Why Failure Isn't Always Good EnoughThere was recently a very interesting piece in
The Guardian ("We must learn to love uncertainty and failure, say leading thinkers", by Alok Jha, Saturday 15th Jan, 2011). In it, the 'planet's biggest brains' responded to
Edge magazine's yearly question, which for 2011 was: "What scientific concept would improve everybody's cognitive toolkit?" In other words, what sexy new idea is going to help us better understand the world? Perhaps surprisingly, a common theme from respondents was 'failure', or rather, the need to abandon the idea that science gives us certainty.
Of course, this is not a new idea, but rather reflects a common public misconception. People think that scientists arrive at certain knowledge of reality, and seem to consider 'scientifically proven' as the ultimate standard of truth. But, as Carlo Rovelli, a physicist at the University of Aix-Marseille, was keen to point out, this is in fact a contradiction in terms, an oxymoron. Science does not
prove things, in any absolute sense, but rather tests hypotheses. As Neil Gershenfeld of the Massachusetts Institute of Technology put it: "The most common misunderstanding about science is that scientists seek and find truth. They don't - they make and test models." These 'models of reality' are therefore provisional, for any new raw data can in theory disprove them.
Whilst it may long have been used as a rule of thumb, the explicit statement of this approach goes back at least to the work of philosopher Karl Popper and his doctrine of 'falsifiability'. The majority of scientific knowledge comes from the process of
induction: by gathering numerous examples (data), I can form a general explanation which fits the evidence. All men are mortal because, as far as we know, there is no way of reversing biological cell death, and all observed human cells seem to be subject to it. One day, perhaps, someone will find a way, but until then, it remains an inductive truth. The 'problem of induction' is therefore that, no matter how much the data fits our hypothesis, there is always the possibility that it may one day be overturned by new data (all men are therefore
provisionally mortal). Popper's answer to this was simply to point out that science was not interested in certainty, but in scientifically testing and rejecting various working hypotheses or models of reality. In doing so, whilst our understanding of reality is never certain, it gets
closer with each 'failure'.
This is sound general scientific method, and I do not wish to undervalue it. Without it, we would not have anywhere near the level of medical or technological knowledge that we possess. We should therefore all be grateful. However, philosophically, this view is problematic. Firstly, as philosopher of science Thomas Kuhn pointed out, Popper's vision of the scientist as the conscientious falsifier of hypotheses is not borne out by history. Often, new conflicting data do not result in the complete abandonment of a theory, but an adaptation of it - sometimes even the rejection of the new data altogether (as 'experimental anomaly', or equipment malfunction, or misobservation). Most science, then, is what Kuhn calls 'normal science': tweaking theories to fit evidence, or massaging evidence to fit theories. Only in exceptional periods does 'revolutionary science' take place, where the underpinning assumptions - the 'world picture' - is itself abandoned. Newtonian mechanics gives way to Einsteinian relativity, and normal science can begin again.
Kuhn is sometimes seen as a radical, but he is not really. He is merely pointing out that underlying assumptions are very rarely directly addressed, and that fundamental shifts are not only quite rare, but often driven by extra-scientific concerns - such as power struggles in society (the battle to switch from Ptolemaic to Copernican cosmology reflecting that between the Catholic Church and opposing secular interests). Kuhn's points may therefore be taken as merely suggesting that scientific progress is not just about the straight-forward progress of rational knowledge. Many scientists would accept this: a pet theory can have ties to non-scientific and non-rational interests, and it's difficult, perhaps impossible, to divorce the two.
However, there is a more fundamental problem with Popper's approach, and with the general notion of science as suggested by those quoted in the
Guardian article. We revise our working models of reality in relation to new or
raw data, as it's often called. But what is 'raw' about it? Scientists often talk of scientific models 'fitting the facts', which, in philosophical terms, assumes a
correspondence theory of truth. My theory is true, because it describes or corresponds to the facts. But, you may say, scientists reject the idea that they search for 'the truth'. Very well: my theory is
false, because it
doesn't fit the facts; therefore I reject it and formulate a new one. All theories are expendable. But there is an assumption here - it is so subtle as to go almost unnoticed, but it is there: there is such a thing as 'the facts' or 'raw data' which are independent of our 'model of reality'. We can set out the problem in these terms:
1. Science attempts to describe (model) reality.
2. Such models are always provisional, but can be revised according to new data (facts).
3. But what determines the form that the facts or raw data take?
4. Answer: Our model of reality does.
This is a basic and simplified form of the argument, but you get the picture: it would seem that in order to test certain aspects of our working model of reality, we must take other aspects for granted. This does seem to lead to a real problem for scientific method - doesn't it?
Let me try to illustrate this. Take ghosts. A scientist might say that ghosts do not exist, because if they did they would show up on scientific equipment and in experiments - and they do not (let's, for argument's sake, say that this is true). Now, there is a fundamental assumption here: if ghosts exist, they will show up in experimental conditions and be detectable using scientific equipment. But this is an assumption that experiments do not - and possibly cannot - test. Failure to find a ghost does not prove that ghosts don't exist, just as failure to find Nessie doesn't affect the Loch Ness tourist trade.
I'm not necessarily arguing for the existence of ghosts or plesiosuars, but merely pointing out that the scientific method is not a neutral way of testing theories. Built into any theory are basic assumptions that the theory requires in order to make sense in the first place. But "the map is not the territory", as Polish philosopher and scientist Alfred Korzybski once put it; our models of reality cannot be used to determine the fundamental nature of that reality - without, we might add, already assuming certain things about it (such as, for instance, that it is material and physical).
But this is not an argument merely against materialism, but against assumptions in general (of which materialism is one). We could pick on others, such as that cause and effect relationships are fixed and regular. This has perhaps been undermined by certain findings in quantum physics, but it is still a central tenet of scientific thinking - and so it should be, for it has served us well. We must of course hold on to such assumptions, but we should also be aware that they are just that. In philosophical terms, we can distinguish between
epistemology (how we know things, how we guarantee knowledge, etc.) and
ontology (what exists). Scientific method - 'failure' - may be an important and useful tool in our search for knowledge and understanding, but, as regards the question of deciding what reality ultimately
is (its
ontology), it is not enough.