Neuroscience
Neuroscience research, evaluated by neuroscientists
How neuroscientific judgment shows outside a department, and the standard the founding cohort will hold it to. Everything below is a draft in public, on purpose.
What counts as evidence of neuroscience research skill
Neuroscience spans scales from single-channel recordings to whole-brain imaging, and its public record has grown to match: bioRxiv preprints, deposited datasets on OpenNeuro and similar repositories, and analysis pipelines shared alongside the paper. A preprocessing pipeline with honest artifact-rejection criteria, or an imaging study whose statistical thresholds are stated before the result, is direct evidence of skill.
Outside universities, neuroscience work runs neurotechnology companies, pharma target-validation teams, and clinical neurophysiology practices. The scientist who validated a BCI decoding pipeline, or who documented why a candidate biomarker did not replicate, is doing neuroscience research whether or not the title on their badge says so.
The field's chronic failure mode is small samples chasing large claims — underpowered imaging studies and circuit findings that do not survive a stricter correction. Reading for that gap between claim and power is a learnable, scoreable skill, and it is one this community needs from its neuroscientists.
Negative and null results carry real weight here: a well-powered study that failed to find a claimed effect, a replication attempt of a widely cited circuit finding, a public reanalysis with a stricter multiple-comparisons correction. Producing them is some of the strongest evidence of judgment the field offers.
The neuroscience evaluation rubric, first draft
This rubric leans hard on the distance between signal and claim, because neuroscience's mix of small samples and high-dimensional data makes that distance the field's central evaluation problem.
- Statistical power and correction
- Sample size is justified before the study, and multiple-comparisons correction matches the number of tests actually run — not a subset chosen after seeing the data.
- Method transparency
- Acquisition parameters, preprocessing pipeline, and exclusion criteria are documented well enough to rerun, and analysis code is inspectable.
- Claim discipline
- Conclusions stay within what the data support; functional or causal language is earned by perturbation or intervention, not borrowed from correlation or coactivation.
- Cross-scale coherence
- Claims made at one level of description — molecular, circuit, systems, behavioral — are checked against what is known at adjacent levels, and inconsistencies are addressed rather than ignored.
Founding neuroscience evaluators will revise this draft against real preprints and datasets before it is used to score anyone.
What founding neuroscience evaluators will do
Begin by stress-testing the rubric against the artifacts neuroscientists actually produce — an imaging study reads differently from a single-unit electrophysiology paper, and the standard has to survive both.
Run calibration rounds on public work: independent scoring of the same preprints and datasets, then structured comparison of where trained readers disagree and why.
Carry a per-discipline calibration record into the open community, seeding the weighting system with neuroscientists whose judgment has a measured track record.
Who this is for
The cohort needs neuroscientists who already read other people's work the hard way, across scales and methods:
- Systems and cognitive neuroscientists who read the preprocessing pipeline before the discussion section.
- Computational neuroscientists whose models and code are their real publication record.
- Neurotech and pharma scientists whose best work lives in validation reports no journal will ever see.
- Clinical neurophysiologists and early-career researchers who learned rigorous design in the shadow of the field's power problem.
Who this is not for
Just as real, and worth reading before you apply:
- Anyone hoping to accredit their own research program — evaluators will score other people's artifacts here.
- Researchers who want a credential faster than a standard can be built honestly.
- Anyone unwilling to have their scoring accuracy measured and visible; calibration is the community's anchor.
- Anyone who needs the platform to be live today — evaluation mechanics are in design, and this page says so on purpose.
Apply to evaluate neuroscience
Neuroscience is pre-selected on the application. Link an ORCID profile, a bioRxiv listing, or a dataset repository we can read.