Ad Brain Scorer
Predict brain activation from your advertisements using Meta AI's TRIBEv2.
Upload a video, audio, or image ad to see which brain regions it activates. Scores are computed across 7 cognitive dimensions relevant to advertising effectiveness.
Score Breakdown
Upload multiple ad files to compare brain activation scores side by side. File type is auto-detected from the extension.
Comparison Table
How Ad Brain Scorer Works
The Model: TRIBEv2
TRIBEv2 is a foundation model from Meta AI that predicts fMRI brain responses to naturalistic stimuli. It processes three modalities simultaneously:
- Visual pathway (V-JEPA2 ViT-G) — encodes video frames
- Auditory pathway (Wav2Vec-BERT 2.0) — encodes audio waveforms
- Linguistic pathway (LLaMA 3.2-3B) — processes speech and text
A Transformer fusion model combines these into predicted brain activation on the fsaverage5 cortical mesh (~20,000 surface vertices).
The 7 Brain Scores
| Score | Brain Regions | Why It Matters for Ads |
|---|---|---|
| Emotional Engagement | Insula, orbitofrontal cortex, anterior cingulate | Emotional ads are shared more and drive purchase intent |
| Visual Attention | Occipital cortex (calcarine, cuneus, lingual) | Higher visual processing = more attention captured |
| Auditory Processing | Superior temporal gyrus, planum temporale | Audio/music quality impacts brand recall |
| Memory Encoding | Parahippocampal, fusiform, temporal pole | Predicts whether the ad will be remembered |
| Reward & Motivation | Medial orbitofrontal, frontal pole | Desire and motivation to act on the ad |
| Language Comprehension | Broca's area, angular gyrus, supramarginal | Message clarity and persuasiveness |
| Social Cognition | TPJ, medial prefrontal, precuneus | Response to people, faces, social narratives |
How Scores Are Computed
- Raw predictions are generated for each brain vertex at each time point
- Vertices are grouped into 7 categories using the Destrieux brain atlas
- Activations are z-score normalized against the global mean
- Scores are mapped to 0-1 range via sigmoid transformation
- Higher scores = stronger activation in that cognitive domain
Interpreting Results
- Scores are relative, not absolute — compare ads against each other
- Timeline view shows which moments in the ad drive the strongest brain response
- 0.5 = average activation — above 0.5 means above-average engagement
- These are computational predictions, not actual brain measurements
Citation
d'Ascoli et al. (2026). A foundation model of vision, audition,
and language for in-silico neuroscience.