Cited by 2,696Functional Brain Connectivity StudiesStatistical Methods and ApplicationsUndergraduate Neuroscience Education and Research

Circular analysis in systems neuroscience: the dangers of double dipping

Open access
Kriegeskorte, Nikolaus, Simmons, W Kyle, Bellgowan, Patrick S F, Baker, Chris I·2009·Nature Neuroscience

Abstract

A neuroscientific experiment typically generates a large amount of data, of which only a small fraction is analyzed in detail and presented in a publication. However, selection among noisy measurements can render circular an otherwise appropriate analysis and invalidate results. Here we argue that systems neuroscience needs to adjust some widespread practices to avoid the circularity that can arise from selection. In particular, 'double dipping', the use of the same dataset for selection and selective analysis, will give distorted descriptive statistics and invalid statistical inference whenever the results statistics are not inherently independent of the selection criteria under the null hypothesis. To demonstrate the problem, we apply widely used analyses to noise data known to not contain the experimental effects in question. Spurious effects can appear in the context of both univariate activation analysis and multivariate pattern-information analysis. We suggest a policy for avoiding circularity.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/nn.2303

Cites

43 of 43 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • An across-run cross-validation approach ensures that data used for defining fROIs were independent of data used for estimating response.

    “For these localizer analyses, an across-run cross-validation approach was used (Nieto-Castañón & Fedorenko, 2012) to ensure that data used for defining fROIs were independent of data used for estimating response (Kriegeskorte, Simmons, Bellgowan, & Baker, 2009).”

    From Functional Specificity and Sex Differences in the Neural Circuits Supporting the Inhibition of Automatic Imitation (2018) · cites this paper as Kriegeskorte, Simmons, Bellgowan, & Baker, 2009

  • Selecting subjects based on the same data used to demonstrate the absence of awareness is comparable to double-dipping and likely to confound the results with regression to the mean.

    “This practice is problematic because selecting subjects based on the same data used to demonstrate the absence of awareness is comparable to “double-dipping” (Kriegeskorte et al. 2009) and likely to confound the results with regression to the mean (Shanks 2017, Stein et al. 2024).”

    From Evidence for unconscious priming using a Bayesian single-subject approach (2026) · cites this paper as Kriegeskorte et al. 2009

  • Many findings of cognitive neuroscience provide a solid basis on which to build.

    “Despite methodological challenges10–11, many of the findings of cognitive neuroscience provide a solid basis on which to build.”

    From Cognitive computational neuroscience (2018) · cites this paper as [11]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. [4]Vul E Kanwisher N Hanson S Bunzl M Begging the question: The non-independence error in fMRI data analysis Foundations and Philosophy for Neuroimaging in press To appear in
  2. [21]Mitchell TM Machine Learning 1997 McGraw Hill

Cited by

Sorted by citations

Authors
Kriegeskorte, N., Simmons, W. K., Bellgowan, P. S. F., & Baker, C. I.
Affiliations
National Institutes of Health (US)
Corresponding author
  • Kriegeskorte, Nikolaus
  • Baker, Chris I
Venue
Nature Neuroscience
Published
2009-04-26
Volume
12
Issue
5
Pages
535-540
ISSN
  • 1097-6256 (ISSN-L)
  • 1546-1726
OpenAlex
W2015866962
Cited by
2,696
References
43
Citations by year
2026: 139 · 2025: 102 · 2024: 116 · 2023: 136 · 2022: 130 · 2021: 166 · 2020: 169 · 2019: 178 · 2018: 172 · 2017: 201 · 2016: 178 · 2015: 164 · 2014: 190 · 2013: 157 · 2012: 152
FWCI
79.69
Type
journal article
Language
en
Open access
Locations
  • Nature Neuroscience (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
Funders
  • National Institutes of Health
  • National Institute of Mental Health (ZIAMH002909)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Functional Brain Connectivity Studies
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Applications
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Undergraduate Neuroscience Education and Research
Keywords
double dipping, data selection, spurious results, statistical inference validity, neuroimaging, electrophysiology, multivariate pattern analysis, noise data
MeSH terms
Animals, Data Collection, Data Interpretation, Statistical, Humans, Image Processing, Computer-Assisted, Magnetic Resonance Imaging, Neurosciences, Reproducibility of Results, Selection Bias, Artifacts, Systems Biology

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review