Feilong Ma

feilonglab.github.ioFeilong Lab @ UofSC — Feilong Lab

7 Nov 2025

🎓Recruiting Ph.D. students for Fall 2026!🎓 The Feilong Lab @sc.edu studies commonalities & differences across brains, and how they relate to cognition, language, & disorders. We develop brain templates, improve hyperalignment, and compare human brains with monkey brains & DNNs. feilonglab.github.io

Feilong Ma reposted

yuqi98.bsky.social

@yuqi98.bsky.social

18 Aug

Our paper is now out in @elife.bsky.social! We asked whether hyperalignment templates work equally well across the lifespan. Across two datasets, age-matched templates gave better alignment and prediction. @haxbylab.bsky.social @feilong.bsky.social Maria Ida Gobbini elifesciences.org/articles/110...

elifesciences.orgBoosting hyperalignment performance with age-specific templatesAge-specific templates improve the quality of functional alignment using hyperalignment algorithms, and the improvement is consistent across age groups, validation analyses, and datasets.

Feilong Ma reposted

yuqi98.bsky.social

@yuqi98.bsky.social

4 Aug

I'll be presenting at #CCN2026 Wed, Aug 5: talk at 10:25am (Skirball Theater) on hyperalignment recovering shared, fine-grained info in monkeys' idiosyncratic fMRI responses. Poster D2 (2–3:45pm) adds parallel human results. Preprint: www.biorxiv.org/content/10.6...

biorxiv.orgHyperalignment Reveals Shared Information in Macaques’ Idiosyncratic Fine-Grained Movie fMRI PatternsWe modeled shared representational spaces in macaque brains using fMRI data from 12 rhesus monkeys obtained during movie-viewing and response hyperalignment (RHA). RHA greatly improved inter-subject c...

Feilong Ma reposted

Rui Xu

@rui-xu.bsky.social

20 May

New preprint as my first post! "Monosynaptic connections link functionally similar regions in human cortex." We use electrical stimulation + fMRI in epilepsy patients to map whole-brain monosynaptic connectivity at 42 cortical sites. doi.org/10.64898/202... 1/n

Feilong Ma reposted

Context Lab

@contextlab.bsky.social

6 May

Check out our latest collab with Manish Saggar, @richardfbetzel.bsky.social, et al. (lead: Chunyin Siu)! We use TDA to show that 2nd-order (edge) brain interactions more distinctly segregate task connections than other kinds of network patterns. Paper/code/data: www.biorxiv.org/content/10.6...

biorxiv.orgGlobal topology of brain-wide co-fluctuations links task states, personality, and behavioral symptom dimensionsFunctional connectivity in network neuroscience is traditionally characterized using time-averaged correlations between brain regions. While these summaries capture stable large-scale organization, they do not fully reflect the temporal structure of moment-to-moment interactions. Here, we investigate how the order of interaction used to represent brain dynamics shapes the organization recovered from neural data. We compare three interaction representations of fMRI dynamics: regional activation (node time series), pairwise co-fluctuations (edge time series), and higher-order triplet interactions (triangle time series); within a common topological framework using Mapper from topological data analysis (TDA). Across task and resting-state data, Mapper representations derived from pairwise co-fluctuations more distinctly segregate task conditions than activation-based or higher-order representations. This organization reflects structured coordination patterns beyond activation polarity and is driven by high-amplitude interaction events. Beyond task states, modularity quality computed across all Mapper representations is highest for edge time series and selectively associated with stable individual differences: higher modularity relates to higher conscientiousness and lower internalizing and externalizing symptom dimensions. Together, these findings suggest that behaviorally relevant information is reflected in the topology of moment-to-moment brain interactions. Topological analysis of interaction-level dynamics therefore provides a complementary and interpretable framework for linking large-scale neural coordination to cognition, personality, and mental health. ### Competing Interest Statement The authors have declared no competing interest. National Institute of Mental Health, https://ror.org/04xeg9z08, MH127608 Stanford Maternal and Child Health Research Institute, https://ror.org/00yt0ea73, Faculty Scholar Award

