Consciousness during cardiac arrest

Clinical evidence, synthesis, and implications for aerospace human factors alongside proposed experimental test programme development

Abstract

This study critically evaluates prospective research into conscious experience reported after cardiac arrest, with particular attention to the distinction between experience, memory, event corroboration and experimentally verified information acquisition. Evidence published by 9 December 2015 supports the occurrence of structured, sometimes enduring experiences among selected survivors, but does not establish experience after irreversible brain death. The first AWARE investigation provides the central clinical case, considered alongside earlier prospective cohorts, studies of covert awareness and physiological accounts of altered experience. A causal measurement model explains why reporting frequency cannot independently identify the prevalence or timing of experience. Three testable hypotheses address memory preservation, temporally restricted information acquisition and the lag between recovery of responsiveness and recovery of reliable control. A proposed multicentre programme combines concealed targets, synchronised physiological records, independent adjudication and explicit falsification criteria. Separate defence and aerospace sections examine casualty awareness, aviation incapacitation, adaptive interfaces, staged handover and functional machine architectures. These applications draw on independently supported neuroscience and human-factors research; they do not depend on accepting a survival interpretation.

Contents

Research orientation

The empirical evidence and competing explanations are evaluated first. Proposed experiments follow. Defence and aerospace applications are then assessed as independently testable research transfers.

1 Research question and intellectual scope

The question examined here is whether conscious experience and acquisition of verifiable information can be demonstrated during cardiac arrest under physiological conditions that challenge ordinary accounts of perception and memory. This question is narrower than whether personal consciousness survives permanent bodily death. A survivor can provide testimony about an episode of reversible circulatory failure; that testimony alone cannot establish experience following irreversible destruction of the brain. Nevertheless, a rigorous demonstration of information acquisition under strongly constrained conditions would materially affect theories of mind and justify further investigation of continuity beyond ordinary physiological boundaries.

Cardiac arrest is scientifically valuable because it generates abrupt changes in perfusion, responsiveness and recovery. It is also unusually difficult to study. Investigators cannot ethically manipulate the dangerous event, clinical care takes priority, measurements are incomplete and only a selected fraction of patients survive and can participate. The resulting evidence has a characteristic asymmetry: the report may be phenomenologically rich while the physiological record is temporally coarse. The apparent conflict between vivid experience and impaired brain function therefore needs to be investigated at the level of measurement and timing, before a metaphysical conclusion is drawn.

The case is developed around prospective studies rather than celebrated individual narratives. Prospective recruitment makes the denominator visible, permits comparison with non-reporting survivors and reduces selection based on extraordinary testimony. It does not automatically solve recall bias, establish cerebral inactivity or exclude information acquired during recovery. The strongest research strategy consequently combines clinical observation with experimental control over information that can be checked.

The aerospace connection is substantive. Flight operations depend on distinguishing an apparently awake operator from an operator who can detect an alert, integrate its meaning and respond appropriately. Acceleration-related loss of consciousness, attentional failure and inappropriate reliance on automation make that distinction operationally consequential. Consciousness research can improve these systems by refining the variables being measured. A claim that consciousness exists independently of the brain is not required for that engineering contribution.

2 Evidence selection and analytical method

This is a structured critical synthesis and a proposed research programme, not a new clinical experiment or an exhaustive systematic review. Searches addressed cardiac arrest and near-death experience; awareness during resuscitation; memory and reported experience; physiological models; covert awareness; consciousness theory; acceleration-induced incapacitation; cockpit alerting; and adaptive automation. Records were checked through PubMed, journal and publisher pages, original papers and author or institutional repositories. Primary clinical and experimental studies receive greatest evidential weight. Scientific reviews and theoretical papers support conceptual interpretation; they do not substitute for independent observations.

Eligibility is determined by the publication date of the cited work, with an absolute cutoff of 9 December 2015. A later repository upload does not change an original article's publication date. Later replications, revised diagnostic guidance and subsequent theory are outside this study. The historical clinical guidance cited below is used to clarify concepts within this evidence window, not to provide a contemporary clinical protocol.

The appraisal separates recruitment, ascertainment, outcome measurement, temporal resolution, information control and external validity. Positive findings are not treated as interchangeable. A reliable questionnaire establishes consistency of measurement; a prospective cohort establishes a frequency within its recruited population; corroborated testimony establishes correspondence with events; a concealed-target result would address access to information. Each supports a different inference. Similarly, a null target result is informative only when the target was actually presented, the exposure window is known and the person could complete the endpoint.

The source record was checked for publication metadata, denominators and the distinction between abstracts and main results. Where only an indexed abstract or publisher excerpt could be checked, assessment is confined to what that record establishes. No patient-level dataset was available for reanalysis. Numerical checks therefore use published aggregate counts and are labelled as descriptive calculations. No pooled prevalence is calculated because definitions, selection and interview procedures differ materially across cohorts.

