Using the lens of the cognitive bias literature, this Article examines and critiques the “reasonable man” standard, focusing on the use of the standard in an extremely fuzzy area of the law: copyright. In copyright, the test for infringement is whether a “reasonable observer” would believe that two works—often involving media that do not lend themselves to precise measurement—are substantially similar. I begin by casting doubt on the usefulness of the reasonable man standard in such a setting. Are judges and juries truly able to determine what an abstract reasonable actor would find substantially similar? What types of cognitive biases will likely cloud this determination? And are biases likely to have a stronger or weaker effect when infringement questions are subjected to group deliberation as opposed to the individual decision making of judges? Next, I address the problems that I uncover in the copyright context by first reviewing some potential solutions, including both a proposal to reduce the role of juries in substantial similarity determinations and the possibility of trial bifurcation. Ultimately, I show that an openly subjective standard that focuses on the works’ intended audience and uses social science surveys as evidence of infringement should replace the prevalent “objective” reasonable observer standard. Implementing such a solution would at least partially acknowledge that we are dealing not with perfectly reasonable, but rather boundedly rational, actors.