If you would like to download a pdf version of this essay, click here.

Abstract

   An infection has, for millennia, sapped the strength of a group of susceptible people, called philosophers. With Hume as a carrier, it reappeared in the west a few centuries ago. Once it has taken hold, the sufferer may go into remission but - like malaria - seems never to be entirely cured. The first exposure is a haunting, experience (Note1) (These numbers refer to footnotes). In recent times the symptoms have become so widespread that displaying them reliably identifies a philosopher. Ordinary people, including scientists, are immune; they find the symptoms puzzling.
   This incurable infection is sceptical doubt . It is time that we quarantine it. All attempts to cure it should henceforth take place in a restricted area. Philosophers outside this area should agree that discussion of attempts to cure it, and, in particular, reference to the inability of others to shake it off , is banned. In order to remember that a cure for the disease has not been found, they should agree that all the relevant key words are conditionalised as /IP, meaning 'given the inductive presupposition'. Any word so marked, such as 'justification/IP', 'support/IP', 'reason/IP', 'evidence/IP', is weakened by the disease, and will remain so until the worker in the quarantine area find a cure.
   Philosophers have noticed that those people who are immune to the infection, are nonetheless unwittingly using the same weakened ideas. Being unaware of the infection, such people regard these ideas as strong and healthy. By quarantining, by making a conscious effort to repress what they suspect, philosophers can rejoin the rest of humanity.
   This, therefore, is an essay on the Methodology of Philosophy of Physics . We propose that attempts to solve the problems raised by sceptical doubt should be isolated from attempts to solve other problems - such that people trying to solve the latter problems are encouraged, having identified the fundamental assumptions that a sceptic might doubt, to accept them as working assumptions - leaving the justification to be worked on independently.
   My aim is to mark off a group of problems in the philosophy of science as potentially insoluble.
   The theme of this essay is not original. To emphasise this, I have linked it to a collection of quotations. Nonetheless, its importance justifies, I hope, a new treatment.
 

   The philosophy of science is characterised by a long-standing problem:

(i) methods M are judged by contemporary physicists to be good ways of discovering the truth about nature.  They have been used for centuries.

But:

(ii) these methods M seem to be completely unjustified.

   How is it possible that an activity widely regarded as the epitome of rationality could depend on completely unjustified methods?

   Our meta-philosophical problem is: "Why is this problem so recalcitrant? Is there any strategy for either solving, or sidestepping, it?"

   We can remind ourself of what is at stake in this problem by quoting a passage from a non-philosophical book about Early Modern Europe George Clark p.160 "At the present time the great majority of those people, all over the world, who think about such matters, believe that among our different kinds of knowledge and ideas about ourselves and the universe we live in, the most reliable are those which we derive from science.  There are so many shades and varieties of this belief that it may be understood in a hundred different ways; but most thinking people, and many unthinking people, agree that science is exact, impersonal, and positive, not distorted by bias, not speculative, in short as authentic and as nearly certain as any acquisition of the human mind can be.  If they were asked to define what they mean by science they would not agree on every point; but they think of it as following the method and spirit that are most familiar in the 'natural' or 'physical' sciences, that is, as based on observation, on experiment, and on the testing of hypotheses by strict reasoning, very often of a mathematical character.  They think of it as progressing from discovery to discovery; as using, so far as it can, all relevant data; as capable of being planned ahead, and as resulting in power to control material objects and living beings, in a conquest of nature.
   This state of mind is so familiar that millions of people have grown up in it and take it for granted; but it has not always been so.  In the Middle Ages..."
   Is this state of mind a delusion?  Do sceptical arguments force us to accept that there is actually no difference in rationality between the methods of science and those of a previous age?
 

