Pith. sign in

REVIEW 2 major objections 4 minor 67 references

SpectralOT aligns whole-brain fMRI across people by mixing functional signal with three Laplace-Beltrami eigenmodes, improving cross-subject decoding while running far faster than prior optimal-transport methods.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

SpectralOT regularizes entropic optimal transport with the first three Laplace-Beltrami eigenmodes of cortical geometry to produce fast, parsimonious whole-brain functional alignments that improve cross-subject decoding.

T0 review reviewed 2026-07-14 challenge →

load-bearing objection Clean, fast geometry-aware OT alignment that actually improves ISC and out-of-subject decoding while being ~30 imes cheaper than FUGW. the 2 major comments →

arxiv 2607.10931 v1 pith:AKXTB7BM submitted 2026-07-12 q-bio.NC cs.LGstat.ML

Fast Whole-Brain, Geometry-Aware Functional Alignment for Cross-Subject Decoding

classification q-bio.NC cs.LGstat.ML
keywords functional alignmentfMRI decodingoptimal transportLaplace-Beltrami eigenmodescross-subject generalizationcortical geometrySpectralOT
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

Brain-activity decoders fail to generalize across people because both anatomy and functional response patterns differ. SpectralOT solves this by building a single composite cost that linearly interpolates functional dissimilarity with geometric dissimilarity derived from the first three Laplace-Beltrami eigenmodes of each cortical mesh. An entropic optimal-transport solver then yields a soft vertex-to-vertex map that can be applied to new data. The method improves inter-subject correlation and out-of-subject decoding accuracy relative to pure anatomical registration and two strong baselines, while requiring only one Sinkhorn pass and running roughly thirty times faster than the previous state-of-the-art. The practical payoff is that population-level decoders become feasible on modest datasets without sacrificing anatomical fidelity.

Core claim

Embedding cortical geometry via the first three Laplace-Beltrami eigenmodes into a linearly weighted functional-geometric cost, then solving a single entropic optimal-transport problem, produces whole-brain alignments that raise both inter-subject correlation and cross-subject decoding accuracy while remaining orders of magnitude faster than fused Gromov-Wasserstein approaches.

What carries the argument

SpectralOT: the composite cost C = (1-α)C_func + α C_geom, where C_geom is the squared Euclidean distance between the first three sign-aligned Laplace-Beltrami eigenmodes of the source and target meshes; this cost is fed once to a Sinkhorn solver to obtain the soft correspondence matrix used for signal transfer.

Load-bearing premise

The first three Laplace-Beltrami eigenmodes, after a simple gradient-based sign correction, already capture enough geometry to keep distant brain regions from matching, provided the two meshes are roughly isometric and similarly oriented.

What would settle it

On a held-out multi-subject fMRI dataset, replace the three-eigenmode geometric cost with pure functional cost (α=0) or pure Euclidean distance; if cross-subject decoding accuracy then falls below the anatomical baseline while SpectralOT stays above it, the geometric embedding is doing essential work; if accuracy stays the same, the claim collapses.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Population-level decoders can be trained after a single, fast alignment step rather than after expensive pairwise Gromov-Wasserstein solves.
  • Because the geometric prior is mesh-intrinsic, alignments can be computed between individual cortical surfaces without first warping them to a common template.
  • The linear α parameter and single Sinkhorn pass make nested cross-validation and integration into deep-learning pipelines practical.
  • The same eigenmode cost extends immediately to volumetric data, opening whole-brain (surface-plus-volume) functional templates.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method supplies a natural building block for a multi-subject functional template: each new subject can be aligned once to a growing average rather than to every other subject.
  • Because the transport plan is differentiable with respect to the cost, SpectralOT can be inserted as a layer inside an end-to-end neural decoder that jointly optimizes alignment and classification.
  • If the three-mode geometric regularizer proves sufficient, many other surface-matching problems outside neuroimaging can replace expensive geodesic distances with the same cheap spectral embedding.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The manuscript introduces SpectralOT, a whole-brain functional alignment method for fMRI that constructs a composite cost matrix as a convex combination of a functional L2 dissimilarity and a geometric cost derived from the first three Laplace-Beltrami eigenmodes of the cortical mesh (after a simple gradient-based sign-flip correction). The composite cost is fed to a single entropic Sinkhorn solver to obtain a soft vertex-to-vertex coupling. Four experiments of increasing complexity (purely anatomical color transfer between fsaverage5 and fsLR, ISC on IBC task contrasts, pairwise out-of-subject decoding on Courtois-Neuromod THINGS, and leave-one-subject-out group decoding on IBC RSVPLanguage) compare SpectralOT against FUGW, ProMises and anatomical registration. The central claim is that SpectralOT improves ISC and out-of-subject decoding accuracy while remaining approximately 30 imes faster than FUGW and easier to tune.