The synthesis uses an explicit evidential ladder. The lowest step is a report of an unusual experience. Higher steps require stable measurement, event corroboration, concealed-target discrimination, temporal localisation and independent replication. Evidence for one step cannot silently be promoted to the next. This approach also guards against the opposite error: treating an unresolved explanation as grounds to disregard an experience or deny its significance to the person reporting it.

3 Definitions and the problem of observation

Experience connectedness and report

Phenomenal consciousness means that there is something it is like to undergo an experience. Connected consciousness means experience involving the external environment. Responsiveness is observable behaviour. Reportability additionally requires comprehension, memory, language and an opportunity to communicate. These dimensions can dissociate. Sanders et al. (2012) distinguish consciousness, connectedness and responsiveness in anaesthesia; their analysis prevents a behavioural definition of unconsciousness from being mistaken for direct access to subjective experience.

Evidence from severe brain injury reinforces the distinction. Owen et al. (2006) demonstrated task-related brain responses consistent with command following in a behaviourally unresponsive patient. Monti et al. (2010) subsequently identified willful modulation of brain activity in five of 54 patients studied. These observations concern living, functioning neural systems. They establish that bedside behaviour can underestimate awareness, not that awareness occurs without neural activity. Their relevance to cardiac arrest is methodological: a negative motor response is an incomplete measurement.

The Greyson scale operationalises features commonly associated with near-death experiences through a 16-item instrument covering cognitive, affective, paranormal and transcendental dimensions (Greyson, 1983). Its validity as a phenomenological measure does not establish the external truth of every item. A high score for altered time or apparent separation from the body records an aspect of experience; it does not locate that experience in the resuscitation timeline or demonstrate perception from an extracorporeal vantage point.

Circulatory arrest and irreversible brain death

Cardiac arrest, absence of a detectable pulse, behavioural unresponsiveness and irreversible cessation of brain function are different propositions. Resuscitation is undertaken precisely because some consequences of circulatory failure may be reversible. Historical neurological criteria for death required a defined clinical assessment rather than inference from a short episode of absent circulation (Wijdicks et al., 2010). The participants in survivor studies were not demonstrated to have recovered from correctly established irreversible brain death.

Equally, a low-amplitude or absent surface EEG signal has a measurement-specific meaning. Electrode coverage, sensitivity, interference, drugs and the relationship between scalp potentials and deeper neural activity all matter. A report of no detectable activity under a particular measurement cannot be expanded into proof that every relevant neural process was absent. Conversely, detecting electrical activity does not establish organised subjective experience. This symmetry is essential when interpreting both suppression and terminal surges.

Survival as a stronger claim

Personal survival requires more than the persistence of information about a person. It implies continuity of a subject or person across bodily death. Near-death experience research primarily addresses the boundary conditions of experience during a recoverable crisis. An experimentally secured anomalous perception would be important, but additional argument would still be needed to connect it to enduring personal identity after death. The present study keeps those questions distinct while retaining the possibility that unexpectedly strong results could change their relative plausibility.

4 Clinical case study and comparative evidence

Prospective cohort findings

The principal quantitative evidence is summarised below. The intervals are descriptive Wilson 95% intervals calculated from the published counts, not estimates adjusted for hospital clustering or missing survivors.

Table 1 Comparative evidence

The Dutch study found no association between NDE occurrence and several measured clinical variables, and documented changes over subsequent years. This challenges simple accounts that predict reports directly from those variables. It does not exclude physiological explanations generally: a mechanism may depend on an unmeasured transition, individual susceptibility or preservation of memory rather than arrest duration alone (van Lommel et al., 2001). Failure to find an association in a heterogeneous observational sample is not equivalent to demonstrating independence from brain function.

The Slovenian study found associations between reported experience and carbon dioxide measures (Klemenc-Ketis, Kersnik and Grmec, 2010). Several causal interpretations remain possible. Carbon dioxide may influence experience; it may track aspects of perfusion or ventilation; or it may relate to the probability that an experience is retained and reported. The study therefore motivates measurement and replication rather than a single established biochemical explanation.

The US cohort further illustrates why apparently similar percentages require careful construction. Schwaninger et al. (2002) separated experiences associated with the monitored arrest from experiences recalled from previous crises. Combining those categories would inflate the estimate for the event under study. Together, the prospective cohorts justify investigation of a recurring phenomenon, while their different interview and inclusion rules prevent a simple pooled prevalence from answering the causal question.

The first AWARE investigation

AWARE is the pivotal case because it attempted to move from subjective description to independently checkable events. In the second-stage interview group, nine of 101 participants reported NDEs and two described explicit awareness of resuscitation events. One account was corroborated against clinical circumstances. Its timing was inferred partly from an automated defibrillator sequence. Neither of the two relevant arrests occurred in a location with a concealed shelf target, so neither supplied a positive hidden-image test (Parnia et al., 2014).

