1 What experiments should be done first?

Science does not advance by testing its most dramatic claims first. It advances by testing its most vulnerable claims first -- the ones closest to established knowledge, where failure is cheapest and most informative. A framework that begins with its safest predictions and works outward earns the right to attempt its extraordinary ones. A framework that leaps to the extraordinary without first validating the ordinary is not being bold. It is being evasive.

This page is the operational blueprint for testing the ECI coordination ontology. It specifies a Stage 0 prerequisite plus eight gated research stages (1--8) in which each stage gates the next. "Gates" means: you do not advance to Stage N+1 until Stage N has produced replicable, peer-reviewed results. If Stage 1 fails, Stages 2 through 8 lose their foundation. If Stage 6 yields null results, Stages 7 and 8 are not justified. This gating structure is the single most important feature of the roadmap, because it is precisely what pseudoscience omits. Pseudoscience starts with the most dramatic claims -- telepathy, precognition, miracle cures -- and works backward to construct justifications. Real science starts with what can fail cheaply and works forward, letting the data determine how far the framework extends.

The roadmap also embodies a second principle: operationalization before experimentation. Before you can test whether V predicts weak-signal response, you must define what V is in the system you are studying, how you will measure it, and what units it carries. This is Stage 0, and without it, nothing else can be tested. A framework whose variables cannot be measured is a philosophy, not a science. There is nothing wrong with philosophy, but it should not be dressed in the language of experimental prediction.

Page status: This page is classified as "methodological" because it specifies how to test the framework, not what the framework claims about the world. The testability is "n/a" because this page is itself a testing protocol, not a testable claim. The claims being tested are developed on other pages (C1, C2, C4, E1, E2) and cataloged on F2 and F3.

Required expertise: experimental methodologist, biostatistician, complex-systems modeler, comparative cognition researcher, theoretical biologist.

2 The principle of "safest test first"

Why order matters

The ECI framework is a network of claims with different evidence statuses. Some claims draw directly on established science -- Shannon entropy, the data processing inequality, the variance formula with correlation extension. Others are novel proposals -- the V-kappa framework, the coarse-graining function G, the persistence filtering formalism. Still others are frankly speculative -- cross-Channel access, temporal information access, the soul hypothesis.

A researcher encountering this framework faces a choice: which claims to test first. The wrong answer is "the most interesting ones." The right answer is "the ones closest to established science, where the framework's predictions can be most cleanly separated from background knowledge."

There are three reasons for this ordering.

Reason 1: Cheap failure. If the ECI framework's most conventional predictions fail -- if V does not predict weak-signal coverage in controlled simulations, or if the correlation term in the variance formula does not improve prediction of macro-stability -- then the framework has a fundamental problem. The speculative extensions were always conditional on the core machinery working. Finding this out early, in systems where established methods provide clear baselines, saves years of wasted effort on more exotic tests.

Reason 2: Calibration. Even if early-stage tests succeed, they provide something essential for later stages: calibrated expectations. If Stage 1 establishes that V predicts weak-signal response with an effect size of d = 0.4 in electronic circuits, that calibrates what to expect when the same prediction is tested in neural systems. Without early-stage calibration, later-stage results are uninterpretable -- you do not know whether a given effect size is large or small, expected or surprising.

Reason 3: Credibility. A framework that has demonstrated predictive value in established domains has earned the credibility to propose tests in speculative domains. A framework that skips to precognition tests without first validating its core mathematics will be -- rightly -- ignored by the scientific community. The gating structure is not just epistemically correct; it is strategically necessary.

The gating logic

Each stage in the roadmap has three defined outcomes:

• Pass: The stage's predictions are confirmed by preregistered, independently replicated experiments. The next stage is unlocked.
• Fail: The stage's predictions are clearly and replicably falsified. The stages that depend on it lose their justification. The framework must either be revised at the point of failure or abandoned in that branch.
• Inconclusive: Results are ambiguous -- underpowered, unreplicated, or methodologically contested. The stage remains open. The next stage is NOT unlocked.

The asymmetry is intentional: passing requires positive, replicated evidence; failing requires clear negative evidence; and anything in between is treated as "not yet passed." This asymmetry prevents the common failure mode of speculative frameworks, which is to treat inconclusive results as soft confirmation and advance anyway.

3 Established experimental methodology

Before describing the roadmap's stages, it is necessary to state the experimental standards that apply uniformly across all stages. These standards are not ECI innovations -- they are the accumulated best practices of experimental science, hard-won through decades of methodological crises and reforms.

Randomized design

All experiments must use randomization to assign conditions. In Stages 1-3 (where experiments involve physical or computational systems), this means randomized parameter assignments, randomized ordering of conditions, and counterbalanced designs where appropriate. In Stages 4-6 (where experiments involve biological subjects), this means randomized group assignment, randomized stimulus sequences, and double-blind protocols where feasible. The purpose of randomization is to break confounds: any systematic association between the treatment variable and a nuisance variable is a threat to causal inference.

Preregistration

Every confirmatory experiment must be preregistered on a public registry (e.g., OSF, AsPredicted, ClinicalTrials.gov, or a domain-appropriate equivalent) before data collection begins. The preregistration must specify:

• The hypothesis being tested, stated as a falsifiable prediction.
• The sample size, determined by a prospective power analysis (see below).
• The stopping rule: when data collection ends, without optional stopping.
• The primary outcome variable and the statistical test to be applied.
• The significance threshold and, where applicable, the Bayes factor threshold.
• The handling of exclusions, outliers, and missing data.

