: Causal processes, causal interactions, and causal laws provide the mechanisms by which the world works; to understand why certain things happen, we need to see how they are produced by these mechanisms.
     From: Wesley Salmon (Scientific Explanation and the Causal Structure of the World [1984]), quoted by David-Hillel Ruben - Explaining Explanation Ch 7
     A reaction: I don't think I've ever found a better quotation on explanation. That strikes me as correct, and (basically) there is nothing more to be said. I'm not sure about the 'laws'. This is later Wesley Salmon.
p.379 Salmon's interaction mechanisms needn't be regular, or involving any systems
     Full Idea: While Salmon's mechanisms are processes involving interactions, the interactions are not necessarily regular, and they do not involve the operation of systems.
     From: comment on Wesley Salmon (Scientific Explanation and the Causal Structure of the World [1984]) by Stuart Glennan - Mechanisms 'hierarchical'
     A reaction: This is why modern mechanistic philosophy only began in 2000, despite Wesley Salmon's championing of the roughly mechanistic approach.
1989 Four Decades of Scientific Explanation
Intro p.3 It is knowing 'why' that gives scientific understanding, not knowing 'that'
     Full Idea: Knowledge 'that' is descriptive, and knowledge 'why' is explanatory, and it is the latter that provides scientific understanding of our world.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], Intro)
     A reaction: I agree, but of course, knowing 'why' may require a lot of knowing 'that'. People with extensive knowledge 'that' things are so tend to understand why something happens more readily than the rest of us ignoramuses.
Pref p.-3 Explanation at the quantum level will probably be by entirely new mechanisms
     Full Idea: My basic feeling about explanation in the quantum realm is that it will involve mechanisms, but mechanisms that are quite different from those that seem to work in the macrocosm.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], Pref)
     A reaction: Since I take most explanation to be by mechanisms (or some abstraction analogous to mechanisms), then I think this is probably right (rather than being by new 'laws').
Pref p.-3 Scientific explanation is not reducing the unfamiliar to the familiar
     Full Idea: I reject the view that scientific explanation involves reduction of the unfamiliar to the familiar.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], Pref)
     A reaction: Aristotle sometimes seems to imply this account of explanation, and I would have to agree with Salmon's view of it. Aristotle is also, though, aware of real explanations, definitions and essences. People are 'familiar' with some peculiar things.
1.1 p.14 We must distinguish true laws because they (unlike accidental generalizations) explain things
     Full Idea: The problem is to distinguish between laws and accidental generalizations, for laws have explanatory force while accidental generalizations, even if they are true, do not.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 1.1)
     A reaction: [He is discussing Hempel and Oppenheim 1948] This seems obviously right, but I can only make sense of the explanatory power if we have identified the mechanism which requires the generalisation to continue in future cases.
1.1 p.24 Deductive-nomological explanations will predict, and their predictions will explain
     Full Idea: The deductive-nomological view has an explanation/prediction symmetry thesis - that a correct explanation could be a scientific prediction, and that any deductive prediction could serve as a deductive-nomological explanation.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 1.1)
     A reaction: Of course, not all predictions will explain, or vice versa. Weird regularities become predictable but remain baffling. Good explanations may be of unrepeatable events. It is the 'law' in the account that ties the two ends together.
1.1 p.24 The 'inferential' conception is that all scientific explanations are arguments
     Full Idea: The 'inferential' conception of scientific explanation is the thesis that all legitimate scientific explanations are arguments of one sort or another.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 1.1)
     A reaction: This seems to imply that someone has to be persuaded of something, and hence seems a rather too pragmatic view. I presume an explanation might be no more than dumbly pointing at conclusive evidence of a cause. Man with smoking gun.
2.2 p.45 A law is not enough for explanation - we need information about what makes a difference
     Full Idea: To provide an adequate explanation of any given fact, we need to provide information that is relevant to the occurrence of that fact - information that makes a difference to its occurrence. It is not enough to subsume it under a general law.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 2.2)
     A reaction: [He cites Bromberger for this idea] Salmon is identifying this idea as the beginnings of trouble for the covering-law account of explanation, and it sounds exactly right.
2.3 p.49 Correlations can provide predictions, but only causes can give explanations
     Full Idea: Various kinds of correlations exist that provide excellent bases for prediction, but because no suitable causal relations exist (or are known), these correlations do not furnish explanation.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 2.3)
     A reaction: There may be problem cases for the claim that all explanations are causal, but I certainly think that this idea is essentially right. Prediction can come from induction, but inductions may be true and yet baffling.
2.4.2 p.55 Good induction needs 'total evidence' - the absence at the time of any undermining evidence
     Full Idea: Inductive logicians have a 'requirement of total evidence': induction is strong if 1) it has true premises, 2) it has correct inductive form, and 3) no additional evidence that would change the degree of support is available at the time.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 2.4.2)
     A reaction: The evidence might be very close at hand, but not quite 'available' to the person doing the induction.
