contact@onbelief.orgThe value of certainty, testability, verification and falsification
"Consistency is indifferent to truth. Once can be entirely consistent and still be entirely wrong"
Steven G. Vick in 'Degrees of Belief'‘‘If we begin with certainties, we shall end in doubts; but if we begin with doubts, and are patient with them, we shall end with certainties.’’ Sir Francis Bacon 1605 ( see source >)
In matters of logic we can often find ourselves in a position where we ask is a proposition about the world true or false. In other words we are faced with a dichotomy, or choice, of opposites. There is of course the possibility that a false-dichotomy is being applied and that neither a false nor true belief has a useful meaning. Ripe oranges are blue or red. No, they are orange.
Opposites can be thought of in terms of numbers (or more precisely as a numerical probabilities). In other words we use numbers to construct arguments of a logical nature. If we examine the outcome probabilities of a logical belief dichotomously we might assign a probability of zero if the idea is false (p=0 or 0%). Conversely if the idea is thought to be true we might award it a probability on one (p=1 or 100%). Take for example the descriptive belief that god exists. The devout atheist, would give that idea a probability of 0 and the fundamentalist believer would give a probability of 1.
Disbelieving Popper
The philosopher of science Karl Popper wrote that it was not possible to definitely prove a theory and its associated hypotheses to be true only to falsify it, i.e. establish that p=0 (see wikipedia article on falsifiability). 'Truth' could then regarded as the set of ideas that have not been falsified despite attempts to do so. In other words Popper suggested that verification and falsifiability are some how different or that there is an asymmetry between them. Logical positivist philosophers, such as AJ Ayer, have long dismissed the asymmetry of falsification, however a major reason for dealing with this matter is the importance placed upon the concept by many practicing scientists of today and it's recent legal relevance in cases related to the teaching of Intelligent Design in American schools. One of the most important objections to the asymmetric reliance on falsifiability is as follows. To state that a theory can be described as scientific because it is falsifiable (i.e. potentially false) does not discriminate scientific theories from ideas that are utter nonsense. (See the section entitled 'Defining as science' in the wikipedia article on 'Intelligent Design'.)
When is a theory scientific ?
A theory can be regarded as scientific if it is based on observation and passes verificational test of prediction. A theory is not seen as scientifically (or predictively) useful if it fails. Science is not the accumulation of a set of least bad theories but the set of current theories that allow best coherent description of current observations and most reliable predictions of future events or experiments, be they pure or applied. For example, the pharmacology of analgesia is not defined as 'the set of tested drugs that do not relieve pain'. No pharmaceutical company would survive if that was the case.
We should of course be suspicious of beliefs where logic suggests that they could not be falsified under any circumstances. Even theories which only seem difficult to test should be suspected. The theories of the Cosmic Big Bang and biological evolution ( for evolution see Links page section 17) fall into the later category but that does not make them unscientific for they rely instead on further observational verification and coherence. To suggest that falsification would be difficult because all current evidence suggests these beliefs are 'true' entirely misses the point. Consistency of current evidence merely stresses the importance of verification and a probabilistic way of believing. Biological evolutionary theory for example relies largely on consistency and coherence as well as the mechanistic explanations of biochemistry and genetics. Nevertheless, some theoretically falsifiable predictions can be made with respect to evolution even if difficult to test. We can for example predict the finding of intermediate forms in the fossil record, for example from fish to land animals and as yet undiscovered types of fossils consistent with different stages of primate evolution. In that sense we are predicting consistency of future observations. In terms of consistency we can predict that we will never find, as JBS Haldlane said, "fossil rabbits in the Precambrian era". To give another more culturally relevant example we can predict that no geologist will ever find human fossil bones formed within the same rock strata as those of the cretaceous period theropod dinosaur Tyrannosaurus rex because evolutionary theory tells us that humans evolved after T. rex became extinct. These examples demonstrate the value of falsification, whiles simultaneously showing the weakness of relying exclusively upon it. In addition they highlight the value of 'consistency of verification'. In terms of supporting the genetic component of the theory of evolution, for example, we can predict viral mutation in diseases like HIV infection and test for its appearance by looking for as yet undiscovered genetic coding sequences.
Asymmetric reliance on 'Isolated falsification' by that analysis seems a largely unproductive view of the sciences and how they progress. Productive descriptions of the world through science rarely if ever emerge purely from falsification. Indeed for the practical scientist and technologists, like the engineer or doctor verification, is often heavily relied upon. The engineer would not regard his theoretical calculations concerning the vibrational moments and torsional stiffness of a bridge as useful if the bridge were to buckle and collapse (see the tacoma narrows bridge failure and a list of bridge disasters at wikipedia). Clearly as bridge users we hope that failures or insufficiency of bridge-making theory will have motivated structural engineers to elaborate or correct their beliefs due to these falsification instances. We then rely on successful verification as we drive across the replacement bridge. Falsification clearly has a part to play both in terms of theory and practice and acts as one stimulus to correction and thus future verification.
