Event:

The past decade has seen rapid growth in the capacity and fidelity of extracellular electrophysiology. Lead by Neuropixels, recordings with hundreds of units across the brain have become routine. I will show our most recent generations of probes, both with 1536 channels. The most advanced of these is small enough to make a 6,000 channel implant in a freely moving, tethered mouse a straight forward experiment while 12,000 channels in a rat will be easier and 18,000 plausible. The very large data sets that result demand more automated analysis pipelines and consideration of system vs. single unit modeling of multi-region brain activity. In addition, probes with 1024 recording channels and 256 stimulation channels are being designed for primate research.
While exciting to imagine, "brain-wide" recordings will create up to 6 TB of data per hour. Digestion of such large data sets precludes the manual curation that has been typical of basic neuroscience electrophysiology. I will discuss the specification of 2 new 1536 channels Neuropixels, both with 4 shanks, and show preliminary data. I would like to engender a discussion of the usefulness of very large scale experiments vs. more focused designs, and the overlap between the research device design needs and the prosthetic opportunities for human health.