Transformer trained on flying bat recordings yields hippocampus 'digital twin' revealing hidden structure
patrickmineault · x · 2026-10-16
A new preprint builds a functional "digital twin" of the hippocampus by training a transformer on Neuropixels recordings and 3D flight trajectories from freely flying bats. Probing the model's latent space revealed previously unknown structure in hippocampal population activity: latent PC1-PC3 trajectories mirror the bats' physical XYZ position, and positive/negative motifs form separable loop-like structures. The work shows large-scale neural recordings can yield a probe-able model of hippocampal computation, offering a new tool for studying spatial coding and population dynamics.
More from Research
- Survey: AI agent evaluation is getting realistic but still fails to measure safety and reliability — burkov · 2026-10-16
- COVA-SFT dataset trains models to solve text problems with visual chain-of-thought — yoavartzi · 2026-10-16
- USTC paper uses speculative sampling to cut latency in interactive video generation — ustc · 2026-10-16
- Tsinghua's CoTranslator distills reverse CoT to boost 4B model math scores — Tsinghua · 2026-10-16
- Nobel laureate Jennifer Doudna on AI's role in science in Big Think interview — ChiaraStellata · 2026-10-16
- DIY benchmark ranks 18 models on subjective judgments, GLM-5.3 tops consensus match — arkuto · 2026-10-16