In Interaction, the Entities row identifies the fundamental objects that constitute a multi-agent world. Unlike Choice, where only one decision-making unit exists, Interaction requires a richer ontology: multiple agents, the influence structures that connect them, and the institutional or rule-based environments that shape how those influences operate.
This section establishes the core building blocks of strategic and relational analysis. The domain must declare which objects—agents, actions, information sets, mechanisms, contracts, networks—exist as primitives at this scale. These entities determine how strategic dependence is modeled, how equilibria are defined, and how behavior is interpreted. If something cannot be expressed using these relational entities or their compositions, it does not belong inside the domain.
Entities here define what the world is made of when outcomes emerge from the interactions among decision-makers.
Science Analysis Template
These are the structural patterns found across all Scientific Disciplines
1. Interaction selects a small set of primitive objects that define its world.
The ontology of Interaction consists of the minimal primitives required to model how multiple agents affect one another.
These primitives include:
- multiple agents, each with independent decision-making capacity,
- actions taken by each agent,
- influence structures connecting agents (who affects whom),
- information sets, including who knows what and when,
- rules or institutions that govern how choices translate into outcomes,
- outcome functions mapping joint actions into results.
These are the atoms of Interaction’s universe.
No system-level variables or aggregate entities exist at this scale; nor do sub-agent psychological or biological mechanisms.
Everything that Interaction explains must be reducible to these primitives or lawful compositions of them.
2. Interaction’s entity-set is mutually exclusive from both Choice and Aggregation, while mutually supporting within itself.
Because Interaction requires interdependence, it cannot assume:
- a solitary agent (Choice),
- a fixed, non-responsive environment (Choice),
- macro-state variables or system dynamics (Aggregation),
- population-wide aggregates or distributions (Aggregation).
Likewise, Interaction assumes nothing below the agent:
no neurons, no subconscious modules, no physiological systems.
Its ontology is the set of agents + relational structures.
Within this set, all primitives support one another:
- influence requires agents,
- rules organize influence,
- information defines strategic choices.
This exclusivity prevents the domain from drifting downward into psychology or upward into system science.
3. Interaction’s ontology determines the correct level of explanation.
Because Interaction defines multiple agents and their relations as the fundamental objects, it implicitly sets:
- what can interact: agents, signals, actions, and rules;
- what processes are meaningful: strategy formation, belief updating, coordination, conflict, and equilibrium;
- what can be measured or modeled: joint action profiles, payoffs, strategic dependencies;
- what kinds of laws apply: equilibrium principles, incentive compatibility, information constraints;
- what counts as a cause: the structure of interdependence, institutional rules, informational asymmetries.
It also sets strict limits:
- outcomes cannot be explained using a single-agent decision rule (Choice),
- outcomes cannot be explained using system-level state evolution (Aggregation).
The entity list filters which explanations are coherent at this scale.
4. Composite entities build upward in Interaction, but only through its own relational rules.
Interaction allows composite objects, but only if they are built from its primitives:
- action profiles (combinations of agent choices),
- information structures (combinations of beliefs and signals),
- institutions (stable rule systems),
- mechanisms (designed mappings from messages to outcomes),
- networks (patterns of influence),
- equilibria (consistent action-belief configurations).
However, composite entities cannot involve:
- aggregates of agents treated as a single unit (Aggregation),
- internal decomposition of agents into subcomponents (lower domains),
- any entity requiring macro-state evolution or system feedback.
All legitimate composite entities must be expressible as structured combinations of agents, actions, rules, and information.
5. Interdisciplinary fields expose joints between Interaction and neighboring domains.
Hybrid fields occur precisely where entity sets intersect:
- Game theory ↔ Psychology:
When modeling belief formation or bounded rationality requires cognitive entities. - Mechanism design ↔ Institutional economics:
When rules enter as entities with formal and behavioral properties. - Network theory ↔ Sociology:
When relational patterns overlap with social structures. - Industrial organization ↔ Aggregation:
When agent interactions scale into market-level or system-level patterns.
These hybrids only work because Interaction’s ontological boundaries remain clear.
The joint operates at the interface, not by blurring the domains.
6. Interaction admits formal entities only as abstract structures tied to interdependence.
Interaction relies heavily on abstract constructs:
- strategy sets,
- information partitions,
- payoff matrices,
- correspondence mappings,
- equilibrium operators,
- mechanism functions.
These are not physical or social entities; they exist only as formal representations of the interdependent structure.
They are anchored to the ontology of “multiple agents + relational rules.”
These constructs lose meaning:
- if reduced to single-agent choice functions (Choice),
- or inflated into system-level laws (Aggregation).
Thus the mathematics of Interaction is scale-specific to multi-agent strategic structure.
7. The resulting ontology of Interaction is non-overlapping, structurally precise, and cleanly bounded.
After defining its entities, Interaction becomes an ontological territory distinct from adjacent domains:
- It contains multiple agents and the structures connecting them.
- It excludes both solitary agents (Choice) and systems/aggregates (Aggregation).
- It prohibits mixing in entities from psychology, biology, or macro-level dynamics.
- It demands that all explanations emerge from relational primitives or lawful expansions of them.
With these boundaries, Interaction is:
- non-overlapping,
- non-redundant,
- and structurally anchored to its proper scale.
Everything within the domain must be decomposable into its entities; everything outside belongs to another field.
