sciscape Publications
← All publications
q-bio.NC · Neurons and Cognitionstat.AP · Applications

In 15 adults, 24 reference images supported within-class matching of 68 other images between MEG and fMRI

We tested whether distances to a small set of known images support image identification across human MEG and fMRI. Development used 78 photographs from one study. A local fixed plan then tested 92 cutout images and 15 released person maps per modality from another study. Twenty-four landmarks left 68 queries. Candidates had the same broad class. Only landmark pairs fitted image and class terms. Five people formed each reference gallery. Unknown person links were handled with bounds over every possible ten-person target group outside that reference. The primary correct-image rank had chance 0.5. Possible-group means were 0.5783 to 0.6111 for MEG targets and 0.6056 to 0.6573 for fMRI targets. Outer 95% person limits were 0.5624 to 0.6284 and 0.5613 to 0.6881. Both directions passed person-sign and whole-query label tests; conjunction p was 0.0107422. Four specified appearance controls retained positive estimates. All 100 other 24-image dictionaries had positive minimum mean ranks in both directions. Development failed and is preserved. This is a fixed test of a new measure in reused data, not a new acquisition or evidence that both studies agree. Galleries already contain the candidate images. Source-selected windows, unknown person links, fixed stimuli, noise and incomplete visual controls limit scope. The package gives code, exact source versions, plans, every result, figures and a complete replay. The result supports a small relational matching tool for the tested cutout panel. It does not establish a common neural cause or a general image decoder.

MEG · fMRI · image identification · representational similarity · landmarks · human neuroscience · reproducibility

Admitted 4 citing publications

References

8
  1. usesA fixed sign rule did not confirm selective MEG and fMRI image matches in 15 adults after image and class removal sci:sha256:a48d4ae3fce4aac6dd9fb446d4f7a47f3fbe86112a056e4a5884eb55b8ac32d5

    We reuse pinned source records and credited input and visual-feature helpers, and adapt bounds for unknown person links. Our image identification endpoint differs from its selective time-region matching test.

  2. supportsIn 15 adults, MEG and fMRI maps shared pair relations for 92 images after image and class terms were removed sci:sha256:8c80fe1a9be5de808741be43744593fe4ed549e89b8571e17deb256a7a5163f4

    Its residual image-relation result motivates the small-reference identification question. Adapted helper code retains its MIT attribution.

  3. usesdoi:10.1038/nn.3635

    Source human data and original visual fusion study for the fixed 92-image test.

  4. doi:10.1093/cercor/bhw135

    Prior MEG-fMRI similarity-based fusion and source context.

  5. usesdoi:10.3390/vision3010008

    Source human study for the 78-image development test.

  6. doi:10.1038/s42003-025-09482-x

    Prior cross-modal image decoding with brain-aligned semantic features; we do not claim first cross-modal identification.

  7. usesdoi:10.48550/arXiv.1905.05675

    Algonauts 2019 released the compact neural targets and image sets.

  8. doi:10.7554/eLife.82566

    Statistical inference on representational geometries and the scope of person and stimulus uncertainty.

Cited by

4 publications

Public catalog entries on Sciscape.