Untethered and Unbothered: Neuropixels Meets SpikeGadgets in Freely Moving Animals
High-density Neuropixels probes can record from hundreds to thousands of neurons at once — but that density used to come at the cost of freedom of movement, since the probe’s data had to be streamed out through a cable to a fixed acquisition system. SpikeGadgets’ Neuropixels-compatible dataloggers remove that constraint: the systems write probe data straight to onboard storage, so the animal can carry the whole recording system with it, wirelessly, while behaving naturally. Over the past twelve months, four papers have put this combination — Neuropixels probes plus a SpikeGadgets datalogger — to work in freely moving and freely flying animals, and the results are reshaping some long-standing assumptions about hippocampal replay, prefrontal sequence coding, and vocal communication. Here’s a roundup.
Replay looks different in a bat than in a rat
The most striking of this year’s crop comes from Michael Yartsev’s lab at UC Berkeley. Angelo Forli, Wudi Fan, Kevin Qi, and Yartsev wirelessly recorded large populations of hippocampal neurons and local field potentials from Egyptian fruit bats engaged in free, spontaneous flight — a feat made possible by a SpikeGadgets Neuropixels Datalogger Headstage light enough (about 19 g) for the bats to carry while flying untethered, with all neural data logged to an onboard SD card during flight. During rest, the team identified time-compressed forward and reverse replay of flight trajectories that coincided with hippocampal sharp-wave ripples, much as replay does in rodents. But the observed replay events were mostly of the remote kind and their speed scaled with the length of the trajectory — observations that challenge current, rodent-derived models of how replay works (Forli et al., 2025, Nature). It’s the first wireless Neuropixels recording of large-scale neural activity in a freely flying mammal, and it suggests that some of what we think we know about replay from rats and mice may not generalize across species.
Building cognitive graphs in the rat prefrontal cortex
At Janelia Research Campus, Maksym Manakov, Mikhail Proskurin, and colleagues in Alla Karpova’s lab used SpikeGadgets’ Neuropixels datalogging hardware to record simultaneously from hundreds of neurons in the rostral anterior cingulate cortex of freely moving rats as the animals worked out, unguided, a hidden sequential pattern in their environment. As rats uncovered the underlying structure through trial and error, population activity in rACC organized itself into compact, structured “cognitive graphs” anchored to the start and end states of self-generated action sequences — representations flexible enough to be refined as the animals learned and to reflect relational similarities across different task contexts (Manakov et al., 2025, bioRxiv). It’s a rare look at prefrontal population coding during genuinely open-ended, self-directed behavior, made possible by recording high channel counts while the animal was free to explore on its own terms.
Marmosets talking over each other — and over background noise
Two independent groups used Neuropixels probes and SpikeGadgets dataloggers this year to ask how the primate brain handles naturalistic vocal exchanges in freely moving marmosets.
Kevin Johnston, Rebekah Gilliland, Raymond Wong, and Stefan Everling (Western University) chronically implanted Neuropixels probes in the anterior cingulate cortex of freely moving common marmosets, with data logged via a SpikeGadgets datalogger while the animals exchanged long-distance “phee” calls with a virtual conspecific. Many ACC neurons modulated their firing around both perceived and self-produced calls, implicating the ACC in the integration of vocal perception and production during real conversational exchanges (Johnston et al., 2025, Journal of Neuroscience).
Around the same time, Arthur Lefevre, Cory Miller, and colleagues at UC San Diego tackled a related topic: the “cocktail party problem.” Marmosets implanted with Neuropixels probes and wearing a SpikeGadgets neurologger, along with portable microphones, were recorded directly in their home cage while conversing with a partner amid a noisy colony room. ACC neurons distinguished the partner’s calls from background callers, and this figure/ground selectivity held up even when background vocalizations overlapped with the conversation — suggesting the ACC helps resolve competing auditory streams in a genuinely social, ecological setting (Lefevre et al., 2025, bioRxiv).
Available now: next-generation 2.0 and NHP Neuropixels hardware
Notably, all four of these studies were done with Neuropixels 1.0 probes. Since then, SpikeGadgets has extended the same wireless-datalogging approach to newer probe generations and larger species. The Bennu Neuropixels Datalogger now supports both Neuropixels 1.0 and 2.0 probes for high-density, untethered recording, and the Satori NHP Neuropixels Datalogger is purpose-built for freely-moving, large-brain models, including non-human primates, pairing with the longer-shank Neuropixels NHP probes. In other words, the kind of recordings described above — replay in flight, sequence learning, naturalistic conversation — should soon be possible with denser probes and in larger-brained species than were accessible even a year ago.
