attractor-basin-cognition 0.1.0

Attractor-basin primitives for cognitive signal routing, field dynamics, and host-neutral mental model substrates.

Attractor-Basin Cognition

Attractor-Basin Cognition (ABC) is a standalone Python library for modeling cognitive signals as attractor-basin dynamics.

Tagline: all cognition is attractor basins when done right.

ABC is not a host application. It does not import or depend on agent servers, graph databases, webhooks, or any project-specific cognitive model. Hosts translate their own objects into ABC's neutral signal protocols.

What It Provides

• Hopfield-backed basin creation, recall, strengthening, and stability checks.
• Oscillatory basin-field dynamics for basin influence, resonance, coupling, and persistence profiles.
• A host-adapter boundary for cognitive systems that need to route signals without coupling this library to their object model.
• Neutral signal protocols for a ladder from impulse-scale activation to packet-like signals, concept fields, and basin graphs.

Signal Ladder

ABC keeps the public API neutral:

• BasinImpulse: a minimal activation plus optional dominant basin.
• PacketLike: a packet-scale signal with content, precision, coherence, and metadata.
• ConceptFieldLike: a field-scale signal with concepts, field energy, and a dominant basin.
• BasinGraphLike: a graph-scale signal with constituent basins, cohesion, and stability.

Application-specific objects belong in host adapters, not inside this package.

Install

pip install attractor-basin-cognition

Quick Start

from abcognition import CoreAttractorService

service = CoreAttractorService(n_units=128)

await service.create_basin(
    "procedural-basin",
    "stepwise task execution and tool use",
)

await service.create_basin(
    "experiential-basin",
    "episodic recall and lived context",
)

result = await service.find_nearest_basin("remember what happened last session")

Field dynamics:

from abcognition.field import BasinFieldEngine

field = BasinFieldEngine(["experiential-basin", "procedural-basin"])
state = await field.step({
    "experiential-basin": 0.8,
    "procedural-basin": 0.2,
})

print(state.influence)
print(state.resonance)

Host-Adapter Boundary

ABC owns basin math and signal contracts. Your host owns identity, persistence policy, memory graphs, agent lifecycle, domain schemas, and product-specific names.

Correct integration shape:

from dataclasses import dataclass
from typing import Optional

@dataclass
class MyHostSignal:
    activation_level: float
    dominant_basin: Optional[str]

Then pass host data through ABC protocols or services without ABC importing the host.

Current Status

This is an alpha extraction. The initial package includes the reusable attractor service, basin state model, field engine, field state model, and neutral protocols. Higher-level mental model helpers should be added only when they remain host-neutral.