Is it so that whenever one gives an explanation to a question starting from 'why' or 'how' the explanation that we receive in return always is true in nature?

From one perspective, an explanation is something that simplifies a set of facts or makes it more understandable and comprehensive. On the other hand, we often use analogies and metaphors for explaining things and those are not always true.

Hence my question, do good explanations have to be true?

  • 1
    NO; true explanations are the "good ones". – Mauro ALLEGRANZA Nov 16 '18 at 12:15
  • 1
    No. In fact half of the problems of the human condition is that any explanation which can be rationalised can be accepted. Why does the world exist. God did it. Why is the economy suffering... Immigrants... And so the world turns. This is at the heart of sophism. Sell an explanation that diverts attention from the truth. – Richard Nov 16 '18 at 23:17
  • 2
    @MauroALLEGRANZA There is more than one true explanation, and some of them are pretty bad, I can think of a number of physics and calculus texts. Moreover, fruitful explanations do not have to be true, e.g. Young's and Maxwell's uses of ether. And the truth can be a mess that does not lend itself to good explanations (much of history). There is no relation between truth and explanatory value. – Conifold Nov 18 '18 at 1:29
  • Since good is qualitative and subjective, therefore a value judgment in terms of one's standards or priorities, a true explanation may be the "good" one for some people, but not for others. Likewise, a plausible explanation which isn't true may not satisfy everyone. en.wikipedia.org/wiki/Value_theory en.wikipedia.org/wiki/Joyce_Mitchell_Cook – Bread Nov 18 '18 at 19:22
  • A counter example, please! Do you have an example of a good explanation which is false? The exercise is interesting, even funny. Worth the effort. – Claude Brisson Nov 20 '18 at 0:38

This is an active topic of debate among professional philosophers of science today. Julian Reiss explains the problem nicely in his paper "The Explanation Paradox". His paper is focused on economics, but we can generalize the problem. Basically, the following statements are all highly plausible:

  1. Explanations must be true.
  2. Many fields of science make heavy use of simplifications, idealizations, and false assumptions in their research.
  3. These fields produce explanations.

But these 3 statements are inconsistent, so at least one must be false. Some professional philosophers think we should stick with #1; they are called "factivists," because they think explanation requires "facts," i.e., true claims. Factivists either reject #2 (these simplifications and idealizations don't play an "essential" role in the explanations) or #3 (Reiss gives the example of Anna Alexandrova and a coauthor, who simply think that economics doesn't produce explanations).

Non-factivists think that explanations don't require truth. Often — though not always — they argue that understanding is a subjective feeling or qualitative grasp of the phenomena. A false explanation can produce this feeling or qualitative grasp, perhaps even better than a true-but-non-explanatory description.

Two recent books worth mentioning are Angela Potochnik's Idealization and the Aims of Science and Kareem Khalifa's Understanding, Explanation, and Scientific Knowledge.

Potochnik is a non-factivist, so she rejects #1. On her view, the primary epistemic aim of science is understanding rather than truth, and idealizations actually promote this aim (because they allow us to cognitively grasp the phenomena). But she also argues that the causal patterns described by an explanation must be "embodied" in the phenomena, perhaps only "imperfectly." So, despite being a non-factivist, she still thinks explanations require some kind of representational relationship to the phenomena. (Unfortunately, I had trouble understanding how this is supposed to work in complex cases with "imperfect embodiment.")

Khalifa is a factivist (or, in his terminology, a "quasi-factivist"). But he argues that the truth requirement only applies to aspects of explanations that we believe to be true. We can adopt a different attitude, called "acceptance," towards the idealizations. We don't believe the false assumptions are literally true; but we accept them as convenient fictions, useful for simplifying calculations or bracketing complications (perhaps only temporarily).

I mention these two books because, while they adopt formally different labels, their views on truth and explanation are very similar. Both think that explanations must be "true-like," at least to some degree; but both also allow false elements to play a role in explanation.



'Good' explanation is ambiguous between 'plausible' and 'true' explanation. An explanation may be both but a plausible explanation is not necessarily a true explanation.

