Research software

P/G Engine
Experimental Platform

P/G Engine is an experimental implementation of specified P/G core mechanisms for controlled studies.

Its frozen computational core includes state updates, relational dynamics, affect, valuation, execution constraints, and decision/search.

A relational architecture for controlled experiments

The frozen P/G core has been implemented end to end, including property and relational updates, affect, valuation, execution constraints, and the specified decision/search path. Closed-loop synthetic verification has been completed under benign and adversarially perturbed conditions.

These results establish implementation coherence under the tested synthetic conditions; they do not constitute human behavioral validation.

Specified state and updates

The experimental core represents declared property and relationship dimensions and computes updates from represented events. Versioned schemas and parameters define the scope of each experiment.

Traceable affect contributions

The affect computation retains signed contributions with their object and structural origin. An aggregate readout can be compared with judgments under a specified observation model.

Trust is a relational dimension; reputation is a property representation. Derived emotions such as gratitude and fear belong to a separate interpretive layer.

Authoring and response layers

The current experimental implementation uses default personality parameter tables. Big Five authoring, personalization layers, the witness-related extension (E4), and deployment hardening and interfaces remain future work.

Implementation checks and empirical evaluation

Specification-based checks ask whether the code realizes the intended dynamics. The completed registered CTT cycle tested computational predictions and model discrimination; R2-2 remains a documented limitation of the frozen mechanism set. This work is separate from the prepared Stage B human protocol, which has not started and would compare specified model predictions with human affect judgments.

Where the framework may be tested

Game agents, human–robot interaction, and economics pose different research questions. Each direction requires its own models and evaluation.

Demo coming soon

A real demo link will be added here when available.