Pith. sign in

REVIEW 4 major objections 4 minor 4 cited by

Towards Neural Foundation Models for Vision: Aligning EEG, MEG, and fMRI Representations for Decoding, Encoding, and Modality Conversion

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that EEG, MEG, and fMRI encoders trained with contrastive learning against frozen CLIP image embeddings create a shared semantic space that supports decoding, encoding, and cross-modality conversion by nearest-neighbor…

desk verdict Solid CLIP-anchored decoding for EEG/MEG/fMRI, but the modality-conversion headline rests on an evaluation that measures CLIP image similarity, not a learned cross-modal neural alignment. read the letter →

arxiv 2411.09723 v1 pith:YRBW57AF submitted 2024-11-14 cs.CV cs.AI

Matteo Ferrante , Tommaso Boccato , Grigorii Rashkov , Nicola Toschi This is my paper · ORCID
classification cs.CVcs.AI
keywords braindecodingencodingneuralmodalityconversioncontrastivelearningelectroencephalographymagnetoencephalographyfunctionalmagneticresonanceimagingsharedrepresentationspace
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that three very different ways of measuring brain activity—electroencephalography (EEG), magnetoencephalography (MEG), and functional magnetic resonance imaging (fMRI)—can be pulled into one shared semantic space anchored by the frozen image encoder of the CLIP model. The authors train a separate neural encoder for each modality with a contrastive loss that pushes the encoder's output toward the CLIP embedding of the image the subject was viewing. In that shared space, all three tasks reduce to nearest-neighbor retrieval: decoding (find images matching a brain signal), encoding (find brain signals matching an image), and modality conversion (find EEG/MEG/fMRI activity that matches activity recorded in another modality). The paper reports 93.8 percent CLIP 2-way accuracy for fMRI decoding, 79.4 percent for EEG, 60.1 percent for MEG, and normalized conversion accuracies from 83.7 to 95.4 percent. If these numbers hold under clean evaluation splits, the framework offers a single non-generative backbone for aligning neural recordings to visual content and to each other.

What carries the argument

The central object is the shared CLIP-anchored representation space. CLIP's frozen image encoder $h$ produces a semantic embedding for every image; each neural modality gets a trainable module $f_n$ that maps raw signals into the same normalized unit hypersphere. The contrastive loss—symmetric cross-entropy with a temperature of $\tau = 1$ between the neural embedding and the CLIP image embedding for the matching image—is what forces the alignment. This single mechanism carries all three experiments: because every modality is anchored to the same image embeddings, nearest-neighbor search in that space implements decoding, encoding, and cross-modal conversion without any direct pairwise training between modalities.

What would settle it

Run modality conversion on a held-out set after training the EEG and fMRI encoders on disjoint image sets, then check whether the CLIP 2-way conversion accuracy stays above chance; if it stays high, the retrieval is carried by image-level CLIP similarity in the candidate set rather than by an aligned neural code.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that aligning each neural modality to CLIP image embeddings is enough to create a joint representation space shared across modalities, without any EEG-MEG or EEG-fMRI training pairs. Each neural module $f_n = g_n \circ a_n$ first maps a subject's raw signal into a common per-subject-aligned space and then refines it to approximate the CLIP image representation $h(\mathrm{img})$. After training, decoding, encoding, and modality conversion are all implemented as cosine-similarity nearest-neighbor search in that space, and the paper evaluates all three with CLIP 2-way accuracy between the images associated with query and retrieved samples. The authors report that fMRI decoding reaches 93.8 percent, EEG 79.4 percent, and MEG 60.1 percent CLIP 2-way accuracy, while conversion accuracies normalized by decoding performance range from 83.7 to 95.4 percent. They explicitly note the model is non-generative and proof-of-concept.

Load-bearing premise

The load-bearing premise is that the shared CLIP-anchored space is dense and semantically consistent enough that nearest-neighbor retrieval from one modality lands on genuinely matching neural activity in another modality, a property the contrastive objective does not by itself guarantee.

