The network theory of psychopathology is increasingly used to describe mental disorders as systems of interacting symptoms. The implications of network theory have mainly been studied using the Ising model, in which the symptom can be absent or present. The recent shift from binary Ising models to ternary models permits symptoms to exist in one of three states: −1, 0, and +1. During my fellowship at IAS, I will focus on using a ternary model to examine the states of depression (-1), euthymia (0), and mania (+1) in bipolar disorder. The main question I will address during this talk is “What is a psychological interpretation of the parameters of the ternary model of bipolar disorder?"

To do so, we developed an opposite-item questionnaire for bipolar disorder and major depressive disorder. Instead of measuring one side of a symptom (e.g., severity of sleep problems) as is done in most questionnaires, our questionnaire contains items with answering options in opposite directions and a neutral midpoint (e.g., hypersomnia – no sleep problems – insomnia). Although I discuss the psychometric properties of our questionnaire, the main motivation is to use this questionnaire to investigate whether the psychological interpretation assigned to the parameters of a ternary network model can be empirically justified. If a parameter is interpreted as reflecting a particular psychological construct, can we use an independent measure of that construct to investigate whether it relates to the estimated parameter as the theory predicts?

During this event, I reflect on how we should interpret estimated ternary network models and the relationship between the parameters and the psychological concepts they represent. Using preliminary work on opposite-item questionnaire of bipolar disorder, I will explore how measurement can help us evaluate and refine these interpretations.