This allows long-term control over months. Notably, we find evidence of degradation of control over long periods. We can restore performance with a brief recalibration procedure.

We find evidence for 'representational drift', evident as changes in the centroid. Decoders that sample multiple days can account for the observed drift.

Using closed-loop training, we can improve decoder performance. This is highly context specific.

graphical abstract:

thanks; definitely a team effort

We used our 'ringtrode' to change phasic excitability and reduce the reactivation rate. We found evidence of significantly lower performance gains after brief breaks.

Really enjoyed the conference a couple of years ago. Look forward to it in 2025!