Article Figures & Data
Figures
- Figure 1.
Schematic drawing of the Neuropixels housing design. Top row from left to right: internal mount (A), external case (B), and tubing (C). Middle row: assembly and relative position of the parts (D–F). Bottom row: a single probe and a headstage housed in half of the housing (G–I). Two halves glued together enable a dual site implant. The tubing, but not the recording cables inside, will be pulled during rat movement.
- Figure 2.
Neuropixels implantation and explantation. A, Single Neuropixels probe secured in an internal mount. B, The internal mount is secured within the housing by four screws. C, Singular probe with lid on the housing (inset shows the lid separating two external cases). D, A close-up view of the headstage and the probe flex when secured in the housing. The flex cable forms an “s” shape. E, Singular probe housing sealed with Vaseline in the bottom, ready for implantation. F, Probe lowering to target, showing anchor screw placement on the rat skull. G, A headstage tucked in the housing when implanted. H, Recording cables protected with metal spring. I, A connected neuropixels probe rat. J, Cone in the chamber ceiling to protect cables. K, A connected rat in the chamber. L, A probe that was just explanted.
- Figure 3.
Paired associate learning task and set shift task. In the PAL task (left), rats were shown a pair of images on a monitor in two of three locations. Touching one of the images led to a sucrose pellet reward and touching the other led to a 10 s time-out. The correct image was determined by the intersection of image identity and location, e.g., the flower is correct only in the left spatial location and the airplane only in the rightmost. In the set shift task, rats needed to learn to nosepoke into an illuminated port (Light rule) or into a port on one specific side of the chamber (Side rule) to earn a reward. The rule shifts without warning and must be discovered through trial and error.
- Figure 4.
LFP recordings from the PAL and set shift tasks. Shown are representative raw traces of every 10th channel along the first bank of the shank (A,B) and power spectra of the signals along the first bank (D,E). C,F Raw traces and power from a reused probe.
- Figure 5.
Spike recordings from new and reused Neuropixels 1.0 probes. Shown are representative putative single units recorded from a new probe and a reused probe. A,B Raw traces, inter-spike intervals, and autocorrelation of four representative putative units. C,D Each shows all the “good” units from a single recording.
- Figure 6.
Signal noise and θ-band power in representative new, reused, and twice reused Neuropixels 1.0 probes. A, Distribution of RMS noise for all 384 channels, for new and reused probes. B, Distribution of θ power over 1/f noise for all channels that had positive power values (suggesting there were clean physiological θ band oscillations), for new (144 channels), reused (333 channels), and twice reused (158 channels) probes. Data were collected from the probes implanted in the prelimbic cortex in three rats. For each probe condition, over 75 min worth of data across multiple sessions was used.
- Figure 7.
Unit counts from 3 consecutive days with new and reused Neuropixels 1.0 probes. Shown are unit counts of every 10 channels along the shank for representative new and reused probes. A, Recordings from days 13, 14, and 15 postimplantation of a new probe. B, Recordings from days 13, 14, and 15 postimplantation of a reused probe. C, Recordings from days 27, 28, and 29 postimplantation of a twice reused probe. D, Summary of A–C, showing total unit counts along the probe for the 3 consecutive days shown in each panel. Although there was a clear drop from these new to reused probes, there was no evident loss across days.
- Figure 8.
Waveforms of representative units are stable within a recording session. Panel A shows the first 100 spikes of selected units from one session. Panel B shows the last 100 spikes of the same units from the same session. The waveforms remain similar over the recording (R = 0.9929).
- Figure 9.
Capture of behaviorally relevant units with a reused probe during the set shift task. A, Spike rates of all units during correct versus incorrect trials. B, All units that had at least 1.5 times higher firing rates during correct versus incorrect trials. C, All units that had at least 1.5 times higher firing rates during incorrect versus correct trials. Data were from the prefrontal cortex probe in a representative set shift session of one rat. Spike rates were averaged over the time period between cue onset and response.
- Figure 10.
Representative units that had similar spiking patterns across days. Units A and D were located near the same channel of the probe and had similar spiking modulation during the incorrect and correct trials. The same is true for putative for units B and E and C and F. Waveforms were highly correlated across days (R = 0.9558). While not conclusive, the concordance of the waveform, anatomic location, and behavioral modulation suggests these are the same cells being tracked across days.
- Figure 11.
Coherence between the PFC and the striatum during the set shift task. A, θ coherence between the PFC and the striatum in the set shift task during Correct trials. B, θ coherence between the PFC and the striatum in the set shift task during Incorrect trials. C, Difference in θ coherence between the PFC and the striatum in correct versus incorrect trials, showing interactions between dorsal and ventral PFC and striatum. Plotted are example data from one task session (∼45 min of recording). The 384 channels were grouped into 38 clusters with 10 channels as one cluster. A sliding window size of 0.5 s and a step of 0.1 s were used to calculate coherence. The corrected coherence was calculated from raw coherence subtracted by coherence from shuffled raw data.
In this issue
