TRIBE v2

TRIBE v2 is an AI brain-modeling tool that predicts fMRI brain responses to video, audio, and text stimuli for neuroscience and AI researchers. Built by Meta as a multimodal brain encoding model, TRIBE v2 lets research teams run in-silico experiments by simulating how the human brain would respond to a given multimodal input, without needing to scan a live participant for every condition. It ingests aligned video, audio, and text signals and maps them onto predicted fMRI activity, making it useful for hypothesis testing, stimulus design, and modeling cross-modal brain function before committing to costly imaging sessions. The model targets neuroscience researchers, AI researchers, and brain-modeling teams who want a computational stand-in for the brain that they can probe programmatically. Because it is multimodal, TRIBE v2 can account for how the brain integrates what people see, hear, and read at the same time, rather than treating each modality in isolation. It fits into a broader shift toward using large encoding models as digital twins of neural activity, where researchers generate and test predictions in silico and reserve scarce fMRI scanner time for the most promising experiments. As a Meta AI demos release, it is oriented toward the research community rather than commercial end users.

Neuroscience and AI researchers use TRIBE v2 to predict fMRI brain responses to video, audio, and text and run in-silico brain experiments.