This distinction preserves the case's significance while specifying its limits. Correspondence between a narrative and resuscitation events is stronger than an entirely private account. However, the study did not provide a contemporaneous patient-specific recording establishing complete cerebral inactivity throughout the proposed experience. Corroboration therefore leaves unresolved the route and exact time of information acquisition. It must not be described as a concealed-target success or proof of consciousness after irreversible death.

The design illustrates an exposure problem. Installing targets is not sufficient; the informative event must happen in a covered location. It also illustrates a selection problem. The arrest denominator, discharge survivors, interview participants and usable verification cases answer different questions. A percentage calculated within interviewed survivors cannot be presented as a prevalence among every person experiencing cardiac arrest.

The evidential meaning of memory quality

Experiences may remain vivid and stable without their external interpretation being settled. Thonnard et al. (2013) found distinctive richness in NDE memories compared with memories of real and imagined events. The study assessed memory characteristics, not independently verified external occurrence. Greyson (2007) found that accounts were not progressively embellished over almost two decades. Stability weakens a simple claim of continuous narrative inflation, but a stable memory can still contain a mistaken attribution of timing or source.

Three dimensions should consequently be measured separately: phenomenological intensity, autobiographical stability and external accuracy. Combining them into a single evidential score would conceal precisely the distinctions an experiment needs to resolve. A moving, consistent account may have no externally testable content; a brief, emotionally neutral observation may contain a decisive concealed target. Research selection should not reward narrative drama over discriminating information.

5 Competing explanations and causal identification

A latent experience is observed through several filters

Let experience be an unobserved event and the subsequent report an imperfect indicator. Among survivors able to participate, the observed report rate depends on the prevalence of the experience, the probability of retaining and reporting it, and the probability of reporting an apparent experience without the hypothesised event. This is a measurement model, not a claim that any participant fabricated an account.

r = aπ + b(1 - π)

Here r is the observed reporting probability, π the probability of experience in the specified physiological window, a the sensitivity of later reporting and b the probability of a report attributed to that window despite its originating elsewhere, all conditional on survival and participation. One observed proportion cannot identify all three quantities. A treatment that improves memory could increase reports even if it leaves the underlying experience unchanged. A sedating exposure could suppress later reporting despite intense experience. These possibilities show why a non-reporting control group cannot automatically be classified as a group with no experience.

Selection also matters causally. Survival, ability to consent and interview completion depend on injury and recovery. If these variables also affect memory, conditioning on them can distort an association between physiology and reported experience. Adjusting only for age and sex does not solve that problem. An adequate analysis needs an explicit account of who becomes observable, measurements of interview eligibility and sensitivity analyses for unobserved reports.

Physiological transitions and residual processing

Borjigin et al. (2013) observed a transient increase in coordinated high-frequency neural activity after experimentally induced cardiac arrest in rats. This demonstrates that the transition toward physiological collapse need not be a monotonic decline in every electrical measure. It does not establish what the animals experienced or validate a direct equivalence with human NDE content. The study is useful because it makes a temporally restricted neural hypothesis testable: some organised activity may occur during a brief transition that routine clinical observation fails to resolve.

Chawla et al. (2009) reported transient increases in monitored electrical activity around withdrawal of life support in a small human case series. The observations did not include later first-person reports and cannot connect a measured surge to a particular conscious episode. Processed indices, sparse raw recordings and the setting of dying patients require particular care. Taken together, these investigations justify better measurement of transitions, rather than either declaring the brain silent from the instant of arrest or equating every surge with awareness.

A residual-processing account has several variants. Information may be acquired just before collapse, during partial perfusion maintained by resuscitation, during an early recovery window or through subsequent reconstruction. These are distinct hypotheses with different predicted temporal signatures. Combining them into an unrestricted explanation would make the account difficult to falsify. A strong study must specify which periods and information channels each model permits before outcomes are known.

Body representation and arousal

Electrical stimulation near the temporoparietal junction produced altered own-body perceptions in a neurological patient (Blanke et al., 2002). This supports a causal contribution of neural mechanisms to aspects of bodily self-location. It does not establish that every reported out-of-body experience has the same cause or explain accurate information inaccessible through ordinary sensory pathways. The explanatory achievement is specific: some core phenomenological components can arise through disturbed multisensory integration.

Nelson et al. (2006) investigated an association between NDE reports and intrusion of rapid-eye-movement sleep phenomena into waking life. A predisposition involving arousal transitions could help explain susceptibility. The design does not show that every NDE is a REM event, and retrospective susceptibility measures do not locate the experience during an arrest. An integrated physiological account must explain which components arise, why individuals differ and how memory survives, rather than invoke a single label as a complete mechanism.

The strongest challenge to an ordinary access explanation

The most discriminating observation would be accurate identification of a concealed, randomly selected stimulus presented only during a verified physiological interval, with no ordinary route of access and with success replicated under independent study. This would directly address information acquisition. It would be stronger than identifying common resuscitation sounds or familiar equipment, because predictable features have high prior probability.