   All humans use our experiencing to induce, abduce, infer, events that we have not experienced. For example, on the basis of our limited experience - limited in space, time, and circumstance, because of our nature - we make claims about the distant past and the future; we generalise to all times, places, circumstances.
   We suppose that some people specially value, as an aim, {Truth }. In particular this aim, as the priority, characterises the investigation called 'physics'. We will therefore consider the uses of indirect evidence, inducing and abducing, to seek true general claims. This is important, because it implies that the only relevant justifications of these methods (called 'justification/IP' below) are those which explain why they help to lead us towards the truth.
   We propose that particular examples of induction and abduction are unified by certain general methods. These are:
M_g: 'Patterns in experienced events also exist in unexperienced events, until you have evidence to the contrary (Note 2).
M_p: 'True theories make experienced events as high-chance as possible '.
These two methods could also be expressed as the theory or presupposition T_p: "Our available evidence is a fair sample; it is not fixed or biassed so as to be misleading. If we generalise from our experience, we will always go less wrong than if we just guess. We can use our experience as a guide to unexperienced events ."

   Devising an exact statement of these assumptions is very hard.
   All common-sense human and animal - even plant - behaviour is based on T_p. (This is not agreed by everyone eg. Watkins (1984?) seems to claim that there is no natural inductive presumption)
   T_p is not usually made explicit; nor are people usually aware of it. It is of special interest only to philosophers, having been noted thousands of years ago.
   It is essential to the justification of our actions and beliefs. Without it we cannot justify any behaviour by reference to evidence; all discrimination between beliefs, and scales of reasonableness, become unjustifiable.
   Yet sceptical arguments, well-known for two thousand years, seem to lead to the conclusion that M_g and M_p are unjustified; our experience may be misleading us; our generalisations may be false; improbability may not guide us to truth. The best efforts of philosophers have failed to find a justification. Perhaps there is none (Note 3).
   Our usual way of justifying claims as true, is to use evidence . We regard intuition, self-evidence, feeling, authority, voting, as inferior (Note 4). But we cannot use indirect evidence to justify the claim T_p, since it specifically provides the justification for using indirect evidence to justify claims. This is a seemingly fatal problem.
   But how else can claims about the most reasonable methods for investigating the world, be justified? No strong solution to the problem has yet been found.

   Our primary concern in this essay is to establish the value of separating the problem of justifying M_g and M_p, in all its manifestations, from other problems whose solution is independent. Therefore we are not here concerned to review every possible justification for them. If they are justified, excellent. We merely comment on one or two suggestions.

   One solution (Peter Lipton; David Papineau Naturalistic Philosophy) uses the evidence of success of inductions in various circumstances (eg. in the past, on the Earth) as inductive evidence for the general method of induction. They argue that this is rule-circular, rather than premise-circular, and that it provides a kind of justification - though only for people who already make inductions; it will not provide reason to persuade someone who is a neutral thinker.
   Another weak possible solution is:- "We might as well use them. They are the best that we can do in the desperate situation in which we find ourselves, because if we don't assume them, we are definitely doomed".
   Another weak solution rationally justifies preference for these methods because we want them to work. This argument works on the idea of a hierarchy of aims, with Truth as the priority but with, say, Intelligibility as a subsidiary aim. When the guidance of Truth underdetermines our decisions, we are justified in using a subsidiary aim to guide us (Note 5). This solution requires that we distinguish three inductive problems:
(Pi) Why is it reasonable to believe that inductive methods guide us towards the truth?
(Pii) Why is it reasonable to rely in practice on inductive methods?
(Piii) Why is it reasonable to prefer to invent , and write down, hypotheses using inductive methods?
   It does not show that the inductive method is a more promising way of discovering the truth than alternatives (Pi); it gives no reason to believe that generalisations formed on this basis are any more likely to be true than random guesswork. It therefore provides no reason for relying on this method in practice (Pii). It does solve (Piii), but so what? By providing no solution to the first two problems, it turns physics, and life, into a theoretical game; (cp. for example, Putnam's criticism of Popper's methodology).
   Furthermore, this solution does not fit with the attitudes of physicists, and other people.    Everyone, except philosophers qua philosophers, assumes that the first two problems are solved. For them, therefore, the third problem does not need a separate solution; of course they invent, and write down hypotheses using inductive methods - "Those are the ones most likely to be true, stupid!". They behave as though they believe that inductive methods give them true, reliable, propositions, though, according to this solution, any degree of confidence in this is irrational; all they are justified in saying is "Wouldn't it be nice if the world behaved like this?".
   This solution is therefore only a rational reconstruction of the third problem of induction; although it provides a possible rationale for the physicist's use of the third kind of induction, it is not actually the reason why the physicists use it.  See also weak justification.