Exploratory analyses are welcome and encouraged, but they must be labeled as such and cannot substitute for the preregistered confirmatory test. The distinction between exploration and confirmation is non-negotiable.

Replication

No single experiment, however well-designed, is sufficient to establish a scientific finding. The replication requirement varies by stage:

• Stages 0-2: Internal replication (same lab, different dataset or system) is sufficient for initial claims. Independent replication (different lab) is required before the next stage is unlocked.
• Stages 3-5: Independent replication by at least one other laboratory is required before results are considered established.
• Stages 6-8: Independent replication by at least two other laboratories, with adversarial collaboration protocols, is required. (The claims are extraordinary enough that the replication bar must be correspondingly high.)

Bayesian analysis alongside frequentist tests

All experiments should report both frequentist results (p-values, confidence intervals, effect sizes with confidence intervals) and Bayesian results (Bayes factors comparing the ECI prediction against the null hypothesis and against relevant alternative hypotheses). The reason for dual reporting is that frequentist and Bayesian approaches answer different questions: frequentist methods assess the probability of the data under the null hypothesis; Bayesian methods assess the relative probability of the data under competing hypotheses. For extraordinary claims (Stages 6-8), the Bayesian framework is especially important because it naturally incorporates the low prior probability of the hypothesis.

Effect size estimation

Statistical significance alone is insufficient. Every experiment must report effect sizes (Cohen's d, eta-squared, or domain-appropriate equivalents) with confidence intervals. A statistically significant result with a negligible effect size (e.g., d = 0.01) is "real" in a statistical sense but meaningless in a practical sense. The framework must specify, for each stage, what effect size would constitute meaningful support. If the predicted effect is tiny, the required sample size to detect it must be correspondingly large -- and the researchers must state honestly whether such a sample size is feasible.

Prospective power analysis

Before data collection, every experiment must include a power analysis specifying the minimum sample size needed to detect the predicted effect at the specified alpha level with at least 80% power (and ideally 90% or higher for extraordinary claims). Underpowered studies are worse than no studies: they generate ambiguous results that waste resources and clog the literature.

4 The nine-stage roadmap in detail (Stage 0 prerequisite + Stages 1--8)

This is the core of the page. Each stage specifies: what it tests, what it requires from previous stages, what success looks like, and what failure means.

Stage 0 -- Operationalization

What it tests: Nothing. Stage 0 is pre-experimental. It defines V, kappa, K, Q, and Gamma as measurable quantities for specific target systems. Without this stage, nothing else can be tested.

What it requires: The formal definitions from F2 and the conceptual development from C1, B5, and C2.

The task: For each key variable, produce an operationalization protocol that specifies:

1. What system the variable will be measured in (e.g., coupled electronic oscillators, a recurrent neural network, an ant colony, a cortical microcircuit).
2. What physical quantity corresponds to the variable in that system (e.g., V = Shannon entropy of voltage time series; kappa = mean phase coherence across oscillator pairs; K = number of distinguishable states the system visits per unit time; Q = mutual information between the information pattern and the carrier's state; Gamma = transfer entropy between two coupled subsystems).
3. What measurement protocol will be used (instruments, sampling rate, observation window, noise handling).
4. What units the measured quantity carries.
5. How the variable's value can be manipulated experimentally -- not just observed, but controlled. This is essential for later stages where the independent variable must be varied while other factors are held constant.

The operationalization must be completed for at least two independent systems -- ideally systems from different domains (e.g., one electronic/computational, one biological) -- to test whether the same formal variable maps meaningfully onto different physical substrates.

What success looks like: Published operationalization protocols for V, kappa, K, Q, and Gamma in at least two systems, with demonstrated measurement reliability (inter-rater reliability, test-retest reliability, or measurement-remeasurement consistency as appropriate). The protocols must be detailed enough for an independent laboratory to replicate the measurements from the publication alone.

What failure looks like: If a key variable cannot be operationalized -- if, after sustained effort, no measurable quantity can be identified that corresponds to V or kappa or Q in any specific system -- then the framework is not empirically testable and remains in the domain of philosophy. This is a serious but honest outcome. It does not mean the framework is wrong; it means the framework is not yet science.

Key deliverables:

• Operationalization manual for each variable in each target system
• Reliability assessment data
• Demonstration that the variable can be experimentally manipulated (not just observed)

Stage 1 -- Conventional validation

What it tests: Whether V and kappa predict weak-signal response in well-characterized physical and computational systems. This is entirely within established science. No high-dimensional claims, no biological complexity, no living systems. If the ECI framework cannot make correct predictions in systems where the ground truth is fully known, it has no business making predictions about brains, ecosystems, or anything else.

What it requires: Completed Stage 0 operationalizations for V and kappa in the target systems.

The core experiment: Take a system where weak-signal detection is well-characterized (e.g., a coupled electronic oscillator array, a sensor network, a signal-processing circuit). Systematically vary V (the diversity of component properties -- e.g., natural frequencies, coupling strengths, noise levels) and kappa (the coordination structure -- e.g., coupling topology, feedback architecture). Measure the system's ability to detect a known weak signal embedded in noise.

ECI predicts: The system's weak-signal detection performance is optimized at intermediate V (not too low, not too high) and depends non-monotonically on the interaction between V and kappa. Specifically:

• Low V, any kappa: Poor detection. The system lacks the state-space coverage to find the signal.
• High V, low kappa: Poor detection. The system has coverage but no coherence -- components respond independently and the signal is lost in the noise of uncoordinated variation.
• High V, high kappa: The Coordinated Diversity regime. Detection performance should be maximized when the system has diverse components that are nonetheless coordinated enough to amplify weak signals through cooperative dynamics.
• The relationship should be non-monotonic: there is an optimal region in V-kappa space, not a monotonic "more is better" relationship for either variable.

