Is there a term for this type of logical fallacy?

One person asserts that theory A would predict outcome X. We have observed, however, that outcome Y occurred and not outcome X. The first person asserts that this is evidence that theory A made a bad prediction. A second person asserts that theory A would not predict outcome X, and therefore outcome Y is not problematic for our faith in theory A.

Is there a type of fallacy where a person dismisses the possible ex ante prediction of A-->X based on ex post knowledge, fitting their interpretation of the theory and its predictions to observed outcomes.

  • 3
    @Mitch We're duplicating effort :-) I'd also suggest various types of ad hoc hypothesis as areas for further research. "No true Scotsman" fallacy - saving a generalization by making a change to a definition. Moving the goalposts - saving a theory by changing it to dismiss contrary evidence. Special pleading - saving a theory by making an unsupported claim that contrary evidence is actually a special case. The choice probably depends on the specific rationale the second person is using to exclude outcome Y, so more information on the rationale would be helpful here.
    – MetaEd
    Dec 11, 2017 at 16:43

3 Answers 3


I'm not sure that it is a logical fallacy. If theory A predicts outcome X, and outcome Y occurs, then there is obviously something wrong with theory A. This is neither an example of post hoc ergo propter hoc, nor denying the antecedent.


An argument that theory A is still correct, even though outcome Y occurred, would be an example of post hoc ergo propter hoc. This is sometimes called the post hoc fallacy (as the Wikipedia article states). However, the fact that outcome X was predicted makes this argument rather silly.

Your scenario is very commonly used by end of the world fantasists (who won't admit their mistakes).

  • Yes, in fact the scenario described by the OP is pretty much how inductive scientific reasoning proceeds in general. If you have a well-established theory, you don't ditch it at the first sign of opposing data - you just tweak the theory accordingly. If you then deny that you've tweaked it, it's not a logical fallacy... it's just a lie.
    – ArchContrarian
    Dec 11, 2017 at 13:57
  • Maybe I wasn't clear enough, the second person is stating that theory A would not predict X, and therefore there is nothing wrong with theory A. The first person would agree with you, not the second. The first person's view suggests that something is wrong with theory X, whereas the second person is saying that there is nothing wrong with theory X.
    – Dr. Beeblebrox
    Dec 11, 2017 at 14:10
  • 1
    For there to be nothing wrong with theory A, it should predict both outcomes. If it doesn't, then it is incorrect. Arguing that it is sill correct if Y occurs would be an example of post hoc ergo propter hoc.
    – Mick
    Dec 11, 2017 at 14:15
  • 1
    ... and a theory that states that X either occurs or it doesn't is not much of a theory.
    – Mick
    Dec 11, 2017 at 14:21

A fallacy is a formal error in argument; there are many different fallacies each of which instantiates a different invalid argument form. There need be no fallacy in the example you describe.

  1. In a particular case no fallacy is necessarily involved. Suppose theory A is true, and on a particular occasion X is predicted from A. A doesn't happpen, Y does. This could indicate merely that a material mistake has been made about what is predictable from theory A. This mistake may be simply an empirical error - a mistake about the facts or the condiitons in which A is applied - and this error need instantiate no invalid argument form (such as 'If p then q; q, therefore p').

  2. However, if we generalise and every prediction made from A is false, and is replaced ex post facto - after the event - by a different outcome (an outcome different from that predicted from A), then (i) A is de facto irrefutable, since nothing on this revisionary ex post facto tactic can falsify it, and a theory that can't be falsified is not (normally) good science or (ii) A is being applied by consistently incompetent practitioners. Again, no invalid argument form need be instantiated.

On fallacies see C.L. Hamblen, 'Fallacies', London : Methuen, 1970 or any logic text available. The reference to falsifiability does not commmit me to Popperian falisificationism. I only said 'normally'. Nor, for the record, do I assume that a complete enumeration of invalid argument forms is possible but among the forms with which I am familiar I do not find one that fits your example. An excellent question, btw.


This appears to me to be a fallacy of Moving the Goalposts.

The scientific method relies heavily on the predictive power of a theory. If the theory is unable to predict and model the real world, it can only be treated as theoretical and unusable for day to day. If proven false, it should be abandoned.

When a theory makes a prediction for outcome X and such an outcome does not happen, it is not uncommon for the scientist to adjust the theory to suit the observed outcome Y. To say that the original theory was correct in spite of this though, would be misleading if not downright dishonest.

You must log in to answer this question.

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