Untethered and Unbothered: Neuropixels Meets SpikeGadgets in Freely Moving Animals
High-density Neuropixels probes can record from hundreds to thousands of neurons at once — but that density used to come at the cost of freedom of movement, since the probe’s data had to be streamed out through a cable to a fixed acquisition system. SpikeGadgets’ Neuropixels-compatible dataloggers remove that constraint: the systems write probe data straight to onboard storage, so the animal can carry the whole recording system with it, wirelessly, while behaving naturally. Over the past twelve months, four papers have put this combination — Neuropixels probes plus a SpikeGadgets datalogger — to work in freely moving and freely flying animals, and the results are reshaping some long-standing assumptions about hippocampal replay, prefrontal sequence coding, and vocal communication. Here’s a roundup.
Replay looks different in a bat than in a rat
The most striking of this year’s crop comes from Michael Yartsev’s lab at UC Berkeley. Angelo Forli, Wudi Fan, Kevin Qi, and Yartsev wirelessly recorded large populations of hippocampal neurons and local field potentials from Egyptian fruit bats engaged in free, spontaneous flight — a feat made possible by a SpikeGadgets Neuropixels Datalogger Headstage light enough (about 19 g) for the bats to carry while flying untethered, with all neural data logged to an onboard SD card during flight. During rest, the team identified time-compressed forward and reverse replay of flight trajectories that coincided with hippocampal sharp-wave ripples, much as replay does in rodents. But the observed replay events were mostly of the remote kind and their speed scaled with the length of the trajectory — observations that challenge current, rodent-derived models of how replay works (Forli et al., 2025, Nature). It’s the first wireless Neuropixels recording of large-scale neural activity in a freely flying mammal, and it suggests that some of what we think we know about replay from rats and mice may not generalize across species.
Building cognitive graphs in the rat prefrontal cortex
At Janelia Research Campus, Maksym Manakov, Mikhail Proskurin, and colleagues in Alla Karpova’s lab used SpikeGadgets’ Neuropixels datalogging hardware to record simultaneously from hundreds of neurons in the rostral anterior cingulate cortex of freely moving rats as the animals worked out, unguided, a hidden sequential pattern in their environment. As rats uncovered the underlying structure through trial and error, population activity in rACC organized itself into compact, structured “cognitive graphs” anchored to the start and end states of self-generated action sequences — representations flexible enough to be refined as the animals learned and to reflect relational similarities across different task contexts (Manakov et al., 2025, bioRxiv). It’s a rare look at prefrontal population coding during genuinely open-ended, self-directed behavior, made possible by recording high channel counts while the animal was free to explore on its own terms.
Marmosets talking over each other — and over background noise
Two independent groups used Neuropixels probes and SpikeGadgets dataloggers this year to ask how the primate brain handles naturalistic vocal exchanges in freely moving marmosets.
Kevin Johnston, Rebekah Gilliland, Raymond Wong, and Stefan Everling (Western University) chronically implanted Neuropixels probes in the anterior cingulate cortex of freely moving common marmosets, with data logged via a SpikeGadgets datalogger while the animals exchanged long-distance “phee” calls with a virtual conspecific. Many ACC neurons modulated their firing around both perceived and self-produced calls, implicating the ACC in the integration of vocal perception and production during real conversational exchanges (Johnston et al., 2025, Journal of Neuroscience).
Around the same time, Arthur Lefevre, Cory Miller, and colleagues at UC San Diego tackled a related topic: the “cocktail party problem.” Marmosets implanted with Neuropixels probes and wearing a SpikeGadgets neurologger, along with portable microphones, were recorded directly in their home cage while conversing with a partner amid a noisy colony room. ACC neurons distinguished the partner’s calls from background callers, and this figure/ground selectivity held up even when background vocalizations overlapped with the conversation — suggesting the ACC helps resolve competing auditory streams in a genuinely social, ecological setting (Lefevre et al., 2025, bioRxiv).
Available now: next-generation 2.0 and NHP Neuropixels hardware
Notably, all four of these studies were done with Neuropixels 1.0 probes. Since then, SpikeGadgets has extended the same wireless-datalogging approach to newer probe generations and larger species. The Bennu Neuropixels Datalogger now supports both Neuropixels 1.0 and 2.0 probes for high-density, untethered recording, and the Satori NHP Neuropixels Datalogger is purpose-built for freely-moving, large-brain models, including non-human primates, pairing with the longer-shank Neuropixels NHP probes. In other words, the kind of recordings described above — replay in flight, sequence learning, naturalistic conversation — should soon be possible with denser probes and in larger-brained species than were accessible even a year ago.