Significance. If the reported gains hold, SpectralOT supplies a practical, geometry-aware alternative to FUGW that removes the need for nested block-coordinate descent and inhomogeneous scaling of Wasserstein versus Gromov-Wasserstein terms. The single free geometric weight α, the linear composite cost, and the public implementation make the method immediately usable for population-scale decoding pipelines and for integration into differentiable deep-learning frameworks. The explicit comparison of blur (Appendix Figure 8), the Nadeau-Bengio-corrected group-level test, and the open code are strengths that raise the bar for subsequent functional-alignment papers.

major comments (2)
  1. Table 1 (THINGS pairwise decoding) shows SpectralOT winning on average, yet two of the six source→target pairs fall below the anatomical baseline and the absolute accuracies remain modest (0.14 average). With only three subjects the pairwise design cannot support a claim of consistent domain-shift reduction; the group-level IBC experiment (Figure 5) is therefore load-bearing, yet it reports only a non-significant difference versus FUGW. A power analysis or bootstrap confidence intervals on the accuracy differences would clarify whether the claimed superiority over FUGW is supported or whether the methods are statistically equivalent.
  2. Methods, Geometric Descriptors and Appendix Figure 7: the decision to truncate to the first three eigenmodes is justified solely by geodesic-error saturation under pure anatomical alignment (α=1). It remains untested whether the same truncation remains optimal once functional cost is present (0<α<1) or on individual (non-template) meshes whose higher-frequency geometry may matter for fine functional topography. A short ablation of k under the ISC or decoding protocols would close this gap.
minor comments (4)
  1. Equation (5) and the subsequent simplification assume equal-mass marginals; the text should state explicitly that the Sinkhorn solver is always initialized with uniform marginals, otherwise the row-normalization step is required.
  2. Figure 4B reports wall-clock times on a single GPU; stating the number of vertices per hemisphere and whether the cost matrices are pre-computed or recomputed would make the 30× claim fully reproducible.
  3. The ProMises low-rank critique in the Appendix is clear, yet the main text still includes ProMises in every comparison; a single sentence noting that the model is retained only for completeness would avoid reader confusion.
  4. Typographical inconsistencies appear throughout (e.g., “computationalefficiency”, “thepredictive”, missing spaces after periods). A careful copy-edit pass is needed.

Circularity Check

0 steps flagged

No circularity: SpectralOT construction, hyper-parameters and held-out evaluations are independent of the claimed ISC/decoding gains.

full rationale

The paper defines a composite cost C_composite = (1-α)C_func + α C_geom from LBO eigenmodes (truncated to k=3 after gradient sign-flip) and paired functional maps, then obtains the coupling via a single Sinkhorn solve (Eqs. 2–5). All performance claims (ISC curves, Table 1 pairwise decoding accuracies, Figure 5 group-level accuracies) are measured on held-out contrasts or held-out subjects after the coupling is fixed; α and ε are free parameters that are either fixed a priori or swept, never fitted to the final decoding scores. Self-citations (Thual et al. 2022/2023, Barbarant et al. 2025 poster) supply only the FUGW baseline and a future-work remark; they do not underwrite any uniqueness claim or close a logical loop. The low-rank critique of ProMises is derived directly from the SVD construction in the appendix and is not circular. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 1 invented entities

The central claim rests on standard spectral geometry and OT theory plus a small set of free hyper-parameters and the modeling choice that three eigenmodes suffice. No new physical entities are postulated; SpectralOT is an algorithmic construct.

free parameters (3)
  • alpha (geometric weight) = 0.5 (decoding); grid-searched for ISC
    Linear interpolation weight between functional and geometric costs; set by grid search or fixed at 0.5 in decoding experiments; directly controls the claimed balance.
  • epsilon (entropic regularization) = 1e-3
    Controls blur of the transport plan; set to 1e-3 for SpectralOT (1e-4 for FUGW) to match effective smoothing; chosen by hand rather than nested CV.
  • number of eigenmodes k = 3
    Truncation of the Laplace-Beltrami basis; fixed at 3 after sensitivity analysis showing diminishing returns.
axioms (3)
  • domain assumption Laplace-Beltrami eigenmodes form an intrinsic, isometry-invariant embedding of mesh geometry (Rustamov 2007).
    Invoked to justify C_geom as a geometry-preserving cost; standard in geometry processing.
  • domain assumption Cortical meshes of different subjects are approximately isometric and share a common orientation after standard surface registration.
    Required for the three-mode embedding and the gradient-based sign-flip procedure to be meaningful.
  • standard math Entropic OT with the composite cost yields a soft correspondence that can be used for barycentric projection of new functional data.
    Standard Sinkhorn theory (Cuturi 2013; Peyré & Cuturi 2018).
invented entities (1)
  • SpectralOT composite cost and pipeline no independent evidence
    purpose: To obtain a single-parameter, single-pass geometry-aware functional alignment.
    The linear mixture of functional L2 and three-mode spectral costs plus the sign-flip heuristic is the paper’s algorithmic contribution; no independent physical existence is claimed.

reviewed 2026-07-14 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Fast Whole-Brain, Geometry-Aware Functional Alignment for Cross-Subject Decoding." pith.science (2026). https://pith.science/paper/AKXTB7BM

@misc{pith2026260710931,
  author       = {Pith},
  title        = {Pith review of: Fast Whole-Brain, Geometry-Aware Functional Alignment for Cross-Subject Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AKXTB7BM}},
  note         = {Machine review of arXiv:2607.10931}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Decoding brain activity is useful for characterizing brain processes and understanding the functional architecture underlying cognition. However, the inter-individual variability in brain response patterns limits the development of decoders that generalize across individuals. A solution to this challenge is functional alignment: aligning functional data across individuals before training population-level decoders. The core issue is to strike the balance between aligning functional features and preserving the anatomical structure, while maintaining computational efficiency. We introduce a new functional alignment method for fMRI, SpectralOT, that embeds cortical geometry into Laplace-Beltrami eigenmodes along functional data to regularize the alignment.

Figures

Figures reproduced from arXiv: 2607.10931 by Bertrand Thirion, Florent Meyniel, Pierre-Louis Barbarant.

Figure 1
Figure 1. Figure 1: Aligning spectral bases across individu￾als. By definition, eigenmodes of the Laplace-Beltrami operator are defined up to their sign. A. This phe￾nomenon may lead to a "sign flip", where corresponding eigenmodes inconsistently describe intrinsic geometries across meshes. B. We rely on a simple procedure to derive aligned bases. For each pair of eigenmodes, we compute the gradient of values in each mesh 3D … view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the SpectralOT method. Our procedure consists of three distinct steps. For each hemisphere: A. We first compute the first three eigen￾modes of the Laplace-Beltrami operator on both source and target mesh and compute the vertex-wise squared L 2 distance across the eigenmodes to derive the ge￾ometric cost matrix Cgeom. B. We then compute the vertex-wise squared L 2 distance between the source and… view at source ↗
Figure 3
Figure 3. Figure 3: Anatomical alignment across different meshes (𝛼 = 1). We compare our method to Gromov￾Wasserstein (GW) solvers in their ability to derive purely anatomical correspondence between two different meshes. We set both OT solvers entropic regularization 𝜀 to 10−4 , with 1000 iterations of Sinkhorn’s algorithm and 10 steps of block-coordinate descent for GW. A. We use left hemispheres of fsaverage5 and fsLR surfa… view at source ↗
Figure 4
Figure 4. Figure 4: Inter-subject alignment on task data. We align IBC dataset’s subject #4 on subject #9. We use paired samples from the Archi task battery to derive alignment transformation. We map left-out testing data from the MathLanguage protocol between subjects and compare the aligned data to the target’s subject test data. A. We report the Pearson correlation between each pair of samples in the aligned and target sub… view at source ↗
Figure 5
Figure 5. Figure 5: Decoding performance on the IBC dataset. Inter-subject decoding accuracy for FUGW, SpectralOT, and the anatomical baseline. Each colored dot rep￾resents classifier accuracy for a left-out subject, aver￾aged across all cross-validation folds. Solid lines con￾nect subjects across methods. The dotted gray line in￾dicates chance level. Statistical comparisons between all method pairs are performed using two-sa… view at source ↗
Figure 6
Figure 6. Figure 6: Anatomical alignment across different meshes (𝛼 = 1, right hemispheres). Replication of experi￾ment 1 on the right hemisphere. We set both OT solvers entropic regularization 𝜀 to 10−4 , with 1000 iterations of Sinkhorn’s algorithm and 10 steps of block-coordinate descent for GW. A. We use right hemispheres of fsaverage5 and fsLR surface templates as source and target respectively. B. Color transfer of the … view at source ↗
Figure 7
Figure 7. Figure 7: Influence of the number of eigenmodes on geodesic error. We evaluate alignment accuracy using the geodesic error metric, defined as the vertex displacement from Thual et al., 2022: for each target vertex j, the displacement is computed as ∑︀ i P s→t i,j Di,j/ ∑︀ i P s→t i,j , where D is the geodesic distance matrix on the mesh. Geodesic error is obtained by targeting the identity transformation (i.e., usin… view at source ↗
Figure 8
Figure 8. Figure 8: Effects of entropic regularization on transformation blurriness. We evaluate transformation blurriness using the vertex spread criterion from Thual et al., 2022: for each source vertex i, q pairs of indices (jq, kq) are sampled from the row-normalized transport plan P˜ i , and the spread is computed as the average pairwise geodesic distance 1 q ∑︀ jq ,kq Djq ,kq (lower is better). We recover the identity t… view at source ↗
Figure 9
Figure 9. Figure 9: Inter-subject functional alignment on task data. We replicate experiment 2 on all pairs of subjects from the MathLanguage task, using paired samples from the Archi task battery to derive the alignment transformation. Left-out MathLanguage test data are mapped from each source subject to each target subject, and alignment quality is assessed by the Pearson correlation between aligned source and target test … view at source ↗
Figure 10
Figure 10. Figure 10: Detailed inter-subject correlations for experiment 2. Difference in Pearson correlation relative to the anatomical baseline, shown for every increasing source/target subject pair in experiment 2, for both ProMises and SpectralOT. Each entry is averaged across pairs of samples, and warmer colors indicate a larger improvement over baseline. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