Even this result would need scrutiny of randomisation, target exposure, clock alignment, accidental reflections, staff knowledge, reconstruction and analysis flexibility. Such scrutiny is not a special burden imposed on an unpopular hypothesis. The same standards apply to a surprising diagnostic test or a flight-critical sensor. An ordinary explanation that relies on unspecified leakage forever is as scientifically unproductive as an extraordinary explanation that accommodates every possible outcome.

6 Consciousness theory and model discrimination

Global access and integrated causal organisation

The global neuronal workspace account links conscious access to information becoming available across distributed systems for flexible use and report (Dehaene and Changeux, 2011). In the present problem, it encourages a distinction between local processing, global access and later recollection. A fragment of sensory processing need not imply a sustained conscious episode; failure to report need not imply the absence of all processing. A temporally adequate experiment should measure dissemination and flexible task use where feasible, rather than rely on raw amplitude alone.

Integrated information theory proposes that experience depends on irreducible causal organisation (Oizumi, Albantakis and Tononi, 2014). It offers a different conceptual emphasis, but neither a high-frequency EEG burst nor an ordinary entropy estimate is a direct measurement of the theory's formal quantity. Substituting a convenient complexity metric for a demanding causal construct would create an appearance of precision without the corresponding inference.

Casali et al. (2013) developed a perturbational complexity index using brain responses to transcranial magnetic stimulation. Its relevance is the methodological move from passive signal magnitude toward differentiated, integrated response. It is not validated by that study as a real-time cardiac-arrest measure or an aviation instrument. During resuscitation, emergency access, electrical interference and clinical priorities may make the original perturbational method impractical. Any alternative measure would require its own validation.

Neural dependence and continuity hypotheses

A neural-production model predicts that experience depends on a sufficient neural organisation. A transmission or filtering model can interpret the brain as constraining access to a wider consciousness. The observed dependency of ordinary cognition on brain state is compatible with many versions of both accounts; labels alone do not generate a discriminating experiment. The scientific question is what each model predicts about specific information under specified conditions.

A useful continuity hypothesis must therefore state its auxiliary assumptions. Does information access persist during absent responsiveness? Is access confined to a local environment? Can the subject encode an unpredictable visual target? What makes recall possible after recovery? If these commitments remain unspecified, a negative target result can always be attributed to lack of attention or failed memory. The model then becomes difficult to distinguish from competitors. Stating auxiliary assumptions openly strengthens the hypothesis by exposing it to a meaningful test.

The same discipline applies to neural explanations. A model cannot appeal simultaneously to every unmeasured interval while claiming that the timing has been explained. A credible account should predict the relationship among physiological state, stimulus availability, target accuracy and subsequent confidence. Model comparison requires prospective predictions with penalties for flexibility, not a retrospective contest over which narrative feels more plausible.

7 Proposed experimental programme

Three hypotheses

The following hypotheses are proposed contributions of this study. They are not established findings or claims of demonstrated historical priority.

H1 Recovery and memory preservation. The occurrence and richness of a later NDE report will depend partly on recovery of memory-supporting function, independently of the severity of the preceding circulatory disruption. Its distinctive prediction is that measures around recovery improve prediction of later reportability beyond arrest duration and baseline characteristics. Support would explain variation in reporting without, by itself, determining whether an earlier experience occurred. Failure of recovery measures to add replicable predictive value would count against this version of the hypothesis.

H2 Temporally restricted information acquisition. Under ordinary sensory access, correct target discrimination will be concentrated in windows with sufficient sensory availability and recoverable neural processing. A competing anomalous-access model predicts reproducible discrimination of strictly concealed visual targets assigned only to other prespecified intervals. These are operational competitors. Success for the latter would challenge the specified ordinary-access model, while remaining short of a demonstration of permanent personal survival.

H3 Recovery of control lags recovery of response. After a reversible impairment, the first behavioural response will precede reliable restoration of situational integration and safe control in at least a subset of individuals. The proposed contribution is a joint measurement programme across clinical recovery and safe aerospace simulation, not a claim that all recovery delays are newly discovered. It predicts that a staged assessment will reduce premature handover relative to a single responsiveness criterion.

Clinical design and target custody

A prospective multicentre study would enrol arrest events through an event registry, preserving counts for every eligibility and follow-up stage. Research equipment must be passive with respect to treatment: no delay to resuscitation, no alteration of ventilation and no prolongation of physiological compromise. A target unit should already be installed or incorporated into equipment brought routinely to the event, with access and electrical safety assessed before recruitment.

The unit would generate and display unpredictable stimuli with synchronised timestamps. Visual targets would face a direction inaccessible from the patient's ordinary viewpoint and would be tested for reflections, camera capture and accidental staff visibility. Auditory targets would serve a different purpose: assessing sensory acquisition under potentially preserved hearing. They must not be interpreted as evidence of non-sensory perception. Multiple modalities allow ordinary processing and concealed access to be examined without conflating them.

