Region-specific neural manifold dynamics along a standardised visual-hippocampal-thalamic recording axis: a multi-lab IBL Reproducible Ephys study

Abstract

The IBL Reproducible Ephys project asked whether different neuroscience laboratories can reliably record from the same brain circuits. We ask a complementary question: does the geometry of neural population dynamics reproduce across laboratories? Using a framework called Noetic Diffusion Theory (NDT), we estimated how strongly neural activity converges toward a repeating pattern (spectral radius) and how much accessible state space the circuit maintains before a stimulus arrives (reachability cone), independently for visual cortex, hippocampus, and thalamus. The good news: these geometry measurements are indeed reproducible across 12 laboratories when we restrict analyses to the correct anatomical targets. The more cautious news: the specific hypotheses we pre-specified — that visual cortex dynamics predict reaction time, and that hippocampal reachability scales with prior expectation — were not supported. This outcome is scientifically valuable: it establishes that the measurement layer itself is valid and portable, while setting clear null boundaries for where these particular dynamics–behaviour links do and do not appear in a large standardised dataset.