A Computational Framework for Brand Perception

Relationship to Established Frameworks

SBT does not replace Aaker, Kapferer, or Keller. It is the microscope they never had. Aaker's four perspectives (brand-as-product, brand-as-organization, brand-as-person, brand-as-symbol) map onto subsets of SBT's eight parametrised dimensions. What was a qualitative checklist becomes a measurable emission policy with per-cohort observer profiles. The classical frameworks are preserved as the limiting case where all observers share the same spectral profile — just as Newtonian mechanics is preserved as the low-velocity limit of relativity.

Aaker drew the anatomy chart. SBT built the instrument. Brand managers have been diagnosing with the naked eye and tacit knowledge — SBT provides the resolution to see that what looks like one Dove is four structurally different organisms in four different minds. Five coherence types where traditional frameworks see one scale. Eight dimensions where Aaker sees four perspectives. Per-cohort perception where Keller sees one equity score.

A 10-property side-by-side comparison of SBT against Aaker, Keller, Kapferer, J. Aaker, Brakus, and Holt — including an honest disclaimer about what the predecessor frameworks do better — is on the research page.

What Is a Brand?

In Spectral Brand Theory, a brand is not an object with fixed properties but a perceptual process: continuous signal emission across eight dimensions, continuous observation through observer-specific spectral profiles, continuous re-collapse into conviction. From noun to verb — brand is something that happens, not something that is.

There is no single brand perception that applies to all observers — each cohort assembles structurally different brand meaning from the same signal environment. This is not fragmentation or inconsistency. It is the fundamental mechanism of perception.

The Eight Dimensions

Every brand signal maps to one of eight perceptual dimensions. Together they form the spectral decomposition — a complete analytical surface for any brand. No signal exists outside these eight categories; no brand escapes this structure.

The Observer Model

Observers are not passive recipients. Each carries a spectral profile — a formal structure that determines which signals they perceive and how those signals cluster into conviction. The same brand emits the same signals, but different observers construct structurally different brands from them.

An observer spectral profile has five components:

1. Spectrum — sensitivity to each dimension (0.0 = invisible, 1.0 = full sensitivity). A Gen-Z consumer may have high social sensitivity (0.9) but low temporal sensitivity (0.2).
2. Weights — importance assigned to each dimension. Spectrum determines what you can see; weights determine what matters. A luxury buyer may weight economic and experiential dimensions at 0.80 combined.
3. Tolerances — how much inconsistency the observer accepts per dimension. Brand employees may have zero tolerance for ideological inconsistency. Casual consumers may tolerate significant narrative contradiction.
4. Priors — existing beliefs about the brand. Brand convictions already collapsed in memory. Shape how new signals are perceived through confirmation bias.
5. Identity gate — the recognition function. Can the observer identify these signals as belonging to this brand? Failure means signals are perceived as noise — no perception cloud forms.

Observer cohorts are perceptual, not demographic. Two observers from different demographics can share a cohort if their spectral profiles converge. Two from the same demographic can belong to different cohorts if their priors diverge. Cohort membership is dynamic — observers drift between cohorts as priors evolve and signals decay.

Observer weights have a direct economic interpretation: the dimensions a cohort weights highest are the dimensions where operational investment yields the greatest return. SBT's demand decomposition — breaking market demand into eight weighted dimensions per cohort — enables spectral resource allocation: invest where target cohorts perceive value, not where the organization habitually spends. When combined with Organizational Schema Theory's top-down cascade, this becomes a demand-validation gate: L0 customer experience contracts must precede L2-L5 operational specification, structurally preventing investment in operations before validating what customers value.

Cloud Formation and Conviction Collapse

Brand perception follows a pipeline: signals arrive, pass the identity gate, cluster into probabilistic hypotheses (perception clouds), and collapse into conviction when evidence crosses a threshold.

The pipeline has four states:

1. Unaware — no signals have passed the identity gate.
2. Forming — signals are clustering but conviction is weak. "I've heard of them."
3. Partial — a cloud has formed with moderate confidence. "I think they're X."
4. Confirmed — conviction has collapsed. "They ARE X." The brand is now a fact in the observer's mind.

Re-collapse is the critical mechanism: when contradicting signals arrive (a scandal, a product failure, a competitor's campaign), conviction dissolves and rebuilds from the surviving evidence set — crystallized priors persist, but unstable beliefs are released for reconstruction. Not an incremental update, but a full rebuild from whatever evidence remains. This explains both brand resilience (strong convictions resist moderate contradiction) and brand crises (overwhelming contradiction forces wholesale re-evaluation).

