Brown Team's Vision Model Groups Objects and Predicts Human Reaction Times

tserre · x · 2026-10-13

The Brown University team (tserre) describes a vision model that groups objects in natural scenes and predicts human reaction times. Early grouping is driven by local boundary signals; later, top-down feedback about object identity lets the grouping signal cross internal edges such as shading or texture changes. The model builds on their prior line of work: hGRU (NeurIPS 2018), GammaNet (ICLR 2020), contractor recurrent backprop (NeurIPS 2020), and a reaction-time metric (NeurIPS 2023).

Related event: Recurrent Neural Network Replicates Brain's Perceptual Grouping and Predicts Human Reaction Times(3 posts)→

Original post →

More from Research

• Oxford shows neural nets can be encrypted inside model weights, surviving 500 SFT steps — zainhas · 2026-10-13
• XPENG's 4B GroundingPI model beats GPT-6 Astra across 34 grounding benchmarks, boosts robot manipulation — XPENG-AI · 2026-10-13
• XPENG's GroundAnything pairs diffusion parallel decoding with precise grounding, 4.51x faster than AR — XPENG-AI · 2026-10-13
• Does RL really generalize worse? Pretraining breadth may beat human transfer — yacineMTB · 2026-10-13
• Natura-1 debuts: 50B-param MoE world model trained on 1M hours of egocentric video — stingning · 2026-10-13
• High schooler clusters two Android phones over USB to run a 27B LLM locally at 30°C — Grouchy-Zebra9780 · 2026-10-13

Research channel →