On Epistemology

In considering knowledge, and particularly scientific knowledge, there is often a presumption that knowledge must be 'true'. People talk of 'validating' a theory or 'proving' a supposition to be true. ('Proof' being used in that extreme sense, rather than the original sense of a 'test').

But we are gradually learning that scientific 'laws' can be improved upon and even supplanted as we gain more effective (some might say 'more correct' - more on this in a moment) understanding and associated descriptions and explanations. (Well, most of us are: I have heard supposed scientists say that today's thermodynamics is fundamentally different from the phlogiston-based theory, in that we won't be throwing it out because we 'know' it to be correct. Hmm.)

Thinking on this, one can get quickly bogged down in the question of how one can know that something is really true. At one extreme position, some would deny that there is such 'stuff' as knowledge. Another stance is to equate knowledge with beliefs. There are many views on this subject, and there is not space to present them all. Here I present a categorisation after Lakatos. ("The problem of appraising scientific theories: three approaches" in "Mathematics, science and epistemology")

Along the way I offer my own opinions, and comment on my interpretation of Lakatos. Since I feel this to be most necessary where I disagree, this may send out an overly negative message about Lakatos. That would be wrong: I have considerable respect for his work. I would not have started with his analysis otherwise. But I do depart over notions of objective 'truth', normative notions of 'progress', and their like. Lakatos was, according to his own definition, a dogmatist, taking "the position that objective knowledge - infallible or fallible - is possible". I find myself much more in sympathy with Feyerabend (more of which, later).

In particular I see, in the enthusiasm of Lakatos for deductive logic, shades of the one problem I perceived with 'Proofs and Refutations' - a failure (as far as I could see) clearly to distinguish between reality and statements about reality. Since all premises about reality to be used in deduction are themselves only, in a sense, expressions of hypotheses about reality, then that is all that any 'conclusion' can be. Deduction cannot say anything about reality itself. Lakatos himself talked of "the epistemologically unbridgeable gap between fact and proposition". (Also cf. Strawson - Introduction to Logic)

I do not quite follow Lakatos's categorisation, and the difference is more than a quibble over categories. Lakatos identified three major schools of thought on this subject: demarcationism, scepticism and elitism. I will summarise his views on each of these. In addition, he considered what he called "four abhorrent philosophical doctrines": psychologism, authoritarianism, historicism and pragmatism.

I am going to consider four categories - the main three of Lakatos plus pragmatism. I will explain why in due course, and discuss the relationships between these belief-systems as I go along.

In summary:

demarcationism: There are objective ways of judging whether one scientific theory is 'better' than another
scepticism: Oh no there aren't
elitism: Oh yes there are - as long as we do the judging
pragmatism: If you say so

Lakatos asserted that he was a demarcationist and considered demarcation to be a fundamentally important issue: "The generalized demarcation problem is, it seems to me, the primary problem of the philosophy of science."

The demarcation with which he was concerned is the demarcation between scientific theories. Lakatos differentiated 'generalised' demarcationism from 'ordinary' demarcationism along the lines that 'ordinary' demarcationists - Popper being the most well-known - concern themselves with demarcation between science and non-science. So they consider how one might decide whether a theory is scientific. 'Generalised' demarcationists concern themselves with whether one scientific theory is better, in some sense, than another.

However, it is important to appreciate that, in the view of Lakatos, the demarcationist's notion of goodness is in some way absolute. "According to the demarcationist one theory is better than another if it satisfies certain objective criteria." [my emphasis]

Demarcationists believe that one can appraise scientific theories in such a way that one can decide whether one theory is more 'advanced' (my word) than another (even if they will argue till Doomsday how). Lakatos seemed to believe in some kind of scientific progress - which raises the question "Toward what?" I think that Lakatos would argue that he was concerned with progress toward the truth. The truth? - pardon my scepticism. But even Lakatos recognised the difficulty with this notion: "There is no ultimate proof that .. we have been heading towards the Truth. We can only (non-rationally) believe or, rather hope, that we have been."

