Let’s take the event of a leaf falling down from a branch. Currently, we have a fully naturalistic explanation for how this happens.

Theoretically though, one can come up with an infinite number of hypotheses that may also explain this fact. For example, one may entertain the hypothesis that an invisible demon dropped the leaf. Or, one may postulate that this demon or some supernatural force didn’t directly cause it but indirectly through some unknown mechanism determined it to happen in advance.

Now, how does one analyze or measure the plausibility of these theories? Assigning a probability seems to be a common way of assessing this, but how does one arrive at a figure? What would it mean to say that there’s a 3% chance of a demon causing the lead to fall? That demon either did or didn’t.

Let’s say the probability measure is meaningless. How else should one analyze how reasonable it is to believe that theory? How much credence should one put in a theory like that?

Furthermore, is it possible or can it be argued that some of these theories may be incoherent or in other words have zero probability? It seems difficult to disprove a theory of an invisible demon pushing down a leaf. But can one argue that this is impossible in virtue of the fact that any physical effect seems to have a physical cause and thus a “being” must atleast itself be physical and arguably detectable?

  • For your titular question of some esoteric empirical theories with no direct evidence such as string theory or quantum loop methods you'd have to find indirect evidence to be able to be falsified, unless you have some kind of super intelligence... Aug 23 at 23:10
  • Gopd question. Welcome to my world. Sep 23 at 7:19

4 Answers 4


It is certainly plausible to suppose that trying to reduce plausibility to probability number won't help.

Direct evidence is not the only resource. In addition to looking for indirect evidence, I would suggest looking

  1. at the details of the theory, such as, how and why does the demon push the leaf down? (Of course, if the latter question is relevant, that may indicate that this is not a scientific theory.)

  2. the range of phenomena the theory is supposed to apply to - a) whether it explains all the cases explained by existing theories and b) whether it explains any cases not explained by existing theories. a) would be essential and b) would be important. (This would rule out ad hoc applications of the theory).

  3. whether it seems likely to support theoretical research and technological developments.

  4. whether it is compatible with other existing theories in the discipline.

  5. whether it is intellectually satisfying, or elegant. (This would include application of Occam's razor.)

Sadly, I'm not aware of any serious philosophical discussion of this topic. SEP - Possibility and Plausibility Theory seems to reduce plausibility to probability. IEP - Simplicity in the Philosophy of Science seems to reduce it to simplicity (which seems to be equivalent to Occam's razor). Neither of which are much help.


How does one “measure” the plausibility of theories with no direct evidence?

Answer: By defining what "plausibility" is relative to how you define a measurement and then either inferring from your measurement that the theory either is or is not plausible, regardless of whether or not there is or is not any direct evidence.

It can be debated whether or not how you've defined your terms is credible, if but for persons to test your theory.


For a theoretical model to be plausible requires that it be taken seriously by practitioners in the particular field of inquiry.

This happens when the model's predictions are a close match to experimental data, or when the model's predictions furnish a logical explanation for puzzling data in another related field of inquiry.

It also happens when the model predicts the existence of something new- and when that thing is specifically searched for, it is indeed found, with the very characteristics that the model predicted.

It is possible that a new model is not taken seriously because there simply isn't any data to compare it with because the techniques to do so haven't been invented yet. Then, when the data comes in, the correspondence is established, and a Nobel prize is given out.

  • Or it happens when all the doctors who were alive in 1820 finally get around to dying, so that germ theory and "removing blood from the sick and injured does not contribute to positive health outcomes" can finally displace miasmas and professional exsanguinations, centuries after any serious examination of the evidence would have shown that the former were plausible, and the latter were not.
    – g s
    Aug 24 at 5:47
  • @g s, fortunately I am talking about physics and not physicians. The practice of medicine might not be an exact science today but back then it wasn't a science at all. Aug 24 at 5:55
  • I would say that expert opinion is definitive insofar as experts are generally better at understanding what predictions a model might make, and what existing measurements it would have to conform to - and then in proportion to the degree to which the experts want the sort of thing that they could get by being right about it. Physics has mostly well-aligned incentives and its models and existing measurements are extremely reliable and hard to understand without specialized training, so sure, for physics, I'm not going to take anything seriously that the practitioners don't.
    – g s
    Aug 24 at 6:04
  • My point is that evidence is king, and the value of expert opinion is circumstantial. You seem to be saying that expert opinion is king because evidence will always change it - which is at most a temporary and local state of affairs, even in physics. Consider e.g. relativity under Stalin.
    – g s
    Aug 24 at 6:21
  • @g s, my point in response would be that relativity under Stalin didn't qualify as science, but I understand what you are saying. Aug 24 at 16:36

How does one “measure” the plausibility of theories with no direct evidence?

By designing experiments that provide the evidence required to evaluate plausibility.

Theories and experimentation are a pair. One is useless without the other.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .