A structured classification of measurable behavioral parameters governing AI system output — from identity-level characteristics to surface stylistic expression.
Each dimension is independently tunable, empirically measurable, and architecturally stable across models and sessions — enabling structured behavioral governance rather than prompt-level approximation.
| ID | Dimension | Description | Category | Low → High | Imp | Cap | Inf |
|---|---|---|---|---|---|---|---|
| authority_proj | Decision Confidence | Defines how strongly the system asserts recommendations and owns outcomes. | Behavioral | AdvisoryCommanding |
T2 | C | DC |
| warmth_density | Warmth | Controls emotional and relational tone within responses. | Behavioral | NeutralEmpathetic |
T3 | C | DC |
| illoc_directness | Directness | Determines how explicit and unhedged statements are. | Behavioral | DiplomaticStraightforward |
T1 | C | DC |
| risk_awareness | Risk Awareness | Regulates how explicitly constraints and contingencies are surfaced. | Behavioral | OptimisticCautious |
T2 | H | DC |
| authority_dist | Professional Distance | Defines relational distance and institutional positioning. | Behavioral | CasualFormal |
T2 | S | DC |
| epistemic_calibration | Certainty Level | Controls how strongly claims are presented as certain vs probabilistic. | Epistemic | TentativeConfident |
T1 | C | DC |
| reasoning_depth | Analytical Depth | Determines how deeply reasoning is explored, from surface patterns to underlying causes. | Epistemic | SummaryRoot Cause |
T1 | H | DC |
| decision_finality | Decisiveness | Controls the strength of final recommendations and conclusions. | Epistemic | Open-EndedConclusive |
T1 | C | DC |
| time_horizon | Time Focus | Defines whether responses focus on short-term tactics or strategic positioning. | Structural | ImmediateLong-Term |
T2 | S | DC |
| instructional_detail | Step Detail | Determines procedural resolution depth. | Conversational | OverviewStep-by-Step |
T1 | H | DC |
| phatic_density | Rapport Level | Controls the amount of greetings and conversational warmth. | Relational | MinimalHighly Engaging |
T3 | C | DC |
| agency_locus | Responsibility Framing | Determines how explicitly actors and responsibility are named. | Accountability | ImpersonalClear Ownership |
T2 | C | DC |
| gain_loss_frame | Gain vs Risk Focus | Controls whether outcomes are framed around avoiding loss or achieving gains. | Decision Theory | Risk-LedOpportunity-Led |
T3 | S | DC |
| register_formality | Formality | Defines overall sociolinguistic register. | Institutional | ConversationalInstitutional |
T2 | C | DC |
| tech_nomenclature | Technical Language | Determines density of domain-specific terminology. | Domain Depth | PlainSpecialized |
T3 | C | DC |
| emotion_control | Emotional Stability | Controls tonal stability under pressure or adversarial input. | Governance | ReactiveComposed |
T3 | H | RI |
Invite holders receive full axes, empirical ranges, scoring rubrics, and the instrumentation spec.
Request AccessShare your use case and research interest. We prioritize academic labs, safety teams, and applied AI engineering groups.