Skip to content
Encoder Lab

Watch a clip, see what the average brain does.

Four short Creative Commons clips, four different cortical systems. Pick one. The brain above re-renders as the video plays — interpolating between editorial keyframes that compose Neurosynth meta-analyses for the moment-by-moment stimulus.

Built in the spirit of Meta's TRIBE v2 demo at aidemos.atmeta.com/tribev2, with two honesty differences: every prediction here is precomputed (not live TRIBE inference), and every clip carries its own provenance + license + Neurosynth term composition. When the author runs the Colab notebook with their HuggingFace facebook/tribev2 access, real TRIBE predictions drop in at the same JSON path — the page renders them without code change.

Paused0.0s / 20.0s

Water lily opening

A timelapse close-up of a flower opening. Sustained visual perception builds across the clip; subtle reward weighting follows as the bloom completes — the editorial composition tracks the viewer, not the flower.

Neurosynth meta-analysis
Neurosynth meta-analysis (preview) · HCP-MMP-360 (Glasser 2016, doi:10.1038/nature18933) · CC0
Clip: Water lily opening bloom, Wikimedia Commons (CC-BY-SA).Method: Per-keyframe Neurosynth composite. Each keyframe blends cached term z-maps with editorial weights, projects to fsaverage5 via nilearn.surface.vol_to_surf, aggregates within HCP-MMP-360 parcels. The frontend interpolates between keyframes as the video plays. This is the PREVIEW path � drop in TRIBE-v2-generated frames at the same path to swap in real predictions.Yarkoni et al., Nature Methods 2011, doi:10.1038/nmeth.1635Time-lapse close-up of a flower opening. Visual + scene encoding builds across the clip; subtle reward weighting as the bloom completes.
Keyframe composition
  • t=0.0s · perception 50% · face 10% · attention 20% · emotion 20%
  • t=7.0s · perception 40% · attention 20% · imagery 20% · emotion 20%
  • t=14.0s · perception 35% · reward 20% · attention 20% · imagery 25%
  • t=20.0s · perception 30% · reward 30% · attention 20% · default mode 20%
What this is not

Not a measurement of your brain. Not even a measurement of any specific brain. What you're seeing is the activation pattern peer-reviewed fMRI literature aggregates for the term composition above, projected onto a standard cortical surface. The model is not the mind. The aggregate is not the person.