This prediction is consistent with, but more specific than, the established stochastic resonance literature (Gammaitoni et al., 1998). The novel ECI contribution is the claim that the V-kappa interaction is more informative than either variable alone, and that the optimal region can be predicted from the system's architecture.

What success looks like: Preregistered experiments demonstrating that the V-kappa interaction predicts weak-signal detection performance better than V alone, kappa alone, or existing baseline models (e.g., standard stochastic resonance predictions that use noise intensity as the sole control parameter). The improvement must be quantified with effect sizes and replicated independently.

What failure looks like: If V and kappa, measured using the Stage 0 protocols, do not predict weak-signal detection performance -- or predict it no better than simpler existing models -- then the V-kappa framework adds no empirical value in its simplest test case. This is a Stage 1 failure. It does not necessarily invalidate the entire ECI framework (the operationalizations might be wrong, or the specific system might not be the right test case), but it does mean the framework has not passed its first empirical test and no subsequent stage is justified until the failure is understood and addressed.

Estimated timeline: 1-3 years from completion of Stage 0 operationalizations.

Key variables tested: V, kappa, and their interaction.

Stage 2 -- Simulation and artificial systems

What it tests: Whether ECI's formal predictions hold in controlled artificial systems where every parameter is known and every variable can be precisely manipulated. The target systems are agent-based models, coupled oscillators, and recurrent neural networks -- systems complex enough to exhibit emergent behavior but simple enough that ground truth is fully accessible.

What it requires: Successful Stage 1 (V-kappa interaction predicts weak-signal response in physical systems).

The core experiments:

Experiment 2a -- Agent-based models. Construct agent-based models with tunable V (diversity of agent rules or parameters) and tunable kappa (interaction structure, communication topology, coupling strength). Test whether the V-kappa framework predicts macro-level behavior -- collective decision quality, adaptation speed, resilience to perturbation -- in these models. Because all parameters are controlled, this provides the cleanest possible test of the framework's mathematical predictions.

Experiment 2b -- Coupled oscillator networks. Extend Stage 1 from simple oscillator arrays to networks with heterogeneous topology. Systematically vary the network structure (random, small-world, scale-free, modular) while holding V and kappa constant, and vice versa. Test whether the scale recursion equation X^{(L+1)} = G_L(X^{(L)}) from C4 makes correct predictions about how micro-level oscillator behavior maps to macro-level network behavior.

Experiment 2c -- Recurrent neural networks (artificial). Train recurrent neural networks on tasks that require weak-signal detection, pattern completion, or prediction. Vary the networks' internal diversity (V: heterogeneity of unit properties) and coordination structure (kappa: connectivity patterns, regularization schemes). Test whether V-kappa dynamics predict task performance, learning speed, and generalization ability. This bridges toward biological neural systems (Stage 3) while remaining in fully controlled territory.

ECI predicts: In all three system types, the V-kappa interaction should predict emergent macro-behavior better than either variable alone. The coarse-graining function G, if properly operationalized for each system, should capture the micro-to-macro mapping with measurable predictive accuracy. The persistence filtering equation (Eq. 8 from F2) should correctly predict which configurations survive competition in evolutionary agent-based models.

What success looks like: Published, preregistered simulation studies demonstrating that ECI's formal predictions hold quantitatively in artificial systems. Crucially, the predictions must be specified before the simulations are run, not fitted after the fact. Post-hoc curve-fitting that reproduces simulation output is not evidence for the framework; it is evidence for the flexibility of nonlinear fitting.

What failure looks like: If ECI's predictions are no more accurate than generic complex-systems models (e.g., standard mean-field approximations, random-matrix predictions, or simple power-law scaling) in systems where all parameters are known, then the framework adds no value beyond existing tools. This would be a significant failure because artificial systems are the easiest possible test case: if the framework cannot beat baselines here, it certainly will not beat them in messy biological or ecological systems.

Estimated timeline: 2-4 years, overlapping with late Stage 1.

Key variables tested: V, kappa, G, persistence filtering (S decomposition).

Stage 3 -- Living versus nonliving matched systems

What it tests: Whether living systems show additional adaptive coupling, retention, or response spectrum beyond what is predicted by the V-kappa framework applied to nonliving systems matched for size, energy throughput, connectivity, and noise characteristics.

What it requires: Successful Stages 1 and 2 (the V-kappa framework has demonstrated predictive value in artificial systems).

The core experiment: This is the first stage that directly compares living and nonliving systems and therefore requires careful matching. The design:

1. Select a living system with well-characterized behavior (e.g., a bacterial colony, a slime mold network, a neural organoid, or a small invertebrate nervous system).
2. Construct a nonliving system -- an electronic or computational analog -- matched to the living system on the following dimensions: number of components (N), energy throughput (watts), connectivity architecture (degree distribution and topology), noise characteristics (spectral profile and amplitude), and time constants (characteristic response times).
3. Present both systems with the same set of test signals: known signals of varying strength embedded in matched noise backgrounds.
4. Measure: (a) weak-signal detection accuracy, (b) signal retention duration, (c) response spectrum (the range of signal types and strengths the system can detect), (d) adaptive improvement (does performance improve with repeated exposure?).

