Brain3D: EEG-to-3D Decoding of Visual Representations via Multimodal Reasoning

TL;DR

Brain3D introduces a multimodal approach to decode EEG signals into 3D visual representations, combining image reconstruction, language reasoning, and 3D mesh generation for improved neural decoding.

Contribution

It presents a novel EEG-to-3D reconstruction pipeline that leverages multimodal reasoning and generative models, advancing beyond traditional 2D decoding methods.

Findings

Achieved up to 85.4% EEG decoding accuracy.

Demonstrated 0.648 CLIPScore in 3D reconstruction.

Validated the approach with comprehensive evaluations.

Abstract

Decoding visual information from electroencephalography (EEG) has recently achieved promising results, primarily focusing on reconstructing two-dimensional (2D) images from brain activity. However, the reconstruction of three-dimensional (3D) representations remains largely unexplored. This limits the geometric understanding and reduces the applicability of neural decoding in different contexts. To address this gap, we propose Brain3D, a multimodal architecture for EEG-to-3D reconstruction based on EEG-to-image decoding. It progressively transforms neural representations into the 3D domain using geometry-aware generative reasoning. Our pipeline first produces visually grounded images from EEG signals, then employs a multimodal large language model to extract structured 3D-aware descriptions, which guide a diffusion-based generation stage whose outputs are finally converted into coherent 3D meshes via a single-image-to-3D model. By decomposing the problem into structured stages, the proposed approach avoids direct EEG-to-3D mappings and enables scalable brain-driven 3D generation. We conduct a comprehensive evaluation comparing the reconstructed 3D outputs against the original visual stimuli, assessing both semantic alignment and geometric fidelity. Experimental results demonstrate strong performance of the proposed architecture, achieving up to 85.4% 10-way Top-1 EEG decoding accuracy and 0.648 CLIPScore, supporting the feasibility of multimodal EEG-driven 3D reconstruction.