67 extracted references · 19 canonical work pages

  1. [1]

    An empirical evaluation of functional alignment using inter-subject decoding , volume =

    Bazeille, Thomas and DuPre, Elizabeth and Richard, Hugo and Poline, Jean-Baptiste and Thirion, Bertrand , year =. An empirical evaluation of functional alignment using inter-subject decoding , volume =. doi:10.1016/j.neuroimage.2021.118683 , journal =

  2. [2]

    Integrating anatomical and functional landmarks for interparticipant alignment of imaging data , volume =

    Jeganathan, Jayson and Paton, Bryan and Koussis, Nikitas and Breakspear, Michael , year =. Integrating anatomical and functional landmarks for interparticipant alignment of imaging data , volume =. doi:10.1162/imag_a_00253 , journal =

  3. [3]

    2025 , copyright =

    MindAligner: Explicit Brain Functional Alignment for Cross-Subject Visual Decoding from Limited fMRI Data , publisher =. 2025 , copyright =. doi:10.48550/ARXIV.2502.05034 , author =

  4. [4]

    Haxby and J

    James V. Haxby and J. Swaroop Guntupalli and Andrew C. Connolly and Yaroslav O. Halchenko and Bryan R. Conroy and M. Ida Gobbini and Michael Hanke and Peter J. Ramadge , title =. doi:10.1016/j.neuron.2011.08.026 , year =

  5. [5]

    Bazeille and H

    T. Bazeille and H. Richard and H. Janati and B. Thirion , title =. doi:10.1007/978-3-030-20351-1_18 , year =

  6. [6]

    A Reduced-Dimension fMRI Shared Response Model , volume =

    Chen, Po-Hsuan (Cameron) and Chen, Janice and Yeshurun, Yaara and Hasson, Uri and Haxby, James and Ramadge, Peter J , booktitle =. A Reduced-Dimension fMRI Shared Response Model , volume =

  7. [7]

    Advances in Neural Information Processing Systems , editor=

    Aligning individual brains with fused unbalanced Gromov Wasserstein , author=. Advances in Neural Information Processing Systems , editor=

  8. [8]

    Aligning human subjects with short acquisition-time fMRI training data , DOI =

    Thual, Alexis and Dehaene, Stanislas and Thirion, Bertrand , year =. Aligning human subjects with short acquisition-time fMRI training data , DOI =. 2022 Conference on Cognitive Computational Neuroscience , publisher =

  9. [9]

    and Slipski, Lukas and Feilong, Ma and Guntupalli, J

    Busch, Erica L. and Slipski, Lukas and Feilong, Ma and Guntupalli, J. Swaroop and Castello, Matteo Visconti di Oleggio and Huckins, Jeremy F. and Nastase, Samuel A. and Gobbini, M. Ida and Wager, Tor D. and Haxby, James V. , year =. Hybrid hyperalignment: A single high-dimensional model of shared information embedded in cortical patterns of response and f...

