How can I find the Truth?

l
We have seen that the best definition of truth is the accurate prediction of sense experiences. In order to find the truth we therefore need a way to test whether a statement or theory is able to accurately predict sense experiences.


Truth test one: direct testing of predictions

The first method is of course to directly test its predictions in a manner that is publicly verifiable i.e. reproducible by anyone following the same methodology. This is the test used in the natural sciences. However, even if the predictions of a theory are verified, scientists do not accept it as true because new sense experiences may always come along in the future that it did not predict and which falsify it. All they can say is that it has not yet been falsified. To put it another way, if a theory is the best predictor of sense experiences, we can say that it is has the highest probability of being true than any other existing theory.

However, it is often not possible to apply this test. For instance, in the social sciences it is much harder to conduct controlled experiments. And the prediction of most theistic belief systems of life after death is very difficult to test directly. Some have argued that this makes such theories ‘unfalsifiable’ and that they should therefore be rejected. But this would violate a principle formulated by William James that “A rule of thinking which would absolutely prevent me from acknowledging certain kinds of truth, if those kinds of truth were really there, would be an irrational rule”. This seems to be a reasonable rule and so we should try to find another test for the truth that can be applied for those theories not amenable to scientific testing.


Truth test two: explanation of sense experiences

A more practical test is to see how well a statement or theory is able to explain existing sense experiences. This is because prediction and explanation are two ways of saying the same thing, since if ‘A predicts B’ then ‘B is explained by A’. The only assumption for this test to work is that the underlying nature of reality is relatively constant over time, so if A predicted/explained B in the past, then it will do so in the future too. As we have seen, this problem is also faced by the scientific method, and means that we can never say that a theory is proved to be true. However, if it is the best at explaining, we can say that it is more likely to be true than any other existing theories. Since we are primarily interested in truth in order to inform our actions, then it would be rational to act on this theory.

So how can we operationalise this test? I believe that the best way to see how well a theory can explain existing sense experiences is to measure its coherence in explaining facts.

By facts, I mean publicly verifiable experiences which are reproducible by anyone, and which you want the theory to explain. For instance, on the question of whether UFOs exist or not, one of the main facts to be explained is that "Many people think they have seen UFOs with the circumstances documented in books X, Y and Z". Anyone can verify this by reading the books mentioned.

By explaining, I mean stating at least the immediate cause of the fact e.g. "Many people have actually seen UFOs" (for the UFO believer) , or "Many people have seen certain things under certain conditions and mistaken them for UFOs" (for the UFO sceptic) .

By coherence, I mean the extent to which all beliefs within the belief system, including the explanations, should be logically consistent and logically justified.

If any of this doesn't make sense then go to Coherent Explanation of Facts for a detailed explanation.


Coherence Quotient (CQ)

If the truth test is the level of coherent explanation of the facts, it would be very useful to have a quantitative measure of it that could be used to objectively assess the coherence of competing belief systems in explaining the same facts. I think that such a measure is relatively straightforward to construct, and call it the Coherence Quotient (CQ) of a theory. I suggest that it be the percentage of the total number of beliefs that are both consistent and justified, where beliefs are weighted according to how important they are in the system. For instance, if there are 10 beliefs, one belief is not adequately justified and that belief is of average importance, then the CQ is 90%. If the belief is more important than average, then the CQ is lower than 90%, whereas if it is less important than average, the CQ would be higher than 90%. If instead, there is an inconsistency between two beliefs, this would be counted as a problem with the belief that is the least important of the two. If it is of average importance, the CQ would again be 90%.

This is very similar to the dissonance ratio used by cognitive dissonance theorists in psychology, except that they only look at the consistency between one's beliefs.

The Belief System Debate method uses the CQ measure to judge between belief systems in a debate. It is therefore important that the measure provide the right incentives and not be open to manipulation. In particular, two issues with this measure will need further development over time:

• How to calculate the weighting: there are actually two issues here. One is whether to measure the importance of a belief to the system in terms of the number of beliefs that directly depend on it as a premise (equal to the number of arrows of logic flowing from it in a Belief Map), or the % of total beliefs that directly depend on it as a premise. Another issue is whether to consider those beliefs that indirectly depend on it as a premise as well. This would seem to make sense. For instance, imagine a belief A that justifies only one other belief B, but where belief B justifies many other beliefs, and compare this to a belief C that justifies only one other belief D, where this belief D only justifies one other belief. We would say that belief A is more important than belief C. Perhaps the ideal solution would be to take the weighted average of beliefs that depend on it as a premise, where the weights decline the further away one goes.

• How to penalise belief systems that provide no explanation of a fact: in the CQ measure mentioned above, it is assumed that explanations are given for all facts. But if belief systems are competing to get the highest CQ score, and they cannot provide a coherent explanation for a particular fact, they have an incentive not to provide any explanation at all. Probably the best response is to say that if a belief system provides an explanation for more facts than another, then it automatically wins the debate irrespective of their CQs. This gives an incentive for all belief systems to try to provide an explanation, even if its justification has low coherence. If no sides in a debate can coherently explain a fact, they can always jointly agree to remove the fact as one to be explained.

Given the complexity of calculating CQ, it would need to be done by a computer. In fact, it would be quite straightforward to add to this feature to the Visual Concept software that is used for creating Belief Maps.

If this makes sense to you and you want a tool to help you to practically apply it, then go to Belief System Analysis.