Editorial extensions

If this is right

  • A single framework can retrieve the image a person was viewing from EEG, MEG, or fMRI, with reported CLIP 2-way decoding accuracies of 79.4 percent, 60.1 percent, and 93.8 percent, respectively.
  • The same encoders run encoding in reverse: given an image, they retrieve neural activity whose associated image is semantically similar, with reported CLIP 2-way accuracies of 85.5 percent (EEG), 58.8 percent (MEG), and 87.8 percent (fMRI).
  • Neural modality conversion is possible without cross-modal training pairs: for example, fMRI-to-MEG conversion reaches 95.4 percent normalized CLIP 2-way accuracy by nearest-neighbor search in the shared space.
  • All three tasks use identical machinery (contrastive alignment followed by cosine-similarity retrieval), so improvements in the alignment step should transfer across tasks.
  • Because the framework is non-generative, decoded or converted outputs are retrieved existing images or neural samples rather than newly generated signals; the authors position the model as a proof-of-concept rather than a full foundation model.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: If the shared space is as dense as the reported accuracies suggest, modality conversion could be used to impute one recording type from another (for example, estimating fMRI-like spatial detail from cheaper EEG recordings) without retraining, though the paper does not test this directly.
  • Inference: The reliance on CLIP as the anchor means the semantic granularity of the model is bounded by CLIP's own representations; images that CLIP cannot separate will likely not be separable in any modality, and that ceiling should be measurable by computing CLIP 2-way accuracy on the image retrieval set itself.
  • Inference: A direct stress test of the shared-space claim would be to train the EEG and fMRI encoders on disjoint image sets and then measure cross-modal conversion on images seen by only one modality; high accuracy there would show genuine neural correspondence rather than image-level similarity leakage.
  • Inference: The paper's privacy discussion implies that as such decoders improve, the same alignment could be used for cross-subject transfer or minimal-cooperation decoding, making neural-privacy safeguards more urgent; the authors raise but do not resolve this concern.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes a framework that trains separate EEG, MEG, and fMRI encoders with a contrastive CLIP loss against a frozen CLIP image encoder, and then uses nearest-neighbor retrieval in the resulting shared space for three tasks: decoding images from neural data, encoding images into neural data, and converting between neural modalities. The reported results include CLIP 2-way decoding accuracies of 79.4% for EEG, 60.1% for MEG, and 93.8% for fMRI, encoding accuracies between 58.8% and 87.8%, and normalized modality-conversion accuracies between 83.7% and 95.4%. The central claim is that this demonstrates a unified semantic representation shared across EEG, MEG, and fMRI. The manuscript is a proof-of-concept study built on public datasets and includes a useful comparison with prior decoding work, but the evidence for the modality-conversion claim is incomplete.

Significance. If the modality-conversion results were properly supported, this would be a valuable step toward a unified multimodal neural representation for vision: one framework handling decoding, encoding, and cross-modal retrieval, with public datasets and attention to known EEG contamination pitfalls. The paper is honest about its proof-of-concept status and non-generative nature. However, the load-bearing conversion claim currently lacks a control separating genuine cross-modal neural alignment from transitivity through CLIP image similarity, and the reported numbers are single point estimates without variability or a defined normalization procedure. The significance is therefore conditional on additional analysis.

major comments (4)
  1. [Sections 2.2-2.3, Table 2]
  2. [Section 2.4, Table 2]
  3. [Tables 1-2, Section 3]
  4. [Section 2.4, Tables 1-2]
minor comments (4)
  1. [Abstract and Section 1]
  2. [Table 1]
  3. [Section 2.1]
  4. [Section 2.2]

Circularity Check

1 steps flagged · score 6.0 of 10

Modality conversion accuracies reduce by construction to CLIP image similarity: each neural encoder is independently trained to match the frozen CLIP image encoder, and conversion is both performed and scored in that same CLIP space.

  1. self definitional [Section 2.2 (Neural Vision Alignment), Section 2.3 (Modality Conversion), Section 2.4 (Modality Conversion Performance), Table 2]
    "we generate a representation zi = f (n, s) ... and, concurrently, we derive the corresponding image representation zj through the image encoder: zj = h(img). ... the contrastive CLIP loss is calculated ... [Modality Conversion:] we encode the sample from our input modality and the target search set from the desired output modality, selecting the top-n matches based on cosine similarity ... [Evaluation:] The performance of this aspect of our model is gauged using the CLIP 2-way accuracy between the images related to the source and target modalities."