   Discussion of the reasonableness of these methods -which will continue - should be confined to quarantine. The problem of their justification is not, of course, solved by so doing.  But we are released from the distraction generated by coming across the problem unexpectedly when tackling other problems, forgetting its apparent insolubility, and wasting time on it when no-one could reasonably criticise us for failing to solve it.
   I am not suggesting that no-one should think about it.  There is a problem concerning the rationality of our investigation of our experiencing. Brave or foolhardy people, haunted by the recurring symptoms of an infection they cannot shake off, will enter the quarantine ward and try to find a cure.  They will work surrounded by the rotting corpses of all those who have tried and failed.

   Assuming fundamental problems away, closing our eyes and pretending that they are not there, seems intellectually disreputable. It is only acceptable in certain circumstances:
(i) We are not pretending that the sceptical problems are solved; we are isolating them from the rest of our endeavours.
(ii) T_p, the anti-sceptical presupposition, must satisfy certain conditions, if those who use it are not to be open to reasonable criticism from those defending alternative activities - alternative ways, perhaps, of seeking truth:
(a) It must not make a specific claim about the world which experience shows to be false
(b) It must be agreed, in word or deed, by all living things on the planet (c) It must do no more than validate learning from experience.
On (a): "Nature is uniform", for example - the theory that all observed patterns can be spatio-temporally generalised - is proved false by relatively direct experience (Note 9).
On (b): We want arguments based on evidence to be convincing to everyone regardless of their specific beliefs. Therefore we have to ensure that T_p is both neutral between all beliefs, and agreed by everyone.
On (c): Even if everyone on the planet believed that the Sun went round the Earth, we must make it possible for Copernicus and Galileo to criticise this belief. Nothing must go into quarantine except the basis for such criticism - the basis for our use of evidence to criticise, to test and generalise.

Descriptive Methodology This is not a worry for the descriptive methodologist. It is only a problem for the complete justificationist whose aim includes showing that physics discovers fully justified truths or near-truths; clearly she cannot accept our cavalier attitude to sceptical doubt. For us, in other words, the lack of justification is interesting but is not a problem; we aim to discover the extent to which physics is justified; here we find a fundamental assumption which is not justified. That is not a crisis, not a problem to be solved, just interesting.

Demarcation: We can, for example, now compare the reasonableness/IP of different approaches to the search for truth, including, for example, fundamentalist religion and science. This is possible because T_p's justification is agreed to be irrelevant in relative terms, when comparing the reasonableness/IP of various human activities.
   Anyone who now criticises another person's area of investigation as irrational, because the other area does not include a justification of T_p, is themselves open to the criticism that they cannot justify it either. And since they are using it every instant of their lives, they are guilty of double standards in objecting to the use of it in another area when they themselves are using it.
   There are grand methods which are not in quarantine: "Only public repeatable evidence should be used in assessing the truth of theories"; "Private revelation can be used in assessing the truth of theories"; "The authority of Holy Books and religious leaders is fundamental to assessing the truth of theories".
   Which of these are more justifiable/IP methods? The question is of practical importance. It is distressing that a thinker like Ernest Gellner is unable to discuss the intellectual merits of these approaches because the foundations of science are in such a muddle - as he says: "surprisingly difficult to justify". It is not only distressing, it is ridiculous. Unjustified inductive decisions we all make every instant of every day are the source of the problem - to which religious believers are not immune. It is a problem endemic to living things, not just to scientists. Yet it saps the will. And it is an illness which ironically has its most visible effects on those who, most earnestly trying to be reasonable, go beyond conditional reasonableness/IP to absolute reasonableness. Certainly the skeleton is in our cupboard; but it is in everyone else's cupboard too. No-one should be ashamed of the skeleton in their cupboard; nor can anyone dare, as any kind of criticism, to point out the skeleton in another group's cupboard (Note 10).
 

