Category: epistemology
Type: thought experiment
Associated thinker(s): Fred Dretske
Original publication or approximate period: Dretske, “Epistemic Operators,” The Journal of Philosophy 67, no. 24 (1970), 1007–1023
Difficulty: Intermediate
Themes: closure, skepticism, knowledge
Related experiments: Fake-Barn Country; Brain in a Vat; Gettier Cases
You are at a zoo, standing before a pen marked “zebras.” The animal in the pen is a zebra, and its appearance and the ordinary setting lead you to believe that it is one. The claim that it is a zebra entails that it is not a mule cleverly disguised to look like a zebra.
Do you know that it is not such a mule? Your zebra-in-a-cage visual evidence would also be present in the stipulated disguise case. In Dretske’s example, the possibility is used to test what that evidence lets you know; it is not a report that zoos actually disguise mules, or a claim that this is a likely event. Whether the mule scenario counts as a relevant alternative is part of the epistemological dispute.
If you know that p, recognize that p entails q, and believe q by inferring it from p, must you know q as well? This is a qualified form of the principle that knowledge is closed under known entailment; unrestricted formulations require further qualifications.
Dretske uses the zebra case as a proposed counterexample to closure. On his relevant-alternatives or tracking approach, you may know that the animal is a zebra while failing to know that it is not a cleverly disguised mule: the same visual basis would remain if the mule hypothesis were true. If closure applies to this inference, however, knowing that it is a zebra would require knowing that it is not a mule.
That tension can be extended into a skeptical argument: if ordinary knowledge entails that a skeptical alternative is false, closure plus the premise that you do not know the alternative is false would imply that you do not know the ordinary claim. The example makes that conflict vivid; it does not by itself prove that closure is false, that the skeptical premise is true, or that the skeptical conclusion follows without those assumptions.
Dretske’s response is to deny closure for some inferences: an alternative may be irrelevant to knowing the ordinary claim but relevant to knowing that the alternative is false. This is an argument about the conditions for knowledge, not a finding about actual zoo practices.
Contextualists offer a different response to skeptical arguments. On a prominent version, the standards expressed by “knows” vary with the context of the person making the attribution; raising a skeptical possibility can raise those standards. Contextualism is a debated account of knowledge attributions, not a demonstration that the mule hypothesis is more likely or epistemically possible.
Moorean-style replies retain closure and reject the skeptical premise that one lacks knowledge of the alternative’s falsity: if one knows the ordinary claim, one can infer that the incompatible skeptical alternative is false. These responses preserve ordinary knowledge at the cost of disputing a premise of the skeptical argument; the example alone does not settle which response is correct.
The scenario is not a claim that zoos paint mules, nor a general instruction to distrust perception. It tests whether knowledge transfers across a particular inference. Rejecting this closure principle does not amount to rejecting all inference; the dispute concerns whether knowledge is preserved across the specified entailment in cases involving alternatives such as the disguised mule.
A brain-in-a-vat hypothesis can be used to construct a structurally similar closure-based skeptical argument: ordinary claims entail that the skeptical hypothesis is false, and the argument then asks whether that falsity is known. The two scenarios differ in the kind of alternative they describe; the analogy does not show that either hypothesis is actually possible or likely.
The Stanford Encyclopedia of Philosophy entry on epistemic closure quotes Dretske’s zebra passage (1970, pp. 1015–1016) and discusses its role in the closure debate. The JSTOR page for Dretske’s article was consulted for bibliographic metadata, but the article’s full text was not accessible in this review. The entry’s summaries of Dretske’s proposal, contextualism, and Moorean replies are concise educational framing; the cited reference discussions present qualifications and objections that are not exhausted here. The comparison with a brain-in-a-vat hypothesis and the questions above are teaching prompts, not claims that either skeptical scenario is actual, likely, or epistemically possible.
Research check: 2026-11-01. This is an evidence-traceability review, not expert review or a judgment that any disputed epistemological position succeeds.