-> Science seeks to explain as well as describe, but empiricists are wary of explanations (e.g. Logical Positivism)
The problem of explanation -> Understanding evidence
The problem of analyzing what it is to have evidence to believe that a scientific theory is true
The problem of analyzing confirmation and evidence (ch 3 & 14)
working out how our theories provide explanations.
The popularity of explanatory inference over inductive suggests a close relation between the problem of analyzing explanation and the problem of analyzing evidence.
Explanatory inference: inference from a set of data to a hypothesis about a structure or process that would explain the data
Inductive inference: Inference from particular cases to generalizations
To reiterate, the Logical Positivists associated explanation with the idea of achieving deep metaphysical insight into the world, and they wanted nothing to do with that idea.
The way Logical Positivists & Logical Empiricists construed explanation in an unassuming way that fit their empiricist picture was the ‘covering law’ theory of explanation.
The Covering Law Theory of Explanation was developed by Carl Hempel and Paul Oppenheim in 1948.
Explanandum: Whatever’s being explained -> “Why X?”
Explanans: The thing doing the explaining -> “Because Y.”
To explain something is to show how to derive it in a logical argument
so a good explanation must also be a good logical argument and the premises must contain at least one statement of a law of nature that contributes to the argument.
Explanandum will be the conclusion of the argument
Explanans will be the premises of the argument
What explanations describe:
particular events
e.g. explaining the fact that the US stock market crashed in 1929 in terms of economic laws operating against the background conditions of the time
general phenomena or regularities
Newton explained Kepler's laws of planetary motions in terms of more basic laws of mechanics plus assumptions about the layout of the solar system.
In both above situations, the CLToE sees them as being expressible as mostly deductively valid arguments with premises and conclusions, although some could also be inductive/non-deductive instead.
If a particular phenomenon can be embedded within an argument whose premises include a law and give high probability on the conclusion, then this yields a good explanation of the phenomenon.
To explain something is to show how to derive it in a logical argument of a kind that makes use of a law in the premises.
To explain something is to show that it's to be expected, to show that it's not surprising given our laws of nature.
So there's no difference between explanation and prediction.
* to predict something is to formulate an argument and argue that it's expected despite uncertainty.
* to explain something is to know that it's already happened, but could have been predicted using an argument containing a law
This was a problem for Logical Empiricists, and everyone else.
A law of nature's supposed to be a kind of basic regularity, a basic pattern in the flow of events.
D-N Theory
D-N Model of explanation
D-N: Deductive-nomological. Only refers to deductive covering law explanations.
-> Main problem = asymmetry problem
-> Most famous illustration of asymmetry problem = case of the flagpole and the shadow
Explanations have a kind of directionality.
* Some arguments can be reversed and remain valid.
* But explanations can't usually be reversed in this way
* So not all good arguments that contain laws are good explanations.
* This objection (asymmetry problem) was given by Sylvain Bromberger in 1966
CLToE sees explanation as extremely similar to prediction – the only difference is what you do and don't know.
* Symptoms can be used to predict but not explain, yet they can be used in a good logical argument along with law to show something's expected.
If you know only that disease D produces symptom S
Then you can make inferences from S to D
In some cases you might be able to predict from D to S
But you can't explain D in terms of S because explanation only runs one way from, D to S, regardless of how many kinds of inferences can be made in the other direction
Good explanations of S can be given in terms of D even if isn't a reliable symptom of D, even if S is not always to be expected when someone has D.
The flagpole case suggests how to build a better theory:
To explain something is to describe what caused it
* a request for an explanation seems to amount to a request for information about what caused the thing in question
-> The whole idea of causation and causal connection is extremely controversial in philosophy and within the empiricist tradition. It's viewed as a suspicious metaphysical concept that should be avoided
In the 1980's, probability theory was used to supplement analyses regarding the idea that explaining something is giving information about how it was caused.
