1 Title and one-sentence summary

Rates, Temporal Structure, and Memory Consequences of Mind Blanking During Wakefulness

Using independently defined events, multimodal recordings, and adversarial model comparison, this project tests whether wakeful mind blanking occurs 15–50 times per hour, exhibits periodic structure, and predicts subsequent inaccessibility of previously encoded information.

Applicant: Future Mind Institute, a US nonprofit; Priority A, consciousness states and attention dynamics. This expanded Project Brief follows the requested length; its typeset version will exceed one page.

2 Scientific question and background

Andrillon et al. (2021) linked local sleep-like slow waves during wakefulness to attention lapses, mind wandering, and blanking reports. Mortaheb et al. (2022) associated blanking reports with a distinctive whole-brain connectivity profile. Neither established a 15–50-events-per-hour law. The approximately 60–120-millisecond EEG microstates reviewed by Michel and Koenig (2018) cannot be equated with blanking events. Duplicate entries in the supplied bibliography represent three unique publications. Andrillon et al., 2021; Mortaheb et al., 2022; Michel and Koenig, 2018.

These 2021–2022 candidate measures make prospective event counting timely in 2026. However, inability to report content does not establish absence of experience. Event frequency, periodicity, and memory consequences therefore require separate tests.

3 Hypothesis provenance

The hypotheses were extracted from the Seth Material, Seth Speaks, §19. Claim altered-states-of-consciousness-4 proposes periodic blank points occurring 15–50 times hourly; alternate-focus-5 repeats this range with activity dependence; waking-consciousness-2 adds an account of subsequent inability to remember. These are overlapping statements from one textual tradition, constituting one hypothesis source rather than three independent evidential contributions. Claims about perceiving other realities are outside this study’s inferential scope.

The applicant identifies the source archive as the Zenodo dataset, licensed CC BY 4.0. Its version and claim mapping will accompany preregistration. Future Mind Institute’s public “Seth Material Research Program” supplies hypotheses only; the material is never treated as scientific evidence.

4 Main hypotheses and falsifiable predictions

H1: Rate. After correcting detection error, the population mean event rate in each load condition falls within 15–50 per hour. Individual distributions will also be reported; this does not assert universal individual compliance. The interval-testing null is a rate at or below 15, or at or above 50.

H2: Temporal structure. A periodic model outpredicts prespecified nonperiodic alternatives. The null is no predictive improvement from periodic terms. Planning assumes intensity modulation of 20% above and below its mean, with periods searched between 72 and 240 seconds. This window is an additional operational assumption: a frequency range alone does not imply periodicity.

H3: Memory. For items with immediate recognition supporting prior encoding, intervening blanking predicts delayed recognition at least five percentage points below matched no-blanking items. The planning standardized within-participant effect is approximately 0.25; the null is no decrement. Each hypothesis receives a separate verdict. Even joint support would establish neither complete interruption of consciousness nor active memory erasure.

5 Research design

An independent 30-person pilot precedes recruitment of 200 adults for the main study, allowing 10% loss and approximately 180 analyzable participants. Each completes three two-hour sessions across three days. Low-, medium-, and high-load blocks have counterbalanced order. High-density EEG, eye tracking, behavioral performance, and randomized experience probes are synchronized. Sleep history, session timing, caffeine, and fatigue are recorded.

The pilot freezes a report-anchored candidate-event algorithm, duration rules, event-merging intervals, and artifact and microsleep exclusions. Thresholds cannot be selected to produce the target rate. Validation holds out participants and estimates sensitivity, specificity, and their uncertainty. Reports distinguish absent reportable content, mind wandering, drowsiness, uncertainty, and missed responses.

Sparse-probe, dense-probe, and probe-free blocks are randomized to estimate measurement interference. Memory items span periods before and after candidate events. Immediate recognition includes confidence; delayed recognition occurs after 20–30 minutes and the following day. Random assignment of items to immediate testing estimates testing effects. Approximately 60 participants, selected through prespecified stratification, receive additional simultaneous EEG–fMRI for localization.

With 180 participants, a simple paired test has approximately 80% power for a standardized effect of 0.25 at two-sided α=.0167. This approximation does not establish power for event counts or item-level memory models. Following the pilot, simulation will incorporate clustering across three days, overdispersion, detection failures, event duration, and usable memory-item counts. Principal tests must achieve at least 80% power with false-positive rates no greater than 5%. Otherwise, the protocol is revised or paused before main-study collection; thresholds cannot be relaxed afterward.

Participants remain unaware of the numerical prediction; acquisition staff remain unaware of the directional memory prediction. Scorers and analysts use masked labels. Preregistration fixes exclusions, outcomes, models, search corrections, and stopping rules.

