Inferring latent self-schemas to advance understanding of mental health problems

Pathological self-schemas are theorized to play a key role in psychiatric disorders. However, measuringself-schemas is challenging because they are defined as latent representations and therefore cannot beadequately captured through self-report. Past work thus employed behavioral tasks in which participantsform actual positive and negative self-judgments. Yet, it is unclear if such tasks elicit rich selfreferentialprocessing and provide information about latent self-schemas distinct from self-report. Toaddress these limitations, we tested a new self-judgment decision-making task in two studies (N = 1008).We found valenced self-referential processing shaped self-judgments while controlling for the effectsof self-reported beliefs, as evident in distinct choice and reaction-time signatures predicted by a driftdiffusionmodel. More depressed (vs. healthier) individuals not only differed in self-reported beliefs,but also valenced self-referential processing—specifically, accumulating less positive (vs. negative)information about themselves. Our paradigm plausibly measures latent self-schemas theorized tounderlie many mental illnesses.