  10. [10]

    Measuring shared responses across subjects using intersubject correlation , ISSN =

    Nastase, Samuel A and Gazzola, Valeria and Hasson, Uri and Keysers, Christian , year =. Measuring shared responses across subjects using intersubject correlation , ISSN =. doi:10.1093/scan/nsz037 , journal =

  11. [11]

    Procrustes Analysis for High-Dimensional Data , volume =

    Andreella, Angela and Finos, Livio , year =. Procrustes Analysis for High-Dimensional Data , volume =. Psychometrika , publisher =. doi:10.1007/s11336-022-09859-5 , number =

  12. [12]

    ArXiv , year=

    Aligning brain functions boosts the decoding of visual semantics in novel subjects , author=. ArXiv , year=

  13. [13]

    Local-Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI , volume =

    Schaefer, Alexander and Kong, Ru and Gordon, Evan M and Laumann, Timothy O and Zuo, Xi-Nian and Holmes, Avram J and Eickhoff, Simon B and Yeo, B T Thomas , year =. Local-Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI , volume =. Cerebral Cortex , publisher =. doi:10.1093/cercor/bhx179 , number =

  14. [14]

    Sinkhorn Distances: Lightspeed Computation of Optimal Transport , volume =

    Cuturi, Marco , booktitle =. Sinkhorn Distances: Lightspeed Computation of Optimal Transport , volume =

  15. [15]

    Gower, J. C. , year =. Generalized Procrustes Analysis , volume =. Psychometrika , publisher =. doi:10.1007/bf02291478 , number =

  16. [16]

    Barycenters in the Wasserstein Space , volume =

    Agueh, Martial and Carlier, Guillaume , year =. Barycenters in the Wasserstein Space , volume =. SIAM Journal on Mathematical Analysis , publisher =. doi:10.1137/100805741 , number =

  17. [17]

    Scientific Data , author =

    Individual. Scientific Data , author =. 2018 , pages =. doi:10.1038/sdata.2018.105 , number =

  18. [18]

    Individual Brain Charting dataset extension, third release for movie watching and retinotopy data , volume =

    Pinho, Ana Luísa and Richard, Hugo and Ponce, Ana Fernanda and Eickenberg, Michael and Amadon, Alexis and Dohmatob, Elvis and Denghien, Isabelle and Torre, Juan Jesús and Shankar, Swetha and Aggarwal, Himanshu and Thual, Alexis and Chapalain, Thomas and Ginisty, Chantal and Becuwe-Desmidt, Séverine and Roger, Séverine and Lecomte, Yann and Berland, Valéri...

  19. [19]

    High-resolution 7-Tesla fMRI data on the perception of musical genres – an extension to the studyforrest dataset , volume =

    Hanke, Michael and Dinga, Richard and H\". High-resolution 7-Tesla fMRI data on the perception of musical genres – an extension to the studyforrest dataset , volume =. 2015 , month = jun, pages =. doi:10.12688/f1000research.6679.1 , journal =

  20. [20]

    and Bellec, Lune and Hebart, Martin N

    St-Laurent, Marie and Pinsard, Basile and Contier, Oliver and DuPre, Elizabeth and Seeliger, Katja and Borghesani, Valentina and Boyle, Julie A. and Bellec, Lune and Hebart, Martin N. , keywords =. CNeuroMod-THINGS, a densely-sampled fMRI dataset for visual neuroscience , publisher =. 2025 , copyright =. doi:10.48550/ARXIV.2507.09024 , url =

  21. [21]

    and Blair, Ross W

    Esteban, Oscar and Markiewicz, Christopher J. and Blair, Ross W. and Moodie, Craig A. and Isik, A. Ilkay and Erramuzpe, Asier and Kent, James D. and Goncalves, Mathias and DuPre, Elizabeth and Snyder, Madeleine and Oya, Hiroyuki and Ghosh, Satrajit S. and Wright, Jessey and Durnez, Joke and Poldrack, Russell A. and Gorgolewski, Krzysztof J. , year =. fMRI...

  22. [22]

    Unbiased nonlinear average age-appropriate brain templates from birth to adulthood , volume =

    Fonov, VS and Evans, AC and McKinstry, RC and Almli, CR and Collins, DL , year =. Unbiased nonlinear average age-appropriate brain templates from birth to adulthood , volume =. doi:10.1016/s1053-8119(09)70884-5 , journal =

  23. [23]

    2013 , MONTH = Dec, DOI =

    Abraham, Alexandre and Pedregosa, Fabian and Eickenberg, Michael and Gervais, Philippe and Muller, Andreas and Kossaifi, Jean and Gramfort, Alexandre and Thirion, Bertrand and Varoquaux, Ga. 2013 , MONTH = Dec, DOI =

  24. [24]

    Nilearn contributors , license =

  25. [25]

    and Cattell, Raymond B

    Hurley, John R. and Cattell, Raymond B. , title =. Behavioral Science , volume =. doi:https://doi.org/10.1002/bs.3830070216 , year =