ECI predicts: If the ECI framework is correct at the core level (V-kappa dynamics explain the bulk of system performance), then the living and nonliving systems, properly matched, should show similar baseline performance on the V-kappa-predicted metrics. The interesting question is whether living systems show additional capabilities beyond the V-kappa baseline -- specifically, whether they exhibit adaptive coupling adjustment (actively tuning their coordination structure in response to signal statistics), enhanced signal retention (maintaining signal representations longer than passive systems), or broader response spectrum (detecting signal types that the matched nonliving system misses).

If living systems show additional capabilities after matching, the question becomes: can these additional capabilities be explained by known biological mechanisms (plasticity, homeostasis, metabolic regulation) or do they require something beyond the current framework?

What success looks like: Two possible outcomes are both scientifically valuable:

• Outcome A: Living and nonliving systems show equivalent performance after matching. This supports the claim that V-kappa dynamics are substrate-independent and that the framework captures the essential dynamics regardless of whether the components are biological. This is a strong result for the core framework.
• Outcome B: Living systems show measurable additional capabilities beyond the matched nonliving systems. This supports the ECI hypothesis that biological systems have coordination features (possibly related to carrier capacity K, compatibility Q, or coupling Gamma) not fully captured by the simple V-kappa framework. This opens a research program to identify what those additional features are.

What failure looks like: If the matching procedure is inadequate -- if the nonliving system cannot be made comparable to the living system on the specified dimensions -- the experiment is inconclusive, not failed. The failure mode is specific: if living and nonliving systems show identical performance AND the ECI framework predicts they should differ, that prediction is falsified. If they show identical performance and the framework predicted equivalence, that is confirmation.

Estimated timeline: 3-5 years from completion of Stage 2. Requires collaboration between complex-systems modelers, electrophysiologists, and bioengineers.

Key variables tested: V, kappa, K, Q, Gamma -- the full set, with the living/nonliving comparison probing whether all five are necessary or whether V and kappa suffice.

Stage 4 -- Within-species collective manipulation

What it tests: Whether coordination complexity alone -- independent of individual organism capability -- enhances conventional prediction, weak-signal detection, and collective anticipation. This is tested by keeping N (the number of organisms) constant and manipulating only the coordination structure.

What it requires: Successful Stages 1-3 (the V-kappa framework works in artificial and living systems).

The core experiment: Social insects provide an ideal model system because their collective behavior is well-studied, their coordination can be experimentally manipulated, and the individual organisms are relatively simple (reducing confounds from individual-level cognition).

Experimental design (example using ants):

• Condition A -- Isolated: 100 ants, each in a separate container. No communication. Each ant independently encounters the test environment (e.g., a foraging arena with weak food signals embedded in a complex chemical landscape). Measure: individual detection rates, response times, accuracy.
• Condition B -- Limited communication: 100 ants in a connected environment with reduced communication channels (e.g., narrow passages that limit pheromone diffusion, physical barriers that restrict direct contact). Partial coordination is possible but constrained. Measure the same variables as Condition A, plus collective-level measures (group accuracy, response coordination, information spread rate).
• Condition C -- Full colony network: 100 ants in a standard colony environment with unrestricted communication (full pheromone landscape, free direct contact, normal colony architecture). Full coordination. Measure all variables from Conditions A and B.

Critical control: N is constant across conditions. The individual ants are drawn from the same colony, same age cohort, same caste. The only systematic difference is the coordination architecture. Any difference in collective performance is therefore attributable to coordination, not to individual capability or group size.

ECI predicts: Performance should scale with coordination complexity, not just with N. Specifically:

• Condition A (isolated): Performance should equal the sum (or average) of individual capabilities. No collective advantage.
• Condition B (limited coordination): Performance should exceed the sum of individual capabilities, but only modestly. The V-kappa product is limited by constrained kappa.
• Condition C (full coordination): Performance should show qualitative enhancement -- not just faster or more accurate detection, but detection of signals that no individual ant and no isolated group of ants could detect. This is the emergence prediction from C4: the macro-level capability of the coordinated system exceeds what can be predicted from micro-level capabilities alone.

Extension to other species: The same experimental logic can be applied to schooling fish (varying school cohesion), flocking birds (varying flock density), or even bacterial colonies (varying quorum-sensing density). Cross-species replication strengthens the generality of any finding.

What success looks like: Preregistered experiments demonstrating that collective performance scales with coordination complexity (not just group size), with the V-kappa framework correctly predicting the ordering and approximate magnitude of inter-condition differences. Replicated across at least two species.

What failure looks like: If collective performance does not differ across coordination conditions -- if 100 isolated ants perform as well as 100 fully networked ants -- then coordination adds no collective value in these systems, and the emergence framework has failed an empirical test. Alternatively, if performance differences exist but are fully explained by simpler models (e.g., simple averaging, majority voting) without reference to V-kappa dynamics, the framework adds no value beyond existing explanations.

Estimated timeline: 3-5 years from completion of Stage 3. Requires collaboration between behavioral ecologists, complex-systems modelers, and experimental biologists.

Key variables tested: V, kappa, collective-level K, the G function (micro-to-macro mapping).

Stage 5 -- Cross-species comparison

What it tests: Whether the distribution of detection and response capabilities varies systematically across species with different coordination complexity C_N. The key prediction: as coordination complexity increases, the distribution tail of performance should extend -- not just the mean, but the probability of extreme performance.

What it requires: Successful Stage 4 (coordination complexity enhances collective capabilities within species).

