Coupled Human Trajectories in Romantic Relationships: An Integrative Causal-Dynamical Framework and Research Agenda

Romantic relationships are commonly studied through partially separated literatures on partner ideals, attraction, mate evaluation, commitment, interdependence, relationship maintenance, dyadic regulation, life-course development, and dissolution. We propose an integrative causal-dynamical framework in which a romantic relationship is represented as the evolving coupling of two previously distinct life trajectories. The framework begins before a specific partner is present. Individuals carry relational representations concerning desirable partners, viable forms of relationship, and anticipated futures. After contact, observation and interaction generate beliefs about a particular partner and dyad. Repeated interaction then modifies individual states, relationship-specific stocks, and structural conditions, progressively changing the extent to which each partner alters the other's future transition possibilities. Search, selection, coupling, coordination, preservation, reassessment, and decoupling are therefore treated not as fixed chronological stages but as decision regimes that may dominate in different regions of relational state space. We organize the system using five state classes: individual adaptive state (\(X\)), generative relational representation (\(G\)), current relational belief (\(B\)), endogenous relational stock (\(Q\)), and structural relational conditions (\(Z\)). Causal coupling and decision architecture are derived properties rather than additional primitive states. The notation is semi-formal: it specifies an architecture and candidate estimands, not a fully parameterized or observationally identified computational model. The framework is explicitly synthetic. Observable-state insufficiency is treated as a general consequence of incomplete state representation; the selection-retention gap is treated as an analytic contrast with antecedents in re-choice, investment, dedication, and constraint research; and post-relationship learning is retained as an open research program rather than as an already identified domain-selective mechanism. The resulting account treats romantic relationships as path-dependent, multi-timescale systems in which interaction can alter both partners' future trajectories, while later decoupling need not erase the causal residues of prior coupling.