CNN-extracted features generate synthetic fMRI responses to unseen images
Bibliographic record
Abstract
Inspired by biological vision, convolutional neural networks (CNNs) have tackled challenging image recognition problems once considered the sole purview of human expertise. In turn, CNNs are now widely used as a framework for studying human vision. The organizational similarity between the layers of CNNs and cortical regions along the visual pathway has been shown in studies using human fMRI data, such that early visual areas' activities are better predicted by the first layers of CNNs while their last layers better predict the response of higher-level visual areas. However, there is a lack of agreement on how well CNN features can predict fMRI responses, particularly in the presence of fMRI noise, which can result in varying brain responses to the repetitions of the same image. Additionally, the utility of these predicted responses to previously unseen images as synthetic fMRI data has not yet been explored. Here we use the BOLD5000 dataset and the AlexNet architecture initialized with the model weights pre-trained on ImageNet to show that features extracted by CNNs can g enerate highly accurate synthetic fMRI responses to images. We demonstrate that synthetic fMRI responses show higher correlations with repetitions of real responses than the real responses themselves, surpassing the quality of real data in the presence of noise. Moreover, we train a decoder with synthetic fMRI data to classify real fMRI data for unseen images and even unseen object categories. Our decoding experiments revealed that the synthetic data outperformed real data, particularly due to the ability to generate larger synthetic datasets. Our findings showcase the high quality of generated synthetic fMRI responses to images based on CNN features, exhibiting both similarities to real data and practical utility in empirical applications.
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 | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| 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".