  26. [26]

    Effective functional mapping of fMRI data with support‐vector machines , volume =

    Lee, Sangkyun and Halder, Sebastian and K\". Effective functional mapping of fMRI data with support‐vector machines , volume =. Human Brain Mapping , publisher =. 2010 , month = jan, pages =. doi:10.1002/hbm.20955 , number =

  27. [27]

    and Singer, Benjamin D

    Sabuncu, Mert R. and Singer, Benjamin D. and Conroy, Bryan and Bryan, Ronald E. and Ramadge, Peter J. and Haxby, James V. , year =. Function-based Intersubject Alignment of Human Cortical Anatomy , volume =. Cerebral Cortex , publisher =. doi:10.1093/cercor/bhp085 , number =

  28. [28]

    and Gobbini, M

    Haxby, James V. and Gobbini, M. Ida and Furey, Maura L. and Ishai, Alumit and Schouten, Jennifer L. and Pietrini, Pietro , year =. Distributed and Overlapping Representations of Faces and Objects in Ventral Temporal Cortex , volume =. Science , publisher =. doi:10.1126/science.1063736 , number =

  29. [29]

    FastSRM: A fast, memory efficient and identifiable implementation of the shared response model , volume =

    Richard, Hugo and Thirion, Bertrand , year =. FastSRM: A fast, memory efficient and identifiable implementation of the shared response model , volume =. doi:10.52294/001c.87954 , journal =

  30. [30]

    and Tootell, Roger B.H

    Fischl, Bruce and Sereno, Martin I. and Tootell, Roger B.H. and Dale, Anders M. , title =. Human Brain Mapping , volume =. doi:https://doi.org/10.1002/(SICI)1097-0193(1999)8:4<272::AID-HBM10>3.0.CO;2-4 , year =

  31. [31]

    doi:10.1073/pnas.0600244103 , year =

    Nikolaus Kriegeskorte and Rainer Goebel and Peter Bandettini , title =. doi:10.1073/pnas.0600244103 , year =

  32. [32]

    Swaroop and Feilong, Ma and Haxby, James V

    Guntupalli, J. Swaroop and Feilong, Ma and Haxby, James V. , editor =. A computational model of shared fine-scale structure in the human connectome , volume =. PLOS Computational Biology , publisher =. 2018 , month = apr, pages =. doi:10.1371/journal.pcbi.1006120 , number =

  33. [33]

    2024 , copyright =

    MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data , publisher =. 2024 , copyright =. doi:10.48550/ARXIV.2403.11207 , author =

  34. [34]

    and Varoquaux, G

    Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V. and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P. and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E. , journal=. Scikit-learn: Machine Learning in

  35. [35]

    and Jiahui, Guo and Halchenko, Yaroslav O

    Feilong, Ma and Nastase, Samuel A. and Jiahui, Guo and Halchenko, Yaroslav O. and Gobbini, M. Ida and Haxby, James V. , year =. The individualized neural tuning model: Precise and generalizable cartography of functional architecture in individual brains , volume =. doi:10.1162/imag_a_00032 , journal =

  36. [36]

    Inference for the Generalization Error , volume =

    Nadeau, Claude and Bengio, Yoshua , booktitle =. Inference for the Generalization Error , volume =

  37. [37]

    Swaroop and Chauhan, Vassiki and Haxby, James V

    Jiahui, Guo and Feilong, Ma and Visconti di Oleggio Castello, Matteo and Guntupalli, J. Swaroop and Chauhan, Vassiki and Haxby, James V. and Gobbini, M. Ida , year =. Predicting individual face-selective topography using naturalistic stimuli , volume =. doi:10.1016/j.neuroimage.2019.116458 , journal =

  38. [38]

    Swaroop and Hanke, Michael and Halchenko, Yaroslav O

    Guntupalli, J. Swaroop and Hanke, Michael and Halchenko, Yaroslav O. and Connolly, Andrew C. and Ramadge, Peter J. and Haxby, James V. , year =. A Model of Representational Spaces in Human Cortex , volume =. Cerebral Cortex , publisher =. doi:10.1093/cercor/bhw068 , number =

  39. [39]

    and Dickter, Adam H

    Hebart, Martin N. and Dickter, Adam H. and Kidder, Alexis and Kwok, Wan Y. and Corriveau, Anna and Van Wicklin, Caitlin and Baker, Chris I. , editor =. THINGS: A database of 1, 854 object concepts and more than 26, 000 naturalistic object images , volume =. PLOS ONE , publisher =. 2019 , month = oct, pages =. doi:10.1371/journal.pone.0223792 , number =