    Each modality encoder f_X is trained, with no cross-modal loss, to make f_X(n) approximate the frozen CLIP image embedding h(img). Modality conversion is nearest-neighbor retrieval in that space, so argmax_j cos(f_X(n_X), f_Y(n_Y)) is approximately argmax_j cos(h(I_X), h(I_Y)). The reported score is CLIP 2-way accuracy between the images I_X and I_Y associated with the source and retrieved target samples. Thus the Table 2 conversions (e.g., normalized 0.954 for fMRI->MEG) are a by-construction measure of CLIP image-to-image retrieval, not of any learned cross-modal neural correspondence. The normalization by target decoding accuracy and the fact that the datasets come from different subjects (no identical stimulus pairs) do not remove this transitivity.

full rationale

The decoding and encoding results are held-out evaluations and are benchmarked against external single-modality works (EEG vs [28,38]; MEG vs [39]; fMRI vs [24,6,10]), so those parts of the central claim have independent content. The manuscript's self-citations [8,10,38] are used only as comparisons, not as load-bearing premises. The problem is localized to the modality-conversion claim: because the encoders are trained independently to reproduce CLIP image embeddings and conversion is retrieved and evaluated via CLIP 2-way accuracy, the conversion accuracies are forced by the training setup and the known semantic structure of CLIP. The paper itself calls the model a proof-of-concept, but it does not acknowledge this transitivity confound. One of the three headline capabilities therefore reduces by construction to the input objective, giving a partial circularity score of 6.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities, forces, or dimensions. Its contribution is a training procedure and shared embedding space, which are not invented entities in the sense of this ledger. The main free choices are the contrastive temperature and standard training hyperparameters, and the main assumptions concern the suitability of CLIP as an anchor and the correctness of inherited dataset preprocessing and split choices.

free parameters (2)
  • Contrastive temperature tau = 1
    Set by hand in Section 2.2; it modulates the sharpness of the contrastive loss and affects alignment quality, though the paper says it could be learned.
  • Training hyperparameters (learning rate, weight decay, batch size, epochs) = 3e-4, 1e-3, 256, 30
    Chosen by hand without search; the paper acknowledges room for improvement through hyperparameter search.
assumptions (4)
  • domain assumption CLIP image embeddings are a suitable semantic anchor for aligning EEG, MEG, and fMRI neural activity.
    Invoked throughout Section 2.2; if CLIP similarity does not reflect neural semantic similarity, both the training objective and the evaluation metric lose validity.
  • domain assumption The conservative EEG data splitting approach described in reference [28] prevents train and test contamination.
    Stated in Section 2.1; the paper relies on this citation rather than reporting the exact split details.
  • domain assumption Subject-specific alignment layers can map all subjects into one common representation space.
    Assumed in Section 2.2; if subject variability is too large or nonlinearly related, the shared space may not be coherent across subjects.
  • domain assumption Preprocessed fMRI betas and trial averaging in the Natural Scenes Dataset preserve the relevant visual information.
    Assumed in Section 2.1; the analysis inherits NSD's GLM denoising and beta estimation choices without re-evaluation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Neural Foundation Models for Vision: Aligning EEG, MEG, and fMRI Representations for Decoding, Encoding, and Modality Conversion." pith.science (2026). https://pith.science/paper/YRBW57AF

@misc{pith2026241109723,
  author       = {Pith},
  title        = {Pith review of: Towards Neural Foundation Models for Vision: Aligning EEG, MEG, and fMRI Representations for Decoding, Encoding, and Modality Conversion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YRBW57AF}},
  note         = {Machine review of arXiv:2411.09723}
}
read the original abstract

This paper presents a novel approach towards creating a foundational model for aligning neural data and visual stimuli across multimodal representationsof brain activity by leveraging contrastive learning. We used electroencephalography (EEG), magnetoencephalography (MEG), and functional magnetic resonance imaging (fMRI) data. Our framework's capabilities are demonstrated through three key experiments: decoding visual information from neural data, encoding images into neural representations, and converting between neural modalities. The results highlight the model's ability to accurately capture semantic information across different brain imaging techniques, illustrating its potential in decoding, encoding, and modality conversion tasks.

Figures

Figures reproduced from arXiv: 2411.09723 by the authors.

Figure 1
Figure 1. Schematic representation of our proposed model, illustrating the alignment of various neural [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The top panel illustrates the ’Decoding’ experiment, where neural data is processed to ’decode’ and [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Comparative Results of Multimodal Neural Decoding. The figure shows the original visual stimuli [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Encoding Experiment Results Displaying Image-to-Brain Activity Correlation. Rows illustrate the results for EEG, MEG, and fMRI modalities. The leftmost images are the encoded stimuli, and the sub￾sequent images represent images related to the brain activities retrieved…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Shifting Attention to You: Personalized Brain-Inspired AI Models

    q-bio.NC 2025-02 conditional novelty 6.0 of 10

    Fine-tuning CLIP with human behavioral embeddings and dynamic MEG responses yields models that better predict human similarity judgments and track individual neural dynamics over time.

