Purpose

This page lists small, falsifiable research tracks that implement the dual view (objective constraints + agent layer). Each item names a question, a minimal protocol, and concrete measurements you can run with standard tools.

How to use this page

• Pick one track in your domain.
• Start with the “Minimal protocol.”
• Log the proposed measurements.
• Report what fails as well as what works.

A. Quantum foundations and control

1. Adaptive measurement as “closing the loop”

• Question: Does real‑time, model‑driven basis selection reduce predictive error relative to fixed bases?
• Minimal protocol: Implement a single‑qubit (or cavity) experiment with measurement‑based feedback; controller updates measurement basis to minimize next‑step negative log‑likelihood.
• Measure: Average predictive log‑loss vs fixed policy; energy/information budget; robustness to noise.
• Tools: QuTiP; Bayesian filters or RL; lab hardware or high‑fidelity simulators.
• Links: 🧿Quantum Foundations, 🎛️Quantum Control & Feedback, ✅Operational Criteria.

2. Coherence → work/information conversion

• Question: What’s the efficiency of converting off‑diagonal coherence into extractable work or mutual information?
• Minimal protocol: Resource‑theory mapping; implement control sequences that diagonalize in stages; record work (or information) per unit coherence consumed.
• Measure: Work/bit; sensitivity to dephasing; ablation on control precision.
• Tools: Resource theory of coherence; stochastic thermodynamics.

3. Phase as context

• Question: Can agent‑controlled phase encode task‑relevant relations that outperform modulus‑only baselines?
• Minimal protocol: Mach–Zehnder or Ramsey interferometry with adaptive phase; optimize task metric (classification/estimation).
• Measure: Task error vs phase control fidelity; degradation when phase is randomized.

B. Thermodynamics of open systems

1. Bounded Maxwell demon

• Question: What net work is achievable with explicit budgets for sensing, memory, and actuation?
• Minimal protocol: Implement a feedback trap or digital demon; cap bandwidth, memory depth, and actuation latency.
• Measure: Net work; mutual information; regret vs oracle; Pareto fronts cost↔benefit.
• Tools: Stochastic thermodynamics; feedback control libraries.
• Links: 🔥Open‑System Thermodynamics, 📊Information: Objective vs Subjective.

2. Structure growth cascade

• Question: How do energy and information gradients propagate across a photon → chemical → computational cascade?
• Minimal protocol: Photosensitive reaction feeding a simple controller that prints a structured artifact; track flows at each stage.
• Measure: Local entropy decrease; dissipation; transfer efficiency; modularity increase.

3. Control‑limited self‑assembly

• Question: How many control bits are needed to reach a target assembly fidelity at fixed energy input?
• Minimal protocol: Compare unguided vs guided self‑assembly under identical thermodynamic budgets; vary control granularity.
• Measure: Yield, defect rate, minimum control entropy for target fidelity.

C. Autocatalysis and proto‑agency

1. Interaction vs control in chemical sets

• Question: Does a network actively restore target ratios under perturbations (closed‑loop control) or just react passively?
• Minimal protocol: Microfluidic platform with periodic perturbations; test restoration policies (feeds/inhibitors).
• Measure: Recovery time, control gain, invariants maintained; counterfactual sensitivity.
• Tools: COPASI/BioNetGen; microfluidics.
• Links: 🧪Autocatalytic Sets & Proto‑Agency.

2. Evolving policies for homeostasis

• Question: Can simple policy search discover control schedules that stabilize autocatalytic networks?
• Minimal protocol: In silico evolutionary algorithm over feed/flush policies; validate best policies in vitro if possible.
• Measure: Stability region size; robustness; policy complexity vs performance.

3. Information budgets in synthetic metabolism

• Question: How do sensing/memory limits bound achievable homeostasis?
• Minimal protocol: Build toy circuits with tunable sensor noise and memory; sweep (information, energy) budgets.
• Measure: Phase diagrams; critical thresholds; trade‑off curves.

D. Neuroscience, cognitive science, AI

1. Agent surprisal vs source entropy

• Question: Does $−log Q_t(o_t)$ converge toward $−log P(o_t)$ as agents learn?
• Minimal protocol: Embodied or simulated agent; log subjective surprisal and estimate source surprisal; intervene on policy.
• Measure: $KL(P \|\| Q_t)$ decay; surprisal drops after actions; energy cost of updates.
• Tools: Active inference/RL; causal estimators.
• Links: 📝Text–Context–Interpretation, 🧭Agency & Delegation.

2. Info‑cloning vs learning

• Question: How do shared priors (“cloned context”) change adaptation speed/diversity?
• Minimal protocol: Two cohorts: scratch vs shared priors; identical tasks and budgets.
• Measure: Convergence time; policy diversity; robustness; transfer.

3. Counterfactual control tests

• Question: Do interventions produce detectably different future Texts consistent with the model’s predictions?
• Minimal protocol: Intervene on policy conditions; predict signatures; compare to realized data.
• Measure: Causal uplift; predictive accuracy improvement; control cost.

E. Collective behavior and cross‑scale inference

1. Individuality–uniformity index

• Question: Does a higher individuality index correlate with adaptability under regime shifts?
• Minimal protocol: Define index over swarms/consortia (variance of policies/trajectories under shared tasks); apply controlled shifts.
• Measure: Performance retention vs index; stability trade‑offs.
• Links: 🧬Hierarchies & Quasi‑Organisms.

2. Cross‑level misinformation costs

• Question: What are the failure modes when lower levels misread higher‑level Texts?
• Minimal protocol: Multiscale simulation; introduce mis-specified coarse‑grainings; apply corrective messaging.
• Measure: Error cascades; minimal message sets; recovery dynamics.

3. Collective closed loops

• Question: When do collectives enact closed loops distinct from individuals?
• Minimal protocol: Multichannel sensing; compute transfer entropy/Granger causality across scales; apply group‑level interventions.
• Measure: Control energy localized at collective modes; effect sizes of group‑targeted actions.

F. Measurement and reporting standards

For any track, please log:

• Agent layer: who/what owns $Q$ (Context), goals, sensors/actuators.
• Budgets: energy, time, bandwidth, memory.
• Baselines: fixed policies, no‑feedback, randomized controls.
• Metrics: predictive log‑loss, KL divergence, mutual information, work, regret, robustness.
• Counterfactuals: what would have happened under an alternative action/model?
• Misses: where the dual view adds nothing or contradicts data—record and refine.

Getting started (low friction)

• Simulation first: reproduce a minimal example in silico; validate with ablations.
• Then instrument: add sensing/control cost accounting.
• Only then scale: move to higher‑dimensional systems or hardware.

Tooling quicklist

• Quantum: QuTiP, qiskit, reinforcement‑learning controllers.
• Thermo/open systems: PySINDy, noneq MD, information estimators (JIDT, dit).
• Chemical networks: SBML, COPASI, microfluidics toolchains.
• AI/ML: PyTorch/JAX, active inference libs, causal inference kits.

Contribute

• Propose a new direction: include a one‑paragraph question, minimal protocol, and 3–4 measurements.
• Share code/data: link a repo or dataset; include a README with metrics and baselines.
• Contact: see 📬Contact & Updates.

Citation

• Book: Alexander Neshmonin, Changing the Paradigm of Life: New Answers to the Old Questions (EN edition), ISBN: 9798316199631.
• Site content: CC BY 4.0. Short quotations and adaptations encouraged with attribution.