Soyuhos et al. (2026)

Soyuhos, O., Scarpa, A., & Baldauf, D. (2026). Distinct resting-state connectomes for face and scene perception predict individual task performance. Human Brain Mapping, 47, e70498. https://doi.org/10.1002/hbm.70498

• Paper gelesen?
• Infos rausgeschrieben?

Soyuhos et al. (2026)

Core argument: The FFA and PPA are embedded in distinct intrinsic resting-state networks whose connectivity profiles predict individual behavioral performance on face-matching and scene n-back tasks (double dissociation). MEG analyses further show this spatial segregation is expressed in beta/gamma amplitude coupling, with dominant incoming influences (ROI→Seed) in higher frequencies and outgoing influences (Seed→ROI) in lower frequencies.

Why this paper matters for the thesis:

• Uses the same CONN-based analysis pipeline, same HCP dataset, same Baldauf lab toolbox (BrainRest), and the same Wilcoxon + FDR statistical framework as this thesis
• Provides methodological precedent for the thesis approach — cite in §3.3 when introducing the connectivity analysis method
• The paper is from the same lab (Baldauf) and therefore the closest methodological reference available

Methods details (fMRI-relevant only — no MEG in thesis):

Step	Soyuhos (2026) detail
Dataset	HCP 1200 Subjects Release; N=55 for connectivity (twin-excluded, MEG-complete); N=371 for CPM
Scanner	Siemens 3T Connectome Skyra, TR=720ms, TE=33.1ms, 52°, 2mm, multi-band=8
Preprocessing	HCP minimal preprocessing pipeline; ICA-FIX denoising
Time series normalization	z-normalized: (x−mean)/stdev before connectivity computation
Run concatenation	All 4 runs concatenated → 1 continuous dataset (~60 min/subject)
Atlas	HCP-MMP1, Connectome Workbench for parcel-wise time series
FC computation	FSLNets toolbox, nets_netmats function, ridgep=0.01 (ridge-regularized partial correlations)
Matrix size	360×360 (whole brain)
Statistical test	Wilcoxon signed-rank test, FDR-corrected q<0.05 for 180 ROIs per hemisphere
CPM network	FFA network: 37 regions (bilateral seeds + significant ROIs); PPA network: 23 regions
CPM feature selection	Edges with negative correlation to RT (p<0.05) selected as predictive features
CPM summarization	Single summary score = sum of partial correlation values of selected edges
CPM model	Linear regression: y = mx + b; slope and intercept from training set applied to held-out subject
CPM validation	LOO cross-validation; permutation testing (1000 iterations); Spearman’s r as performance metric
CPM sample	N=350 (face-matching), N=352 (scene tasks) after motion exclusion (>0.15mm FD)

Key results:

• FFA network: preferential connectivity with lateral occipitotemporal, inferior temporal, temporoparietal regions
• PPA network: preferential connectivity with ventral medial visual, posterior cingulate, entorhinal-perirhinal regions
• Double dissociation in CPM: FFA network predicts face-matching RT, PPA network predicts scene n-back RT (not vice versa)
• MEG: spatial segregation reflected in beta (13–30Hz) and gamma (30–100Hz) amplitude coupling; dominant incoming influences in high frequencies

Relation to thesis methods:

• Same statistical framework (Wilcoxon + FDR) ✓
• Same HCP dataset and preprocessing ✓
• Same CPM approach (LOO, Spearman r, permutation) ✓
• Difference: Soyuhos uses data-driven ROI selection (Neurosynth contrast); thesis uses theory-driven ROI selection based on dual-stream literature
• Difference: Soyuhos uses FSLNets ridgep=0.01; check with Daniel whether BrainRest uses same regularization
• Difference: Soyuhos N=55 for connectivity; thesis N=812 (larger, 4-run completeness criterion)

BibTeX key: soyuhos2026

@article{soyuhos2026,
  author    = {Soyuhos, Orhan and Scarpa, Aurelia and Baldauf, Daniel},
  title     = {Distinct resting-state connectomes for face and scene perception
               predict individual task performance},
  journal   = {Human Brain Mapping},
  year      = {2026},
  volume    = {47},
  pages     = {e70498},
  doi       = {10.1002/hbm.70498}
}

PDF: /Users/maxmacbookpro/Downloads/Human Brain Mapping - 2026 - Soyuhos - Distinct Resting‐State Connectomes for Face and Scene Perception Predict Individual.pdf

see also

Soyuhos, O., & Baldauf, D. (2022)
3.3 Matrix Construction
3.4 Brain Behaviour Correlation
Tags: neuroscience science source