Lakatos also believed that deductive logic had a central role to play in demarcation ("without deductive logic there can be no genuine criticism, no appraisal of progress") I have already expressed my scepticism about that idea. However, for our present purpose this is not important. What matters is the notion of an absolute measure of goodness, which typically is something to do with proximity to 'the truth', or a movement toward 'the truth'.

To a sceptic, according to Lakatos, scientific theories are 'just' belief systems. That 'just' has a pejorative tone because Lakatos himself believed that science, in its pursuit of and approximation to truth, is different from just any belief system - astronomy being different in this sense from astrology. "One belief system is no more 'right' than any other; although some have more might than others. ... None has epistemological superiority."

From this Lakatos infers that sceptics deny the possibility of progress in belief-systems. And he then goes on to state that the sceptic 'thus' denies the possibility of producing any acceptable solution to the problem of appraising scientific theories - because, I suppose, he presumes that sceptics have no yardstick against which to appraise them. I think that this is a matter of definition.

If we define a sceptic to be one who really cannot perceive any relative values in theories, then Lakatos's description is tautologically - i.e. trivially, emptily - correct.

But what of other values? Aren't some theories more useful than others? Aren't some theories more elegant than others? Cannot bodies of knowledge 'progress', by refinement, become more useful - or more elegant - or more ...?

Well - yes. But those who are prepared to accept such discriminants are not sceptics, according to the strict definition: they are pragmatists. We will return to this later.

This is the only other school to which Lakatos gave serious attention. According to him, elitists say that good science can be distinguished from bad science, but only by scientists themselves. No wonder he calls it elitism. So he sees elitists as aligned with demarcationists with respect to acceptance of being able to make a judgement on the relative goodness of theories, but with sceptics in not perceiving any objective criteria for such judgement

And what of differing views within the elite? And what of multiple competing elites? As Lakatos says, "But what if the insiders disagree?" He then uses this problem (amongst other things) to invalidate the elitist stance.

But since I do not know anyone in science who would take such a stance, this all smacks of the stuff which gets philosophy a bad name - invent a fundamentally unsound idea; find, not surprisingly, problems with it; discuss the problems.

There are many flavours of pragmatism. The most prevalent is acceptance of usefulness as a discriminant. 'Usefulness' has already quite a broad scope, but I will use the even more general 'value'. And I do not mean just economic value. For me, elegance, simplicity and explanatory power are all acceptable 'values'. Even - whisper it soft - political expedience is a 'value'. One theory can be more valuable than another in many different ways; a body of knowledge can 'progress' by becoming more valuable.

It may be that others would restrict the application of the label of pragmatism - perhaps to utilitarian values. That is no problem. I would be happy to accept another label to mean what I intend to mean. If you would prefer, I could use 'anyvaluism', but for now I will stick with pragmatism.

It is interesting that Lakatos is implicitly pejorative about pragmatism because of the multiplicity of values which might be used in appraisal. He quotes F C S Schiller "there are as many pragmatisms as pragmatists". Yet for a pragmatist that is fine. If this is cultural relativism, so be it. One may not like it, but I do not see how one can deny it.

What is abhorrent about pragmatism?
Even theories about theories are value-laden, it seems. I believe that the basis of the 'abhorrence' which Lakatos (and I believe Popper) felt for pragmatism - and all the other non-dogmatic 'isms' - arose from his fear of the (ab)use of power and politics to pervert science.

We must remember the era - late sixties, early seventies, Chomsky liberalism, etc. Lakatos was a refugee from a culture in which science was used to justify and support an ideology and political policies based upon that ideology. And only scientific theories which did provide that support were accepted. Lysenkoism is perhaps the best known example in the West. Scientific knowledge was, in effect, determined by politicians. Dissenters were not tolerated. (Lest we think that such a situation pertains only in totalitarian states, consider whether Chomsky's views on language are independent of, or influenced by, his liberal beliefs.)

Lakatos was, I believe, wanting instead a 'pure' science, independent of politics, governed only by facts and deductive logic. And, being based on a notion of objective knowledge, Lakatos (I believe) saw both generalised demarcationism and scientific elitism as in some way pure and beyond political manipulation.

