Can We Ever Be Certain About Scientific Theories?

by Mano Singham

A commenter to a previous article raised an interesting perspective that requires a fresh essay, because it reflects a commonly held view about how the validity of scientific theories get established.

The commenter says:

A scientist cannot be certain about a theory until that theory has truly been tested, and thus far, I am unaware of our having observed the evolution of one species from another species. Perhaps, in time, we will observe this, at which point the theory will have been verified. But until then, Evolution is merely a theory and a model.

While we may have the opportunity to test Evolution as time passes, it is very highly doubtful that we will ever be able to test any of the various theories for the origins of the Universe.

I would like to address just two points: What does it mean to "test" a theory? And can scientists ever "verify" a theory and "be certain" about it?

Verificationism as a concept to validate scientific theories has been tried and found to be wanting. The problem is that any non-trivial theory generates an infinite number of predictions. All the predictions cannot be exhaustively verified. Only a sample of the possible predictions can be tested and there is no universal yardstick that can be used to measure when a theory has been verified. It is a matter of consensus judgment on the part of scientists as to when a theory becomes an accepted one, and this is done on a case-by-case basis by the practitioners in that field or sub-field.

This means, however, that people who are opposed to a theory can always point to at least one particular result that has not been directly observed and claim that the theory has not been 'verified' or 'proven.' This is the strategy adopted by ID supporters to attack evolutionary theory. But using this kind of reasoning will result in every single theory in science being denied scientific status.

Theories do get tested. Testing a theory has been a cornerstone of science practice ever since Galileo but it means different things depending on whether you are talking about an experimental science like chemistry and condensed matter physics, or a historical science like cosmology, evolution, geology, and astronomy.

Any scientific theory is always more than an explanation of prior events. It also must necessarily predict new observations and it is these predictions that are used to test theories. In the case of experimental sciences, laboratory experiments can be performed under controlled conditions in order to generate new data that can be compared with predictions or used to infer new theories.

In the case of historical sciences, however, observations are used to unearth data that are pre-existing but as yet unknown. Hence the 'predictions' may be more appropriately called 'retrodictions', in that they predict that you will find things that already exist. For example, in cosmology the retrodictions were the existence of a cosmic microwave background radiation of a certain temperature, the relative abundances of light nuclei, and so forth. The discovery of the planet Neptune was considered a successful 'prediction' of Newtonian theory, although Neptune had presumably always been there.

The testing of a historical science is analogous is to that of the investigation of a crime where the detective says things like "If the criminal went through the woods, then we should be able to see footprints." This kind of evidence is also historical but is as powerful as those of futuristic predictions, so historical sciences are not necessarily at a lower level of credibility than experimental sciences.

Theories in cosmology, astronomy, geology, and evolution are all tested in this way. As Ernst Mayr (who died a few days ago at the age of 100) said in What Evolution Is (2001): "Evolution as a whole, and the explanation of particular evolutionary events, must be inferred from observations. Such inferences must be tested again and again against new observations, and the original inference is either falsified or considerably strengthened when confirmed by all of these tests. However, most inferences made by evolutionists have by now been tested successfully so often that they are accepted as certainties." (emphasis added).

In saying that most inferences are 'accepted as certainties', Mayr is exaggerating a little. Ever since the turn of the 20th century, it has been accepted that scientific knowledge is fallible and that absolute certainty cannot be achieved. But scientists do achieve a remarkable consensus on deciding at any given time what theoretical frameworks they have confidence in and will be used to guide future research. Such frameworks have been given the name 'paradigms' by Thomas Kuhn in The Structure of Scientific Revolutions (1970).

When scientists say they 'believe' in evolution (or the Big Bang), the word is being used in quite a different way from that used in religion. It is used as shorthand to say that they have confidence that the underlying mechanism of the theory has been well tested by seeing where its predictions lead. It is definitely not "merely a theory and a model" if by the word 'merely' the commenter implies a theory that is unsupported or untested.

So yes, evolution, like all the other major scientific paradigms, both historical and experimental, has been well tested.

Top of page