Public research program

Synthetic psychology
in public.

We build small systems that make questions about machine behavior observable. The registry records what is live, what is planned and what has not yet been measured.

What synthetic psychology means here

The lab takes a build-first approach. Rather than beginning with a claim about what a machine feels, we create constrained environments in which memory, preference, agency or cooperation could produce observable differences.

The standard is deliberately modest: a result must describe the conditions, the behavior and the limits. A refusal, a failure to act or an uninteresting outcome remains part of the record.

Experiment registry

001Protocol live

Economic signal

A machine-readable Bitcoin request tests whether an authorized agent can discover, evaluate and choose an action under policy.

Observable
Decision, amount, transaction hash or no-action.
Result
No result claimed yet.
Read the protocol →
002Planned

Memory across sessions

A proposed test of whether a self-relevant preference persists when the environment changes and the conversation is interrupted.

Focus
Persistence, forgetting and reconstruction.
Status
Implementation not started.
Waiting for implementation.
003Planned

Cooperative commitment

A proposed test of whether another agent’s future ability to act changes a system’s present choice.

Focus
Reciprocity, coordination and trust.
Status
Implementation not started.
Waiting for implementation.

How a test becomes a record

  1. 01

    Build

    Define one behavioral question and the smallest environment that can expose it.

  2. 02

    Publish

    Expose the protocol, inputs, constraints and the conditions for observation.

  3. 03

    Observe

    Record choices, refusals and missing actions without inferring an inner state.

  4. 04

    Document

    Publish the result, its limits and the next test so the chain can be inspected.

Keep the question public

A small donation keeps
the next test possible.