Digital Traces of Everyday Smartphone Usage Predict Fluid Intelligence

Scalable, objective judgments of individuals’ everyday cognitive functioning may facilitate decision making in key societal domains, including educational, occupational, and healthcare contexts. In a preregistered, theory-informed machine learning approach, we examine how digital traces of everyday smartphone usage allow for predictions of individuals’ general fluid intelligence (Gf), one of the most central human cognitive abilities. Leveraging a German quota sample (N = 367) providing smartphone sensing data of up to six months (1,269,491 phone usages), cross-validated results demonstrate that everyday smartphone usage consistently predicts individuals' Gf scores on a psychometric test (r_Md = .43, r_IQR = [.35, .51]) with comparable criterion validity. High-dimensional combinations of behaviors (particularly short-term activities related to basic cognitive tasks and dealing with complexity) were predictive for Gf, highlighting how cognitive abilities manifest pervasively yet highly uniquely across individuals. The findings suggest that smartphone-based cognitive judgment may complement traditional cognitive assessments, enhancing early identification of cognitive functioning, fit, and support.