6 Data and analysis plan

Primary outcomes are corrected event rate, held-out predictive improvement from periodic structure, and delayed-recognition differences for previously encoded items. Secondary outcomes include duration, load dependence, slow waves, pupil measures, and connectivity patterns.

A hierarchical latent-event model incorporates duration, missed detections, false positives, and post-probe disturbance in its observation process. Analyses report valid waking exposure and sensitivity to uncertain detection parameters. Inhomogeneous Poisson, nonperiodic renewal, and state-switching models compete with periodic models while retaining fatigue and load covariates. Period searching receives multiplicity correction. Time-rescaling diagnostics and cross-day prediction assess adequacy; a spectral peak alone cannot establish periodicity.

An item-level mixed logistic model estimates memory differences while accounting for immediate performance, confidence, load, retention interval, and testing effects. A joint encoding model examines selection bias. Microsleep, mind wandering, and failed encoding provide explicit competing explanations. Immediate correct recognition supports encoding without proving it with certainty; an event association does not establish causation. Holm correction controls the three primary tests.

Deidentified data, code, dictionaries, and all outcomes will be released. Imaging will be defaced, with controlled access for data retaining substantial identification risk.

7 Milestones and duration

Months 1–3 cover ethics, the adversarial agreement, pilot acquisition, and power simulation. Month 4 freezes event definitions and preregisters the protocol. Months 5–11 cover the main cohort and imaging. Months 12–15 cover masked quality control and model comparison. Months 16–18 cover unmasking, independent verification, data release, and submission. Completion in 12 months depends on recruitment and scanner availability; 18 months is the planning horizon.

8 Budget scale

The proposed total is US$650,000, within a US$450,000–750,000 implementation range:

• Personnel: $260,000.
• Participant compensation: $70,000.
• EEG and eye-tracking facilities and consumables: $80,000.
• Simultaneous EEG–fMRI: $90,000.
• Data management and statistical work: $50,000.
• Independent verification and open science: $20,000.
• Indirect costs and contingency: $80,000.

The estimate includes the pilot and assumes existing equipment. A reduced award would first reduce exploratory imaging. Any change affecting primary tests requires renewed power calculations.

9 PI profile and candidate teams or institutions

The PI should have demonstrated experience in waking attention, sleep physiology, experience sampling, and preregistered multimodal research. The team also requires a memory researcher and an independent statistical lead.

Potential collaborators are Thomas Andrillon at the Paris Brain Institute for local-sleep methods; Athena Demertzi at the University of Liège’s Physiology of Cognition Lab for state localization; and Naotsugu Tsuchiya at Monash University for consciousness measurement. These are candidates, without commitments. Advocates and skeptics will jointly sign adjudication rules, while an independent statistical group executes the analysis.

10 Risks and abandonment criteria

Overinclusive definitions, reporting failures, and probe-induced state changes are principal risks. If held-out pilot sensitivity or specificity falls below 80%, or the latent rate is unidentifiable, numerical adjudication pauses. This is a measurement failure, not confirmation or refutation of the underlying account.

If the corrected confidence interval lies wholly outside 15–50 per hour, the interval prediction is abandoned for that condition; a boundary-crossing interval is inconclusive. If the upper confidence bound on periodic modulation is below 20% and there is no held-out advantage, the specified periodic model is abandoned. If the upper bound on memory decrement is below five percentage points, the prespecified meaningful memory effect is abandoned. Adequate explanation by competing mechanisms removes grounds for an additional mechanism. Post hoc subgroups cannot rescue primary hypotheses. fMRI localization cannot independently determine brief event boundaries.

11 Ethics and compliance

Ethics approval, informed consent, voluntary withdrawal, and MRI safety screening precede participation. The protocol does not induce sleep deprivation or provide unvalidated diagnostic feedback. Cross-border data transfers require appropriate legal agreements. Source disclosure remains neutral, and consent explains research uses. Funding, relevant beliefs, and conflicts of interest are disclosed. Funders cannot veto publication of negative findings.

12 Alignment with existing funders

Future Mind Institute provides core funding and governance as a neutral consciousness-and-life-sciences funder. BIAL’s healthy-human psychophysiology remit offers the most direct fit for partial cofunding. Fetzer’s mission provides a more indirect connection; suitability for this basic-research proposal should be confirmed before application.

Templeton World Charity Foundation’s COGITATE model offers a governance precedent for preregistration, open science, and adversarial collaboration. BICS, UVA DOPS, and IONS are potential scientific exchange contacts, not interchangeable funding bodies or committed partners. These alignment assessments do not imply an open funding call or an award commitment.