   We now spot, capture, and quarantine, some of the various disguised forms of the sceptical infection.

Bayesianism is underpinned/IP We argued above that Bayes' theorem provides no justification for T_p; not surprisingly, a formalism which is designed to capture everyday intuitions cannot then justify the intuitions.
   Rather than probability theory justifying T_p, T_p establishes probability theory's link, beyond human use, to the world; it establishes the right of investigators to use Bayes' theorem to provide truth-credit for claims. To claim T_p is true, is to claim that probability theory can sensibly/IP be applied to our experience of Nature. Without an assumption equivalent to T_p, probability theory retains its importance as a logically structured description of a whole system of human action and beliefs, linked by the use of such everyday words as 'probable' and 'chance'. T_p is the presupposition that probability theory works in the interface between our experiencing and nature; it is the presupposition, equivalently, that, given certain experiences of the world, certain betting quotients, certain beliefs, and actions consequent on them, will turn out to be true.
 

   Bacon's, Mill's, Whewell's, Popper's, and Lakatos', identification of the methods judged by the present consensus of Physicists to be the best for investigating Nature - for seeking the truth - suffer from a familiar problem of justification (Note 11). We do not have to be Bayesians to notice the repetition of the words 'probability' and 'chance' in these methods, especially as described by physicists. Since the justification for each has been stymied only , specifically, by sceptical doubt, every one should from now on be considered justified/IP:
The method of Inductive Generalisation : If we have observed an A which is also a B , we can take this as evidential support of the claim that all A s are Bs , as long as we have no reason to think otherwise. Further observation of A s which are Bs therefore does not increase this support. Diverse evidence becomes significant only when we have other supported claims (background theories) which suggest that an A is not a natural kind - that in certain specified circumstances, A s are likely not to be Bs . This is based on the presumption that, unless atlas are Bs , it is highly improbable that we happened to come across one that was (Note 12). In this way investigators claim that general laws are probably true.
   The methods of Similarity and Difference (eg. Bacon): The observation that every appearance of B has been associated with the appearance of A , and that when A was not present, even if the other factors were, then A was not present, indicates that the two are somehow linked - that the observation is too improbable to be just chance . This is justified/IP.
   The method of Concomitant Variation (eg. Mill): If evidence of the concomitant variation of B when A is varied can acceptably be judged as just chance, a coincidence, an improbable event, then it gives no indication of a causal link. But this kind of improbable event does not happen, so the method is justified.
   The method of "Consilience of Inductions " (eg. Whewell, Mill): the improbability, the low chance of a hypothesis successfully having detailed low-observability consequences in several areas of phenomena is evidence that it contains some truth (is not entirely a human fiction). This cannot be justified, but the improbability of the event, if the hypothesis was entirely false, justifies/IP the claim that it has some truth in it.
   The method of Novel Fact Prediction (eg. Whewell, Popper, and Lakatos): If a theory, or a research programme, makes successful predictions of events which were very improbable on the basis of background theories, then this is indirect evidence that the theory or research programme has some truth in it - is not just a human fiction (Note 13). This method is justified/IP (Note 14). If a research programme fails to have improbable successes , rests on its laurels, adjusts to new evidence, and becomes increasingly complex as a result, then (a) it is not collecting any more truth-credit (b) it may be losing it, if we can justify/IP the use of structural simplicity as a truth-indicator/IP.
   The method of Inference To The Best Explanation (IBE) The idea of this (Peter Lipton (1991) p.1) is that investigators in the present consensus claim truth credit for a hypothesis, such as a low-observability claim, if it provides the best explanation of some available evidence. Let us assume, following Lipton, that it is a descriptive fact that they regard being a good (Note 15) explanation as an independent truth-indicating factor, extra to the indirect evidence methods such as those listed above, which are themselves extra to empirical adequacy (saving the phenomena). To what extent are they justified in using this factor?
If 'best' is taken to mean loveliest, providing most understanding (most unified, most elegant, simplest), we need to assess the extent of justification for claiming that these qualities are "symptoms of likeliness" (p.62-3). If it is zero, explanatoriness is just a truth-irrelevant, pleasant, feature of a theory (pp.125-32 The "Voltaire objection") (Note 16). The only coherent suggestion seems to be because we have evidence that they have tended to be features of theories which have then demonstrated empirical success - either directly, such as when a conjectured entity is later more or less directly observed, or indirectly, where success is demonstrated by one of the above methods (Note 17). This, as Lipton states (pp.125-6 & 187-188), explaining clearly that the destructive force of scepticism in the case is no more disqualification than in all others, is a justification/IP.
   The meta-method of Pessimistic/Optimistic/Not-sure Induction : We are justified/IP in trying to use evidence from the history of physics to support a general claim that, say, claims concerning the existence of very low-observability entities, supported/IP by a certain amount of indirect evidence, have typically turned out badly/well/not-sure.
   The Problem Of Demarcation : If justification/IP is all that we require, we can now set about assessing whether one group of investigators - such as physicists - are more rational/IP than another - such as astrologers. The task is difficult, but it is no longer bedevilled by the destructive complaint that all investigations are unjustified.