Target sequences, randomisation seeds and presentation logs would be sealed before outcome collection and held by a team separate from clinical interviewers. A cryptographic commitment could establish that a sequence was not changed after an account was obtained, but the physical display still requires independent verification. A perfectly sealed file does not show that a screen was operating, correctly oriented or invisible to staff. Each event therefore needs an exposure record, a hardware-quality record and an access log.

Whenever available through routine care, ECG, arterial pressure, resuscitation events, ventilation, administered drugs and neurological observations should share a common time base. Research EEG would need prespecified artifact rejection around chest compressions and defibrillation. A missing or uninterpretable channel is recorded as uncertainty, not converted into neural silence. Physiological windows should be assigned by adjudicators unaware of reported experience and target correctness.

The temporal test should use distinct target sequences across the arrest, resuscitation and recovery windows. A later response could then be compared against each sequence, creating a prespecified matrix of target time and reported time. Concentration of matches in the recovery column despite reports of earlier perception would support temporal misattribution. Reliable matches to a concealed sequence available only during the claimed interval would challenge that account. Every comparison, including lagged matches, must be declared in advance so that searching across many time windows cannot create an apparent success.

Interview and adjudication

An initial free narrative would be recorded before target-specific questioning or discussion of clinical details. The interviewer would then use the established phenomenological instrument and a fixed set of neutral prompts. The protocol must separate a spontaneous target description from a forced-choice selection. All eligible participants who can complete the task should receive the same target procedure, including those who report no NDE, so that the endpoint is not selected after learning the phenomenology.

A four-alternative forced-choice task provides a simple primary chance probability of one quarter when the alternatives and randomisation are genuinely exchangeable. The alternatives should be matched for salience, cultural familiarity and verbal describability. Confidence and response time are secondary outcomes. A participant selecting several alternatives, refusing or completing only part of the task must be handled according to rules specified before data collection. Narrative similarity should not become a substitute endpoint after exact discrimination fails.

Two adjudication teams would work independently. One reconstructs physiological timing without seeing the narrative. The other scores narrative specificity and correspondence without seeing the clinical state or target identity. A third custodian links the locked results. Disagreements should remain visible and be resolved by a declared rule. A conclusion about unusual access should require integrity of both the physiological and informational chains.

Statistical design and falsification

The primary analysis would test the prespecified target endpoint using the actual randomisation structure. Repeated observations from the same patient, targets shared within equipment or correlated hospital practices require cluster-aware analysis. Phenomenological scores, physiology and memory outcomes would have a separate prespecified model. Flexibility in inclusion, outcomes and stopping can produce misleading significance even without intentional misconduct (Simmons, Nelson and Simonsohn, 2011).

A planning calculation illustrates the scale required. For an independent four-choice task, detecting an increase from 25% to 35% correct with 90% power and a two-sided significance level of 0.005 requires approximately 334 evaluable participants under a normal approximation. This is an assumption-based calculation, not an estimated effect from the clinical literature. Clustering, discreteness, multiplicity and missing data require simulation and may increase the sample. If only 4% of registered events yield an evaluable interview and 80% of those have valid target exposure, approximately 10,440 events would be needed before additional allowances.

A null result would constrain a model only to the degree that its predicted effect and exposure assumptions were tested. A wide interval around chance would remain inconclusive. A well-powered interval excluding the minimum effect specified by the anomalous-access model would count against that model under the tested conditions. Conversely, a small p-value would justify replication and an integrity audit; it would not by itself assign the result to a particular ontology.

A parallel programme without life threatening exposure

Memory, attention and recovery hypotheses can also be studied in non-hazardous tasks, existing clinical datasets and appropriately supervised research settings. High-fidelity simulation can examine attentional overload, interruptions and handover without deliberately producing loss of consciousness. Archival acceleration data can inform the temporal structure of impairment. The clinical and aerospace branches should share definitions and analysis principles while retaining separate physiological models. A simulator's workload manipulation is not a substitute for cerebral hypoperfusion.

8 Defence adjacent implications

Casualty assessment and recovery

The defence-adjacent value begins with avoiding premature conclusions about awareness. Behaviourally unresponsive casualties may not provide a reliable negative signal of experience. Covert-awareness and anaesthesia research support respectful communication and careful distinction between observation and inference (Owen et al., 2006; Sanders et al., 2012). This is a research and human-factors implication, not a replacement for established trauma or neurological protocols.

A prospective recovery study could test whether memory of treatment, comprehension and task competence recover on different trajectories after a critical incident. Outcomes should include accuracy of later recall, ability to understand instructions and performance on non-hazardous tasks. Such work could improve communication during evacuation and subsequent investigation of an event. It should not use metaphysical beliefs or the occurrence of an NDE as a proxy for operational fitness.

Aviation incapacitation and return of control

Whinnery and Whinnery's (1990) analysis of 500 acceleration-induced loss-of-consciousness episodes is directly relevant to defence aviation. It distinguishes the loss of consciousness from subsequent incapacitation and records complex associated phenomena. The episodes do not establish equivalence with cardiac-arrest NDEs, but they demonstrate why recovery cannot be reduced to a single instant when the eyes open or movement resumes.