Feilong Ma reposted

Gagan Wig

@gaganwig.bsky.social

6 May

We're hiring! FT research assistant @ UT Dallas Join our team to work on a longitudinal neuroimaging study of midlife brain aging. Strong opportunity for those interested in brain networks, aging, and Alzheimer’s disease risk. App. deadline now May 15. Details: jobs.utdallas.edu/postings/31680

Feilong Ma reposted

Ev Fedorenko

@evfedorenko.bsky.social

6 May

So excited about Eghbal's (@eghbal-hosseini.bsky.social) and Jack King's work, which is building on Eghbal's earlier work on linguistic prediction as trajectory straightening!

Eghbal Hosseini@eghbal-hosseini.bsky.social

How is uncertainty in LLMs output reflected in internal representations? In our new work (to appear at ICML 2026), we show that the shape of internal token trajectories provides a direct geometric link to behavioral uncertainty (output entropy). 🧵(1/n)

Feilong Ma reposted

Chris Baldassano

@chrisbaldassano.bsky.social

7 May

The final typeset version of this paper is online now! Check out the new analysis approach that Narjes developed to track changes in event structure at multiple timescales, allowing us to see how event dynamics in a film clip change with repeated viewing

This post is unavailable.

Feilong Ma reposted

Thomas Yeo

@bttyeo.bsky.social

9 Apr

We develop a new TMS targeting algorithm and test it in an open label trial in a treatment-resistant depression population with high comorbidities. Preprints by @rubykong92.bsky.social Phern-Chern Tor 1. doi.org/10.1101/2025... 2. doi.org/10.64898/202... Our new approach ...

Feilong Ma reposted

Rutvik Desai

@rutvikdesai.bsky.social

Seven faculty positions in language/brain health at University of South Carolina! sc.edu/study/colleg...

Feilong Ma reposted

OpenNeuro @openneuro.bsky.social just hit a huge milestone: 1500 datasets! Congrats to the team on making this project so successful over the last 7 years.

youtu.beFeilong Ma - Individual differences in fine-grained brain functional architectureYouTube video by Dartmouth

It was such a great pleasure to be back at Dartmouth and share our research on modeling individual differences in fine-grained brain functional organization using hyperalignment: youtu.be/fgkRwfNpaCw

dartmouth.eduTalks at the 2025 CCN workshop

Check out the Dartmouth Center for Cognitive Neuroscience website for more videos from an amazing lineup of speakers! www.dartmouth.edu/ccn/workshop...

sciencedirect.comReliable individual differences in fine-grained cortical functional architectureFine-grained functional organization of cortex is not well-conserved across individuals. As a result, individual differences in cortical functional ar…

Feilong et al. (2018, NeuroImage) doi.org/10.1016/j.ne... Feilong et al. (2021, @elife.bsky.social ) elifesciences.org/articles/64058 Busch et al. (2024, J Neurosci @sfnjournals.bsky.social) www.jneurosci.org/content/44/6... Feilong et al. (2023, @imagingneurosci.bsky.social) doi.org/10.1162/imag...

Feilong Ma reposted

Thomas Yeo

@bttyeo.bsky.social

17 July 2025

1/11 Excited to share our @Naturestudy led by @leonooi.bsky.social @csabaorban.bsky.social @shaoshiz.bsky.social AI performance is known to scale with logarithm of sample size (Kaplan 2020), but in many domains, sample size can be # participants or # measurements... doi.org/10.1038/s415...

nature.comAdvancing neural decoding with deep learning - Nature Computational ScienceA recent study introduces a neural code conversion method that aligns brain activity across individuals without shared stimuli, using deep neural network-derived features to match stimulus content.

11 July 2025

Excited to share our News&Views on Kamitani Lab's NatComputSci paper! Their neural code converter enables transformation of brain activity patterns across individuals, and it doesn't need shared stimuli or connectivity information! www.nature.com/articles/s43...

