Science, truth, and honesty
Andy Wills
This lecture series starts by considering two fundamental questions — ‘what is truth?’ and ‘what is science?’. We start with the nature of truth.
Truth as a property of claims
It’s not possible to get anywhere if one considers truth as some kind of general ineffable property of the universe. Truth is, at heart, one property of a claim. Someone makes a claim and someone (the same person, or someone else) attributes the property ‘true’ to it. Or the property ‘not true’ (false). It is also possible to attribute some level of belief, perhaps using probability. Finally, it is possible to ‘opt out’ and say you have insufficient information to make a sensible judgement.
Subjective and objective claims
Truth is a property of claims, but there are different types of claims — a recurring theme in this lecture. The first distinction we need to make is between subjective and objective claims.
Subjective claims are those whose truth differs for different people. So, for example, the truth of the statement ‘My favourite colour is blue’ differs depending upon who says it. Objective claims are those whose truth value is not affected by who says it. ‘Charlotte Smith’s favourite colour is blue’. This is true, or false, irrespective of whether I say it, you say it, or Charlotte Smith says it.
Note that the term ‘subjective’ is among the most widely mis-used, particularly when referring to psychology. For example, the claim ‘I think chocolate tastes better than cabbage’ is subjective, but the claim ‘chocolate tastes better than cabbage’ is objective, and is amenable to scientific test (for example, taste ratings taken over a representative sample of individuals).
Vague claims are subjective
Critically, a claim is not objective if it is too vague to be tested. For example, ‘smoking is wrong’ - not objective because ‘wrong’ has a number of different meanings.
Scientific claims
At its heart, science is the process of making scientific claims and then attempting to determine whether those claims are true or false. Scientific claims are objective rather than subjective, see earlier. Science also limits, or should limit, the claims that it makes to those whose truth or falsity can, at least in principle, be clearly determined.
In general, this means scientific claims are descriptive rather than prescriptive.
Descriptive versus prescriptive claims
Science’s focus on descriptive claims does not exempt scientists from having moral or ethical principles, or from wanting to change the world in particular ways. But scientists choose to investigate the issues inherent in prescriptive claims by converting them to descriptive claims that are relevant to the prescriptive point.
For example, a scientist who wishes to make the claim ‘abortion should be made illegal’ looks for a descriptive claim that would support the prescriptive claim. For example, if one wished to support the outlawing of abortion, one might make the claim ‘feotal pain receptors have developed by eight weeks gestation’, in order to support your prescriptive claim. If one wished to argue against that prescriptive claim, you might then make the descriptive claim ‘foetal pain receptors are not connected to the brain until at least 20 weeks’.
Similarly, in order to support the prescriptive claim about drink-driving, you would make some kind of descriptive claim. For example, ‘Risk of car accidents doubles at 80mg/100ml blood alcohol (UK drink-driving limit)’
So, science is not about avoiding societally difficult questions. It is about making claims that can be reliably examined, in order to help answer difficult questions.
Absolute versus contextual claims
To summarize, scientific claims are objective, and they are usually descriptive. However, scientific claims are seldom absolute.
Absolute claims are invariant. They hold always. Their truth value is not conditional on circumstances. They are not conditional on time or place.
Contextual claims hold under a defined set of conditions. Scientists should define those conditions, but do not always do this. For example, consider the claim, ‘the leadership positions that women occupy are less promising than those of their male counterparts’. This is not intended to be an absolute claim. It is, rather, a claim about a current state of affairs. If this claim is true now, but no longer true in 20 years, perhaps due to increased awareness of the issue, this does not undermine the truth value of the original claim – as long as it is seen in the appropriate context.
In general, scientists wish to investigate claims that are, as far as possible, context independent. In other words, as close to being absolute claims as is achievable. Indeed, it is hard to think of claims as scientific in the broader sense if they are too context-specific. For example:
‘The leadership positions that women in 2003 in UK FTSE100 companies occupy are less promising than those of their male counterparts’
Such a claim is technically scientific, but the truth being claimed is so contextual that it is not possible to, for example, make any further predictions.
Scientific claims and truth
Scientific claims are objective, descriptive, and relatively context independent. They also have a number of other properties. First, some scientific claims are true and others are false. A claim that is false can still be scientific if it meets the other criteria.
Observable, measurable states
Second, scientific claims are based on observable, measurable states. Making sure your claim is descriptive rather than prescriptive helps to make it observable and measurable, but doesn’t always get you all the way there. For example, consider the claim ‘impulsive people are more likely to be criminals’. Being a criminal is a state that is observable and measurable in a number of relatively uncontroversial ways. For example, if you have been convicted of a crime then you are a criminal in an observable and measurable way.
Impulsivity is a fairly vague concept that has to be somehow translated into something directly measurable. One of the contributions of modern psychology has been to develop explicit measures that can be clearly assessed for reliability and validity. In the case of impulsivity, there is a 30-item standard questionnaire called the Barratt Impulsivity Scale (BIS). So, our claim becomes:
‘People with one or more criminal convictions score higher on the BIS than people without a conviction’
Independent replication
Third, scientific claims must be expressed in such a way that they permit independent replication. For example, claiming that Willsian Therapy reduces depression is not a scientific claim if Wills is the only person who can perform Willsian Therapy.