The proposed engineering implication is a staged restoration of responsibility. A recovered operator first establishes reliable responsiveness, then demonstrates access to relevant flight information, then resumes an appropriate level of control. The thresholds would need empirical validation against recovery outcomes. No novel monitoring signal should independently override a pilot or certify recovery merely because it correlates with an experimental condition.

Research governance

Neural and physiological records can reveal sensitive information about fatigue, stress and health. A defence research programme should separate voluntary research participation from routine personnel assessment, limit access to identifiable records and define in advance how an incidental finding is handled. These are proposed governance provisions for the research design. They support valid recruitment and reduce incentives to conceal symptoms or produce a socially preferred account.

9 Aerospace engineering neuroscience and interfaces

Measuring the state that matters

A flight-deck system should not ask only whether a pilot is conscious. The relevant questions include whether the pilot can hear a critical alert, integrate it with the aircraft's state, understand automation mode and execute an appropriate response. Dehais et al. (2014) demonstrated inattentional deafness to critical alerts in a simulated cockpit. The example shows that the presence of an alert and an apparently engaged operator does not guarantee conscious access to its meaning.

The proposed operator model therefore distinguishes arousal, access and action capability. Arousal is estimated from suitable physiological and behavioural measures. Access is tested through context-specific acknowledgements or task evidence. Action capability concerns correct, timely control. These variables are related but should not be collapsed into a single opaque readiness score. A score can conceal a serious failure in one component behind strong performance in another.

Passive monitoring and adaptive interfaces

Pope, Bogart and Bartolome (1995) investigated EEG-derived engagement measures in an adaptive automation context. Zander and Kothe (2011) subsequently described passive brain-computer interfaces that use neural information without requiring an explicit control command. These provide legitimate foundations for studying interfaces responsive to cognitive state. They do not validate a universal consciousness detector or establish that one EEG ratio remains reliable during vibration, acceleration, hypoxia and emergency manoeuvres.

A candidate system would combine eye behaviour, task performance, response timing and carefully qualified physiological measures. Each input should have its own quality estimate. When an electrode loses contact, the system should become less certain, rather than interpret artifact as a sudden change in the operator. Validation must hold out entire pilots and sessions; random splitting of adjacent signal windows would overstate generalisation because neighbouring observations share person-specific and temporal structure.

Adaptation should initially affect information presentation rather than flight authority. Possible interventions include reducing nonessential display content, coordinating redundant alerts and presenting a brief explanation of automation state. Benefits should be measured through error recovery, missed-alert rates and response appropriateness, with nuisance alerts recorded as harms. An interface that detects workload accurately but increases distraction has failed its operational purpose.

Control allocation and recovery hysteresis

Parasuraman, Sheridan and Wickens (2000) distinguish automation of information acquisition, analysis, decision selection and action implementation. This decomposition allows a consciousness-informed system to support an impaired component without assuming that every function should become more automated. The proposed design may simplify information acquisition while preserving the pilot's authority, or provide temporary stabilising assistance under an independently established safety envelope.

Recovery hysteresis means that the evidence needed to return control can differ from the evidence that triggered assistance. It avoids rapid switching when a noisy physiological measure fluctuates around a threshold. The concept becomes particularly useful when responsiveness returns before sustained comprehension. The hypothesis is tested by comparing premature handovers, unnecessary assistance, recovery time and pilot understanding across fixed and staged policies.

Trust must also be calibrated. Lee and See (2004) show why reliance depends on the relationship between system capability, context and the operator's understanding. A monitoring interface should express uncertainty and its basis in terms the pilot can use. A message that requests confirmation of a missed checklist item is more actionable than an unsupported declaration that the pilot is unconscious. Transparent, task-specific feedback also makes false alarms easier to recognise and correct.

Commercial aviation and spacecraft

In commercial aviation, relevant research settings include abnormal-procedure simulation, high workload during approach, automation mode changes and recovery after interruption. Evaluation should compare the proposed monitoring and interface policy against a conventional baseline, keeping aircraft dynamics and alert content constant. The central endpoint is improved handling of the situation, not merely successful classification of a laboratory workload label.

For spacecraft and long-duration mission interfaces, the transferable design problem is continuity of situational understanding across interruptions, delayed assistance and shared human-machine control. A proposed interface would retain the evidence behind a decision, identify unresolved uncertainty and present a concise reconstruction when a crew member resumes a task. The supporting foundations are human-automation and cognitive-architecture research, not a demonstrated spaceflight application of NDE findings. Environmental transfer would require dedicated testing.

Table 2 Research applications

Machine consciousness and functional architecture

Shanahan (2006) implemented a cognitive architecture combining internal simulation with a global workspace in a simulated robot. The work supports an engineering distinction between incorporating functions inspired by consciousness research and demonstrating subjective experience. For aerospace systems, global availability of relevant state, internal simulation of candidate actions and explicit handling of competing goals may be useful regardless of whether the machine is conscious.

