a philosophy of science primer - part II
Continued from part I…
The Problems With Logical Empiricism
The programme proposed by the logical empiricists, namely that science is built of logical statements resting on an empirical foundation, faces central difficulties. To summarize:
- it turns out that it is not possible to construct pure formal concepts that solely reflect empirical facts without anticipating a theoretical framework,
- how does one link theoretical concepts (electrons, utility functions in economics, inflational cosmology, Higgs bosons,…) to experiential notions?
- how to distinguish science from pseudo-science?
Now this may appear a little technical and not very interesting or fundamental to people outside the field of the philosophy of science, but it gets worse:
- inductive reasoning is invalid from a formal logical point of view!
- causality defies standard logic!
This is big news. So, just because I have witnessed the sun going up everyday of my life (single observations), I cannot say it will go up tomorrow (general law). Observation alone does not suffice, you need a theory. But the whole idea here is that the theory should come from observation. This leads to the dead end of circular reasoning.
But surely causality is undisputable? Well, apart from the problems coming from logic itself, there are extreme examples to be found in modern physics which undermine the common sense notion of a causal reality: quantum nonlocality, delayed choice experiment.
But challenges often inspire people, so the story continues…
Critical Rationalism
OK, so the logical empiricists faced problems. Can’t these be fixed? The critical rationalists belied so. A crucial influence came from René Descartes’ and Gottfried Leibniz’ rationalism: knowledge can have aspects that do not stem from experience, i.e., there is an immanent reality to the mind.
The term critical refers to the fact, that insights gained by pure thought cannot be strictly justified but only critically tested with experience. Ultimate justifications lead to the so called Münchhausen trilemma, i.e., one of the following:
- an infinite regress of justifications,
- circular reasoning,
- dogmatic termination of reasoning.
The most influential proponent of critical rationalism was Karl Popper. His central claims were in essence
- use deductive reasoning instead of induction,
- theories can never be verified, only falsified.
Although there are similarities with logical empiricism (empirical basis, science is a set of theoretical constructs), the idea is that theories are simply invented by the mind and are temporarily accepted until they can be falsified. The progression of science is hence seen as evolutionary process rather than a linear accumulation of knowledge.
Sounds good, so what went wrong with this ansatz?
The Problems With Critical Rationalism
In a nutshell:
- basic formal concepts cannot be derived from experience without induction; how can they be shown to be true?
- deduction turns out to be just as tricky as induction,
- what parts of a theory need to be discarded once it is falsified?
To see where deduction breaks down, a nice story by Lewis Carroll (the mathematician who wrote the Alice in Wonderland stories): What the tortoise Said to Achilles.
If deduction goes down the drain as well, not much is left to ground science on notions of logic, rationality and objectivity. Which is rather unexpected of an enterprise that in itself works amazingly well employing just these concepts.
Explanations in Science
And it gets worse. Inquiries into the nature of scientific explanation reveal further problems. It is based on Carl Hempel’s and Paul Oppenheim’s formalisation of scientific inquiry in natural language. Two basic schemes are identified: deductive-nomological and inductive-statistical explanations. The idea is to show that what is being explained (the explanandum) is to be expected on the grounds of these two types of explanations.
The first tries to explain things deductively in terms of regularities and exact laws (nomological). The second uses statistical hypotheses and explains individual observations inductively. Albeit very formal, this inquiry into scientific inquiry is very straightforward and commonsensical.
Again, the programme fails:
- can’t explain singular causal events,
- asymmetric (a change in the air pressure explains the readings on a barometer, however, the barometer doesn’t explain why the air pressure changed),
- many explanations are irrelevant,
- as seen before, inductive and deductive logic is controversial,
- how to employ probability theory in the explanation?
So what next? What are the consequences of these unexpected and spectacular failings of the most simplest premises one would wish science to be grounded on (logic, empiricism, causality, common sense, rationality, …)?
The discussion is ongoing and isn’t expected to be resolved soon. See part III…
carol, causality, circula reasoning, critical rationalism, deduction, deductive nomological, delayed choice experiment, descartes, experience, explanations, falsification, hempel, induction, inductive statistical, leibniz, logical empiricism, nonlocality, oppenheim, philosophy of science, problems, pseudo science, science, science overview, the big picture, theory, Thisnthat, trilemma, verification