2.5 p.59 Statistical explanation needs relevance, not high probability
     Full Idea: Statistical relevance, not high probability, is the key desideratum in statistical explanation.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 2.5)
     A reaction: I suspect that this is because the explanation will not ultimately be probabilistic at all, but mechanical and causal. Hence the link is what counts, which is the relevance. He notes that relevance needs two values instead of one high value.
3.2 p.74 Think of probabilities in terms of propensities rather than frequencies
     Full Idea: Perhaps we should think of probabilities in terms of propensities rather than frequencies.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.2)
     A reaction: [He cites Coffa 1974 for this] I find this suggestion very appealing, as it connects up with dispositions and powers, which I take to be the building blocks of all explanation. It is, of course, easier to render frequencies numerically.
3.2 p.78 Why-questions can seek evidence as well as explanation
     Full Idea: There are evidence-seeking why-questions, as well as explanation-seeking why-questions.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.2)
     A reaction: Surely we would all prefer an explanation to mere evidence? It seems to me that they are all explanation-seeking, but that we are grateful for some evidence when no full explanation is available. Explanation renders evidence otiose.
3.2 p.86 Ontic explanations can be facts, or reports of facts
     Full Idea: Proponents of the ontic conception of explanation can say that explanations exist in the world as facts, or that they are reports of such facts (as opposed to the view of explanations as arguments, or as speech acts).
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.2)
     A reaction: [compressed] I am strongly drawn to the ontic approach, but not sure whether we want facts, or reports of them. The facts are the causal nexus, but which parts of the nexus provide the main aspect of explanation? I'll vote for reports, for now.
3.6 p.103 Flagpoles explain shadows, and not vice versa, because of temporal ordering
     Full Idea: The height of the flagpole explains the length of the shadow because the interaction between the sunlight and the flagpole occurs before the interaction between the sunlight and the ground.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.6)
     A reaction: [Bromberger produced the flagpole example] This seems to be correct, and would apply to all physical cases, but there may still be cases of explanation which are not causal (in mathematics, for example).
3.6 p.103 Can events whose probabilities are low be explained?
     Full Idea: Can events whose probabilities are low be explained?
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.6)
     A reaction: I take this to be one of the reasons why explanation must ultimately reside at the level of individual objects and events, rather than residing with generalisations and laws.
3.8 p.114 Does an item have a function the first time it occurs?
     Full Idea: In functional explanation, there is a disagreement over whether an item has a function the first time it occurs.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 3.8)
     A reaction: This question arises particularly in evolutionary contexts, and would obviously not generally arise in the case of human artefacts.
4.1 p.121 Explanations reveal the mechanisms which produce the facts
     Full Idea: I favour an ontic conception of explanation, that explanations reveal the mechanisms, causal or other, that produce the facts we are trying to explain.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 4.1)
     A reaction: [He also cites Coffa and Peter Railton] A structure may explain, and only be supported by causal powers, but it doesn't seem to be the causal powers that do the explaining. Is a peg fitting a hole explained causally?
4.1 p.121 The three basic conceptions of scientific explanation are modal, epistemic, and ontic
     Full Idea: There are three basic conceptions of scientific explanation - modal, epistemic, and ontic - which can be discerned in Aristotle, and that have persisted down the ages.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 4.1)
4.3 p.127 Understanding is an extremely vague concept
     Full Idea: Understanding is an extremely vague concept.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 4.3)
     A reaction: True, I suppose, but we usually recognise understanding when we encounter it, and everybody has a pretty clear notion of an 'increase' in understanding. I suspect that the concept is perfectly clear, but we lack any scale for measuring it.
4.3 p.132 For the instrumentalists there are no scientific explanations
     Full Idea: There is a centuries-old philosophical tradition, sometimes referred to by the name of 'instrumentalism', that has denied the claim that science has explanatory power. For the instrumentalists there are no scientific explanations.
     From: Wesley Salmon (Four Decades of Scientific Explanation [1989], 4.3)
     A reaction: [He quotes Coffa] Presumably it is just a matter of matching the world to the readings on the instruments, aiming at van Fraassen's 'empirical adequacy'. If there are no scientific explanations, does that mean that there are no explanations at all? Daft!
1998 Causality and Explanation
3.1 p.7 Salmon's mechanisms are processes and interactions, involving marks, or conserved quantities
     Full Idea: For Salmon mechanisms are composed of processes and interactions. The interactions are identified in terms of transmitted marks and statistical relations, or (more recently) exchanges of conserved quantities.
     From: report of Wesley Salmon (Causality and Explanation [1998], 3.1) by Machamer,P/Darden,L/Craver,C - Thinking About Mechanisms 3.1
     A reaction: They say that Salmon has too little to say about the activities that constitute a mechanism. A 'mark' doesn't sound too promising, but I quite like the exchange of conserved quantities, which gets into the guts of what is going on.