A proposed architecture would separate perception, a shared task representation, simulated consequences and authorised action. Evidence entering the shared representation would retain its source and uncertainty. Simulated outcomes would be labelled as predictions rather than observations. A supervisory component would reject actions outside the validated operating envelope. These are design proposals derived from functional architecture and automation research; they are not demonstrated consequences of survival of consciousness.

An experiment could compare this architecture with a simpler controller on delayed information, conflicting cues and recovery from interruption. Performance measures would include constraint violations, error detection, handover quality and the fidelity of explanations to actual internal state. A flexible controller outperforming a baseline would establish a functional advantage. It would not establish phenomenal consciousness. Conversely, the absence of a consciousness claim would not diminish a useful, reproducible engineering result.

10 Interpretation and research priorities

The strongest warranted conclusion within this literature window is that some survivors report structured experiences whose content, stability and occasional correspondence with events merit prospective study. The evidence does not establish their occurrence after irreversible brain death. The key unresolved variables are the timing of acquisition, the relationship between experience and subsequent memory, and the adequacy of the information controls.

Priority should therefore be assigned to complete exposure records, synchronised timing and independently scored outcomes. A smaller study with trustworthy target custody and clear physiological windows may be more informative than a larger collection of retrospective narratives. For exceptionally rare endpoints, a multicentre registry remains necessary, but scale should follow demonstrated feasibility and a defensible analysis plan.

The proposed contribution is a common discipline for consciousness research and aerospace human factors: infer only the state actually measured, retain uncertainty and test the transition from observation to consequential action. A replicated concealed-information result would create a new explanatory problem. A sufficiently precise null result would constrain a specific proposed capability. Either could coexist with useful advances in recovery assessment, cockpit alerting and human-machine cooperation. The research is most valuable when its outcome remains open and its evidential requirements are specified in advance.

Appendix A Quantitative audit

The intervals in Section 4 use the Wilson score method with z = 1.959964, the stated numerator and denominator, and no continuity correction. For a sample proportion p and size n, the centre is [p + z²/(2n)]/[1 + z²/n], and the half-width is z times the square root of [p(1 − p)/n + z²/(4n²)], divided by [1 + z²/n]. These intervals describe uncertainty conditional on the observed sample. They do not correct ascertainment, loss to follow-up or differences in case definition.

For AWARE, the descriptive intervals are 4.8-16.1% for NDE reports and approximately 0.5-6.9% for explicit-awareness reports. Its main results describe 55 of 101 participants with some awareness or memory, whereas its abstract uses a 46% memory figure. The broad recollection percentage is therefore not used as an unqualified estimate here (Parnia et al., 2014, pp. 1799-1802).

The planning approximation in Section 7 uses a null proportion of 0.25, alternative proportion of 0.35, two-sided alpha of 0.005 and power of 0.90. The standard normal approximation gives approximately 334 independent evaluable observations. Dividing 334 by 0.04 × 0.80 gives 10,437.5 registered events, rounded to approximately 10,440. Every rate in this planning example is an assumed design input; none is presented as a predicted recruitment yield or a demonstrated anomalous effect.

References

Blanke, O., Ortigue, S., Landis, T. and Seeck, M. (2002) 'Stimulating illusory own-body perceptions', Nature, 419, pp. 269-270. https://doi.org/10.1038/419269a

Borjigin, J. et al. (2013) 'Surge of neurophysiological coherence and connectivity in the dying brain', Proceedings of the National Academy of Sciences, 110(35), pp. 14432-14437. https://doi.org/10.1073/pnas.1308285110

Casali, A.G. et al. (2013) 'A theoretically based index of consciousness independent of sensory processing and behavior', Science Translational Medicine, 5(198), 198ra105. https://doi.org/10.1126/scitranslmed.3006294

Chawla, L.S., Akst, S., Junker, C., Jacobs, B. and Seneff, M.G. (2009) 'Surges of electroencephalogram activity at the time of death: a case series', Journal of Palliative Medicine, 12(12), pp. 1095-1100. https://doi.org/10.1089/jpm.2009.0159

Dehaene, S. and Changeux, J.P. (2011) 'Experimental and theoretical approaches to conscious processing', Neuron, 70(2), pp. 200-227. https://doi.org/10.1016/j.neuron.2011.03.018

Dehais, F., Causse, M., Vachon, F., Régis, N., Menant, E. and Tremblay, S. (2014) 'Failure to detect critical auditory alerts in the cockpit: evidence for inattentional deafness', Human Factors, 56(4), pp. 631-644. https://doi.org/10.1177/0018720813510735

Greyson, B. (1983) 'The near-death experience scale: construction, reliability, and validity', Journal of Nervous and Mental Disease, 171(6), pp. 369-375. https://doi.org/10.1097/00005053-198306000-00007

Greyson, B. (2007) 'Consistency of near-death experience accounts over two decades: are reports embellished over time?', Resuscitation, 73(3), pp. 407-411. https://doi.org/10.1016/j.resuscitation.2006.10.013

