Realignment of representational drift in mouse visual cortex via flexible electrode arrays

The ability to stably decode brain activity is crucial for brain-computer interfaces, which are often compromised by recording instability due to immune responses and probe drifting. In addition, many brain regions undergo intrinsic dynamics such as 'representational drift', in which neural activities associated with stable sensation and action continually change over time. Here we demonstrate that tissue-like flexible electrode arrays enable stable, months-long tracking of the same neurons in the mouse visual cortex and hippocampus during visual stimulation. Using these arrays, we recorded single-unit action potentials over extended periods and characterized representational drift in neuron activities and population-level latent dynamics to repeated stimuli. We further modelled these latent dynamics using geometric transformation to capture shifts in neural representations across sessions. Using this approach, we built a decoder that maintains high performance across long time spans and demonstrates the potential to generalize across different animals when applied to the aligned latent dynamics. These findings suggest a path towards neural recording and decoding frameworks that remain reliable despite continuous changes in neural activity.