ViT texture representations align with each other and with human psychophysics better than VGG-19 representations, suggesting architecture drives texture coding more than training objective.
the visual cortex
9 Pith papers cite this work, alongside 13,833 external citations. Polarity classification is still indexing.
9
Pith papers citing it
13.8k
external citations · OpenAlex
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
MINE uses mechanistic interpretability on language-aligned image representations to generate per-voxel feature descriptions, validated via image generation and counterfactual edits that causally shift brain activation.
Feature visualization on TRIBE v2 brain encoders recovers the known ventral visual hierarchy from V1 to V4 and produces distinctive patterns for MT, FFA, and PPA, with optimized stimuli driving ~4x higher activation than natural images.
Direction maps and pinwheel structures in MT emerge spontaneously when a spatiotemporal deep network is trained on videos with contrastive self-supervised learning and spatial regularization.
Stimulus symmetries render many neural representations functionally equivalent yet produce qualitatively different RSMs, including drifting ones from SGD or regularization in image-encoding networks.
Mathematical analysis shows sparse linear regression mitigates output dimension collapse in brain-to-image reconstruction at small data scales by exploiting sparsity in the brain-to-feature mapping.
Dual-stream EEG decoder separates identity and orientation to support 3D reconstruction from neural signals via circular regression and conditioned diffusion.
Hybrid neuromorphic-ANN models outperform standard deep learning on few-shot benchmarks and under occlusion/impulse noise via astrocytic modulation and spiking dynamics.
Joint sparse coding and temporal dynamics in mPFC and computational networks reduce cross-context interference and enhance separability, enabling better retention in lifelong learning without extra heuristics.
citing papers explorer
-
Texture Representations in Deep Vision Models: Comparing CNNs, Vision Transformers, and Human Perception
ViT texture representations align with each other and with human psychophysics better than VGG-19 representations, suggesting architecture drives texture coding more than training objective.
-
Mechanistically Interpretable Neural Encoding Reveals Fine-Grained Functional Selectivity in Human Visual Cortex
MINE uses mechanistic interpretability on language-aligned image representations to generate per-voxel feature descriptions, validated via image generation and counterfactual edits that causally shift brain activation.
-
Feature Visualization Recovers Known Cortical Selectivity from TRIBE v2
Feature visualization on TRIBE v2 brain encoders recovers the known ventral visual hierarchy from V1 to V4 and produces distinctive patterns for MT, FFA, and PPA, with optimized stimuli driving ~4x higher activation than natural images.
-
Self-organized MT Direction Maps Emerge from Spatiotemporal Contrastive Optimization
Direction maps and pinwheel structures in MT emerge spontaneously when a spatiotemporal deep network is trained on videos with contrastive self-supervised learning and spatial regularization.
-
Stimulus symmetries can confound representational similarity analyses
Stimulus symmetries render many neural representations functionally equivalent yet produce qualitatively different RSMs, including drifting ones from SGD or regularization in image-encoding networks.
-
Overcoming Output Dimension Collapse: When Sparsity Enables Zero-shot Brain-to-Image Reconstruction at Small Data Scales
Mathematical analysis shows sparse linear regression mitigates output dimension collapse in brain-to-image reconstruction at small data scales by exploiting sparsity in the brain-to-feature mapping.
-
Dual-Stream EEG Decoding for 3D Visual Perception
Dual-stream EEG decoder separates identity and orientation to support 3D reconstruction from neural signals via circular regression and conditioned diffusion.
-
The Neuromorphic Supremacy
Hybrid neuromorphic-ANN models outperform standard deep learning on few-shot benchmarks and under occlusion/impulse noise via astrocytic modulation and spiking dynamics.
-
Joint sparse coding and temporal dynamics support context reconfiguration
Joint sparse coding and temporal dynamics in mPFC and computational networks reduce cross-context interference and enhance separability, enabling better retention in lifelong learning without extra heuristics.