Klemenc-Ketis, Z., Kersnik, J. and Grmec, S. (2010) 'The effect of carbon dioxide on near-death experiences in out-of-hospital cardiac arrest survivors: a prospective observational study', Critical Care, 14, R56. https://doi.org/10.1186/cc8952

Lee, J.D. and See, K.A. (2004) 'Trust in automation: designing for appropriate reliance', Human Factors, 46(1), pp. 50-80. https://doi.org/10.1518/hfes.46.1.50_30392

Monti, M.M. et al. (2010) 'Willful modulation of brain activity in disorders of consciousness', New England Journal of Medicine, 362(7), pp. 579-589. https://doi.org/10.1056/NEJMoa0905370

Nelson, K.R., Mattingly, M., Lee, S.A. and Schmitt, F.A. (2006) 'Does the arousal system contribute to near death experience?', Neurology, 66(7), pp. 1003-1009. https://doi.org/10.1212/01.wnl.0000204296.15607.37

Oizumi, M., Albantakis, L. and Tononi, G. (2014) 'From the phenomenology to the mechanisms of consciousness: Integrated Information Theory 3.0', PLOS Computational Biology, 10(5), e1003588. https://doi.org/10.1371/journal.pcbi.1003588

Owen, A.M., Coleman, M.R., Boly, M., Davis, M.H., Laureys, S. and Pickard, J.D. (2006) 'Detecting awareness in the vegetative state', Science, 313(5792), p. 1402. https://doi.org/10.1126/science.1130197

Parnia, S., Waller, D.G., Yeates, R. and Fenwick, P. (2001) 'A qualitative and quantitative study of the incidence, features and aetiology of near death experiences in cardiac arrest survivors', Resuscitation, 48(2), pp. 149-156. https://doi.org/10.1016/S0300-9572(00)00328-2

Parnia, S. et al. (2014) 'AWARE-AWAreness during REsuscitation-A prospective study', Resuscitation, 85(12), pp. 1799-1805. https://doi.org/10.1016/j.resuscitation.2014.09.004

Parasuraman, R., Sheridan, T.B. and Wickens, C.D. (2000) 'A model for types and levels of human interaction with automation', IEEE Transactions on Systems, Man, and Cybernetics Part A, 30(3), pp. 286-297. https://doi.org/10.1109/3468.844354

Pope, A.T., Bogart, E.H. and Bartolome, D.S. (1995) 'Biocybernetic system evaluates indices of operator engagement in automated task', Biological Psychology, 40(1-2), pp. 187-195. https://doi.org/10.1016/0301-0511(95)05116-3

Sanders, R.D., Tononi, G., Laureys, S. and Sleigh, J.W. (2012) 'Unresponsiveness ≠ unconsciousness', Anesthesiology, 116(4), pp. 946-959. https://doi.org/10.1097/ALN.0b013e318249d0a7

Schwaninger, J., Eisenberg, P.R., Schechtman, K.B. and Weiss, A.N. (2002) 'A prospective analysis of near-death experiences in cardiac arrest patients', Journal of Near-Death Studies, 20(4), pp. 215-232. https://doi.org/10.17514/JNDS-2002-20-4-p215-232

Shanahan, M. (2006) 'A cognitive architecture that combines internal simulation with a global workspace', Consciousness and Cognition, 15(2), pp. 433-449. https://doi.org/10.1016/j.concog.2005.11.005

Simmons, J.P., Nelson, L.D. and Simonsohn, U. (2011) 'False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant', Psychological Science, 22(11), pp. 1359-1366. https://doi.org/10.1177/0956797611417632

Thonnard, M. et al. (2013) 'Characteristics of near-death experiences memories as compared to real and imagined events memories', PLOS ONE, 8(3), e57620. https://doi.org/10.1371/journal.pone.0057620

van Lommel, P., van Wees, R., Meyers, V. and Elfferich, I. (2001) 'Near-death experience in survivors of cardiac arrest: a prospective study in the Netherlands', The Lancet, 358(9298), pp. 2039-2045. https://doi.org/10.1016/S0140-6736(01)07100-8

Whinnery, J.E. and Whinnery, A.M. (1990) 'Acceleration-induced loss of consciousness: a review of 500 episodes', Archives of Neurology, 47(7), pp. 764-776. https://doi.org/10.1001/archneur.1990.00530070058012

Wijdicks, E.F.M., Varelas, P.N., Gronseth, G.S. and Greer, D.M. (2010) 'Evidence-based guideline update: determining brain death in adults', Neurology, 74(23), pp. 1911-1918. https://doi.org/10.1212/WNL.0b013e3181e242a8

Zander, T.O. and Kothe, C. (2011) 'Towards passive brain-computer interfaces: applying brain-computer interface technology to human-machine systems in general', Journal of Neural Engineering, 8(2), 025005. https://doi.org/10.1088/1741-2560/8/2/025005