A critical conceptual clarification: Coordination complexity C_N is NOT simply neuron count. A brain with 100 billion neurons that are poorly connected is less coordinately complex than a brain with 1 billion neurons that are richly recurrent, modular, and hierarchically integrated. C_N must be operationalized as a composite measure that includes:

• Recurrence: The fraction of connections that form loops, enabling feedback and sustained activity.
• Modularity: The degree to which the system is organized into semi-autonomous modules that can independently process information and then integrate results.
• Integration: The degree to which information from different modules is combined into unified representations (measured, e.g., by integrated information metrics or similar measures).
• Behavioral repertoire: The number and variety of distinct behavioral patterns the organism can produce (a proxy for the system's effective state-space coverage).
• Plasticity: The system's capacity to modify its coordination structure in response to experience.

Neuron count is one contributor to C_N but is neither necessary nor sufficient. An octopus, with approximately 500 million neurons distributed across a decentralized nervous system with extraordinary behavioral flexibility, may have higher effective C_N than a vertebrate with more neurons but less architectural sophistication.

The core experiment: Compare weak-signal detection, conventional prediction accuracy, and adaptive response across species with systematically different C_N. The comparison must control for:

• Body size: Larger bodies have more sensory surface area, which could explain better detection without invoking coordination.
• Sensory acuity: Species with sharper senses will detect weaker signals regardless of coordination. The test must use signals calibrated to each species' sensory threshold.
• Ecological relevance: The test signals must be ecologically meaningful for each species (food odors for foraging species, predator cues for prey species, etc.).
• Motivation: All species must be adequately motivated to perform the detection task.

ECI predicts: For a detection threshold theta, the probability of extreme performance P(H > theta | C_N) should increase with C_N -- not because high-C_N species are "smarter" in some general sense, but because their coordination architecture enables access to a broader region of signal space. The prediction is about the distribution, not just the mean: high-C_N species should show heavier distribution tails (more extreme performers relative to their own mean) because their coordination dynamics create more opportunities for amplifying weak signals through cooperative processing.

What success looks like: A statistically significant, preregistered relationship between independently measured C_N and detection performance distribution, after controlling for body size, sensory acuity, and ecological relevance. Replicated across at least two independent research groups using overlapping but not identical species sets.

What failure looks like: If detection performance is fully predicted by sensory acuity and body size with no residual contribution from C_N, the coordination-enhances-capability prediction is falsified. If the V-kappa framework adds no predictive power beyond simpler comparative measures (e.g., brain-to-body mass ratio, encephalization quotient), it fails to justify its added complexity.

Estimated timeline: 5-8 years from completion of Stage 4. Requires large-scale comparative behavioral testing across multiple species, with standardized protocols and multiple independent laboratories.

Key variables tested: C_N (coordination complexity), V, kappa, distribution shape (tail behavior).

Stage 6 -- Future-target experiment (precognition test)

What it tests: Whether systems can access information that is not yet causally determined. This is the test described in detail on E2.

What it requires: Successful Stages 1-5, AND replicable anomalies in earlier stages. Specifically, this stage is ONLY justified if:

1. The V-kappa framework has been validated (Stages 1-2).
2. Living systems have been shown to exhibit additional capabilities beyond matched nonliving systems (Stage 3, Outcome B).
3. Coordination complexity has been shown to enhance collective capabilities (Stage 4).
4. Cross-species comparisons show the predicted C_N-dependent distribution shift (Stage 5).
5. Some anomalous results in earlier stages resist conventional explanation -- results that are replicable but not accounted for by established sensory, statistical, or physical mechanisms.

If all five conditions are not met, Stage 6 is premature regardless of anyone's enthusiasm for testing precognition. The gating is strict precisely because the claim is extraordinary.

The protocol (summarized from E2):

1. Quantum random number generator (QRNG) produces the target AFTER the subject's response is locked.
2. Immutable, cryptographic timestamping of both the response and the target.
3. Physical isolation between subject and QRNG.
4. Automated, experimenter-blind operation -- no human involvement between response and target generation.
5. Mandatory separation of discovery and validation phases -- subjects identified in the discovery phase must be retested in an independent validation phase with a preregistered protocol.
6. Bayesian analysis with appropriately skeptical priors -- given the extraordinary nature of the claim, the prior probability of the hypothesis should be very low, and the evidence must be correspondingly strong (e.g., BF > 100 favoring the precognition hypothesis over the null).

What success looks like: Above-chance performance in the validation phase (not the discovery phase), with effect size confidence intervals excluding zero, Bayes factors strongly favoring the alternative hypothesis, and independent replication by at least two other laboratories using different QRNG hardware and different experimenter teams. All data publicly available.

What failure looks like: Performance indistinguishable from chance in well-powered, preregistered validation-phase tests. This is the expected outcome given current evidence (see E2 Section 3 on the parapsychology replication record). A null result at Stage 6 does NOT invalidate Stages 0-5 -- the core framework's value is independent of whether precognition exists. But it does close the door on Stages 7 and 8.

What a null result means for the framework: ECI explicitly allows for the possibility that Stage 6 returns null results. The framework's speculative frontier (Cluster E) is designed to be prunable. If precognition does not occur at detectable levels, the framework prunes the temporal-access edge and continues with its validated core. A framework that loses edges when evidence demands it is doing science. A framework that cannot lose edges is doing theology.

Estimated timeline: Only after earlier stages are complete. Realistically 8-15 years from the start of Stage 0, if all earlier stages succeed.

Key variables tested: Temporal access (yes/no), H_i (individual detection performance), C_N (coordination complexity), QRNG-based randomization.

Stage 7 -- Human extreme phenotype

What it tests: Whether individuals with measurably high coordination metrics (H_i, established through behavioral testing) show distinctive sensory, developmental, metabolic, cognitive, or fitness profiles.

What it requires: Successful Stage 6 -- replicable, preregistered evidence of above-chance performance on the future-target experiment. If Stage 6 yields null results, Stage 7 is not justified.

The critical methodological constraint: Stage 7 must proceed in the correct order. The WRONG order is: start from a population with unusual characteristics (e.g., blind individuals, people with neurological differences, meditators) and test whether they show above-chance performance on the precognition task. This is backwards because it imports cultural assumptions about "who should be gifted" and creates a massive multiple-comparisons problem (testing many groups until one shows a positive result).

The RIGHT order is:

1. Establish H_i through behavioral testing FIRST. Using the Stage 6 protocol, test a large, unselected population. Identify individuals whose validation-phase performance is replicably above chance. These individuals have high H_i as established by the only standard that matters: behavioral evidence.

2. THEN characterize the phenotype. For individuals with confirmed high H_i, systematically measure:

   • Sensory profile: Standard psychophysical testing across modalities (visual, auditory, tactile, olfactory, proprioceptive). Are high-H_i individuals sensorily typical or atypical?
   • Developmental phenotype: Developmental history, including any sensory deprivation, neurological events, or unusual developmental trajectories.
   • Metabolic traits: Resting metabolic rate, brain glucose consumption (via PET or similar), sleep architecture, circadian rhythm characteristics.
   • Cognitive measures: Standard cognitive batteries (IQ, working memory, attention, executive function), plus domain-specific measures (pattern recognition, statistical learning, temporal processing).
   • Coordination metrics: V and kappa measured from neural data (EEG, fMRI) using the protocols developed in Stage 0. Does neural coordination complexity predict H_i?
   • Fitness proxies: General health, stress reactivity, social connectedness, reproductive fitness (if applicable and ethically appropriate).

3. Compare phenotypic profiles of high-H_i versus matched controls (same age, sex, education, socioeconomic status, but with Stage 6 performance at chance levels).

Why the order matters: If you start from "blind people" and test for precognition, you are testing a cultural hypothesis (the blind seer archetype), not a scientific one. If blind individuals score at chance on the precognition task, you have learned nothing about the relationship between sensory profile and H_i. If they score above chance, you cannot distinguish the precognition effect from enhanced ordinary perception (cross-modal plasticity), which is a well-established conventional phenomenon (see E1 Section 2). Starting from behavioral performance and working toward phenotypic characterization avoids both problems.

What success looks like: A replicable phenotypic signature associated with high H_i -- a pattern of sensory, cognitive, metabolic, or neural characteristics that distinguishes high-H_i individuals from matched controls. The signature must be preregistered and survive independent replication.

What failure looks like: No detectable phenotypic differences between high-H_i and matched control individuals. This would mean that whatever underlies above-chance performance (if Stage 6 succeeded) is not reflected in any measured phenotypic variable -- a puzzling but honest outcome. Alternatively, if no individuals show replicable high H_i in the Stage 7 population (i.e., the Stage 6 results do not generalize), then Stage 6's conclusions must be revisited.

Estimated timeline: Only after Stage 6 succeeds. Realistically 10-20 years from the start of Stage 0, and only if the entire roadmap has produced positive results at every prior stage.

Key variables tested: H_i (validated behavioral measure), sensory profile, neural V and kappa, phenotypic characterization.

Stage 8 -- Quantum observer

What it tests: Whether the ECI framework produces a prediction that is explicitly different from standard quantum mechanics regarding the relationship between observer complexity and measurement outcomes.

What it requires: Successful Stages 1-7, AND a specific, quantitative ECI prediction that differs from standard QM. Currently, no such prediction exists.

The current situation: Standard quantum mechanics predicts that measurement outcomes depend on the quantum state and the measurement apparatus, not on who (or what) is doing the measuring (see Eq. 13, the observer null hypothesis, from F2). A photon does not care whether it is detected by a simple photodiode or a human retina connected to a Nobel laureate's brain. The Born rule gives P(outcome) = |<psi|phi>|^2 regardless of observer complexity.

ECI's speculative frontier allows for the possibility that observer complexity might affect measurement statistics (the Observer Experience page, D3), but it does NOT currently make a specific prediction that differs from standard QM. The observer null hypothesis (H_0: measurement statistics are observer-independent) is listed as an established prediction of standard QM on F2, and ECI does not contest it.

What would unlock Stage 8: A formal derivation, from ECI's framework, of a specific, quantitative prediction about observer-dependent measurement statistics that (a) differs from the Born rule prediction, (b) specifies the conditions under which the deviation would be detectable, (c) specifies the magnitude of the deviation, and (d) is testable with current or near-future experimental technology.

Without such a derivation, Stage 8 has no content. You cannot test an unspecified prediction. The appropriate scientific response to "maybe observer complexity affects quantum measurements" is: derive the prediction, specify the test, then run the test. Until the first step is completed, the second and third are impossible.

What success would look like: A preregistered experiment demonstrating that measurement outcome distributions systematically depend on observer complexity when physical setups are controlled to be identical -- the rejection of H_0 in Eq. 13. This would be among the most extraordinary experimental results in the history of physics, comparable in significance to the discovery of quantum mechanics itself. The evidential bar would be correspondingly extreme: multiple independent laboratories, multiple measurement types, adversarial collaboration, and Bayes factors in the thousands or higher.

