In Popper’s account of justifiable scientific theory, one of the key ways that scientific theories are distinguished from metaphysical theories is in terms of their falsifiability. However, as we have seen (see 3.l), there are sceptical problems with this requirement. No observation can be identified with certainty as a falsification of a theory.

Nevertheless I have suggested that falsification is a criterion that can be used imprecisely by individuals uncertain about their commitment to a provisional theory. Rather than either abandoning a promising theory or making an unconditional commitment to it, they should determine their own criteria for falsification in their own time scale. In this way an approximate attempt at provisional belief is made that is open to later refinement, and a theory capable of such treatment can still be differentiated from an obviously metaphysical belief that requires either absolute commitment or nothing.

I can now clarify this point with reference to agnostic foundationalism as one of the two necessary criteria for justification. Agnostic foundationalism requires a theory to be held in a provisional way, a requirement that can only be described psychologically rather than in terms of the content of the theory or its justification. I now want to suggest that agnostic foundationalism is equivalent to falsifiability applied in the way I have just summarised.

Agnostic foundationalism consists of an awareness that the theory one is provisionally committed to could be wrong. To be practically effective, this recognition needs to go beyond a merely abstract labelling of a belief as fallible. The Buddhist tradition provides plenty of cautionary examples of the ineffectiveness of the mere abstract labelling of beliefs as fallible, where beliefs that are said to be ‘ultimately empty’, like the belief in enlightenment, are still nevertheless often practically treated as absolute metaphysical beliefs[1]. In order to make the labelling effective we need to ensure that a belief we take as provisional is in some way falsifiable for our purposes.

The terms of the falsification must suit the nature of the belief and of the person concerned and their context. For example, a provisional belief that your cat likes fish is fairly well falsified by you placing dish of fish in front of the cat and the cat just walking off – however, perhaps you might try once more to make sure the cat is hungry at the time. A provisional belief that meditation could be helpful to you might require a few months of practice, after which if you have not experienced any helpful developments, you may decide to drop it. A provisional belief that the Middle Way is good may require several years of exploration to falsify, as you understand more deeply what is meant and try out implications of this broader belief. If you dismissed such a complex theory too soon, it might be because you got an auxiliary hypothesis wrong, or you simply didn’t understand the theory well enough. Nevertheless, though these possibilities always continue, there is a point where you have to judge them improbable and, if you can find a better alternative theory, move on to that.

Those conditioned into the fact-value distinction may still be wondering here how a value claim can be falsified. I would suggest three possible ways, which may lead to the same destination:

1. Value claims are always associated with factual claims, so you can falsify the associated factual claims

2. Value claims can be tested (imprecisely) against our experience of values, by considering whether they are implied by values we have already accepted through experience. If they are not, they may be falsified for that reason. See 7.c below for ways that existing values found in experience can imply the value of the Middle Way.

3. Values are desires, so they can be tested in terms of whether those desires are wholly or partly fulfilled. So, for example, if you value happiness more than other possible values, you could test this value by acting so as to prioritise happiness for a given period of time and seeing whether you experienced more happiness as resulting for yourself or others. If you did not, the approach of valuing happiness would be falsified for you. Such an approach assumes that we should not value ‘moral’ ends exclusive of means, and that a pursuit of happiness should produce happiness:  if it does not, the pursuit of happiness will not be best adapted to conditions according to its own standard – that of self-fulfilment. The desire concerned would be better fulfilled by being integrated with other desires which will help it to gain fulfilment in the longer term. See II.2 for a fuller discussion of the background assumptions made here.

Neither facts nor values are easy to falsify. Falsification is a complex matter, full of incremental judgements, ambiguity and imprecision. This is the nature of human experience in relation to judgements. If you want precision, then go back to metaphysics, but observe where that takes you. However, an imprecise and individually-judged falsification is far better than none at all, and it gives a specific bite to agnostic foundationalism that it might not otherwise have. The imprecision of the process does not prevent us from distinguishing provisional beliefs that are capable of going through that process from metaphysical beliefs that are not: indeed it is precisely because provisional beliefs are imprecise that we can falsify them, so it would be unrealistic to expect a precise falsification given this point.

The justification of our beliefs, as of our values, should not be thought of in terms like that of a mathematician deriving a precise proof. Instead, a better analogy would be with a sculptor. Seeing what our experience is roughly shaped like, and driven by a desire to shape it in a particular way, we chip away at our experience like a stone block to produce a gradually more refined set of beliefs – first crudely and then with gradually more precise tools. We need to adapt both to the structure of the stone as we encounter it and to our vision of what the sculpture will look like, but allow each to interact with the other in the way we create.

[1] See Ellis (2011a) ch.6