   Philosophers of science have distorted perfectly reasonable/IP methods for investigating Nature, as they have struggled to make them immune to sceptical criticism. In the process they have not only ended up with oddly unrealistic theoretical methodologies, but also, by setting up an impossible standard and then failing to reach it, have opened the way to anti-rationalists. Such attempts should be regarded in the way that attempts to create a perpetual-motion machine are regarded by physicists - as almost certainly doomed, and only to be attempted by a few eccentrics .
   Physicists, like most other people, are not interested in philosophical sceptical doubts; physicists are not philosophers. Therefore the kind of philosophy of physics which revolves around these doubts - extensive analysis of simple generalisations, for example - has been found completely uninteresting by physicists.  Philosophers who continue to tackle the above problems in the quarantine ward should stop hoping that physicists, and other ordinary people, will be interested; they never will be.
   By quarantining sceptical doubts, the philosopher has a chance of regaining the interest of physicists, and other people.

   I am not suggesting that T_p is true. I am not suggesting that there is any evidence for T_p, or even that there could be evidence for it. At any moment, events could begin to indicate to us that it is horribly wrong.

   "Hang on!", the reader may think, "Does this mean that all philosophy of science problems in the unquarantined area are dissolved?" Certainly not...

   There are plenty of problems in the philosophy of physics which are not in quarantine:

