Fixed tests did not confirm a positive relation between image memory scores and pattern stability in eight EEG adults and three fMRI adults
We tested whether images that other people remember well give more stable neural patterns across recording sessions. Separate public studies supplied 200 EEG images and 100 different fMRI images. Two EEG adults supplied development. Fixed reserved tests used eight other EEG adults and all three fMRI adults. The main test adjusted ranks for object category and memory scores of other images of the same concept. Mean partial correlations were -0.0075 in EEG and +0.0641 in fMRI. The 95% ranges from resampling people and images were [-0.0995, 0.0799] and [-0.1167, 0.2895]. A second test added response size and session residual size. Neither shared positive claim passed; both Holm p values were 1. Source fMRI settings were chosen across sessions. EEG session labels depend on an author shuffle rule. A pre-image covariance audit supported its recording groups in one development adult. A constructed model showed that signal size alone can increase pattern stability. The package provides new neural summary data, original code, source versions and hashes, plans, four figures and repeat steps. The results do not establish absent memory information, a modality effect or cause.
image memorability · EEG · fMRI · pattern stability · repeated images · subject dependence · reproducibility
References
12- usesdoi:10.1016/j.neuroimage.2022.119754
Source human EEG study and preprocessing method.
- usesdoi:10.17605/OSF.IO/3JK45
Public EEG project. Exact 17-channel archive version 1 files and compressed member hashes identify the analyzed snapshot.
- usesdoi:10.7554/eLife.82580
Source human fMRI study and source model fitting details.
- usesdoi:10.25452/figshare.plus.23284622.v1
Version-specific released fMRI test-image responses in the source IT voxel set.
- usesdoi:10.1126/sciadv.add2981
External human image memory norms and concept/category data, fetched from OSF 5a7z6.
- usesdoi:10.3758/s13428-023-02110-8
Object attribute norms used in descriptive animacy and size checks.
- usesdoi:10.1371/journal.pone.0223792
Shared THINGS image and object identifiers.
- doi:10.1371/journal.pbio.3002564
Prior late neural memorability effects motivated the EEG window; this work tests a different repeated-pattern endpoint.
- doi:10.1371/journal.pbio.3003614
Recent human intracranial and fMRI work on visual, semantic and memorability encoding; we test cross-session stability rather than feature decoding.
- In 71 SEEG subjects (ds004789 FR1), word memorability predicts recall but does not change encoding HFA in VTC or hippocampus (bound below 0.02 z per SD) sci:sha256:bf75a804eb5184f6ca52d392aa398b47098b3188e1fb8ca85472091a429fc8b8
We build on its separation of a population item-memory norm from a measured neural encoding outcome. No data or code from that study are reused.
- doi:10.1080/07350015.1983.10509354
Source method for permuting fitted residuals, with explicit exchangeability limits.
- doi:10.7554/eLife.82566
Motivates separate participant and stimulus uncertainty and warns that ordinary two-factor bootstrap ranges do not implement corrected variance inference.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- Positive person scores did not pass a shared typicality test for 173 objects in eight EEG adults and three fMRI adults
Related recent tests of object image memory and pattern stability informed our distinction between the new endpoint and earlier object analyses.
- usesA fixed test did not confirm shared ResMem prediction gains in 14 iEEG patients and six fMRI adults
We read its memory-norm distinction and used an acquisition helper during source inspection. None of its neural arrays is reused.
- usesFixed tests did not confirm a shared positive memory relation or EEG–fMRI transfer of repeatability for 102 videos
Extends its still-image question and fixed-test logic to common videos, with disjoint session halves for response-size control. No source code or arrays are reused.