  40. [40]

    Through their eyes: Multi-subject brain decoding with simple alignment techniques , volume =

    Ferrante, Matteo and Boccato, Tommaso and Ozcelik, Furkan and VanRullen, Rufin and Toschi, Nicola , year =. Through their eyes: Multi-subject brain decoding with simple alignment techniques , volume =. doi:10.1162/imag_a_00170 , journal =

  41. [41]

    and Naselaris, Thomas and Benjamini, Yuval and Yu, Bin and Gallant, Jack L

    Nishimoto, Shinji and Vu, An T. and Naselaris, Thomas and Benjamini, Yuval and Yu, Bin and Gallant, Jack L. , year =. Reconstructing Visual Experiences from Brain Activity Evoked by Natural Movies , volume =. Current Biology , publisher =. doi:10.1016/j.cub.2011.08.031 , number =

  42. [42]

    and Brincat, Scott L

    Barbosa, Joao and Nejatbakhsh, Amin and Duong, Lyndon and Harvey, Sarah E. and Brincat, Scott L. and Siegel, Markus and Miller, Earl K. and Williams, Alex H. , year =. Quantifying Differences in Neural Population Activity With Shape Metrics , url =. doi:10.1101/2025.01.10.632411 , publisher =

  43. [43]

    The representational dynamics of the animal appearance bias in human visual cortex are indicative of fast feedforward processing

    Chiu-Yueh Chen and Ga \"e lle Leys and Stefania Bracci and Op de Beeck, Hans. The representational dynamics of the animal appearance bias in human visual cortex are indicative of fast feedforward processing. Imaging Neuroscience. 2023

  44. [44]

    and Nishimoto, Shinji and Gallant, Jack L

    Naselaris, Thomas and Kay, Kendrick N. and Nishimoto, Shinji and Gallant, Jack L. , year =. Encoding and decoding in fMRI , volume =. NeuroImage , publisher =. doi:10.1016/j.neuroimage.2010.07.073 , number =

  45. [45]

    Individual Comparisons by Ranking Methods , volume =

    Wilcoxon, Frank , year =. Individual Comparisons by Ranking Methods , volume =. Biometrics Bulletin , publisher =. doi:10.2307/3001968 , number =

  46. [46]

    Hyperalignment: Modeling shared information encoded in idiosyncratic cortical topographies , volume =

    Haxby, James V and Guntupalli, J Swaroop and Nastase, Samuel A and Feilong, Ma , year =. Hyperalignment: Modeling shared information encoded in idiosyncratic cortical topographies , volume =. doi:10.7554/elife.56601 , journal =

  47. [47]

    and Coalson, Timothy S

    Glasser, Matthew F. and Coalson, Timothy S. and Robinson, Emma C. and Hacker, Carl D. and Harwell, John and Yacoub, Essa and Ugurbil, Kamil and Andersson, Jesper and Beckmann, Christian F. and Jenkinson, Mark and Smith, Stephen M. and Van Essen, David C. , year =. A multi-modal parcellation of human cerebral cortex , volume =. Nature , publisher =. doi:10...

  48. [48]

    Van Essen, D. C. and Glasser, M. F. and Dierker, D. L. and Harwell, J. and Coalson, T. , year =. Parcellations and Hemispheric Asymmetries of Human Cerebral Cortex Analyzed on Surface-Based Atlases , volume =. Cerebral Cortex , publisher =. doi:10.1093/cercor/bhr291 , number =

  49. [49]

    A sulcal depth-based anatomical parcellation of the cerebral cortex

    Destrieux, C and Fischl, B and Dale, AM and Halgren, E , year =. A sulcal depth-based anatomical parcellation of the cerebral cortex. , volume =. doi:10.1016/s1053-8119(09)71561-7 , journal =

  50. [50]

    ArXiv , year=

    A Survey on fMRI-based Brain Decoding for Reconstructing Multimodal Stimuli , author=. ArXiv , year=

  51. [51]

    A Primer on Pattern-Based Approaches to fMRI: Principles, Pitfalls, and Perspectives , volume =

    Haynes, John-Dylan , year =. A Primer on Pattern-Based Approaches to fMRI: Principles, Pitfalls, and Perspectives , volume =. Neuron , publisher =. doi:10.1016/j.neuron.2015.05.025 , number =

  52. [52]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Pai, Gautam and Ren, Jing and Melzi, Simone and Wonka, Peter and Ovsjanikov, Maks , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2021 , pages =

  53. [53]

