Abstract: This review develops an integrative process-oriented framework for the psychological spillover effect, with the aim of addressing three persistent challenges that have long impeded progress in the field: conceptual ambiguity, mechanistic fragmentation, and insufficient responsiveness to digital-age transformations.
The central source of conceptual confusion in spillover research lies in its indiscriminate overlap with related constructs such as compensation, crossover, and ripple effects. To resolve this issue, this article reconstructs the conceptual architecture of spillover by proposing a four-dimensional discriminative framework anchored on transmission subject, transmission direction, intentionality of outcome, and diffusion structure. Within this framework, spillover is operationalized as an intraindividual, same-direction, unintentional, point-to-point transfer process, thereby distinguishing it from compensation, which is characterized as motivation-driven reverse complementarity; from crossover, which involves inter-individual transmission; and from ripple effects, which entail network-based multi-directional diffusion. By establishing clear discriminant criteria, this framework not only enables more standardized and rigorous use of the spillover construct, but also facilitates the integration and comparison of empirical findings across diverse subfields.
Existing literature has treated cognitive dissonance, self-perception, social learning transfer, and other theoretical accounts as parallel explanatory mechanisms, yet a unifying meta-mechanism has remained absent. To fill this gap, this article draws upon Lewin's field theory and the S-O-R framework to propose a process mechanism model composed of four functional components distributed across three sequential stages: structural prerequisites, trigger-transformation stage, diffusion-transmission stage, and effect-manifestation stage.
The structural prerequisite, psychological boundary permeability, functions as the channel architecture that enables cross-contextual transfer. This article elaborates on its dual expression: baseline width, which refers to trait-level individual differences in boundary strength, and dynamic openness, which denotes state-level fluctuations under conditions of resource depletion or resource gain. Notably, the hypothesis of asymmetrically permeable boundaries is incorporated to demonstrate that directional asymmetry in permeability directly determines spillover probability, activation threshold, and the associated self-regulatory costs. The trigger-transformation stage converts external events into transferable psychological content through a sequential chain of event encoding, meaning association, and state activation. Processing elasticity theory is introduced to account for variability in processing depth: high-intensity negative events may compress meaning association into automated schema activation, whereas deliberative contexts tend to extend controlled processing. The diffusion-transmission stage addresses the critical question of how psychological states originating in Domain A maintain their influence upon entering Domain B. Cross-contextual reactivation is proposed as the unifying meta-mechanism, encompassing three subprocesses: state residue, which represents persistent background activation; contextual transition, which involves decoupling from the prior context and recoupling to the new one; and state reactivation, which refers to foreground activation triggered by Domain B cues. Three specific reactivation pathways are identified and shown to operate as interconnected cascades rather than independent routes: cognitive reactivation operates through consistency principles, emotional reactivation proceeds via affect carryover, and behavioral reactivation occurs through observational learning and generalization. Finally, the effect-manifestation stage captures observable psychological and behavioral changes in Domain B, which are categorized according to influence direction, distinguishing constructive from obstructive effects, and according to influence content, differentiating emotional, cognitive, and behavioral outcomes. Particular attention is given to the mechanistic asymmetry between negative spillover, which is driven by resource depletion and consequent boundary breakdown, and positive spillover, which emerges from resource gain and facilitates boundary extension.
The proposed process model is further extended to human-computer interaction contexts without disrupting its core architecture. Digital technologies fundamentally reshape each stage of the process. Psychological boundaries evolve from managing switches between discrete domains to regulating interference among concurrently active threads. The trigger-transformation stage shifts from a dyadic individual-event interaction to a triadic individual-algorithm-event interaction, wherein algorithmic curation substitutes for autonomous selective attention and manufactured meaning resonance replaces self-constructed association. The diffusion-transmission stage transforms from single residue decay to algorithm-anchored multi-residue competition accompanied by fragmented transitions. The effect-manifestation stage exhibits amplified asymmetry: positive spillover becomes superficial and fragile due to fragmented content consumption, whereas negative spillover becomes covert yet cumulatively explosive through persistent micro-arousal. Novel moderators specific to human-AI interaction are further identified, including the degree of anthropomorphism, perceived interaction control, and consistency of interaction history.
This framework advances spillover research in three significant ways. First, it provides a unified conceptual grammar that replaces ambiguous construct usage with actionable discriminant criteria. Second, it transforms spillover from a static phenomenological label into a testable temporal process with clearly specified mechanistic targets at each phase. Third, it extends the model to digitally mediated contexts without abandoning its core architecture, thereby equipping future research to address hybrid reality, algorithm-infused environments, and human-AI boundaries. Future empirical priorities should include validating the three-phase model through experimental and longitudinal designs, establishing phase-specific moderators via meta-analytic synthesis, and translating mechanistic insights into targeted interventions across organizational, environmental, and mental health domains.

Key words: spillover effect, psychological boundary, behavioral spillover, digital age, human-computer interaction