Investigating the FBS coding method in the analysis of design protocols (case study: comparing behavior change the design of architecture students in relation to complex design problems)

Background and objectives: Idea generation is a critical stage in architectural design, moving from initial concepts to refined solutions through an iterative process of development and revision. To study the cognitive aspects of design, protocol analysis—which examines verbalised thought during design—has been a trusted method for over fifty years, providing insights into designers’ strategies, reasoning, and shifts in focus. While various coding schemes classify design statements, their diversity limits cross-study comparability.
The Function–Behaviour–Structure (FBS) model, developed more than three decades ago, offers a structured, versatile framework for analysing design protocols. It organises design reasoning into Function (F)—defining goals; Behaviour (B)—including expected (Be) and structure-derived behaviours (Bs); and Structure (S)—the physical/formal aspects of design. Two additional categories, Requirements (R) and Documentation (D), capture contextual and representational elements. The model also identifies eight transformation processes, such as formulation, synthesis, analysis, evaluation, and reformulations, that map the pathways from problem definition to solution development.
Widely applied internationally but less so in Iran, FBS is here examined within Iranian architectural education. This study explores its effectiveness through a case-based analysis, focusing on how undergraduate students’ design behaviours vary when tackling complex problems driven either by diversity of data or by ambiguity in problem definition.
 
Materials and Methods: This exploratory case study used protocol analysis to examine how architectural design students respond to different forms of complexity. Participants were fourth-semester undergraduates who undertook two separate 30-minute team-based design exercises. The first involved creating a large commercial complex with over 20 distinct functional spaces on a challenging site, representing complexity from data diversity. The second required designing a neighbourhood library, marked by ambiguity in interpreting terms like ‘neighbourhood’ and ‘social acceptability.’ A short pre-task—designing a table—was conducted to familiarise students with collaborative work in a think-aloud format.
Working in pairs, students verbalised their thoughts continuously, following Ericsson and Simon’s think-aloud method. All their utterances were recorded, transcribed verbatim, and segmented into atomic units, each containing a single design-related idea. These were coded according to FBS schema. Nonconforming statements were labelled ‘O’ (Other). Coding was performed twice by the same coder, with a ten-day gap between iterations, and reconciled to produce a final dataset.
Analysis occurred in several stages. Code distributions were calculated for each exercise to compare the prominence of categories under each complexity type. Next, sequences of codes were examined to identify the eight recognised FBS transformation processes, classified as individual or group depending on whether transitions occurred within or between participants. The ratio of problem-focused codes (R, F, Be) to solution-focused codes (Bs, S, D) was computed to assess attentional orientation. Finally, temporal analysis divided each session into five equal segments to track shifts in problem–solution emphasis over time. All statistical processing used SPSS, enabling comparison of behavioural patterns across the two design problem types.
 
Results and conclusion: The comparative analysis revealed notable differences in how students approached the two types of complex design problems. In the ambiguous problem (Exercise 2), students devoted more attention to Requirements (R), Functions (F), and Expected Behaviours (Be)—elements tied to problem framing—than in the data-diverse problem (Exercise 1). This supports the idea that ambiguity encourages deeper engagement with goals, interpretation, and desired outcomes. In contrast, the data-diverse exercise drew stronger focus to Structure (S)—the material and formal characteristics of solutions—and to related representational processes.
The analysis of transformation processes showed that ambiguity prompted more individual-level formulation, reflecting continuous interpretation and problem definition. Data diversity, however, generated more analysis processes—linking structures to inferred behaviours—supporting solution evaluation. The data-diverse context also led to greater group-level synthesis—team-based generation of new structures from behavioural ideas—and more group reformulation processes exploring variations on proposed solutions.
Considering the ratio of problem-oriented to solution-oriented codes, ambiguous tasks yielded higher problem focus (0.17 vs. 0.08), indicating students stayed longer in the problem space. Temporal analysis reinforced this pattern: in ambiguous tasks, focus shifted steadily from problem to solution across the session. In data-diverse tasks, students began with solution focus, reverted mid-session to problem clarification—likely due to cognitive overload—then returned to refining solutions.
Group–individual process comparisons revealed that data diversity fostered more collaboration, possibly because diverse constraints require integrated viewpoints. Ambiguity, by contrast, increased individual activity, perhaps due to differing personal interpretations that hinder consensus.
These findings confirm that both ambiguity and data diversity distinctly shape design cognition. Ambiguity sustains problem-oriented thinking, fosters reinterpretation of goals, and encourages individual reflection. Data diversity channels attention toward tangible solutions, supports collaborative creativity, and promotes iterative returns to the problem space.
From a teaching perspective, manipulating problem complexity can be a purposeful instructional tool. Ambiguity can be used to prolong engagement in defining objectives and functions, while data diversity is well-suited to stimulating intense work on form and collaborative solution-building. The FBS framework’s ability to chart temporal and categorical shifts offers educators a means to time interventions, such as encouraging reconsideration of problem goals.
Although the limited sample size constrains generalisation, the study demonstrates that the FBS coding method is effective and revealing in Iranian architectural design education. It provides a structured vocabulary for rich, detailed analysis of design reasoning. Broader studies in varied domains and with more participants could reinforce and expand these insights, contributing to both design research and pedagogy.