Implicit Behavioral Decoding from Next-Step Spike Forecasts at Population Scale

2026-05-13

AI Summary

This work proposes the first unified Mamba-based framework for closed-loop brain–computer interfaces that jointly addresses neural activity prediction and behavioral decoding within a single forward pass. Trained solely to predict Neuropixels spike counts at the next time step, the model implicitly learns rich behavioral representations sufficient to outperform linear baselines that directly use raw spikes. Augmented with lightweight session-specific readout heads, the architecture efficiently processes large-scale neural data within a 500 ms context window. On the Steinmetz visual decision task in mice, it achieves 75.7% and 66.1% decoding accuracy for choice and stimulus side, respectively—surpassing baseline performance by 4–6 percentage points. Notably, near-optimal decoding is attainable with only 100–150 calibration trials, and the entire pipeline operates with an end-to-end latency under 50 ms.

Abstract

Closed-loop brain-computer interfaces often require both a forecast of upcoming neural population activity and a readout of the animal's behavioral state. A single Mamba forecaster, trained only on next-step spike counts at Neuropixels scale, can deliver both in one forward pass. A lightweight per-session linear head reading the model's predicted rates decodes behavior better than the same linear classifier reading the raw spike counts, under matched temporal context. We test on the Steinmetz visual-discrimination benchmark, which spans 39 sessions, roughly 27,000 neurons, and 1,994 held-out trials. Across three training seeds, Mamba's predicted rates decode mouse choice at 75.7$\pm$0.2% trial vote, roughly 2.3 times chance level, and stimulus side at 66.1$\pm$0.6%, about twice chance. Compared to a matched 500 ms-context linear decoder on the raw spike counts, Mamba wins at trial vote by 4-6 pp on response and 4-6 pp on stimulus side. A session-start calibration block of about 100-150 trials brings the readout within 1-2 pp of asymptote, and the full pipeline fits inside the 50 ms bin budget on workstation-class GPUs typical of tethered chronic Neuropixels recordings.

Problem

• brain-computer interface
• neural decoding
• spike forecasting
• behavioral state
• population neuroscience

Innovation

• Mamba
• neural population decoding
• spike forecasting
• closed-loop BCI
• behavioral state inference