@algonautsproject.bsky.social

16 May 2025

I love the work, not only because it speed up FIC models a lot, but also how it saves poor students from grad student descent 🤣🤣

Thomas Yeo@bttyeo.bsky.social

While the world burns, we cook up a new preprint! doi.org/10.1101/2025... Biophysical modeling is a key tool to derive mechanistic insights into the brain. These models are governed by biologically meaningful parameters (unlike artificial neural networks), but the dirty secret ... 1/N

Feilong Ma reposted

The Algonauts Project

6 Jan 2025

(1/4) The Algonauts Project 2025 challenge is now live! Participate and build computational models that best predict how the human brain responds to multimodal movies! Submission deadline: 13th of July. #algonauts2025 #NeuroAI #CompNeuro #neuroscience #AI algonautsproject.com

algonautsproject.comThe Algonauts Project 2025homepage

nature.comA cortical surface template for human neuroscience - Nature MethodsOnavg is a surface template of the human cortex. In contrast to existing templates, the cortical surface is uniformly sampled, which has advantages in numerous applications.

19 July 2024

Our cortical surface template, onavg (short for OpenNeuro Average), is published in Nature Methods. Onavg evenly samples different parts of the cerebral cortex, leading to improved performance of various analysis and reduced computational time. www.nature.com/articles/s41...

feilong.github.ioThe onavg (OpenNeuro Average) template — tpl-onavg

19 July 2024

The code and data of onavg are openly available. See additional information (including code) at our website feilong.github.io/tpl-onavg/ and template files at TemplateFlow github.com/templateflow...

Surface registration methods are also rapidly developing, such as MSM by @metricsemma.bsky.social , SUGAR by Jianxun Ren et al. Onavg can potentially be used in combination with these methods to further improve the workflow of surface-based analysis. www.sciencedirect.com/science/arti...

sciencedirect.comSUGAR: Spherical ultrafast graph attention framework for cortical surface registrationCortical surface registration plays a crucial role in aligning cortical functional and anatomical features across individuals. However, conventional r…

Our functional connectivity was inspired by Ciantar et al. (2022) doi.org/10.1007/s004... which was recently replicated by Jeganathan et al. www.biorxiv.org/content/10.1... See also related work on volumetric data: www.um.edu.mt/library/oar/...

doi.orgGeometric effects of volume-to-surface mapping of fMRI data - Brain Structure and FunctionIn this work, we identify a problem with the process of volume-to-surface mapping of functional Magnetic Resonance Imaging (fMRI) data that emerges in local connectivity analysis. We show that neighbo...

The creation of onavg has greatly benefited from open science. We used 1,031 high-quality brains openly available from OpenNeuro, and we deeply appreciate the participants and researchers who generously shared their data. We also thank DataLad, StudyForrest, and Center for Open Neuroscience.

In densely sampled regions, there are more vertices in each searchlight and more searchlight centers in the region, leading to prolonged computations. By uniformly sampling the cortex, onavg reduces the computational time of various commonly used algorithms.

These benefits, in theory, generalizes to any analyses that rely on sampling density, uniformity, or spatial patterns. In Extended Data Figs 7–9, we showcased how onavg benefits resting-state functional connectivity, localizer contrast maps, and modeling individual differences.

As a result of uniform sampling, the quality of spatial patterns is greatly improved, leading to increased performance of various MVPA algorithms.

We optimized the location of cortical vertices by penalizing a pair of vertices if they were too close on the anatomical surface. After the optimization, we obtained the onavg template, where vertices are evenly distributed on the anatomical surface.

Traditional templates choose their vertices based on the inflated spherical surface. Vertices are mostly uniform on the spherical surface. However, on the actual anatomical surface, vertices are often sparsely distributed on gyri and densely distributed on sulci.

Feilong Ma reposted

Floris de Lange

@predictivebrain.bsky.social

15 May 2024

🚨Job alert!🚨 Are you looking for a PhD position, and interested in neuroAI, the predictive brain and visual perception/cognition? Then this job is for you: www.ru.nl/en/working-a... Sharing/re-posting is very much appreciated!

ru.nlPhD Candidate: Computational Cognitive Neuroscience at the Predictive Brain Lab | Radboud UniversityWorking as PhD Candidate at the Predictive Brain lab? Check our vacancies here!