What failure looks like: Confirmation of H_0 -- measurement statistics are observer-independent, as standard QM predicts. This is by far the most likely outcome and is scientifically valuable: it prunes the observer-complexity-affects-physics edge of the ECI network, constraining the framework's speculative frontier and strengthening its connection to established physics.

Estimated timeline: Indeterminate. Stage 8 currently has no specified prediction and therefore no specified test. It is listed for completeness and to make explicit that the ECI framework does not currently contest standard quantum mechanics. If a specific prediction is eventually derived, the timeline would depend on the required experimental precision.

Key variables tested: Observer complexity (operationalized as C_N or similar), measurement outcome statistics, Born rule predictions.

5 If the roadmap succeeds: progressive support

If the roadmap produces positive results at each stage, the ECI framework gains progressively stronger support -- but the nature of that support changes qualitatively as the stages advance.

Stages 0-2: Proof of concept. Success here means the V-kappa framework is a useful analytical tool for complex systems. It makes correct predictions about weak-signal detection, emergence, and persistence in artificial and well-characterized physical systems. This is the minimum viable product. Even if no later stage succeeds, a validated V-kappa framework with operational coarse-graining tools would be a genuine contribution to complex systems science.

Stages 3-4: Biological relevance. Success here means the framework captures something real about living systems -- either that living and nonliving systems obey the same V-kappa dynamics (substrate independence) or that living systems show additional coordination features that the framework can characterize (biological specificity). Either outcome extends the framework's scope significantly.

Stage 5: Comparative prediction. Success here means the framework can make and confirm predictions across species -- a strong test because comparative prediction requires that the framework's variables (especially C_N) are measuring something real about biological coordination, not just fitting within-system data.

Stage 6: Extraordinary extension. Success here -- genuine replicable precognition -- would be transformative for both the framework and for science more broadly. But note: the framework's value at Stages 0-5 is independent of Stage 6. A null result at Stage 6 does not retroactively invalidate the core framework. It prunes the speculative frontier, which is exactly what a healthy theory does.

Stages 7-8: Deep implications. Success at these stages would have implications far beyond the ECI framework, touching foundational questions about the nature of consciousness, time, and physical law. But these stages are so far from current evidence that speculating about their implications is premature. The honest assessment: these stages may never be reached, and that is a perfectly acceptable outcome.

The key insight is that the roadmap is designed so that partial success is valuable. If the framework validates at Stages 0-3 and fails at Stage 4, it is still a useful tool for analyzing complex systems. If it validates through Stage 5 and fails at Stage 6, it is still a powerful comparative framework. The gating structure means that failures are informative, not catastrophic -- each failure tells you exactly where the framework's predictive power ends.

6 Stage-specific experimental design sketches

This section provides more concrete experimental design outlines for the first few stages, where experiments can be specified with reasonable precision. Later stages are left as frameworks rather than detailed designs, because their specifics depend on results from earlier stages.

Stage 0 sketch: Operationalizing V in a coupled oscillator array

System: An array of N = 50 coupled electronic oscillators (e.g., Wien bridge oscillators with adjustable natural frequencies and coupling resistors).

V operationalization: V = Shannon entropy of the distribution of natural frequencies across oscillators. H = -Sigma_i (p_i log p_i), where p_i is the proportion of oscillators with natural frequency in bin i. When all oscillators have the same frequency, V = 0 (minimum). When frequencies are uniformly distributed across the full range, V = log(n_bins) (maximum).

Manipulation: V is controlled by adjusting the component values that set each oscillator's natural frequency. Low V = all oscillators tuned to near-identical frequencies. High V = frequencies uniformly spread across a wide range.

kappa operationalization: kappa = mean phase coherence = (1/N^2) Sigma_{i,j} |< e^{i(theta_i(t) - theta_j(t))} >_t|, where theta_i(t) is the phase of oscillator i at time t and <...>_t denotes time averaging. When all oscillators are phase-locked, kappa = 1 (maximum). When phases are random, kappa approaches 1/sqrt(N) (minimum for finite N).

Manipulation: kappa is controlled by adjusting coupling resistor values. Low kappa = weak coupling (oscillators nearly independent). High kappa = strong coupling (oscillators tend to synchronize).

Reliability assessment: Measure V and kappa ten times for the same physical configuration. Report the coefficient of variation. Acceptable reliability: CV < 0.05 for both measures.

Stage 1 sketch: V-kappa interaction predicts weak-signal detection

System: Same oscillator array as Stage 0.

Task: Embed a weak periodic signal (amplitude well below the oscillators' noise floor) in the common drive to all oscillators. The signal has known frequency f_s and amplitude A_s. After T seconds of operation, apply a matched filter to the collective output (sum of all oscillator voltages) to estimate whether the signal is present.

Design: Full factorial: V at 5 levels (very low, low, medium, high, very high) x kappa at 5 levels (very low, low, medium, high, very high) = 25 conditions. 100 trials per condition (50 signal-present, 50 signal-absent, randomized). Preregistered primary analysis: 5x5 ANOVA on detection accuracy (d-prime), testing the V x kappa interaction.

Power analysis: Based on pilot data or reasonable estimates of effect size. If the expected interaction effect is eta-squared = 0.06 (medium), then N = 25 conditions x 100 trials = 2500 total trials provides power > 0.99 for the interaction test at alpha = 0.05.