Scientific claims
Finally, scientific claims have to be falsifiable. This basically means that it must be possible to imagine an outcome of an experiment or other study that would lead to the conclusion the claim was false. If this is not possible, the claim is unfalsifiable and hence unscientific. Note the difference between falsifiable (one can imagine a set of circumstances in which it could be shown to be wrong), and falsified (one has demonstrated that it is wrong). Good, productive, science consists of the collection of claims that are falsifiable but not (yet) falsified.
Science and dishonesty
To summarize, science is about making scientific claims, and we’ve looked at what makes a claim scientific (or unscientific). But science, like all human endeavours, is about more than this. It is also about being part of a culture that holds certain behaviours and processes in high regard. Transgressing these rules leads to the normal cultural sanctions — isolation, exclusion and, sometimes, expulsion. In some cases, these cultural aspects represent A WAY of doing science, rather than THE ONLY WAY it could be done.
The first cultural norm of modern science is honesty. This is most easily defined by its opposite — dishonesty. It is dishonest to partially report your results if your intention is to obscure results that are inconsistent with your claim. It is dishonest to choose a form of data analysis because it gives you the result you want. It is dishonest to publish the same data more than once without acknowledging you have done this. And it is dishonest to say you collected some data when actually you just made it up!
The reason honesty is such an important cultural norm in science is that dishonesty gets in the way of reliably evaluating claims, and wastes peoples’ time. Science is a highly social cooperative activity.
Failures to replicate
Another important cultural norm in science is scepticism. Scientists should welcome it when others are skeptical about their claims, because such scepticism can lead to close examination of the evidence, which in turn advances our understanding. Essentially, being scientific is about soliciting and welcoming criticism. This is something that many of us, including me, still find emotionally quite difficult.
One recent development in scepticism is the increasing emphasis on not only scientific claims permitting independent replication, but the importance that such independent replication is actually conducted. Recent meta-analyses suggest that a half to two-thirds of results in psychology fail to replicate.
As homework, you were asked to investigate whether Ludmer et al. (2011) had been successfully replicated. Some answers to that question can be found here.
Further materials
The following information is relevant to the topic of the lecture, but was not discussed within it. Reading to the end should broaden and deepen your understanding of the topics we covered.
Relativism
When one asks ‘what is truth?’, one generally is asking what it means to say that an objective claim is true. One possibility is to avoid the question by asserting that there are no objective claims – all claims are subjective and hence the truth value of all claims depends on who is speaking them.
This position about truth is called relativism. It’s not an uncommon view, even amongst academics, but it’s pretty problematic.
First, any definition of relativism is self-defeating. For example, relativism is the belief that ‘all truth is relative’. This claim is, in itself, universal and so proves that not all truth is relative! The same problem exists for expressions such as ‘No one can say what is right or wrong’.
Another real problem with relativism is that it seems to preclude the possibility of being wrong. If truth is simply what you believe, then your beliefs cannot be in error (they can however be incoherent, see below).
One can avoid these paradoxes by saying, for example ‘some truth is relative’. This seems undeniably true, but it still leaves us with the question - what does it mean to say an objective claim is true?
Correspondence and coherence
The answer is really quite prosaic. Truth can be defined both in terms of correspondence and coherence. Anyone who has ever watched a crime programme on TV knows this.
Correspondence Theory states that Truth is determined by the correspondence between what is claimed and what is observed. An murder suspect claims to have been at home between 10pm and midnight, but he was seen by a reliable witness some miles from his house at that time. The murder suspect’s objective claim is judged to be false because it does not correspond to observations.
Coherence Theory states that Truth is determined by the coherence between new claims and other justified claims. The Defence claims that the apparent murder victim actually committed suicide. A number of Prosecution witnesses testify that the victim had no history of depression or other mental illness, and appeared happy in the days and weeks leading up the attack. The Defence’s claim seems unlikely to be true, not on the basis of anyone having observed how the injuries were inflicted, but on the lack of coherence between this claim and other justified claims.
Open-access publication
Science is highly social and cooperative. Collecting data, and evaluating claims, and then keeping the results of that evaluation to yourself is generally frowned upon. There are two important recent developments that have emerged from this belief in the importance of publication.
The first is a realisation that the traditional method of academic publishing is getting in the way of the reasons we should be publishing in the first place. Generally, scientists publish in scientific journals. Subscriptions to these journals cost thousands of pounds a year, and are generally only held in university libraries, which are often not open to the public.
The solution is open-access publishing. This means making your publications freely available at no charge. There are a number of ways to achieve this. For example, every scientific paper I have ever written is available for free download from my website.
The second recent development related to publication is the realisation that the internet now permits scientists to make not only their publications, but also their raw data, freely, publicly, and permanently available. This is really incredibly important. It is one thing to run a study again because you want to independently replicate it. It is quite another to have to run it again because the original author didn’t analyse the data properly and the raw data no longer exists.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International Licence.