Understanding intelligence as something that unfolds through interaction.

Co-creative AI needs representations that explain how interaction reorganizes, stabilizes, regulates, and develops through time—not only records of observable actions. This paper introduces Cognitive Trajectory Modeling (CTM), which treats cognition, interaction, and creative activity as temporally organized trajectories unfolding across cognitively meaningful attractor landscapes. It formalizes the Cognitive Trajectory Principle, distinguishes cognitive trajectories from interaction traces, and situates CTM within a hierarchy of cognitive, interaction, and domain dynamics. The framework provides a foundation for studying temporally extended dynamics in co-creative AI and human–AI interaction.

Artificial intelligence is commonly evaluated as computation contained within an individual model or agent. This paper instead proposes interaction as the primary unit of analysis for co-creative AI. Drawing on distributed and embodied cognition, enaction, participatory sense-making, HCI, and computational creativity, it argues that intelligence can emerge through changing relations among agents, environments, and sociotechnical systems. Interaction-Centered Intelligence emphasizes trajectories of coordination, participatory engagement, adaptive regulation, and interactional drift, with implications for explainable co-creative AI, hybrid intelligence, and future human–AI systems.

As generative AI makes co-creation an increasingly important interaction paradigm, prior co-creativity research needs a structured way to inform contemporary system design. This paper introduces the Co-Creative Design Framework (CCDF), a systematic framework that formalizes human–AI co-creation through cognitive and interaction principles. It identifies dimensions of variation that define the co-creative interaction space, with particular attention to agency and interaction dynamics. The framework was developed iteratively by synthesizing co-creativity and AI research and is intended to support the design and analysis of co-creative AI and hybrid-intelligence systems.

This paper proposes enaction as a theoretical foundation for co-creative AI, emphasizing that meaning emerges through an agent’s situated interaction with its environment. It develops a descriptive framework around five pillars—autonomy, sense-making, embodiment, emergence, and experience—and applies them to the analysis of twenty co-creative AI systems, including contemporary generative systems. The paper uses the framework to categorize existing systems systematically and concludes with design recommendations for building more enactive forms of co-creative artificial intelligence.

This paper introduces Creative Sense-Making, a method for quantifying interaction dynamics during open-ended co-creation such as collaborative drawing and pretend play. The framework synthesizes cognitive science theories and empirical research on improvisation to represent cognitive states and interaction types through time. It is applied across human collaboration and human–AI drawing studies, with validity examined through cross-domain use and inter-rater reliability. The resulting framework combines a qualitative coding method, interaction-coding software, and a cognitive account of their interpretation.

This paper presents the design and evaluation of Drawing Apprentice, a real-time co-creative partner for improvisational abstract drawing. The system was designed as a casual creator: its primary aim is to sustain enjoyable creative engagement rather than merely improve the quality of a finished artifact. The study examines whether participatory sense-making can emerge between users and a computational agent through mutually responsive drawing interaction. Its findings support the value of interaction dynamics and reciprocal adaptation as central considerations in the design and evaluation of co-creative systems.

This paper reports the theory, design, and implementation of Drawing Apprentice, an artistic computer colleague that improvises with human users in real time. The system draws on theories of art, creative cognition, and collaboration synthesized into an Enactive Model of Creativity. The paper describes the system implementation and presents early collaborative artworks, positioning enaction as a theoretical framework for creative technologies based on continuous improvisational collaboration between people and computers.

This paper develops the beginnings of a Cognitive Theory of Creativity Support focused on novice creators. The theory identifies barriers that can prevent novices from beginning or sustaining creative work, including fear of failure, limited time, and insufficient skill. The authors use the theory to analyze creativity-support tools from several domains and describe StorySketch, an initial system designed from these principles to help storytellers without advanced graphical skills engage in visual storytelling.

Using grounded-theory analysis, this paper investigates the creative practices of machinima filmmakers working with real-time 3D game engines. The study finds that filmmakers develop underspecified mental images and refine them through immediate experimentation with camera position, lighting, character placement, and other scene parameters. These tools distribute part of the evaluative work between person and system, enabling ideas to be generated, explored, and assessed through active manipulation. The paper develops this account as a model of distributed exploratory visualization in digital filmmaking.