It Knows My Emotions Better Than I Do: Digital Phenotyping and the Contest Over Symptom Interpretation

Digital phenotyping is increasingly presented as a means of detecting mental distress through smartphones, wearables, and other sensor-rich personal devices. In this article, we argue that the central philosophical and clinical issue raised by digital phenotyping extends beyond predictive accuracy to interpretive authority, which concerns who gets to name what a feeling, behavioral pattern, or physiological fluctuation means. Psychiatry has always involved uncertainty, proxy reasoning, and the negotiation of testimony, but digital phenotyping intensifies these dynamics by translating lived states into behavioral and physiological signals that are then modeled as indicators of psychopathology. This translation promises earlier detection and more continuous monitoring, yet it also risks displacing first-person understanding, narrowing the available interpretive vocabulary, and recoding distress through pattern recognition in ways that can obscure meaning. Drawing on work in digital phenotyping, philosophy of technology, phenomenological psychopathology, and epistemic injustice, we develop a theoretical account of how digital systems may compete with patients and clinicians for the right to interpret symptoms. We show that the most consequential conflicts concern credibility, categories, temporality, and subject formation. The result is a new diagnostic tool embedded in a new social arrangement of knowledge in which persons may come to mistrust their own experience unless it is confirmed by data. We conclude by proposing a dialogical and clinically modest model for digital mental health, one that treats algorithmic outputs as revisable prompts within a larger interpretive practice while preserving space for human interpretation.