Background: Psi research is a controversial area of science that examines telepathy, clairvoyance, precognition, and psychokinesis (mind over matter). Central to the debate over the existence of psi is of whether independent investigators can replicate reportedly successful psi experiments. One important variable involves the beliefs of experimenters and participants. A preregistered experiment is presented that sought to replicate and extend previously published parapsychology experiments suggestive of precognition by examining implicit beliefs. Methods: On each trial of the standard (non-psi) priming task, a pleasant or unpleasant word (the "prime") is briefly shown on computer screen, followed immediately by a pleasant or unpleasant picture. Trials on which the image and the priming word have different valences are termed “Incongruent”; trials on which the picture and the priming word share a common valence are termed “Congruent”. Participants in such experiments typically respond more slowly on Incongruent trials than on Congruent trials. In this "time-reversed" psi version of the experiment, the presumed cause-effect sequence is reversed so that the prime is not flashed until after the participant has already recorded his or her judgment. The experimental hypothesis remains the same: response times will be longer on trials with Incongruent prime/picture pairs than on trials with Congruent prime/picture pairs. Additionally, the study assesses expectations of success on the psi task of 32 experimenters—each testing 12 participants—using self-report questionnaires and the Implicit Association Task (IAT). Results: A significant correlation was found between the Implicit Association Test (IAT) effect and the participants’ reported beliefs in psi, with the effect in the direction opposite to the hypothesized correlation. Conclusions: This study offers an innovative approach to the role of beliefs in psi in a precognition study and speaks to the challenges of replication in controversial science.