This “Future Directions in Music Cognition” talk is inspired and informed by Philip Ewell’s recent keynote, “Music Theory’s White Racial Frame” (SMT’s 2019 Annual Meeting), blogposts, and essay, “Music Theory and the White Racial Frame” (MTO 26.2.4). It focuses on methodological problems of mainstream music theory, a largely inductive practice that has problems of (a) confirmation-biased population sampling (often driven by the Theorist’s implicit bias), (b) small sample sizes in general (including cases where n = 1), and (c) overfitting to these small, unrepresentative samples. Specifically, our basic conceptions of harmony, melody, phrase structure, tonality, and form are largely based on a narrow subset of the works of a handful white, male European composers (Bach, Haydn, Mozart, and Beethoven). It has been suggested that these methodological problems can be solved by expanding the analytic canon to include a more representative and historically grounded range of composers of the Baroque, Classic, and Romantic eras, and to include more composers from marginalized groups (BIPOC and women, specifically). It will be argued that this will not solve the problem, in so far as the problems of confirmation bias and overfitting are not addressed, that is, if certain “masterworks” are still held up as exemplars and are thus given privileged status in the inductive practice of music theory. Researchers in music cognition, many of whom are members of the music theory community, have important roles to play in building an antiracist future for music theory, as we are aware of these methodological pitfalls and potholes. Music Cognition can provide a non-inductive alternative to the definition of key musical concepts. Moreover, as music cognition relies upon music theory for key concepts which underlie its own research program (harmony, tonality, and form), we have a substantial stake in doing so.