  2. Artificial Neural Networks for Magnetoencephalography: A review of an emerging field

    q-bio.NC 2025-01 conditional novelty 6.0 of 10

    A systematic review of 119 ANN-MEG studies shows rapid growth across decoding, BCI, clinical, modeling, and source-localization applications, with recurring reproducibility gaps.

  3. WorldWeaver: Generating Long-Horizon Video Worlds via Rich Perception

    cs.CV 2025-08 unverdicted novelty 5.0 of 10

    WorldWeaver reduces temporal drift in long-horizon video generation by jointly modeling RGB and depth perceptual conditions with segmented noise scheduling.

  4. Foundation Models for Cross-Domain EEG Analysis Application: A Survey

    cs.HC 2025-08 conditional novelty 4.0 of 10

    A survey that organizes EEG foundation-model research into five output-modality categories: native EEG, text, vision, audio, and multimodal fusion, with a claim to be the first such comprehensive taxonomy.

Reference graph

Works this paper leans on

39 extracted references · 16 canonical work pages · cited by 4 Pith papers

  1. [1]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sas- try, A. Askell, P. Mishkin, J. Clark, G. Krueger, I. Sutskever, Learning transferable visual models from natu- ral language supervision (2021). arXiv: 2103.00020

    arXiv 2021

  2. [2]

    Naselaris, K

    T. Naselaris, K. N. Kay, S. Nishi- moto, J. L. Gallant, Encoding and decoding in fMRI, NeuroIm- age 56 (2) (2011) 400–410. doi: 10.1016/j.neuroimage.2010.07.073. URL https://linkinghub. elsevier.com/retrieve/pii/ S1053811910010657

    doi:10.1016/j.neuroimage.2010.07.073 2011

  3. [3]

    Zafar, A

    R. Zafar, A. S. Malik, N. Kamel, S. C. Dass, J. M. Abdullah, F. Reza, A. H. Abdul Karim, Decoding of vi- sual information from human brain activity: A review of fMRI and EEG studies, Journal of Integrative Neu- roscience 14 (02) (2015) 155–168. doi:10.1142/S0219635215500089. URL http://www.worldscientific. com/doi/abs/10.1142/ S0219635215500089

    doi:10.1142/s0219635215500089 2015

  4. [4]

    VanRullen, L

    R. VanRullen, L. Reddy, Reconstruct- ing faces from fmri patterns using deep 14 generative neural networks, Commu- nications Biology 2 (1) (2019) 193. doi:10.1038/s42003-019-0438-y . URL https://doi.org/10.1038/ s42003-019-0438-y

    doi:10.1038/s42003-019-0438-y 2019

  5. [5]

    Horikawa, Y

    T. Horikawa, Y. Kamitani, Generic decoding of seen and imagined ob- jects using hierarchical visual features, Nature Communications 8 (1) (2017) 15037. doi:10.1038/ncomms15037. URL http://www.nature.com/ articles/ncomms15037

    doi:10.1038/ncomms15037 2017

  6. [6]

    Takagi, S

    Y. Takagi, S. Nishimoto, High-resolution image reconstruction with latent diffu- sion models from human brain activity, bioRxiv (2023). arXiv:https://www. biorxiv.org/content/early/2023/ 03/11/2022.11.18.517004.full.pdf, doi:10.1101/2022.11.18.517004. URL https://www.biorxiv.org/ content/early/2023/03/11/2022.11. 18.517004

    work page doi:10.1101/2022.11.18.517004 2023

  7. [7]

    Z. Chen, J. Qing, T. Xiang, W. L. Yue, J. H. Zhou, Seeing beyond the brain: Conditional diffusion model with sparse masked modeling for vision decoding (2022). arXiv:2211.06956

    arXiv 2022

  8. [8]

    Ferrante, T

    M. Ferrante, T. Boccato, N. Toschi, Se- mantic brain decoding: from fmri to conceptually similar image reconstruc- tion of visual stimuli (2023). arXiv: 2212.06726

    arXiv 2023

Show all 39 references
  1. [9]

    Ozcelik, R

    F. Ozcelik, R. VanRullen, Brain-diffuser: Natural scene reconstruction from fmri signals using generative latent diffusion (2023). arXiv:2303.05334

    2023 arXiv

  2. [10]