(a) How do physicists choose which part of the experimental and theoretical chain of deduction is to blame when a prediction goes wrong? And how is the choice justified/IP?
(b) What is the difference between artificial, merely verbal, simple theories and real conceptual simplifications?
(c) How do physicists justify/IP their choice of the basic language in which laws, and theories, are couched?
(d) How do physicists justify/IP preferring explanatory, elegant, structurally simple theories? Why do they prefer mathematically simple laws?
(e) Is physics value-neutral?
(f) Should physics include concepts which are essentially untestable, or should it limit itself to those which lead more or less directly to experiments? (Are, for example, such concepts as 'absolute time', 'the exact position and momentum of an electron', and 'the exact positions and momenta of all the molecules in a box' (Arrow of Time) 'meaningless', or is this mindless positivism masquerading as tough-mindedness?)
(g) Are all physicists realists? Should they be? Could evidence cause us to doubt the possibility of realistic theories of hidden causes of phenomena?
(h) Does science have an aim, or only scientists? Is the philosophy of physics actually the psychology of physicists?
(i) Is there a completely theoretically neutral factual base, provided by experiments, on which theories are then built?
(j) Are some human activities less justifiable/IP than physics - psychoanalysis, religious fundamentalism, astrology, Marxism......?
(k) What happens when the characteristic aims of physicists underdetermine their behaviour?
(l) Does physics proceed by normal science and revolutions? Are successive theories incommensurable?
(m) Can we identify the characteristic aim(s) of physicists? Can we then identify the methods used by physicists? To what extent are these methods justifiable/IP given the aims?
(n) What are the guiding theories at the top of the theoretical structure of physics? Universal causation? Structural simplicity? Local intense causation?
(m) Can philosophers sensibly try to explain the history of physics? Should they? Can a methodologist be normative? What is the relationship between methodology, sociology, and history?

   This incomplete list answers the criticism that quarantining sceptical doubt leaves no worthwhile questions for the Philosopher.

   How are M_p (Probability) and M_g (Generalisation) related?
   M_p includes M_g. If we experience several As which are also Bs, and no As which are not Bs, we can use M_g to propose that all As are Bs. Equally, if all As were Bs then our finding of As which are Bs is a typical event, it is highly probable; if almost all As in the universe, randomly distributed, are not Bs, (and we have no evidence to think that we are in one of special areas of the universe) then our experience is highly improbable. Using M_p we should prefer the possibility that all As are Bs. So both methods lead to the same conclusion in this case.
   M_g could be true without M_p being true . M_p also justifies/IP our preference for theories which have passed severe tests; M_g has nothing to do with theories. For example, we could imagine a universe in which generalisations (laws), believed because our evidence fits with them, are true - reliably the same everywhere in the universe for ever, but theoretical speculations, believed because they have successfully predicted novel facts, are false - all completely on the wrong track. We could imagine living in such a universe, and gradually coming to this conclusion, on the basis of historical evidence of repeated theoretical failure.
So we need M_g and M_p as two distinct methods.

   Not everyone, of course, supports these views; Bas Van Fraassen, for example, seems to cast doubt on T_p (p.35). Understandably anti-realists will not be keen on it, since it is used, as we have seen, as the justification/IP the abduction - the use of indirect evidence to support the suggestion that low-observability claims, including ones concerning the existence of entities, have some truth-credit.

(See also philosophers who have accepted quarantining; the weak justification for the inductive presupposition)

Move on to discussion of the methods used by physicists, and the extent of their justification (assuming that sceptical doubt is quarantined)

Notes:

