Fig 2. The DREDge motion-correction algorithm achieves recording stability even under extreme circumstances. In this experiment, the electrode was continuously driven without stopping—i.e. the opposite of the typical goal of seeking mechanical stability. The diagonal banded lines, sloping upward with time, correspond to ‘hot spots’ on the electrode array, created by neurons with particularly large action potentials. As the electrode drives down, but the neurons remain in place, these hot spots appear to move upwards relative to the physical electrode. DREDge leverages the close spacing of Neuropixels probe recordings, which ensures that a given action potential is visible on multiple nearby channels. Stability is achieved via post-hoc analysis (the equivalent of processing-based image stabilization rather than physical lens-based image stabilization). The resulting ‘registered’ data shows horizontal bands, indicating that each neuron now correctly maintains a consistent vertical location. Bands last for a fixed amount of time, as the neuron is tracked from when it first emerges on the bottom-most electrodes to when it exits the top-most electrodes (as the electrode moves continuously downwards). This is both a demonstration of how well the algorithm is able to achieve stability, and a possible new recording technique where one records from more neurons (over a greaterdepth) but views each neuron for a shorter period of time.