Reflexa: Uncovering How LLM-Supported Reflection Scaffolding Reshapes Creativity in Creative Coding

TL;DR

This paper investigates how to scaffold reflection in LLMSupported creative coding. It introduces Reflexa, a system that combines dialogic reflection (Reflexa Core), version-based exploration (Reflexa Flow), and visual inspiration (Reflexa Spark) to structure ongoing reflective practice. Grounded in Fleck & Fitzpatrick’s reflection levels and Schön’s reflection-in-action, Reflexa is evaluated through formative design studies and a within-subject user study (n=18) showing that structured reflection mediates the relationship between AI interaction and creative outcomes. Findings reveal that Reflexa enhances controllability, exploration breadth, originality, and aesthetic quality, reframing reflection as a system-mediated, co-creative process. The work offers design guidance for building LLM-based tools that support richer human–AI co-creativity and durable reflective practices in creative coding.

Abstract

Creative coding requires continuous translation between evolving concepts and computational artifacts, making reflection essential yet difficult to sustain. Creators often struggle to manage ambiguous intentions, emergent outputs, and complex code, limiting depth of exploration. This work examines how large language models (LLMs) can scaffold reflection not as isolated prompts, but as a system-level mechanism shaping creative regulation. From formative studies with eight expert creators, we derived reflection challenges and design principles that informed Reflexa, an integrated scaffold combining dialogic guidance, visualized version navigation, and iterative suggestion pathways. A within-subject study with 18 participants provides an exploratory mechanism validation, showing that structured reflection patterns mediate the link between AI interaction and creative outcomes. These reflection trajectories enhanced perceived controllability, broadened exploration, and improved originality and aesthetic quality. Our findings advance HCI understanding of reflection from LLM-assisted creative practices, and provide design strategies for building LLM-based creative tools that support richer human-AI co-creativity.

Figures (17)

• Figure 1: Study procedure across three stages.
• Figure 2: The creative coding process with LLMs unfolds through three stages: ideation, function built, and feature refinement.
• Figure 3: Overview of Reflexa system features grounded on Fleck & Fitzpatrick's reflection framework (R0-R4) fleck_reflecting_2010 (see Section \ref{['sec:rw-reflection']}).
• Figure 4: The Reflexa interface is an LLM-based creative coding tool supporting reflection. Users begin in the chatbox of Reflexa Core, select a reflection-supported mode R1-R3 (a.1), enter a prompt (a.2), and click "send" (a.3) to receive generated code with reflective suggestions in the dialogue area of Reflexa Core. This code can be copied (b.1) into the editor (c.1), executed (c.2), and previewed in the code editor area (c.3). The "collect" (c.4) button saves code as nodes in the Reflexa Flow, each supporting a visual preview (d.1), modification (d.2), or merging (d.3) of two versions, which can initiate quick iteration in Reflexa Spark above chatbox. Selecting a node synchronizes the dialogue panel and p5.js editor with the chosen version. Spark entered by modify in the Flow.
• Figure 5: Interaction flow in Reflexa Core with three modes.