1 Peter Lipton (1991) p. ix. See also Nick Maxwell (1980). Back
2 "Nature is uniform" is an (even more) unsatisfactory expression of this assumption. Back
3 Perhaps, for example, humans merely have inherited an instinctive generalising instinct, an instinctive expectation that our experience is a fair sample, because it has been useful, or not harmful, to our ancestors' survival. Back
4 Unless we have T_p, we actually do not seem to have any reason for preferring to justify general claims by evidence , rather than by these other methods. This does not imply that we can use other methods, it implies that without T_p we have no way of judging which ways of supporting general claims are better. Back
5 Nick Maxwell's 1980 explains this particularly clearly. Back
6 Making Bayes' theorem one of the axioms, and deriving, say, the usual axiom of conditional probability as a theorem, demonstrates the absence of justification. Back
7 Howson and Urbach (1989) p.296: The choice, the assessment, of prior probabilities, and indeed conditional probabilities, is human matter of judgement. But once done, once claimed as 'probabilities', the probability calculus supplies "quite objective standards of consistency in reasoning involving such probabilities. The probabilities might be personal, but the constraints imposed on them by the condition of consistency are certainly not". p. 1: "According to the Bayesian view, scientific, and indeed much of everyday reasoning, is conducted in probabilistic terms ... (it) gives an insight into the nature of scientific reasoning". However (p.1) "why it is reasonable" is more dangerous, unless they mean only their 'consistency' point. Back
8 Nunan (1993) is cautious on this point. He is trying to include an aspect of novel fact prediction in the calculus. He writes (p.35): "Novelty can be justified in terms of Bayesian confirmation theory", but he also writes (p.21 n.8): "(it) should never be construed as anything other than a formalisation of an intuitively plausible model of how theory evaluation ought to be conducted when it is being done rationally". In other words, the formalism systematises evaluations which are already intuited to be justified ; it does not provide any further justification of its own. The intuition is primary; if a certain kind of novel fact prediction seems intuitively justified, then it is. All the formalism can expose for criticism is inconsistency (p.20). Back
9 However it has been a common code used by people trying to explain the justification for scientific methods; they realised that some kind of general presupposition was needed. Back
10 In my experience, religious believers, for example, are well-prepared to accept the quarantining of sceptical doubt; they realise that otherwise the way is open to the destruction of knowledge, which is in no-one's interests. Back
11 Some of them, such as the idea that low-observability claims of high generality can be reliably discovered by studying tables of recorded observations, simply do not work - scepticism or not. Back
12 The puzzle of 'grue' (Goodman predicates) is can perhaps be solved/IP - in the way Goodman expected, by "embedding". Our property language has been selected partly on the basis of which properties have been found/IP to play successful/IP rôles in the creation of a pattern of causal relationships - eg. hence 'emerald'. 'Green' is a name for a sensed quality (a determinable determinate). Until we have some reason/IP to doubt that this is a quality of things, we retain it in our descriptions and laws. A change at some specified time does not figure in our concept of the property, because we have presupposed that the phenomena in question are time-independent.
The paradox of confirmation (Hempel's Raven) is similarly tackled. Back
13 The argument is that if the theory was just a human fiction, then the probability that it would have anticipated nature - had correct consequences for which it was not designed - is negligible. Therefore this claim , that the theory is a human fiction, is unlikely to be true. Back
14 If the truth of the consequence is not known to the deviser of the theory, or indeed to anyone, (because, say, its testing is in the future), then its correct prediction is improbable - if the theory is entirely false. But it is almost equally improbable that many known consequences can be drawn together into a logically simple structure, linked by, say, a single realistically-interpretable model, or a single, mathematically simple, law such as Newton's law of gravitation. Thus the success of Newton's theory in having the tides, and the precession of the equinoxes, as consequences, is almost as improbable, whether he knew about them or not, as he was devising the theory.
This also explains how researchers in a programme can learn from contrary evidence. It may encourage them to make changes in the hard core which then make the theory improbably more simple ,for example, than it was before; it may now relate more varied. Back
15 'Explanation' has to be taken to mean more than the deductive-nomological 'logical derivation from the theory' - since otherwise this method would (Lipton p. 2&58) "reduce to the equally familiar hypothetico-deductive model of confirmation". Similarly, 'best' has to be taken not to mean likeliest, most probably true (best supported by the evidence). since "this would push the method towards triviality" (p.62).Back
16 The "Hungerford objection" (pp.122-124) that the qualities mentioned are too subjective, Lipton rightly rejects. Subjective judgements can be well-tested guides (for example in the case of wine-tasters, used by Bill Newton-Smith). Physicists, some more than others, could have learned to detect some subtle aspect of a theory, which evidence subsequently indicates is a truth-predictor, too often to be due to chance (using T_p). Back
17 The plausibility of this factor is increased, once one realises (Lipton pp. 131 & 188) that the kind of factors preferred will have altered as physicists learned which ones seemed to be truth-indicating. In other words, they constitute a high-level guiding theory. Back
 

Back to philosophy of physics map page