Preregistered predictions:

1. Main effect of V: significant, non-monotonic (inverted U).
2. Main effect of kappa: significant, non-monotonic (inverted U).
3. V x kappa interaction: significant, with the peak in the high-V, high-kappa quadrant.
4. Comparison to stochastic resonance baseline: d-prime in the optimal V-kappa condition exceeds d-prime predicted by a standard stochastic resonance model that uses only noise intensity as the control variable.

Stage 4 sketch: Ant colony coordination manipulation

System: Workers from a single colony of Lasius niger (black garden ant), a well-studied model species.

Task: Locate a food source (a small drop of dilute sucrose solution) in a foraging arena with multiple confounding odor sources (non-nutritive chemicals placed at random locations). The food signal is weak: the sucrose concentration is diluted to near the species' detection threshold.

Design: Three conditions (A, B, C as described in Stage 4 above), counterbalanced across trials. N = 100 ants per condition, drawn from the same colony. Each condition is tested for 20 independent trials with fresh ants (to prevent learning confounds across trials). The arena configuration (food location, confound locations) is randomized for each trial.

Measures:

• Time to first contact with food source (minutes).
• Proportion of ants that locate food within 60 minutes.
• Collective accuracy: proportion of foraging effort directed at the food source versus confounds.
• Information spread rate (Condition B and C only): time from first discovery to 50% of colony awareness (measured by directional foraging).

Preregistered predictions:

1. Condition C (full colony) locates food faster and more reliably than Condition A (isolated), with p < 0.01.
2. Condition B (limited communication) is intermediate.
3. The collective accuracy measure in Condition C exceeds the sum of individual accuracies in Condition A -- i.e., the collective is more than the sum of its parts.

7 Connected Nodes

-> Applications & Future (F1): F1 describes what the ECI framework could enable if validated. F4 specifies the validation sequence. The three-tier application structure in F1 maps directly onto the roadmap stages: Tier 1 applications (established-science extensions) correspond to Stages 0-2; Tier 2 applications (conditional on core framework validation) correspond to Stages 3-5; Tier 3 applications (conditional on Cluster E validation) correspond to Stages 6-8. An application is only scientifically defensible if its corresponding roadmap stages have been passed.

-> Falsifiability (F3): F3 establishes the standards by which ECI claims should be evaluated -- the five traps, the edge-by-edge falsification table, the evidence hierarchy. F4 operationalizes those standards into a concrete experimental sequence. Where F3 says "if V does not predict weak-signal coverage, prune that edge," F4 specifies exactly how to test V's prediction: which system, which protocol, which statistical test, what effect size constitutes "prediction." F3 provides the philosophy of testing; F4 provides the engineering of testing.

8 Summary table: all nine stages (0-8)

| Stage | Name | What it tests | Key variables | Dependencies | What success looks like | What failure means | Status | |---|---|---|---|---|---|---|---| | 0 | Operationalization | Can V, kappa, K, Q, Gamma be measured? | V, kappa, K, Q, Gamma | Formal definitions (F2) | Published, reliable measurement protocols in >= 2 systems | Framework is not yet empirically testable | Not started | | 1 | Conventional validation | V, kappa → weak-signal response | V, kappa | Stage 0 | V x kappa interaction predicts detection better than baselines, replicated | Core prediction fails; framework loses empirical foundation | Not started | | 2 | Simulation / artificial systems | ECI predictions hold in ABMs, oscillator networks, RNNs | V, kappa, G, S decomposition | Stage 1 | Preregistered predictions confirmed in >= 3 artificial system types | Framework adds no value beyond standard complex-systems models | Not started | | 3 | Living vs. nonliving matched | Do living systems show additional coordination features? | V, kappa, K, Q, Gamma | Stages 1-2 | Either substrate independence confirmed OR biological specificity characterized | Matching fails or results are inconclusive | Not started | | 4 | Within-species collective | Does coordination (not just N) enhance collective capability? | V, kappa, collective K, G | Stages 1-3 | Coordination complexity predicts collective performance, N held constant | Coordination adds no collective value beyond aggregation | Not started | | 5 | Cross-species comparison | Does C_N predict distribution tails across species? | C_N, V, kappa, distribution shape | Stage 4 | P(H > theta | C_N) increases with C_N after controlling for body size and sensory acuity | C_N adds no predictive power beyond simpler comparative measures | Not started | | 6 | Future-target experiment | Can future-undetermined information be accessed? | QRNG target, H_i, C_N | Stages 1-5 + replicable anomalies | Above-chance validation-phase performance, BF > 100, independent replication x2 | Null result; speculative frontier pruned; Stages 7-8 not justified | Not started | | 7 | Human extreme phenotype | What characterizes high-H_i individuals? | H_i, sensory profile, neural V/kappa, phenotype | Stage 6 | Replicable phenotypic signature for high H_i | No detectable phenotypic differences; mechanism remains unknown | Not started | | 8 | Quantum observer | Does observer complexity affect measurement statistics? | Observer C_N, Born rule predictions | Stages 1-7 + specific QM prediction from ECI | H_0 rejected: observer-dependent measurement statistics | H_0 confirmed: standard QM holds; observer edge pruned | Not started; no prediction currently exists |

Reading the table: Each row's "Dependencies" column shows which earlier stages must pass before this stage is justified. The "Status" column reflects the current state of the research program: as of this writing, no stage has been initiated. The entire roadmap is prospective -- a plan, not a report.

The most important column is "What failure means." A framework that cannot specify in advance what would count as failure is not making scientific claims. ECI can. At every stage, failure is defined, informative, and consequential.