Neurotassel: mouse learning and chronic recordings, 2019
Lab or organization
No atlas brief linked yet.
Mouse recordings with Neurotassels
The 2019 study uses Neurotassel filament hardware for medial prefrontal cortex recordings in mice. Recording experiments were performed at the Chinese Academy of Sciences’ Institute of Neuroscience; histology was performed at China’s National Center for Nanoscience and Technology. This is secondary coverage alongside the US-first catalog.
Learning cohort: six mice, 13 training days
Six mice received 16-channel arrays and began training one week after implantation. In an olfactory delayed pair-association task, head-fixed mice retained a sample odor across a five-second delay and licked for a water reward when the odor pair matched. Training lasted 13 days, rather than 13 uninterrupted days of continuous acquisition.
The study recorded 121 neurons across training. Of these, 21% were repeatedly recorded for more than five days and 12% for more than ten days. Stable tracking was therefore partial, not every neuron surviving across every session. Waveform clustering supported the authors’ same-neuron assignments; this is not independent cellular labeling.
The paper compares the tracking distribution with microwire-tetrode recordings, including 730 neurons from eight mice. These are separate recordings, not the same six animals receiving both devices in a randomized crossover trial.
Chronic cohort: seven mice, weeks 3-6
A separate group of seven mice received 16-channel devices. Recordings isolated 36 neurons at week 3 and 28 at week 4. At week 5, two mice lost spike signals; 25 neurons were recorded from the remaining five mice. Week 6 yielded 19 neurons from five mice, nine tracked across the entire reported interval. The paper does not identify the cause of the two animals’ signal loss.
The endpoint is six weeks after implantation, with weekly recordings beginning at week 3. It is not six weeks of uninterrupted same-neuron recording in all seven animals. Five-week histology compared Neurotassels with contralateral silicon probes and reported minimal neuronal loss around the filaments. That sampled comparison does not prove absence of tissue response everywhere.
Separate optical experiment
61-channel arrays on sharpened optical fibers recorded during 473 nm optogenetic stimulation in Thy1-ChR2 mice. The paper shows spontaneous and light-evoked spikes, seven recorded neurons in its summarized response and onset/offset optical transients handled with median subtraction. This is optical activation in transgenic mice, not electrical stimulation through a human implant.
Limits
The methods say experiments were not randomized, investigators were not blinded and no statistical method predetermined sample sizes. Learning, chronic recording and optical experiments are separate evidence groups. The preliminary 1,024-channel claim does not enlarge the chronic 16-channel cohort or establish 1,024 simultaneously useful units.
No human clinical outcome, at-home BCI or lifelong implant reliability is established. No atlas link is assigned to an unrelated existing laboratory brief.