Stance Architecture
Every act of reasoning begins from a particular orientation toward a situation — called a stance. Stance Architecture is the formal system within Lucid Theory that describes how these orientations are structurally configured and how they operate within and across reasoning processes.
Rather than assuming a single neutral viewpoint, Lucid Theory recognizes that reasoning unfolds through the interaction of multiple perspectives — each with its own configuration of assumptions, priorities, and interpretive tendencies. Stance Architecture is the formal treatment of that structure.
A stance is a structured configuration of five interpretive tendencies — not a viewpoint or a bias, but a formally defined configuration that shapes how reasoning moves through an epistemic field. These five components form a coherent whole; a change to any one changes the stance.
What a stance treats as given or unquestioned — the background conditions that precede interpretation and shape which questions can even be formed.
What a stance attends to first. The ordering of attention within a reasoning episode — which features of a situation are perceived as salient and which recede.
What a stance treats as evidence. The standards by which claims are assessed, weighted, and accepted or set aside.
The organizing concepts through which a stance structures a domain of inquiry — the categories and distinctions it brings to a situation before interpretation begins.
Whether a stance explains through causes, structures, probabilities, or narratives — the form that understanding takes within this interpretive position.
Within the Epistemic Field Model — which maps the interpretive landscape of a reasoning process — stances act as positions within that landscape. Each stance highlights certain relationships while downplaying others. Two stances applied to the same situation will perceive different features as salient, weight different evidence as relevant, and arrive at different interpretations — not because one is correct and the other mistaken, but because each is positioned differently within the field.
When multiple stances are applied to the same domain, their interaction produces one of four outcomes — each informative rather than merely problematic:
Consider two stances applied to the same user behavior — a researcher repeatedly revisiting the same documents. An efficiency stance (evaluative criteria: speed, task completion) perceives this as friction — a workflow problem to be resolved. A depth stance (evaluative criteria: comprehension, conceptual integration) perceives it as a signal of meaningful engagement. Neither interpretation is wrong; both are generated by the structure of the stance, not the data alone. Together they reveal that ‘revisiting’ is an ambiguous signal requiring further resolution — a finding neither stance could have produced in isolation.
Divergent-Convergent Reasoning describes the movement of reasoning through two alternating phases. Stances operate differently in each — and the difference is structurally significant.
In the divergent phase, stances are treated as multiple sources of insight. The goal is to expand the interpretive space — to explore what different positions perceive, not to determine which is right.
Each stance contributes a perspective that the others cannot fully generate. Divergent reasoning proceeds by letting these perspectives coexist and accumulate, rather than converging prematurely on any single account.
In the convergent phase, insight emerges from integrating elements across stances — not from selecting a single winning stance. Integration is synthesis, not elimination or compromise.
Integration preserves the genuine insight each stance contributed while resolving or explicitly holding the tensions between them. The result is an interpretation that no single stance could have produced alone.
Artificial systems could explicitly represent multiple stances and explore their interactions — rather than arriving at interpretations through a single implicit perspective. This makes stance a design principle, not merely a descriptive concept. Four capabilities follow from it:
Approaching a problem through multiple explicitly configured stances rather than a single implicit orientation.
Comparing the interpretive outputs of different stances directly — surfacing where they converge and where they diverge.
Recognizing when two stances produce interpretive tensions — treating those tensions as information rather than errors to be suppressed.
Synthesizing elements from multiple stances into a more complete interpretation — preserving insights from each without forcing premature consensus.
These capabilities form part of the reasoning architecture described in mykungfu.ai — where the theoretical system becomes an engineered implementation.
Together, EFM, Stance Architecture, and DCR form the spatial, positional, and dynamic dimensions of Lucid reasoning — three interlocking models that describe the structure, position, and movement of reasoning within an epistemic field.
← Lucid Theory