Misunderstanding idealization, truth, and understanding (& 2).

In my past entry I commented on the problems that Angela Potochnik’s Idealization and the Aims of Science contains regarding her analysis of idealization. Now I will look at the question of truth. My main argument is that neither the widespread use of blatant falsities in science, nor the existence of other goals different from ‘literal truth’, entail in any way that the pursuit of truth has to be discarded as one essential goal of the kind of practice we call ‘scientific research’, and more seriously, that forcing ourselves to see science as an endeavour totally disconnected from the pursuit of true knowledge about the world prevents us to rightly understand the methods and the accomplishments of science.

Replacing models

First of all, the inference from the premise that ‘most scientific models contain radically false assumptions’ to the conclusion that ‘being in some relevant sense closer to the truth is not an important goal of science’ is prima facie as doubtful as the inference from ‘almost all human beings through history have lived in misery’ to ‘the pursuit of economic wealth is not an important goal for humans’. Perhaps most scientific models are indeed very far from the truth, but this does not mean that scientists wouldn’t often be happier if they knew how to replace those models with some that were substantially closer to the relevant truths. It is even conceivable that in some cases scientists may consider that the passing from some old model to a new one that is recognisably ‘less true’ in all the relevant aspects is a case of scientific progress, because other values different from ‘truth’ are better exemplified in the second model; but in order to show that this makes truth an irrelevant value in science one should have to demonstrate that this type of examples are not something occasional, but systematic, or at least, that we cannot just explain them as cases in which one of the multiples values is given preference over another value without entailing that this second value is ‘unimportant’.

Let’s illustrate this argument with a different goal in mind, one that (though non-epistemic in nature) is obviously very important in scientific practice: cheapness. A research team may opt for using a calculator that is known to commit more mistakes than another one, if the second is extraordinarily more expensive than the first. From this we should not infer that exactitude in the calculations is ‘not relevant at all as a scientific goal’, only that, as most kind of goals human beings have, there may be trade-offs between them. Hence, we may also say that truth and understanding (or, in the terms employed a few pages above: credibility and enlightening) can be in a trade-off relation, without this entailing that some of the two goals is irrelevant just because the other happens to have more weight in some, or even in most cases.

Substantially close to the truth

Second, and more importantly, it is not even that the pursuit of (closeness to the) truth can in principle be taken as an important goal of science even in spite of most scientific models containing blatant falsities: I think that we can argue for the much stronger thesis that science has actually been considerably (and often spectacularly) successful in providing us with knowledge of the world that is substantially close to the truth, and that history shows, without the need of any kind of Whiggism, that in many areas we have made a lot of progress in getting more and more detailed knowledge of the furniture and working of the world. In some cases, this may have been done even at the cost of having less understanding as we (thought we) had before: often what happens is that we transit from a vision of some segment of nature that provides both a neat small collection of elements and a simple explanation of its mutual interconnections, to a view that recognises the existence of a plethora of very different entities but simultaneously a much messier and less intelligible causal or taxonomic network between them (think, for example, in the evolution of the catalogues of astronomical entities, or of the groups of living beings at different levels). In cases like this, it is absurd to require scientists that they renounce to the big amounts of mundane truths they have discovered, just because the previous vision of the field gave them a stronger feeling of ‘understanding’. But, of course, in many other cases the progress in truth has fortunately gone hand in hand with a parallel progress in understanding, and we end both knowing much more things about the world, and understanding them in a much better way.

Potochnik’s anti-veritism

In the third place, I think that the (for me, rather bizarre) anti-veritist attempt of dispensing with the basically trivial claim that science has very often been considerably successful in the pursuit of truth has an easy explanation: the confusion of the possible shortcomings of some philosophical theories about the nature of truth or representation, with failures in the run-of-the-mill understanding that ordinary scientists may have of the properties and virtues of their models when they themselves use representational or veristic language to discuss a lot of things about those models and their connection with the world. In the case of Potochnik, the confusion probably derives from the supposition that the concept of ‘truth’ must refer to something like an absolute point-by-point metaphysical correspondence between our statements and an absolutely precise ontological scafolding of the world in itself, or something like that, and hence, that the scientific acceptance of anything that fails to be exactly identical to such a ‘literal, absolute, and eternally unchanging truth’ should be considered as a refutation of the idea that scientists pursue in some interesting sense ‘true knowledge about the world’. But if we understand the concepts of truth and approximate truth in a more trivial sense, as just expressive tools of ordinary scientific language (rather than as a philosophical relation –whatever that could mean– between language and the world), we can easily see that a scientific model being successful in the sense of being approximately true (or ‘close enough to the truth’ for the relevant purposes) is not something requiring an ontological analysis (probably doomed to be engulfed by conceptual paradoxes), but just one of the relatively trivial things real scientists say of their models when they evaluate them. Once this success has been established strongly enough in the course of empirical research, the fact that some aspects of the model are not exactly, nor even remotely, ‘like’ their possible ‘analogues’ in the target systems is in many cases no argument against the conclusion, for scientists’ conjectures do not affirm that the world have to be literally or exactly like the model system in all respects, only that it was ‘similar enough in the relevant ways’.

Changing the discussion from whether science tries to discover a true description of the world, to whether it tries to understand real causal patterns (as Potochnik defends) doesn’t move a millimetre the argument in favour of Potochnik’s anti-veritism, for our deflationary view of truth helps us to be agnostic about the ‘right transcendental stuff’ the world may be made of, inviting us to concentrate just on scientists’ assertions or claims, taking ‘truth’ as just another expressive tool with which to formulate those same assertions: if scientists claim that one model captures better the causal patterns of a target system than another model, then what a deflationist infers from this is that scientists consider that it is true that the first model captures better those causal patterns than the second, and that’s all the truth that is relevant in the discussion about whether scientists pursue the truth or not.

Hence, scientific models and scientific theories being filled with idealizations ‘that radically depart from the truth’ is no reason at all to put into doubt science’s capacity of getting an increasingly approximate knowledge of the truth about the systems it studies, for many of these idealizations are, on the one hand, not ‘mere falsities’, but approximately accurate descriptions of some real things (like point-masses in astronomy may be ‘accurate enough’ for many purposes), and on the other hand, because even if some idealizations are not justifiable as ‘approximations’ in this loose sense, this doesn’t go against the fact that the models containing them can succeed in saying many right things about the world thanks in part to the working of those fictional elements. Paraphrasing Teresa of Ávila, we can say that very often science discovers the truth by means of false idealizations.

Popper said it first

I cannot finish these comments on Potochnik’s book without mentioning the surprise it caused me to realize that she failed to even mention the author that has been probably most influential in promoting the idea that all interesting scientific hypotheses are basically false (and in introducing the debate on whether this fact can be nevertheless coherent with the idea that scientists try to discover theories that are closer and closer to the truth). I’m referring, of course, to Karl Popper. Potochnik didn’t also refer at all (as we saw she does with formal approaches to scientific idealization) to the rampant philosophical literature on truthlikeness or verisimilitude stemming from Popper, probably disdaining it by assuming that, for those philosophers, scientific progress is, in her own words, “a lockstep pursuit of the truth”… which of course it is not what those authors affirm.

References

Popper, K. R., 1935/1959, The Logic of Scientific Discovery, London: Routledge.

Potochnik, A., 2017, Idealization and the Aims of Science, The University of Chicago Press.