AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Towards Interpretable Visual Decoding with Attention to Brain Representations

Pinyuan Feng, Hossein Adeli, Wenxuan Guo, Fan L. Cheng, Ethan Hwang, Nikolaus Kriegeskorte

ICLR 2026 · Vision-language and multimodal › Human and embodied modeling · approach: fMRI visual decoding · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add parcel-wise fMRI token dropout (ablation)NeuroAdapter · NSD Subject 1CLIP/Incep/PixCorr: Incep 61.2% -> 92.15%; CLIP 64.8% -> 92.03%; PixCorr .038 -> .133Improved
add parcel-wise linear mapper (ablation)NeuroAdapter · NSD Subject 1CLIP/Incep/PixCorr: CLIP 78.11% -> 92.03%; PixCorr .073 -> .133Improved
keep 200 highest-SNR parcels rather than 500 (ablation)NeuroAdapter · NSD Subject 1CLIP: 88.66% -> 92.03%Improved
keep 200 highest-SNR parcels rather than 1000 (ablation)NeuroAdapter · NSD Subject 1CLIP: 84.89% -> 92.03%Improved
use 200 parcels rather than 100 (ablation)NeuroAdapter · NSD Subject 1CLIP: 90.41% -> 92.03%Improved
rank 8 decoded candidates with brain encoder vs single prediction (ablation)NeuroAdapter · NSD Subject 1CLIP: 90.8% -> 92.03%Improved
pick highest-correlation candidate vs lowest (ablation)NeuroAdapter · NSD Subject 1AlexNet(2) 2WC: 80.88% -> 88.02%Improved
pick highest-correlation candidate vs random (ablation)NeuroAdapter · NSD Subject 1Inception 2WC: 92.05% -> 91.91%No clear effect
vary condition dimension 192 to 1024 (ablation)NeuroAdapter · NSD Subject 1CLIP: 1024: 90.33%; 768: 92.03%; 192: 90.78%Mixed
vary denoising steps 20-80 (ablation)NeuroAdapter · NSDmetricsNo clear effect
remove VDVAE low-level pathway from Brain Diffuser (ablation)Brain Diffuser · NSD Subject 1PixCorr: PixCorr .305 -> .143Hurt
NeuroAdapter without embedding alignment vs MindEye1NeuroAdapter · NSDCLIP 2WC: 94.1% -> 90.97%Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy