- The paper introduces a dual-purpose inductive methodology, 'recherche surjective', to study interaction mechanisms beyond static interfaces.
- It employs targeted parameter alteration in prototypes to establish replicable property studies, addressing HCI's noted replication deficit.
- The work advocates unifying engineering and theory in HCI, paving the way for a foundational physics of human–computer interaction.
Observing Interaction Rather Than Interfaces: Toward an Inductive Methodology for HCI
Introduction
This paper presents a formal critique of the prevailing methodologies in Human–Computer Interaction (HCI), diagnosing the chronic proliferation of isolated, non-generalizable empirical findings as stemming from a misplaced emphasis on interfaces rather than interaction itself. By proposing a dual-purpose, inductive methodology—termed "recherche surjective"—that leverages both theoretical and engineering efforts, the work outlines a unified framework for observing, consolidating, and theoretically framing interactional phenomena. This approach explicitly seeks to facilitate the development of cumulative HCI knowledge in the context of technological evolution and diverse end-user requirements.
Critique of Current Empirical Practices in HCI
The analysis situates itself within historical debates on scientific rigor and theory-building in HCI, citing canonical discussions and critiques on the lack of general theories, risky hypotheses, and, crucially, systematic replication (2510.06156). The empirical status quo is characterized as overdetermined by task-specific, technology-bound results, routinely generated through isolated prototypes with minimal methodological or theoretical continuity. Meta-analytic evidence is presented illustrating that both theoretical and replication studies are severely underrepresented in flagship HCI venues, with empirical and engineering contributions dominating research output. The result is a research ecosystem in which iterative engineering consistently outruns theoretical synthesis, undermining both validation and explanatory adequacy.
A Shift of Focus: Interaction as the Object of Study
The paper advocates for a paradigmatic shift: making "interaction"—conceived as the reciprocal, temporally extended behavior between user and system—the primary object of empirical study, decoupled from transient interface technologies. By reconceptualizing interaction as a dialogue comprising cyclic exchanges mediated by perceptual and action channels, the author aligns with recent definitional work in HCI that foregrounds interaction “prisms” over static interface typologies. This reorientation is supported by theoretical models that decompose interaction into analyzable loops and channels, such as action-perception-cognition cycles (2510.06156).
Methodological Proposal: Experimental Diffraction and Property Studies
Key to operationalizing this perspective is the introduction of the "interactional prism"—a conceptual analogue to the modulation of variables in statistical experimentation but specifically tied to interactional feedback loops. Experimental methodology proceeds via targeted alteration (diffraction) of a single parameter within a task-relevant prototype, measuring its effect on pertinent interaction properties. These property studies are strictly parameter-homogeneous, varying only one interface parameter at a time, and are carried out in the context of real user needs and representative tasks. This approach is characterized by an "inductive and opportunistic" logic: parameter selection is driven by the optimization of genuine user applications rather than preconceived theoretical conjectures, leading to the cumulative discovery and eventual reconciliation of interactional properties.
Integration, Replication, and the Physics of Interaction
The framework is explicitly structured to address the historical replication deficit in HCI. Property studies, by design, produce observations that can be replicated across diverse technological and task contexts, supporting both partial/conceptual and direct replications. When aggregated—via meta-analysis or hypothesis-verification studies—these results yield more robust, generalizable properties and boundary conditions. The long-term ambition is the articulation of systematic relationships between these properties, culminating in the prospective development of an empirically grounded "physics of human–computer interaction." Notably, the term is not used poetically but denotes the aspiration to formulate generalizable, predictive regularities governing interaction phenomena.
Implications and Prospects
By fusing technological demonstration and theoretical exploration, the methodology promises enhanced economic and scientific incentives for replication and property-oriented studies. Empirical outputs are immediately applicable to prototype optimization, thereby amortizing the substantial engineering resource investment typical of HCI, while simultaneously feeding systematic theoretical generalization. The proposal is seen as scalable to community-wide research coordination, using standardized experimental protocols (potentially leveraging platforms like Touchstone/Touchstone2) to facilitate cumulative evidence gathering, condition specification, and meta-analytic synthesis. Potential extensions include consideration of non-ergonomic interaction properties and transplantation of the surjective research logic to allied research domains.
Conclusion
This paper articulates a methodologically rigorous, inductively-driven pathway for HCI centered on the systematic observation of interaction rather than mere interfaces. By experimentally probing the structure and effects of interaction loops and integrating opportunistically gathered evidence, the work illuminates a trajectory toward cumulative science in HCI. Future directions include formalization of interactional properties, their contextual boundaries, and the discovery of underlying principles governing complex human–system behavior, thus laying groundwork for a foundational physics of interaction.