“Philosophy of science is philosophy enough”, wrote WVO Quine, one of the twentieth century’s greatest philosophers. The single most important insight of twentieth-century philosophy of science is known as holism. We might reasonably call holism “the” idea of the twentieth century, as it was first discussed explicitly by Pierre Duhem in 1903, and later explored in the minutest detail by Quine, who died on Christmas Day 2000. The term ‘holism’ is used in different ways in different contexts, so let’s be clear at the outset what I mean in the present context.
By holism I mean the idea that hypotheses get tested in groups, rather than individually. Let’s take a quick look at the logic of testing to see how this works in practice. First, a scientist somehow comes up with a hypothesis. In science, hypotheses usually describe things that cannot be seen directly, such as the behaviour of electrons, viruses or force fields, or the evolutionary emergence of lungfish from water when vertebrates began to walk on land, or the invisibly slow movement of drifting continents.
Next, a scientist deduces an observational consequence of such a hypothesis. This is where holism comes in. Hardly any individual claim logically implies any other claim in the manner ‘it’s a rainy Monday’ implies ‘it’s raining’. Very few simple implications of that sort are of any use in science. Instead, a scientific claim or hypothesis works in concert with many other assumptions to imply something that can be observed. For example, the hypothesis that the universe is expanding implies that the light from faraway objects will be red-shifted – but only in conjunction with a wide range of other hypotheses and assumptions about such things as the Doppler Effect, the fact that light does not get “tired” by losing energy over very long periods of time, the fact that elements have distinct emission spectra, and so on.
We can write this logical situation as follows:
If H1 & H2 & H3 & H4 & … then O
Once an observational consequence O of a hypothesis H1 has been deduced or computed, someone looks to see if O can actually be observed as the hypothesis predicted – or rather, as was predicted by the hypothesis in question along with its penumbra of other hypotheses and assumptions.
If O is actually observed as predicted, all is well (for now). But if it isn’t, something has gone wrong. And now we can see why holism is so important. If O is false, then the conjunction of H1 & H2 & H3 & H4 &… must be false as well. But we can’t say which of these individual hypotheses is false. Something has gone wrong, but we can’t reliably narrow things down to locate a single culprit.
So Popper’s famous idea that a single unfavourable observation “falsifies” a hypothesis is mistaken. Things are much less clear-cut than that.
The empirical evidence for a hypothesis consists of the observations made when the hypothesis passes tests. And there are other forms of evidence than purely “empirical” evidence (a simple hypothesis is better than a complicated one, and so on). Holism does not change any of that. But because each hypothesis only passes tests in concert with many other hypotheses and assumptions, the passing of any test counts as evidence for all of them together. Observations do not imply or narrow down the possibilities to the hypothesis currently under scrutiny – at best they can be considered to “corroborate” it rather than confirm it, to be “consistent” with it rather than imply it.
With holism comes pragmatism. A hypothesis is worth believing if it works well in practice, embedded as it always is in a larger theory or still larger “paradigm” (i.e. an even broader range of assumptions and ways of doing things). This sort of pragmatism is reminiscent of Burke’s political conservatism, which rejects basic principles and instead judges any political system by how well it actually works in practice, given the circumstances and traditions that are an integral part of it.
With holism also must come the rejection of foundationalism. Foundationalism is the epistemological theory that supposes some of our beliefs have a privileged status (such as being “self-evident”) and that these beliefs work as a basis for the rest of our beliefs. Typically, these privileged beliefs are thought to be about conscious experiences, the sort of things we “cannot be wrong about” such as “I’m having an experience of blue in my visual field”.
In the context of scientific evidence, it used to be believed (by Francis Bacon, and somewhat embarrassingly by as great a scientist as Newton) that observations implied scientific claims, in other words that they worked like “units of evidence” or “data” supporting theory. But natural philosophers such as Galileo and Robert Boyle realised that hypothesis (i.e. guessing) and testing (i.e. observational checks on the consequences) were essential. Even so, despite that important correction, before holism there was still the temptation of thinking that individual observations supported individual hypotheses in a weaker than strictly logical way, so the image of science “resting on a basis of data” lived on. All that is over with holism.
As an account of empirical knowledge in general, foundationalism is mistaken. Yet it is incredibly influential. People who do not have a training in philosophy (and alas some who do) widely assume that scientific hypotheses “rest” on a “foundation” of “data”, in much the same way as they suppose, equally wrongly, that empirical knowledge “rests” on a “foundation” of “experience”. Experience and observation are still vital, of course, but they don’t work as a foundation.
Using these ideas, in an upcoming post I will explain why, if we are prepared to bend over backwards far enough, we can literally believe anything we like. We manage this by embracing ideology, or rather by allowing ideology to embrace us. I shall also point the way out of its deathly grip.