Understanding how network structure influences system dynamics is essential for advancing psychological modeling. This tutorial introduces the causalnet R package, which enables researchers to systematically enumerate candidate directed networks by orienting a user-specified undirected or partially directed adjacency template. Users can impose directional constraints—such as those derived from prior theory or time-series models (e.g., graphical vector autoregressive models)—to restrict the space of admissible directed network configurations. The package supports dynamic simulations on these networks using either a theoretically grounded nonlinear model (Park et al., 2025) or a simplified linear alternative. Researchers can simulate system behavior and compare dynamic outcomes across structural configurations, parameter sets, or modeling assumptions. The primary audience is applied psychological and behavioral scientists who wish to evaluate competing theoretical accounts of symptom and behavior dynamics when causal direction is uncertain. Importantly, causalnet is not intended to identify a unique causal network from cross-sectional data; instead, it supports theory- and evidence-constrained enumeration of candidate directed structures and simulation-based screening of their dynamic implications against empirical targets. We illustrate how this workflow can be used to adjudicate competing psychological theories by linking structural assumptions to predicted dynamic signatures such as persistence and recovery. This approach facilitates a systematic exploration of how causal architecture and interaction dynamics give rise to the emerging dynamics of psychological processes over time.
