cneuromod-things : a large-scale fMRI dataset for task- and data-driven assessment of object representation and visual memory recognition in the human brain
Bibliographic record
Abstract
Understanding how the brain represents objects is a transdisciplinary endeavor that benefits from large and comprehensive datasets. The THINGS initiative is a global effort that aims to collect large-scale datasets with diverse neuroimaging techniques and in multiple species to advance our understanding of object processing in the mind and brain. At its core lies the THINGS database, which includes a thoroughly annotated set of images that are unique for their broad and systematic sampling of natural and man-made objects. Contributing to this growing initiative, we present cneuromod-things, an fMRI dataset acquired while four participants each completed between 33 and 36 sessions of a continuous recognition paradigm on thousands of THINGS images. The same ~4k unique images were shown three times to every participant over the course of the experiment (18 repetitions for each of 720 image categories), providing stable representations for a wide range of systematically sampled images. In contrast to existing fMRI datasets using THINGS, our design is suitable for data-driven analyses at the image level and for investigating visual memory across diverse semantic categories. All four participants are part of the Courtois Project on Neural Modelling (CNeuromod), for which they have completed hundreds of hours of both controlled and naturalistic fMRI tasks, including TV watching and video game playing. This massive dataset also includes extensive anatomical scans, tractography, resting state, functional localizers, eye-tracking and physiological data. The cneuromod-things dataset thus offers the opportunity to model object representations with images from a broad set of semantic concepts using subject-specific models trained on data from the most extensively characterized neuroimaging participants to date.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".