Spurious reconstruction from brain activity

Advances in brain decoding, particularly visual image reconstruction, have sparked discussions about the societal implications and ethical considerations of neurotechnology. As these methods aim to recover visual experiences from brain activity and achieve prediction beyond training samples (zero-shot prediction), it is crucial to assess their capabilities and limitations to inform public expectations and regulations. Our case study of recent text-guided reconstruction methods, which leverage a large-scale dataset (NSD) and text-to-image diffusion models, reveals limitations in their generalizability. We found decreased performance when applying these methods to a different dataset designed to prevent category overlaps between training and test sets. UMAP visualization of the text features with NSD images showed limited diversity of semantic and visual clusters, with overlap between training and test sets. Formal analysis and simulations demonstrated that clustered training samples can lead to "output dimension collapse," restricting predictable output feature dimensions. Diversifying the training set improved generalizability. However, text features alone are insufficient for mapping to the visual space. We argue that recent photo-like reconstructions may primarily be a blend of classification into trained categories and generation of inauthentic images through text-to-image diffusion (hallucination). Diverse datasets and compositional representations spanning the image space are essential for genuine zero-shot prediction. Interdisciplinary discussions grounded in understanding the current capabilities and limitations, as well as ethical considerations, of the technology are crucial for its responsible development.