    Ferrante, T

    M. Ferrante, T. Boccato, F. Ozcelik, R. VanRullen, N. Toschi, Multimodal decoding of human brain activity into images and text, in: UniReps: the First Workshop on Unifying Representations in Neural Models, 2023. URL https://openreview.net/ forum?id=rGCabZfV3d

    2023

  3. [11]

    LeBel, L

    A. LeBel, L. Wagner, S. Jain, et al., A natural language fmri dataset for voxelwise encoding models, Scientific Data 10 (2023) 555. doi:10.1038/ s41597-023-02437-z

    2023

  4. [12]

    J. Tang, A. LeBel, S. Jain, et al., Semantic reconstruction of contin- uous language from non-invasive brain recordings, Nature Neu- roscience 26 (2023) 858–866. doi:10.1038/s41593-023-01304-9

    2023 doi

  5. [13]

    Antonello, A

    R. Antonello, A. Vaidya, A. G. Huth, Scaling laws for language encoding mod- els in fmri (2023). arXiv:2305.11863

    2023 arXiv

  6. [14]

    Caucheteux, J

    C. Caucheteux, J. King, Brains and algorithms partially converge in nat- ural language processing, Communica- tions Biology 5 (2022) 134. doi:10. 1038/s42003-022-03036-1

    2022

  7. [15]

    Caucheteux, A

    C. Caucheteux, A. Gramfort, J. King, Evidence of a predictive coding hi- erarchy in the human brain listen- ing to speech, Nature Human Be- 15 haviour 7 (2023) 430–441. doi:10. 1038/s41562-022-01516-2

    2023

  8. [16]

    D´ efossez, C

    A. D´ efossez, C. Caucheteux, J. Rapin, et al., Decoding speech perception from non-invasive brain recordings, Nature Machine In- telligence 5 (2023) 1097–1107. doi:10.1038/s42256-023-00714-5

    2023 doi

  9. [17]

    S. R. Oota, M. Gupta, R. S. Bapi, G. Jobard, F. Alexandre, X. Hinaut, Deep Neural Networks and Brain Align- ment: Brain Encoding and Decoding (Survey), arXiv:2307.10246 [cs, q-bio] (Jul. 2023). URL http://arxiv.org/abs/2307. 10246

    2023 arXiv

  10. [18]

    Conwell, J

    C. Conwell, J. S. Prince, K. N. Kay, G. A. Alvarez, T. Konkle, What can 1.8 billion regressions tell us about the pressures shaping high-level visual representation in brains and machines?, bioRxiv (2023). arXiv:https://www. biorxiv.org/content/early/2023/ 07/01/2022.03.28.485...

    2023 doi

  11. [19]

    A. T. Gifford, B. Lahner, S. Saba- Sadiya, M. G. Vilas, A. Lascelles, A. Oliva, K. Kay, G. Roig, R. M. Ci- chy, The algonauts project 2023 chal- lenge: How the human brain makes sense of natural scenes (2023). arXiv: 2301.03198

    2023 arXiv

  12. [20]

    Adeli, S

    H. Adeli, S. Minni, N. Kriegesko- rte, Predicting brain activity us- ing transformers, bioRxiv (2023). arXiv:https://www.biorxiv. org/content/early/2023/08/05/ 2023.08.02.551743.full.pdf, doi:10.1101/2023.08.02.551743. URL https://www.biorxiv.org/ content/early/2023/08/05/2023...

    2023 doi

  13. [21]

    Nguyen, X

    X.-B. Nguyen, X. Liu, X. Li, K. Luu, The algonauts project 2023 challenge: Uark-ualbany team solution (2023). arXiv:2308.00262

    2023 arXiv

  14. [22]

    H. Yang, J. Gee, J. Shi, Memory Encoding Model, arXiv:2308.01175 [cs] (Aug. 2023). doi:10.48550/arXiv. 2308.01175. URL http://arxiv.org/abs/2308. 01175

    2023 doi arXiv

  15. [23]

    Choksi, M

    B. Choksi, M. Mozafari, R. VanRullen, L. Reddy, Multimodal neural networks better explain multivoxel patterns in the hippocampus, Neural Networks 154 (2022) 538–542. doi:https://doi. org/10.1016/j.neunet.2022.07.033. URL https://www.sciencedirect. com/science/article/pii/ S089...

