Should “scientific knowledge” be aimed at what is true or what works?

In order to address what scientific knowledge should be aimed at we must first establish what scientific knowledge is. It seems to me there are broadly two ways to understand the term. The first is that ‘scientific knowledge’ refers to items of ‘knowledge’ (we shall look at different definitions of knowledge later) that related to the fields we class as ‘sciences’ (which for the purposes of this essay we shall assume to the physical sciences as well as the social sciences - and avoid the question of what it means for a field to be scientific). Alternatively, we can view ‘scientific knowledge’ as a term in its own right that refers to various statements associated with scientific fields and does not depend on the definition of knowledge (the reasons why we might consider this odd approach will become clear as we progress).

Let us start by considering the first definition. The most obvious question to ask under this definition is what does it mean for something to count as knowledge. One definition that has been popular historically is knowledge as justified true belief (JTB). Under this definition for a person, P, to know a statement, S, there are three criteria: firstly, S must be true; P must believe S; and finally P must have justification for their belief in S.

The first proposition of this definition - S must be true - seems to give us the answer to our original question: ‘Should “scientific knowledge” be aimed at what is true or what works?’. Given that statements must be true to count as knowledge (at least under the JTB) scientific knowledge must be aimed at what is true, otherwise scientific knowledge is not knowledge at all. However, this leads us to have some problems when trying to class statements as scientific knowledge - primarily how do we know if a given statement is true?

We might be tempted to argue that we can establish a scientific statement as true by performing a series of experiments; however by doing so we run into two problems. Firstly, we run into the problem of induction which points out that all we have is past observations (albeit in some cases quite a lot of them) and this cannot establish a more general truth. To see why, consider the statement ‘All swans are white’. We might come to this conclusion based on us going out to a lake and seeing one thousand swans all of which of white. However, this not does not rule out the possibility of a black swan and in fact since people have observed black swans our original statement is false - no matter how many white swans we observe.

Science operates in much the same way - while experiments may provide favourable evidence for a particular theory - no matter how many we do we cannot establish the truth of the statements. Secondly, we can even take this argument further and consider skepticism. Skepticism aims to argue that we cannot establish the truth of certain (and in some cases, all) statements. Various examples include dream skepticism (mentioned by Descartes in Meditations) or Putnam’s Brain in a Vat argument. For our case let us consider a point that underlies many skeptical arguments: our senses can be deceived.

An example of this are hallucinations that some people experience: people claim to see or hear things that the rest of us can’t which suggests that at least someone’s senses are being deceived. Or an example that might be more familiar to us is when you have a dream so realistic you’re not sure if you’re awake or actually dreaming. Now, if we cannot trust our senses then this poses a challenge for science because even our evidence in favour of our theories cannot be trusted, and as such it seems nearly, if not completely, impossible to establish, the truth of scientific statements.

While there are arguments to tackle this problem of skepticism, instead let us assume that we have a successful solution and we have even solved the problem of induction hence we can establish certain scientific statements as true. Now by considering the consequences of this we may reach some unexpected conclusions.

By way of an example let us consider two theories of gravitation: Newton’s theory of gravitation and Einstein’s General Relativity (GR). And let us assume that General Relativity is completely true and absolutely correct. Now while Newton’s theory and GR agree at small scales they differ when things get very big or really fast. So this means that Newton’s theory is not true (at least not without various caveats) which means it cannot count as scientific knowledge.

Does this not seem peculiar? We have a theory that is still practically useful and was the best theory we had for gravitation for several hundred years but at no point did it count as scientific knowledge (since it was never true). We cannot say that Newton had knowledge of gravitation. This seems to be very odd and not in keeping with the way we talk about scientific statements - we often talk about knowing various scientific facts even though they may be superseded by another theory in the future.

Perhaps this is just arrogance and a misuse of language but it might also suggest that we need a different definition of scientific knowledge - one that does not depend on truth.

To aid with the development of this idea let us consider Kuhn’s use of paradigms and paradigm shifts. For Kuhn, a paradigm is a set of theories and beliefs that lay out the puzzles to be solved by other scientists. Occasionally, when problems begin to accumulate within a paradigm - either with puzzles unable to be solved by the greatest minds or the build up of unexplainable empirical evidence - a new paradigm must be found and the transition is known as a paradigm shift.

Now let us consider the use of the term ‘scientific knowledge’ within a given paradigm. It seems the term makes most sense as something that refers to the statements that establish the paradigms and the solutions to the puzzles solved within it; since these are statements that are considered to be scientific in a given context. This allows us to say that both Newton and Einstein (from the earlier example) both had scientific knowledge of gravitation as within their respective paradigms as they knew the solutions and established practice for the paradigms in which they worked.

However, we do need to establish some more strict criteria for what does and doesn’t count as scientific knowledge within a given paradigm. We must include what Kuhn describes as the ‘Theory Choice’ - the underlying theories that describe that paradigms. But what of the other statements - what makes one of them count as scientific knowledge? I would argue that if the statement is a solution to a puzzle posed by the paradigm then it should count as scientific knowledge as it is the aim of scientists within a given paradigm to solve these puzzles.

This leads us to respond to “Should “scientific knowledge” be aimed at what is true or what works?” with the answer of what works within a given paradigm. This is because scientific knowledge is relative to a paradigm so cannot be absolutely true. Now this relativist approach to scientific knowledge has a couple of unusual properties when talking about knowledge. Firstly, it does not depend on an absolute truth (which was our intention), however, we have also ended up with a social component of knowledge. Since a given paradigm is decided upon by the community of scientists at the time, whether or not a statement counts as scientific knowledge depends on their choice and as such our definition includes a social component.

While some may it find it odd that knowledge has this social component, I would argue this is in keeping with how many people use the term knowledge in their day to day lives. For instance, a religious group may talk about having knowledge of God, whereas an atheist group may not consider those statements as knowledge.

Overall, it is clear that scientific knowledge must be aimed at what works within a given paradigm. Furthermore, the relativist and social aspects of scientific knowledge that have been mentioned may provide us with techniques that could be expanded into a more general theory of knowledge (perhaps based on some form of coherentism)