Imagine you have an idealized, 'complete' explanation containing everything in the causal history of the event in question, specified in total detail
People don't want the complete explanation
In any context of discussion in which a request for explanation is made, only some piece or pieces of the complete explanation will be relevant
We're often able to know and describe these relevant pieces of a total causal structure
So to give a good explanation all you need is a description of these relevant pieces of the whole.
-> Was developed after the demise of the CLToE by Michael Friedman and Philip Kitcher as an alternative to a theory of unification in science that's tied to the idea of deriving phenomena from laws.
Unificationist Theory: Explanation in science is a matter of connecting a diverse set of facts by subsuming them under a set of basic patterns and principles.
We try to develop general explanatory schemata (schemes) that can be applied as widely as possible to reduce the number of things we must accept as fundamental.
Kitcher argued that Darwin's theory of evolution and Newton's theory of matter were compelling in their early stages despite not making many new specific predictions because they promsied to explain so much.
Explanatory promise: the ability of a theory to unify a great range of phenomena with a few general principles.
Summary of Two Main Proposals Against CLToE
Causal Theory
Unification Theory
But a variety of different relations can be explanatory.
Whether they were all for absolute causality or absolute unification, philosophers have ultimately agreed that there seems to be a need for both when it comes to a theory of explanation – a pluralism.
-> Contextualism: The standards for good explanation are partially dependent on the scientific context
* The idea of explanation operates differently within different parts of science, and within the same part but at different times.
* So the standards for a god explanation in field A don't necessarily apply to field B.
Kuhn argued something similar when he said that:
different theories/paradigms bring with them their own standards for what counts as a good explanation.
standards about whether a relation counts as 'causal' also depend on paradigms.
Kuhn argued that the idea of explanation will evolve as our ideas about science and about the universe change.
Scientific traditions usually have good reasons for their treatment of the idea of explanation dependent for instance on views of what the world contains.
Technical sense
e.g. mathematical measurements of 'the amount of variation explained'
Rhetorical use
e.g. 'your theory accommodates this result, but doesn't really explain it.'
A low-key application in science
e.g. describing how things work, description of mechanisms and processes
Like Van Fraassen, Godfrey-Smith denies that explanation is a single, special relation common to all of science.
Van Fraassen denied that explanation is something inside science, and denied that scientific reasoning includes the assessment of the explanatory power of theories. -> explanation is something that people do when they take scientific theories and use them to answer questions external to science
Godfrey-Smith contends that explanation is thoroughly internal to science, but variously so.
-> "Inference to the best explanation"
Explanatory Inference: Inference from a set of data to a hypothesis about a structure or process that would explain the data.
Suggests there's only one kind of relationship between data and hypothesis that's involved in explanatory inference.
Godfrey-Smith views explanatory inference as a matter of devising and comparing hypotheses about hidden structures that might be responsible for data.
-> Delving into the intersection of philosophy of science and metaphysics.
What are laws of nature? What is causation?
* both cases involve concepts aimed at picking out a special kind of connection between things in the world.
Recent philosophical attitudes have been to revise rather than reject these concepts.
* Philosophers have tried to analyze both in terms of patterns in the arrangement of things, rather than some extra connection between things -> a.k.a. Humeanism
David Hume -> the first philosopher to develop a really focused suspicion about concepts of connection between events in nature.
A philosopher with Humean views will construe laws of nature as no more than regularities/basic patterns in the arrangement of events. Which leads to a problem...
Law: Something that directs, guides, or governs in some way
Laws are seen as responsible for the regular patterns we see, rather than being identical with those patterns.
* Humean philosophers, like Logical Empiricists, regard this 'governing' conception of laws as something to be avoided
In 1983, Nancy Cartwright argued that what people call 'laws of physics' don't usually describe the behavior of real systems, but of highly idealized fictional systems.
Philosophers are no longer as obsessed with natural laws as the goal of scientific theorizing.
Genuine science doesn't have to contain hypothesized laws with ideas organized via that concept.