TCS and Fallibilism

by David Deutsch

This article was first published in Taking Children Seriously 23

Over two millennia ago, the idea that human beings are inherently fallible was introduced into the foundations of the theory of knowledge by Pre-Socratic philosophers. Ever since, everyone who has taken this insight seriously – and who has therefore denied the possibility of revealed knowledge, certain knowledge, or justified knowledge – has been accused by defenders of tyranny of denying the possibility of knowledge (true theories) itself. ‘You've got to build on solid foundations or you'll never get anywhere’, they claim.

They are mistaken. We (the likes of Xenophanes, Socrates, William Godwin, Karl Popper and TCS supporters) are rationalists as well as fallibilists. We believe that it possible for human beings, through conjecture, reason and criticism, to come to know and understand truths about the world, including truths about the human condition and about specific people, and including truths about matters that are not experimentally testable. We do not believe that we possess the final truth about any of these matters, but we do believe that our successive theories can become objectively truer – with more true implications and fewer errors. But because we are fallible, it is not possible for us to know which of the ideas that we believe to be true are in fact true, or in which cases we are right when we believe that we have obtained a truer theory than we had before. History is full of examples of ideas – the flatness of the Earth, Newton's laws, the subservience of women – that were once the epitome of certain, unquestionable truth but are in fact severe errors. We hold it to be true that many of the ideas that we now believe to be true, including some of those that we believe most strongly to be true, are in fact riddled with errors.

That is why the frameworks that we endorse for science, politics and interpersonal relationships are fundamentally different from those of non-fallibilist world views. Our frameworks – protocols, rules, maxims etc. – are all about what to do in the face of opposing theories, ideas, values etc., which may be true, given that we start from the premise that our own may be mistaken. Other frameworks are all about how to obtain ‘justified’ (revealed, certain, etc.) knowledge – i.e. theories for which you can know in advance that anyone who contradicts you will be wrong – and how to behave when you have it. The former is invariably a fraud; the latter a recipe for tyranny.

Those who think that believing that one may be mistaken, even when one is sure that one’s theory is true, is self-contradictory, are mistaken. They have not understood the fallibilist conception of objective knowledge. As Sir Peter Medawar said in Advice to a Young Scientist:

I cannot give any scientist of any age better advice than this: the intensity of the conviction that a hypothesis is true has no bearing on whether it is true or not.

In the fallibilist scheme of things, arguments typically take the form of criticisms of theories that contradict the one that is being argued for. In science, for instance, an experimental test may establish that a range of previously plausible theories is false because their predictions were not borne out, but it cannot prove that any theory is true, because it may yet make many false predictions in some – or even all – other situations. Quite generally, you cannot hope to find evidence that some future criticism, of a form you do not yet know, will not reveal a fatal flaw in your favorite theory.

There is often a moment of understanding, when you assimilate an explanation of why something is so, rather than merely that it is so. And this often comes along with an increased conviction that the idea is true. But remember the Medawar quote. Even when you get this conviction, it does not mean that the idea is true. If anything, it would mean that you should be especially careful to criticise the idea.

So in regard to the theory of TCS philosophy, we believe it to be true even though it is not experimentally testable, but we could be wrong. We believe that it is true for the following reasons. First, what passes for rival educational theories all depend on structures of arbitrary authority (parents or teachers have the last word) and mechanical visions of human nature (such as Behaviorism), which have already been rejected for excellent reasons in every field other than education – and the arguments for retaining them in education are crudely ad hoc or simply circular. Secondly, conventional educational practices are notoriously ineffective even in their own terms: everyone knows that most adults could not begin to pass a test in most of the subjects over which they slaved for most of their time in their school lessons; yet almost no one draws the obvious conclusion that forcing such lessons on children is insane. Thirdly, TCS educational theory is consistent with wider philosophical ideas – fallibilism among them, and others that are fundamental to our society such as the idea of human rights – that we hold for independent reasons. Does anyone really find it satisfactory to espouse lofty principles such as freedom of speech, freedom of association, freedom from involuntary servitude, the right to due process and so forth, while simultaneously justifying locking children up in a classroom all day doing things they hate and then giving them ‘detentions’ for speaking disrespectfully to a teacher when he forcibly prevents them from leaving the room?

Experiment could not refute the theory of TCS, but argument and criticism might. For instance, it might show that the theory contradicts some principles that we have independent reasons to retain. Or it might show that (contrary to what we currently believe) one of the conventional systems of education is consistent with fallibilism and our wider philosophical beliefs, or that TCS theory is not consistent with them, or that some of the wider philosophical beliefs are themselves flawed, or whatever.