Scaling Up Multichannel Electrophysiology: SpikeGadgets and the Push Toward Higher Channel Counts
For decades, the central bottleneck in systems neuroscience wasn’t a lack of good questions — it was a lack of good hardware. You could record from a handful of neurons in one region, or a few hundred in one probe, but the brain doesn’t work one region at a time. Understanding how sensory information becomes a decision, or how one hemisphere’s worth of circuitry supports cognition, means recording simultaneously across many sites, many regions, and — ideally — many thousands of channels at once. Three recent papers show how SpikeGadgets’ multichannel electrophysiology hardware is being pushed toward exactly that kind of large-scale recording, and what that scale makes possible.
Ten brain regions at once
Tian-Yi Wang and colleagues wanted to know how a network of brain regions — not just one — reorganizes while an animal learns a visual discrimination task. To do this, they chronically implanted electrode dense ultra-flexible microelectrode arrays across 10 cortical and subcortical regions in mice, summing to 1024 channels connected to our Modular Stacking Headstage. Tracking single units across all 10 regions simultaneously, and across many days of learning, let the team watch functional connectivity itself change shape. As mice learned to make correct responses, a subnetwork linking visual and frontal regions emerged, and a region’s rank within that network predicted when it would start encoding visual information (Wang et al., 2026, eLife). This is the kind of network-level result that’s essentially invisible if you can only record from one or two regions at a time — it requires genuinely large scale, chronic, multichannel electrophysiology across the whole brain.
A thousand channels, one primate hemisphere
Another recent channel-hungry application of SpikeGadgets hardware comes from Tobias Teichert’s lab at the University of Pittsburgh. Their MePhys (“mesoscopic electrophysiology”) platform tessellates an entire macaque hemisphere with 992 electrode contacts spread across 62 chronically implanted multi-electrode shafts, all routed through a SpikeGadgets’ Modular Stacking Headstage. That’s enough simultaneous coverage to examine functional interactions across more than 300,000 pairs of electrodes at once — something previously only possible, in a much coarser form, with fMRI or scalp EEG. The team used this scale to show that a subanesthetic dose of ketamine, a drug that models aspects of psychosis, produces a marked state of functional disconnection across the hemisphere (Teichert et al., 2025, Journal of Neurophysiology). MePhys is effectively a new imaging modality sitting between microscopic electrophysiology and whole-brain imaging — made possible only because a single headstage could handle over a thousand channels.
Building the next order of magnitude: 5,376 channels on one chip
A new preprint from Rice University shows where channel counts are headed. Yingying Fan, Yuhang Ma, and colleagues — working with SpikeGadgets co-founders Mattias Karlsson and Magnus Karlsson, who developed the FPGA backend — describe a custom application-specific integrated circuit (ASIC) supporting 5,376 simultaneous recording channels, each sampling at 20 kilosamples per second, with over 1.3 Gb/s of total data throughput and just 5.5 μVrms of input noise. The chip integrates in-pixel amplification, time-division-multiplexed analog-to-digital converters, and on-chip stimulation, and the team demonstrates it paired with a flexible µECoG array using a gold-bump-bonding interconnect strategy (Fan et al., 2026, bioRxiv). Every order-of-magnitude jump in channel count changes what questions are even askable, and SpikeGadgets is paving the path.
Scaling Up Multichannel Electrophysiology: SpikeGadgets and the Push Toward Higher Channel Counts
For decades, the central bottleneck in systems neuroscience wasn’t a lack of good questions — it was a lack of good hardware. You could record from a handful of neurons in one region, or a few hundred in one probe, but the brain doesn’t work one region at a time. Understanding how sensory information becomes a decision, or how one hemisphere’s worth of circuitry supports cognition, means recording simultaneously across many sites, many regions, and — ideally — many thousands of channels at once. Three recent papers show how SpikeGadgets’ multichannel electrophysiology hardware is being pushed toward exactly that kind of large-scale recording, and what that scale makes possible.
Ten brain regions at once
Tian-Yi Wang and colleagues wanted to know how a network of brain regions — not just one — reorganizes while an animal learns a visual discrimination task. To do this, they chronically implanted electrode dense ultra-flexible microelectrode arrays across 10 cortical and subcortical regions in mice, summing to 1024 channels connected to our Modular Stacking Headstage. Tracking single units across all 10 regions simultaneously, and across many days of learning, let the team watch functional connectivity itself change shape. As mice learned to make correct responses, a subnetwork linking visual and frontal regions emerged, and a region’s rank within that network predicted when it would start encoding visual information (Wang et al., 2026, eLife). This is the kind of network-level result that’s essentially invisible if you can only record from one or two regions at a time — it requires genuinely large scale, chronic, multichannel electrophysiology across the whole brain.
A thousand channels, one primate hemisphere
Another recent channel-hungry application of SpikeGadgets hardware comes from Tobias Teichert’s lab at the University of Pittsburgh. Their MePhys (“mesoscopic electrophysiology”) platform tessellates an entire macaque hemisphere with 992 electrode contacts spread across 62 chronically implanted multi-electrode shafts, all routed through a SpikeGadgets’ Modular Stacking Headstage. That’s enough simultaneous coverage to examine functional interactions across more than 300,000 pairs of electrodes at once — something previously only possible, in a much coarser form, with fMRI or scalp EEG. The team used this scale to show that a subanesthetic dose of ketamine, a drug that models aspects of psychosis, produces a marked state of functional disconnection across the hemisphere (Teichert et al., 2025, Journal of Neurophysiology). MePhys is effectively a new imaging modality sitting between microscopic electrophysiology and whole-brain imaging — made possible only because a single headstage could handle over a thousand channels.
Building the next order of magnitude: 5,376 channels on one chip
A new preprint from Rice University shows where channel counts are headed. Yingying Fan, Yuhang Ma, and colleagues — working with SpikeGadgets co-founders Mattias Karlsson and Magnus Karlsson, who developed the FPGA backend — describe a custom application-specific integrated circuit (ASIC) supporting 5,376 simultaneous recording channels, each sampling at 20 kilosamples per second, with over 1.3 Gb/s of total data throughput and just 5.5 μVrms of input noise. The chip integrates in-pixel amplification, time-division-multiplexed analog-to-digital converters, and on-chip stimulation, and the team demonstrates it paired with a flexible µECoG array using a gold-bump-bonding interconnect strategy (Fan et al., 2026, bioRxiv). Every order-of-magnitude jump in channel count changes what questions are even askable, and SpikeGadgets is paving the path.