Inferences based on the above methods need to be connected to make a meaningful whole. Reasoning can serve this purpose. To preserve its independence and allow an unbiased verification, it must be governed by its own internal criteria. Four of them are suggested below. A number of examples, further in the text, show that existing scientific, religious and philosophical interpretations occasionally breach one or more of these (they are not brought up here in order to avoid repetition).
Congruence means that reasoning should not contradict accepted facts (facts are already statements, which is why congruence can be an internal criteria). This is not to say that facts are rock solid; they can change too. Congruence allows facts to be challenged, but not ignored. Therefore, if a statement contradicts commonly accepted facts, a valid justification needs to be provided. In this context, besides observable facts, plausible deductive and phenomenological ones are also recognised. For example, mathematical inferences and some historical events for which there may not be direct evidence (e.g. the existence of a proto-language deduced from similarities in existing languages) can be considered facts. On the other hand, theories based on an interpretation of facts do not have the status of a fact. This does not mean that every claim has to be empirically proven. As already mentioned, proofs based on induction are not always reliable and can be limiting. So, no claim should be rejected outright unless and until it can be refuted. In this way congruence resembles Popper's falsification method although it could not be identified with it: disregarding automatically an assertion just because it is not empirically falsifiable can be in some cases premature. In any case, it is sometimes difficult to pinpoint incongruences, and it is also possible to interpret facts incorrectly without contradicting them. For these reasons, other criteria are needed.
Consistency means that individual statements that are part of the same explanatory structure should not be in conflict with each other. In other words, an interpretation must not contradict itself. Circumventing this criterion by claiming, for instance, that a ‘higher state of consciousness' transcends the contradictions is not considered justified because it is an external validation that cannot be challenged.
Completeness requires that an interpretation should be able to account for all the cases relevant to the subject. This does not mean that every detail needs to be addressed, but that no fact at the same level of interpretation can be ignored. ‘At the same level' is a caveat implying that if an interpretation is general, it is not necessary to discuss some anomalies and aberrations that may be a result of specific circumstances or features. However, even small exceptions need to be taken into account if they cannot be explained at the lower interpretative levels. If this had not been the case in the past, we would not have the theory of relativity or quantum physics nowadays. The criterion of completeness is important because without it congruence and consistency can be achieved within a limited scope, by simply excluding those phenomena that do not fit.
Cohesiveness means that all the parts of an interpretation should be meaningfully connected. This implies that all its elements should relate to each other and are necessary. Nothing can be redundant; every part should have its place, purpose and function within the whole (it must effect the whole somehow). Cohesiveness is similar to Ockham's razor. Its main purpose is to prevent deux ex machina explanations. It also guards against jumping to conclusions, when the path from the given premises is unclear. Cohesiveness can compensate for not requiring material evidence to prove that something exists. Rather than asking for proofs, this criterion demands that something may be included only if doing so provides a more cohesive or more probable explanation than if it is not. For instance, the existence of unicorns cannot be refuted (only evidence for their existence can be). However, because there are no consequences that cannot be explained without involving unicorns, their existence can be considered irrelevant, and therefore unnecessary (until shown otherwise). Science and common sense already use this criterion. For example, the physical forces, energy, or even human thoughts are not self-evident and their existence cannot be falsified. But, they offer the best explanation for certain phenomena at the moment, so they are widely accepted.
- . Silver makes this point clear: ‘...most scientists believe in the theory [of evolution], but it has not been proved. Facts may be regarded as indisputable; theories are not' (1998, p.19).
- . The Principle of Credulity may be relevant here: ‘... it is a sound principle of reasoning to suppose that things are as they seem to be, unless and until proved otherwise' (Swinburne, 1991, p.145).
- . The difference between the Ockham razor and cohesiveness is in a degree: cohesiveness allows the introduction of a new element if an explanation which includes it is substantially more likely, even if it is not absolutely necessary. For example, a meaningful word made of small stones could be the result of chance, a random falling and rolling of the stones, but this is extremely unlikely. A far more plausible explanation requires a new factor that may not be present or detectable any more in a direct way: an intelligent being that deliberately made the word from the stones.