The Infamous Philosophical Swans
Knowledge is not advanced by the negation of falsely certain propositions. Take for example, the now famous philosophical example of the false assertion of certainty that "all swans are white". This descriptive proposition cannot of itself be an advancement of knowledge. Whereas the description that some are black is a contribution to knowledge since it distinguishes black swans from white and other possibilities such as pink or green. The finding of black swans is of value in itself. (Indeed science distinguishes itself from the unproductive ruminations of modern philosophers on classical texts by constantly producing novel and increasingly elaborate descriptions of the world and new explanatory theories.) In addition the finding of black swans says nothing about the previous existence of extinct pink, flamingo-like swans or the future evolution of finch-like green swans. The value of the description of black swans as a falsification of the unreasonably certain hypothesis that 'all swans are presently white' is almost, if not totally, nil.
Consider also the possibility that all black swans might have recently died. The constancy of falsely certain beliefs can therefore be another reason for their rejection. Certainty of either verification or falsification of hypotheses in the absence of complete and definitive evidence should be rejected. Even then, what appears to be 'definitive' evidence should always be regarded as provisional.
By contrast the formulation of non-constant less than certain propositions concerning non-constant complex systems such as living things can have value. Some swans might presently be a colour other than white, or might have been so in the past or might be in the future under different conditions. The verification of that hypothesis comes by finding swans of a different colour. Equally, rather than assert that no human can live to an age of 180 years it is more productive to state that the probability is extremely unlikely by analysing the distribution of age at death in well described populations. Better still, in formulating a probability density function about the age at death, useful testable predictions can be made about the occurrence of death in population under study.
'Luminiferous ether' and astronomical ideas
In 19th century physics it was thought that light required a media through which it could be propagated. Light would thus be behaving like sound in the air or waves on water. 'Empty space', it was thought, was not enough for the carrying or propagation of light. This hypothesized medium was known as the 'Luminiferous ether'. A famous experiment using light beams and mirrors now known as Michelson-Morley experiment was carried out and then refined and developed by other scientists. Despite all attempts to provide evidence of a Luminiferous ether all experiments failed to confirm its existence. Indeed what data were obtained could be explained by a newer and entirely different theory, special relativity. In that context the result of the 'failed' prediction was then reformulated as verification of another newer belief. The initial 'failure' cannot however be legitimately viewed as falsification because, as Dr Carl Sagan said "absence of evidence is not evidence of absence" when invoking the fallacy of the 'argument from ignorance'.
In the modern world of the physical sciences the astronomer would not be expected to assert with false certainty that all galaxies must take one of the geometric forms already discovered. Similarly it would not be beneficial to postulate with false certainty that the sun was created in a 'day' and will last for ever. It is however useful to observe evidence of red giant stars, dwarf stars, supernovae, pulsars, galaxy shape, cosmic microwave background radiation, and the red shift of star light. Then it is possible to formulate Hubble's law and postulate the existence of relativistic Doppler effects, black holes, gravitational lensing, dark matter and the Big Bang. In so doing we can understand something of the probable evolution of stars and galaxies. We are not diminished by uncertainty in these matters but can instead celebrate the descriptive achievements of modern astronomy.
Falsifiability and testability
Falsifiability should not be confused with testability. The value of a descriptive proposition cannot be described as scientifically usefully if it is not testable. Prediction and testing can result in either refutation or support for a belief. Both make contributions to understanding. It therefore seems better if the 0 and 1 states of probability are regarded as symmetrical with regard to observationally based theories. In other words absolute certainty, in the world of objects rather than logical propositions, has the same meaning when considering either the truth or falseness of a descriptive theory. [For an alternative and perhaps self-contradictory view and summary of falsification see Out of Error: Further Essays on Critical Rationalism by David Miller page 254 and subsequent pages]
There are of course dangers in viewing the universe as we know it entirely by purely symmetric propositions. We believe that in the macroscopic world there is 'an arrow of time'. [See Thermodynamic Asymmetry in Time in the Stanford Encyclopedia of Philosophy]. The direction or arrow of time has important consequences for all descriptive propositions and predictions. In a world where time could be reversible cause and effect would not be distinguishable and prediction would not be meaningful. Indeed it is because of the arrow of time that verification has meaning.
It seems sensible to reject the universality of descriptive propositions as this is an assertion of certainty under all circumstances even those that are unknowable. For example, consider the belief that 'if I drop an apple it will always fall downwards'. Would you believe the universality of this descriptive proposition if you were orbiting the earth in a spacecraft and experiencing conditions of microgravity? (The technical answer would of course be yes it is falling at the same speed as the earth is turning and so continues to orbit, however that would not make any sense within your frame of measurement or reference) In a related fashion we can say that, when taken to the extremes of analytical precision, most if not all complex systems should be regarded as non-linear.
The Undesirability of Certainty
Certainty is both unnecessary in theoretical terms and undesirable for current and future development of theory and observation. The making of new observations drives uncertainty and results in elaboration of theory. In the next section, concerning 'strength of belief', the case will be examined for thinking of descriptive, rather than logical, ideas as deserving of the status of being probably true or probably false. In the real world we often have incomplete or imprecise evidence concerning the nature of some aspect of the world. In such circumstances its seems wise to be less than certain.
Reader responses
For 2 reader responses go to response page 6 >
On the Nature of Belief
www.onbelief.org
Scotland, 12th October 2007 and thereafter
Copyright 2007 onwards