Five Coherence Types

Brand coherence is not a single variable from low to high. It has qualitative types with structurally different resilience properties. A 7/10 signal coherence and a 7/10 ecosystem coherence describe fundamentally different brands. These five types are theoretically derived from the observer model — they are candidate mechanisms, not empirically validated categories. The case studies below are illustrative analyses that demonstrate the framework's diagnostic reach, not empirical validation of the typology.

The grade assigned to each coherence type reflects disruption resilience — the brand's structural capacity to absorb contradicting signals without conviction collapse. A higher grade means more resilient architecture, not better coherence quality.

Non-Ergodic Perception Dynamics

Brand perception is non-ergodic: the time average (one observer's experience over time) diverges from the ensemble average (many observers at one point in time). This has profound consequences for measurement and strategy. The ergodicity framework serves as an organizing analogy here — it precisely captures the structural mismatch between aggregate brand tracking and individual cohort trajectories, though the formal mathematical properties of ergodic systems are not claimed to transfer wholesale to perception dynamics.

Signals compound multiplicatively, not additively. Sequence matters. A consumer who first encounters a brand through a negative news story, then sees a positive ad, forms a different conviction than one who encounters them in reverse order — even though the signal set is identical. This makes brand perception path-dependent.

Negative conviction can become an absorbing state: once an observer's negative belief crosses a threshold with no experiential friction to challenge it, no future positive signals reach them. The observer with the least evidence can have the most stable conviction. Brand crises confirm the negative conviction rather than challenging it.

The ergodicity coefficient (epsilon) measures, per dimension, how reliably ensemble statistics predict individual cohort trajectories. When epsilon approaches zero, aggregate brand tracking surveys become meaningless. You must track cohort trajectories longitudinally.

The TrajectoryRisk validator operationalizes R6's drift vector mu(X_t, t) as a discrete approximation: per-dimension velocity (signed rate of change per period), direction classification (rising/falling/stable), acceleration from three or more snapshots, and linear time-to-absorption estimates for dimensions approaching irreversible decline. Conformal prediction provides distribution-free calibrated uncertainty bands around velocity estimates, with finite-sample coverage guarantees (R6, Proposition 6).

Two Ways to Read a Brand

The L1 output is the complete spectral tensor: eight dimension scores per cohort, emission weights, tolerance thresholds, prior states, and cloud formation modes. This is the machine-readable ground truth — AI-queryable, structurally precise, lossless.

The L2 output is a rendered projection: a coherence type label, a grade, a narrative summary. Human-readable and actionable — but lossy by design. The grade is a projection of disruption resilience, not a score of coherence quality. Two brands can share the same L2 grade while having structurally different L1 profiles — the same way different light spectra can produce the same perceived colour. This is spectral metamerism applied to brand analysis.

Note: L1 and L2 here refer to SBT output layers (spectral profile and rendered summary), not the OST specification cascade levels with the same numbering. In Organizational Schema Theory, L1 denotes Signal Requirements and L2 denotes Process Contracts. The two frameworks share layer notation but use it independently.

Why Mathematical?

Every brand strategist already thinks in dimensions. A positioning map has two axes, brands as dots, and distance between dots equals differentiation. That is geometry. SBT uses eight axes instead of two.

Why does that matter? Because every time you pick two axes for a positioning map, you discard six dimensions of information. Brands that look identical on the two you chose might be completely different on the six you left out. A nutritional label has thirteen dimensions and nobody calls it "too mathematical." A brand has eight perceptual dimensions. Measuring all eight instead of picking two is the difference between a positioning map and a positioning system.

The math does not replace intuition. It extends intuition into the territory where working memory runs out — like a telescope extends vision, or a spectrograph disperses light into its component wavelengths, revealing hidden spectral signatures your eyes alone cannot detect.

Mathematical Foundations

The framework's core mechanisms are backed by a series of mathematical research papers (R1-R7) that formalize spectral distance as a proper metric, characterize metamerism conditions, derive cohort boundary theorems, prove channel capacity limits, establish trajectory impossibility results, and model signal diffusion dynamics. These results are not merely theoretical — they are enforced as runtime constraints via a Python validation module with seven validators and 102 passing tests. Every LLM-generated spectral profile can be checked against proven mathematical bounds before use.

Traditional Concepts in Spectral Terms

Go Deeper

Apply the framework to any brand with the seven-module audit toolkit. Browse all 65 terms in the glossary. See it applied to Hermès, Tesla, IKEA, Patagonia, and Erewhon in the case studies. Read the article series that builds SBT from first principles.