    2022 doi

  16. [24]

    P. S. Scotti, A. Banerjee, J. Goode, S. Shabalin, A. Nguyen, E. Cohen, A. J. Dempster, N. Verlinde, E. Yundler, D. Weisberg, K. A. Norman, T. M. Abraham, Reconstructing the mind’s eye: fmri-to-image with contrastive 16 learning and diffusion priors (2023). arXiv:2305.18274

    2023 arXiv

  17. [25]

    W. Xia, R. de Charette, C. ¨Oztireli, J.- H. Xue, Dream: Visual decoding from reversing human visual system (2023). arXiv:2310.02265

    2023 arXiv

  18. [26]

    A. K. Robinson, T. Grootswagers, T. A. Carlson, The influence of image masking on object representations during rapid serial visual presentation, Neuroimage 197 (2019) 224–231

    2019

  19. [27]

    A. T. Gifford, K. Dwivedi, G. Roig, R. M. Cichy, A large and rich EEG dataset for modeling human vi- sual object recognition, Neuroimage 264 (119754) (2022) 119754

    2022

  20. [28]

    Palazzo, C

    S. Palazzo, C. Spampinato, J. Schmidt, I. Kavasidis, D. Giordano, M. Shah, Cor- rect block-design experiments mitigate temporal correlation bias in eeg classi- fication (2020). arXiv:2012.03849

    2020 arXiv

  21. [29]

    Spampinato, S

    C. Spampinato, S. Palazzo, I. Kavasidis, D. Giordano, M. Shah, N. Souly, Deep learning human mind for automated vi- sual classification (2019). arXiv:1609. 00344

    2019

  22. [30]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, L. Fei-Fei, Imagenet: A large-scale hierarchical image database, in: 2009 IEEE conference on computer vision and pattern recognition, Ieee, 2009, pp. 248– 255

    2009

  23. [31]

    R. Li, J. S. Johansen, H. Ahmed, T. V. Ilyevsky, R. B. Wilbur, H. M. Bharad- waj, J. M. Siskind, Training on the test set? an analysis of spampinato et al. [31] (2018). arXiv:1812.07697

    2018 arXiv

  24. [32]

    M. N. Hebart, O. Contier, L. Teich- mann, A. H. Rockter, C. Y. Zheng, A. Kidder, A. Corriveau, M. Vaziri- Pashkam, C. I. Baker, Things-data, a multimodal collection of large-scale datasets for investigating object rep- resentations in human brain and behavior, eLife 12 (2023) ...

    2023 doi

  25. [33]

    M. N. Hebart, A. H. Dickter, A. Kid- der, W. Y. Kwok, A. Corriveau, C. Van Wicklin, C. I. Baker, THINGS: A database of 1,854 object concepts and more than 26,000 naturalistic object im- ages, PLoS One 14 (10) (2019) e0223792

    2019

  26. [34]

    E. J. Allen, G. St-Yves, Y. Wu, J. L. Breedlove, J. S. Prince, L. T. Dow- dle, M. Nau, B. Caron, F. Pestilli, I. Charest, J. B. Hutchinson, T. Nase- laris, K. Kay, A massive 7t fmri dataset to bridge cognitive neuroscience and artificial intelligence, Nature Neuroscience 25 (1...

    2022 doi

  27. [35]

    Y. Bai, X. Wang, Y. pei Cao, Y. Ge, C. Yuan, Y. Shan, Dreamdiffusion: Gen- 17 erating high-quality images from brain eeg signals (2023). arXiv:2306.16934

    2023 arXiv

  28. [36]

    Palazzo, C

    S. Palazzo, C. Spampinato, I. Kavasidis, D. Giordano, M. Shah, Generative ad- versarial networks conditioned by brain signals, 2017, pp. 3430–3438. doi:10. 1109/ICCV.2017.369

    2017

  29. [37]

    Kavasidis, S

    I. Kavasidis, S. Palazzo, C. Spampinato, D. Giordano, M. Shah, Brain2image: Converting brain signals into im- ages, in: Proceedings of the 25th ACM International Conference on Multimedia, MM ’17, Association for Computing Machinery, New York, NY, USA, 2017, p. 1809–1817. doi:1...

    2017

  30. [38]

    Ferrante, T

    M. Ferrante, T. Boccato, S. Bargione, N. Toschi, Decoding visual brain repre- sentations from electroencephalography through knowledge distillation and la- tent diffusion models (2023). arXiv: 2309.07149

    2023 arXiv

  31. [39]

    Benchetrit, H

    Y. Benchetrit, H. Banville, J.-R. King, Brain decoding: toward real-time re- construction of visual perception (2023). arXiv:2310.19812. 18

    2023 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.