Before the Result: What Preregistration Would Require of an NSI Study

The research question: was the prediction specified before the evidence?

A compelling explanation can be created after a pattern has been observed. Testing a prediction requires a different sequence: specify the claim and the relevant analysis before learning the result that will evaluate it. The research question is whether an NSI study can make that sequence visible to an independent reader. Preregistration provides a record of planned decisions, but it does not certify that the theory is correct, the measures are valid or the study has enough information to answer its question. Nosek and colleagues described preregistration as a way to distinguish prediction from postdiction and to make analytical commitments inspectable. [1] The distinction does not make exploratory work illegitimate. Exploration can generate valuable questions. The problem arises when a question developed from the observed data is presented as though it had been tested independently on those same data. A transparent report identifies which analyses were planned and which emerged later, allowing readers to evaluate their evidential roles separately.

A title and a hypothesis are not a complete protocol

A useful registration would specify the participant population, recruitment approach, inclusion rules, stopping rule, measurement procedures and primary outcome. It would define the predictor and the expected relationship precisely enough that another researcher could reproduce the intended analysis. For NSI, a phrase such as 'nervous-system state predicts decisions' leaves too much unspecified. Which state measure, which decision, which time interval and which comparison would determine whether the prediction was supported? The protocol should also state how missing data, artifacts and exclusions will be handled. These decisions can alter the sample and therefore the result. A researcher may have legitimate reasons to revise a procedure after discovering a technical problem, but the change should be recorded and explained. The purpose is not to pretend that studies never encounter surprises. It is to prevent necessary changes from becoming invisible opportunities to select the most favorable result.

Registered reports add a different review stage

Chambers introduced Registered Reports at Cortex as a publishing initiative that evaluates the question and proposed methods before the results are available. [2] This differs from simply depositing a plan in a registry. A registered report can involve peer review and an in-principle publication decision before the main findings are known, subject to the relevant journal's requirements. The distinction matters when describing a project's status: a private draft, a public registration and an accepted registered-report protocol are not equivalent achievements. OSF's own documentation describes registrations and preregistrations as timestamped records with defined properties and templates. [3] The platform can preserve the plan, but it cannot supply missing scientific judgment. A registered use of an unsuitable instrument remains unsuitable. The discussion of heart-rate variability measurement illustrates the point: declaring an analysis in advance does not turn HRV into a specific measure of fear. The construct, measurement and interpretation must still be defended independently.

Specify what would count against the proposed framework

A strong protocol includes outcomes that would fail to support the claim. It should not define success as any association among a large set of measures. For example, a proposed study might ask whether one prespecified physiological measure adds predictive information beyond a baseline containing task conditions and previous behavior. The precise variables and comparison must be settled before evaluation. If the added measure does not improve the declared outcome, the result should remain informative rather than being replaced by a newly chosen target. This connects directly to the standards for validating NSI. Independent testing, appropriate baselines and replication are still required after preregistration. The framework is proposed, not clinically or financially validated. Registration can improve the interpretability of a test, but it is not evidence that the framework has already passed one. Public communication should distinguish a research plan, collected data, an initial analysis and an independently replicated finding rather than treating them as successive marketing labels.

A practical publication record could include the registration link, a concise account of deviations, the final analysis code where sharing is appropriate, and an explanation of data-access limits. Participant privacy may restrict open release of physiological or financial records. Transparency does not require exposing identifiable information; it requires describing what can be checked and providing justified alternatives where feasible. A synthetic demonstration or controlled access arrangement should be labelled accurately rather than presented as the original dataset. The conclusion is that preregistration makes a research promise inspectable. It asks the researcher to specify the question, the evidence and the analysis before the answer is known, then to explain what changed. Its value lies in improving the interpretation of results, including null or disappointing ones. For NSI, that would be a concrete methodological step, while the scientific validity of its measures and claims would remain a separate question to be answered by evidence.