Appeal to Ignorance


Taxonomy: Logical Fallacy > Informal Fallacy > Appeal to Ignorance2

Affirmative Negative
There is no evidence against p.
Therefore, p.
There is no evidence for p.
Therefore, not-p.


[Joe McCarthy] announced that he had penetrated "Truman's iron curtain of secrecy" and that he proposed forthwith to present 81 cases… Cases of exactly what? "I am only giving the Senate," he said, "cases in which it is clear there is a definite Communist connection…persons whom I consider to be Communists in the State Department." … Of Case 40, he said, "I do not have much information on this except the general statement of the agency…that there is nothing in the files to disprove his Communist connections."3


An appeal to ignorance is an argument for a conclusion based on a lack of evidence. There are two forms of the argument―see the Forms, above―depending on whether the argument is affirmative or negative:


There are a few types of reasoning which resemble the fallacy of Appeal to Ignorance, and need to be distinguished from it:

  1. Presumptive Reasoning: It is reasonable to argue from a lack of evidence when there is a presumption for or against a proposition. For instance, in American criminal law there is a presumption of innocence of a defendant, and a corresponding burden of proof on the prosecution. If the prosecution fails to provide evidence of guilt, it has failed to meet its burden and the jury is supposed to conclude that the defendant is not guilty, even if the defense has presented no evidence of innocence.

    Similarly, the burden of proof is usually on a person making an unusual or improbable claim, and the presumption may be that such a claim is false. For instance, suppose that someone claims that the president is really a reptilian alien shape-shifter from another dimension, but when challenged can supply no evidence for this strange claim. It would not be a fallacious appeal to ignorance for you to reason that, since there is no evidence that the president is an alien, he probably isn't.

  2. The Closed World Assumption4: We sometimes have meta-knowledge—that is, knowledge about knowledge—which can justify inferring a conclusion based upon a lack of evidence. For instance, schedules—such as those for buses, trains, and airplanes—list times and locations of arrivals and departures. Such schedules usually do not attempt to list the times and locations when vehicles do not arrive or depart, since this would be highly inefficient and an unending task. Instead, there is an implicit assumption that such a schedule is complete, that all available vehicle departures and arrivals have been listed. Thus, we can reason using the following sort of enthymeme:

    There is no departure/arrival listed in the schedule for location L at time T.
    Suppressed Premiss: All departures and arrivals are listed in the schedule.
    Therefore, there is no departure/arrival for location L at time T.

    This kind of completeness of information assumption is called the "closed world assumption" in artificial intelligence research. When it is reasonable to accept this assumption it is not a fallacy of appeal to ignorance to reason this way.

  3. Auto-Epistemic Reasoning5: Another type of reasoning is called "auto-epistemic"―"self-knowing"―because it involves reasoning from premisses about what one actually knows and what one would know if something were true. The form of some such reasoning is:

    If p were true, then I would know that p.
    I don't know that p.
    Therefore, p is false.

    For instance, one might reason:

    If I were adopted, then I would know about it by now.
    I don't know that I was adopted.
    Therefore, I wasn't adopted.

    Similarly, when extensive investigation has been undertaken, it is often reasonable to infer that something is false based upon a lack of positive evidence for it. For instance, if a drug has been subjected to lengthy testing for harmful effects and none has been discovered, it is then reasonable to conclude that it is safe. Another example is:

    If there really were a large and unusual type of animal in Loch Ness, then we would have undeniable evidence of it by now.
    We don't have undeniable evidence of a large, unfamiliar animal in Loch Ness.
    Therefore, there is no such animal.

    Auto-epistemic reasoning does not necessarily commit the fallacy of appeal to ignorance since it is actually appealing to knowledge―specifically, self-knowledge.


Reader Response:

When reading up on Appeal to Ignorance I found I don't agree with your remarks around auto-epistemic reasoning. I hope you can clear this up.

What I don't get is how auto-epistemic reasoning is not a case of Appeal to Ignorance. In the examples, it looks like the circumstances are such that the reader should somehow feel that it's okay to break the rule. It seems to me that the rules of logic don't change if something is particularly unlikely or insane.

The reasoning in the example is inconclusive: even though reasonable people should assume there is no Loch Ness monster for lack of evidence, this lack of evidence is not conclusive proof of the monster's nonexistence.

The premise that all large and unusual types of animal result in undeniable evidence of their existence is false. For example, the existence of large deep-sea squids has long been denied for lack of undeniable evidence. Loch Ness is much smaller than the world's oceans and it has been researched far more than reasonable, but still finding evidence of a monster is a matter of chance. I'm convinced that the chances that Nessie is still eluding us are so small that it would be insane to operate under the assumption that she might be real, but that doesn't prove anything absolutely.―Iwein Fuld

You've argued yourself right up to the solution to this quandary, and there's only one further step to take. You say that reasonable people should reject the existence of a monster in Loch Ness based on a lack of evidence, but that this lack is not proof of its non-existence.

The step you need to take here is to realize that auto-epistemic reasoning is a type of inductive8 reasoning, rather than deductive. That is, auto-epistemic reasoning is not conclusive, and does not prove anything; at best, it makes it probable that its conclusion is true. However, as in the case of Nessie, sometimes the reasoning makes the conclusion so overwhelmingly probable that only "insane" or unreasonable people would refuse to accept it. So, while everything you say about Loch Ness and deep-sea squids is correct, all it shows is that there is still a bare possibility that a large, unknown animal lives in the Loch. Granted, but it's extremely improbable.

A further, important point concerns your remark that auto-epistemic reasoning "breaks the rule" against appealing to ignorance. A general fact about logical fallacies is that there are always exceptions to the rule, and this is especially true for informal fallacies, such as appeal to ignorance. That a logical fallacy is informal means that one cannot tell simply from the form of an argument that the fallacy has been committed; instead, one has to pay attention to the argument's content. In other words, there are non-fallacious arguments which have the same form as the fallacy.

I included the discussions of presumptions, the closed-world assumption, and auto-epistemic reasoning because these are all exceptions to the rule against appealing to ignorance. They are all types of argument that have the form of an appeal to ignorance, but do not commit the fallacy because of their content.

I hope that clears it up!


  1. Translation: "Argument to ignorance", Latin.
  2. S. Morris Engel, With Good Reason: An Introduction to Informal Fallacies (6th Edition) (St. Martin's, 2000), pp. 245-7. For advanced discussions, see:
    • Eric C.W. Krabbe, "Appeal to Ignorance", in Fallacies: Classical and Contemporary Readings, edited by Hans V. Hanson and Robert C. Pinto (Penn State Press, 1995), pp. 251-264.
    • Douglas Walton, "The Appeal to Ignorance, or Argumentum ad Ignorantiam", Argumentation 13 (1999), pp. 367-377.
  3. Richard H. Rovere, Senator Joe McCarthy (Methuen, 1960), pp. 106-107.
  4. Ellen Thro, The Artificial Intelligence Dictionary (1991).
  5. See: Thro, under "autoepistemic logic".
  6. M. H. Abrams & Geoffrey Galt Harpham, A Glossary of Literary Terms (8th edition, 2005), p. 281.
  7. God is Not Great, (2007), p. 150.
  8. That is, it's non-deductive reasoning: "inductive" is sometimes used in a more limited way to refer to probabilistic reasoning.