For instance, a plausible explanation might be this. I enter a room. A vase has fallen from a table. There are cat prints leading to the table; cat prints on the table, and cat prints leading away from the table. As well, there is a cat sitting in the room. A plausible explanation, indeed the best explanation available given the evidence, is that the cat has jumped on to the table and knocked the vase over. This is a perfectly rational, sensible inference; it provides a plausible explanation.

But it may be false. It may be that the cat walked up to the table, jumped up onto it when there was no vase, jumped down and walked away to the position in which it was sitting when I entered the room. What actually happened is that after the cat had walked, jumped, walked and sat down, someone had put the vase on the table, knocked it over, quit the room and left the cat to take the 'blame'. This is the true explanation, quite different from the plausible explanation which, given the data, is the best available to me. My data don't include any knowledge of the miscreant who put the vase on the table, knocked it over, and quit without letting anyone know what he or she had done.

  • Your example is wrong. Because even after understanding the situation, anyone can still approve to the fact that the first explanation is statically good, hence true at 99%. – Claude Brisson Nov 20 '18 at 0:48
  • @Claude Brisson. Sorry, I don't see your point. I never denied that the first explanation was good and remained good if no extra data showed up. I only said it was not true, which ex hypothesi it isn't - though there might never be any evidence of any agent other than the cat. That is, the extra data may never show up. I hope I have not misunderstood your point, – Geoffrey Thomas Nov 20 '18 at 9:11

If you wish an answer from the philosophy of sciences, perhaps this view of Stephen Hawking (1942-2018) might help:

We find ourselves in a bewildering world. We want to make sense of what is around us and ask: what is the nature of the universe? (...) To try to answer these questions we adopt some 'world picture'. Just as an infinite number of tortoises supporting the flat earth is such a picture, so is the theory of the superstrings. Both are theories of the universe, though the latter is more mathematical and precise than the former. Both theories lack observational evidence: no-one has ever seen a giant tortoise with the earth on its back, but then, no one has found a superstring either. However, the tortoise theory fails to be a good scientific theory because it predics that people should be able to fall off the edge of the world. [A brief history of time, "Conclusion"]

Truth, in a litteral sense, is not what mainstream (hard) science is after. It more modestly tries to find answers to questions that explain as many phenomena as possible and -- more importantly -- are not contradicted by already known facts.

For the second perspective, metaphors and simplifications (or diagrams, etc.) are not answers by themselves. They are tools used to help us grasp the answers. The function of those metaphors is pedagogic (i.e. as a tool for pedagogy).

At some point, however, one may need to drop them, to get at the essence of the theory. It is a matter of intellectual honesty. As physicist Feynman (1918-1988) pointed out:

We look at [the world]. That's the way it looks!... If I am going to tell you honestly how to world looks like, to human beings who have struggled as hard as they can to understand it... I am not going to simplify it, I am not going to fake it... make it [look] like a ball bearing on a spring. It isn't!

[Richard Feynman, Sir Douglas Robb Lectures, University of Auckland (1979); lecture 1, "Photons: Corpuscles of Light"]


If I don't understand a particular situation, then a good explanation will help me understanding, while a bad explanation will not help me or even confuse me.

An explanation can be true and not help me understanding, in which case it is a bad explanation. An explanation can be not true but help me understanding, in which case it is a good explanation.

Of course it depends how far away an explanation is from the truth. If it is too far away then it can make me think I understand the situation, but I don't actually understand it correctly.

For example: "The weight displayed by your bathroom scales depends on your body mass, multiplied by Earth's gravitational force. That's why if you used the scales on the moon with much less gravitational force, the displayed weight would be much less". That's close enough to being true, close enough to make you understand, but it leaves out the effects of centrifugal force, the force by sun, moon and planets etc. So it's not true, but a good explanation.


To me it seems clear that the answer is "it depends on who's asking". The way that I would explain the molecular basis of disease to my mom would be very different than I would explain it to my thesis defence committee. Why? Because in the my case of my mom it is sufficient for my mom understand that disease has a physical basis and not voodoo witch magic. While for my thesis defence committee it is sufficient that I demonstrate my understanding of molecular biology and biochemistry.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.