Abstract

Most of psychology and cognitive neuroscience is based on cross-sectional investigations, that is, participants are generally measured once. A key limitation of cross-sectional data is that associations between brain and behaviour may reflect a variety of underlying processes that cannot readily be distinguished. We will discuss two emerging techniques, network analysis and latent change models, that are instead tailored to the study of change over time.

Psychopathology, such as depression or anxiety, is traditionally characterized as an underlying illness that causes symptoms. A novel and rapidly emerging view on psychopathology called network analysis focuses on the mutual influences between symptoms. Rather than “I am sad, sleep poorly and can’t concentrate because I’m depressed”, we might say “I slept poorly, therefore I can’t concentrate, and that makes me sad, which causes me to sleep poorly, etc.”. Here we will explore how changing network parameters may underlie the onset of depression in adolescence, and how they may change again during successful psychotherapy treatment.

In a second talk we will focus on a powerful class of structural equation models known as latent change score (LCS). This general modelling approach links longitudinal data to the analytical strengths of the SEM framework, which include model testing, model comparison, measurement invariance, and measurement error. We will illustrate how these models capture how our cognitive abilities emerge (in adolescence and early adulthood) and decline (in old age).