So why did Lakatos let in scepticism? Because, I think, that by being value-free, scepticism is also, in a way, pure, in that it is not open to political manipulation.

I'm with Feyerabend "I don't believe that charlatans can be banned just by tightening up the rules." (Against method, p.163)

Does pragmatism include authoritarianism, etc?
No. A pragmatist accepts that elitists and even those taking 'abhorrent' positions have a valid point of view in that each is using their own criterion of 'value', however abhorrent that may be to some. But the pragmatist does not accept that any of these positions is 'right' except for those taking that position. If one takes just one of these positions, and disallows others, then one is not a pragmatist.

Would a pragmatist consider demarcationism to be as valid as any other view?
If being absolutely better has some value, and it seems reasonable to suppose it would, then I suppose so. But this is, for me, like angels on a pin. I find it so hard to think myself into the notion of absolute rightness that I cannot consider this properly. Luckily, for our present purposes, it is not necessary to decide. What matters is that demarcationists seek objective knowledge, while pragmatists are more catholic.

The story so far
I started by recognising four schools of thought on knowledge. I think that for the purposes of consideration of the nature of any 'contribution to knowledge' from research we can now discard two. Strict scepticism seems at most uninteresting, and probably empty. Elitism is - to me at least - a nonsensical idea. This leaves generalised demarcationism and pragmatism, and the critical difference between them is that demarcationists have faith in the notion of 'objective' knowledge.

Does any of this matter?
Not much - but a bit. I do think that research objectives should be formulated so that we may make 'a contribution to knowledge'. But we should not get hung up over seeking absolute truth. If you think you can do it - fine. But 'being useful' or 'adding value' will suffice - though the nature of that value must be identified.

Now, just having some kind of conceptual framework can be valuable. It might be arbitrary, rather than having any intrinsic value (or 'truth'). But as long as it gives a set of people a common language for a class of situations, then they may benefit. But if that 'set of people' is restricted to either an individual or a particular research group, then their conceptual framework does not, I think, qualify as scientific knowledge.

In general, the value of the framework should be perceived by more than those immediately involved in the research. 'Knowledge', however defined, should be 'transferable' in some way. It may be transferred through a 'law', such as e=mc˛, together with a more-or-less agreed way of mapping observations of reality to the operands of the law. Or it may be transferred through sharing of a conceptual framework in which the sharers model the world in a similar way. (I see the former as a particular instance of the latter.)

Aside ... This all relates to my own definition of objectivity - shared subjectivity. This allows different groups to have different objective knowledge. If there are some things on which all humans can agree, then that is as near to absolute objectivity as we can get. (Note that this is distinct from discussion of whether there is objective reality - which I will not go into here.)


Further reading

"The problem of appraising scientific theories: three approaches" in "Mathematics, science and epistemology"

In Against Method, Feyerabend argues that successive theories that apparently concern the same subject cannot in principle be subjected to any comparison that would reveal the truer explanation. According to this notion of incommensurability, there is no neutral or objective standpoint and therefore no rational way in which one theory can be chosen over another. Instead, scientific progress is, he claims, the result of a range of sociological factors working to promote politically convenient notions of how nature operates.

It is interesting that in Against Method (Chapter 15, Appendix 1) Feyerabend made an impassioned attack on the idea that any group, even those believing in the (beguiling?) power of reason, telling others how to run their lives. This includes Lakatos and Popper.

Lakatos and Feyerabend
Feyerabend had, with Lakatos, an extended exchange of views on 'method'. In For and Against Method : Including Lakatos's Lectures on Scientific Method and the Lakatos-Feyerabend Correspondence, Matteo Motterlini presents the (friendly) argument between them, based upon published work and correspondence.

P F Strawson
Introduction to Logic is no 'mere' introduction. Many a mathematical modeller and computer scientist would do well to read, mark, learn and inwardly digest Strawson's drawing out of the distinction between reality (however you understand it), statements about reality, and the importance, for reasoning, of the distinction.

GO TO .... | ideo home | SII home | EPSRC SI programme home | SI proposers' guidance