    Functional maps: a flexible representation of maps between shapes , volume =

    Ovsjanikov, Maks and Ben-Chen, Mirela and Solomon, Justin and Butscher, Adrian and Guibas, Leonidas , year =. Functional maps: a flexible representation of maps between shapes , volume =. ACM Transactions on Graphics , publisher =. doi:10.1145/2185520.2185526 , number =

  54. [54]

    , title =

    Rustamov, Raif M. , title =. Proceedings of the Fifth Eurographics Symposium on Geometry Processing , pages =. 2007 , isbn =

  55. [55]

    and Aquino, Kevin M

    Pang, James C. and Aquino, Kevin M. and Oldehinkel, Marianne and Robinson, Peter A. and Fulcher, Ben D. and Breakspear, Michael and Fornito, Alex , year =. Geometric constraints on human brain function , volume =. Nature , publisher =. doi:10.1038/s41586-023-06098-1 , number =

  56. [56]

    and Bronstein, M

    Kovnatsky, A. and Bronstein, M. M. and Bronstein, A. M. and Glashoff, K. and Kimmel, R. , year =. Coupled quasi‐harmonic bases , volume =. Computer Graphics Forum , publisher =. doi:10.1111/cgf.12064 , number =

  57. [57]

    Progressive Entropic Optimal Transport Solvers , booktitle =

    Parnian Kassraie and Aram-Alexandre Pooladian and Michal Klein and James Thornton and Jonathan Niles-Weed and Marco Cuturi , year =. Progressive Entropic Optimal Transport Solvers , booktitle =

  58. [58]

    PyTorch: an imperative style, high-performance deep learning library , year =

    Paszke, Adam and Gross, Sam and Massa, Francisco and Lerer, Adam and Bradbury, James and Chanan, Gregory and Killeen, Trevor and Lin, Zeming and Gimelshein, Natalia and Antiga, Luca and Desmaison, Alban and K\". PyTorch: an imperative style, high-performance deep learning library , year =. Proceedings of the 33rd International Conference on Neural Informa...

  59. [59]

    and Jbabdi, Saad and Glasser, Matthew F

    Robinson, Emma C. and Jbabdi, Saad and Glasser, Matthew F. and Andersson, Jesper and Burgess, Gregory C. and Harms, Michael P. and Smith, Stephen M. and Van Essen, David C. and Jenkinson, Mark , year =. MSM: A new flexible framework for Multimodal Surface Matching , volume =. doi:10.1016/j.neuroimage.2014.05.069 , journal =

  60. [60]

    and Dale, Anders M

    Fischl, Bruce and Sereno, Martin I. and Dale, Anders M. , year =. Cortical Surface-Based Analysis , volume =. NeuroImage , publisher =. doi:10.1006/nimg.1998.0396 , number =

  61. [61]

    Mazdak and Hoopes, Andrew and Hoffmann, Malte and Magnet, Robin and Ovsjanikov, Maks and Z\"

    Abulnaga, S. Mazdak and Hoopes, Andrew and Hoffmann, Malte and Magnet, Robin and Ovsjanikov, Maks and Z\". Unified Brain Surface and Volume Registration , publisher =. 2025 , copyright =. doi:10.48550/ARXIV.2512.19928 , url =

  62. [62]

    and Kietzmann, Tim C

    Lu, Zejin and Doerig, Adrien and Bosch, Victoria and Krahmer, Bas and Kaiser, Daniel and Cichy, Radoslaw M. and Kietzmann, Tim C. , title=. Nature Human Behaviour , year=. doi:10.1038/s41562-025-02220-7 , url=

  63. [63]

    POT: Python Optimal Transport , journal =

    R. POT: Python Optimal Transport , journal =. 2021 , volume =

  64. [64]

    Computational Optimal Transport , publisher =

    Peyré, Gabriel and Cuturi, Marco , keywords =. Computational Optimal Transport , publisher =. 2018 , copyright =. doi:10.48550/ARXIV.1803.00567 , url =

  65. [65]

    Computer Graphics Forum (SGP) , volume=

    Nicholas Sharp and Keenan Crane , title=. Computer Graphics Forum (SGP) , volume=

  66. [66]

    2025 , MONTH = Aug, KEYWORDS =

    Barbarant, Pierre-Louis and Meyniel, Florent and Thirion, Bertrand , URL =. 2025 , MONTH = Aug, KEYWORDS =

  67. [67]

    Differentiable Ranking and Sorting using Optimal Transport , url =

    Cuturi, Marco and Teboul, Olivier and Vert, Jean-Philippe , booktitle =. Differentiable Ranking and Sorting using Optimal Transport , url =

This paper was first reviewed by grok-4.5 on July 14, 2026.