Abstract: Trademark dilution is a cause of action for interfering with the uniqueness of a trademark. For example, consumers would probably not think that "Kodak soap" was produced by the makers of Kodak cameras, but its presence in the market would diminish the uniqueness of the original Kodak mark. Trademark owners think dilution is harmful but have had difficulty explaining why. Many courts have therefore been reluctant to enforce dilution laws, even while legislatures have enacted more of them over the past half century. Courts and commentators have now begun to use psychological theories, drawing on associationist models of cognition, to explain how a trademark can be harmed by the existence of similar marks even when consumers can readily distinguish the marks from one another and thus are not confused. Though the cognitive theory of dilution is internally consistent and appeals to the authority of science, it does not rest on sufficient empirical evidence to justify its adoption. Moreover, the harms it identifies do not generally come from commercial competitors but from free speech about trademarked products. As a result, even a limited dilution law should be held unconstitutional under current First Amendment commercial-speech doctrine. In the absence of constitutional invalidation, the cognitive explanation of dilution is likely to change the law for the worse. Rather than working like fingerprint evidence--which ideally produces more evidence about already-defined crimes--psychological explanations of dilution are more like economic theories in antitrust, which changed the definition of actionable restraints of trade. Given the empirical and normative flaws in the cognitive theory, using it to fill dilution's theoretical vacuum would be a mistake.