Brain-based predictive modeling of mind wandering and involuntary thought: The idiographic approach

A substantial portion of waking life consists of mental experiences that arise involuntarily, including mind wandering, spontaneous thought, and rumination. Given that these experiences are fundamental to cognitive function and mental health, there has been growing scientific and clinical interest in using measures of brain activity to model and predict the momentary occurrence of these various forms of involuntary thought. In the most common machine learning approach in neuroimaging and electrophysiology research, model training data include neural features derived from a population of individuals, and testing is performed on held-out data in one or more individuals. However, sources of idiosyncrasy, such as individual differences in the nature of thought and brain functional organization, raise critical questions regarding whether population-derived neural models can generalize to individuals. In this article, we describe how an idiographic (person-specific) approach has the potential to improve theory and practice in predictive neural modeling of involuntary thought. This approach emphasizes dense sampling of individuals so that a diverse set of brain states and mental experiences can be modeled within a single individual, and rigorous tests of cross-subject generalizability can be performed. We review the advantages and shortcomings of both nomothetic (population-based) and idiographic predictive modeling approaches, including recent insights from dense-sampling neuroimaging studies that demonstrate person-specific brain-based predictions of mind wandering. We discuss implications for developing personalized clinical biomarkers, strategies to overcome the practical challenges of an idiographic approach and neuroethical concerns that must be considered as person-specific models may enhance the potential for accurate brain-based predictions of involuntary thought.