Interhemispheric competition during sleep

Abstract

Our understanding of the functions and mechanisms of sleep remains incomplete, reflecting their increasingly evident complexity^1,2,3. Likewise, studies of interhemispheric coordination during sleep^4,5,6 are often hard to connect precisely to known sleep circuits and mechanisms. Here, by recording from the claustra of sleeping bearded dragons (Pogona vitticeps), we show that, although the onsets and offsets of Pogona rapid-eye-movement (REM_P) and slow-wave sleep are coordinated bilaterally, these two sleep states differ markedly in their inter-claustral coordination. During slow-wave sleep, the claustra produce sharp-wave ripples independently of one another, showing no coordination. By contrast, during REM_P sleep, the potentials produced by the two claustra are precisely coordinated in amplitude and time. These signals, however, are not synchronous: one side leads the other by about 20 ms, with the leading side switching typically once per REM_P episode or in between successive episodes. The leading claustrum expresses the stronger activity, suggesting bilateral competition. This competition does not occur directly between the two claustra or telencephalic hemispheres. Rather, it occurs in the midbrain and depends on the integrity of a GABAergic (γ-aminobutyric-acid-producing) nucleus of the isthmic complex, which exists in all vertebrates and is known in birds to underlie bottom-up attention and gaze control. These results reveal that a winner-take-all-type competition exists between the two sides of the brain of Pogona, which originates in the midbrain and has precise consequences for claustrum activity and coordination during REM_P sleep.

Main

In mammals, cortical electroencephalograms during sleep can be decomposed into rapid-eye-movement (REM) sleep, which is characterized by desynchronized electroencephalographic signals accompanied by rapid eye movements^7,8,9, and non-REM (NREM) sleep, which is characterized by slow-wave activity (and hence is also called slow-wave (SW) sleep). In rodents and humans, NREM sleep has been implicated in the reactivation and consolidation of certain forms of memories^10,11,12,13. The possible functions of REM, however, remain largely speculative (for example, unlearning¹⁴), although some evidence^15,16,17 suggests a possible link with emotional memory. Biphasic sleep also exists in birds¹⁸, reptiles^19,20,21 and fish²², suggesting the possibility of common roots (at least as old as the vertebrate lineage (500 million years)). (Biphasic sleep patterns have also been observed in some invertebrates²³). If so, comparative approaches in systems that represent diverse animal lineages might help us to better understand not only the evolution of sleep, but also its functions and mechanistic underpinnings.

In Pogona vitticeps, a diurnal agamid lizard, the two phases of a regular electrophysiological sleep rhythm (as recorded in the dorsal-ventricular ridge or DVR) consist of local field potentials dominated by sharp-wave ripples that occur irregularly every 0.5–2 s for about a minute, followed by faster awake-like activity that co-occurs with rapid eye movements¹⁹, also for about a minute. These two activity modes alternate regularly throughout the night¹⁹. Because of the similarities between these two electrophysiological sleep phases and mammalian SW and REM sleep, we identify them as SW-like and REM-like (thereafter respectively named SW and REM_P). (Note that our nomenclature is descriptive, and does not necessarily imply, for lack of knowledge at this point, functional or mechanistic identity with mammalian sleep states). The dominant features of SW activity in the DVR—the sharp-wave ripples—are produced in the claustrum, and are detectable in the adjacent DVR as a propagating wave²¹. Here we set out to examine the nature of REM_P signals in Pogona, using electrophysiological recordings in vivo. Recording from the left and right claustra simultaneously, we observed differences in the nature of interhemispheric coordination between SW and REM_P sleep. These differences, in turn, revealed an ongoing competition between the two hemispheres during REM_P sleep—but not during SW sleep—and identified a role for a pair of midbrain pre-isthmic nuclei, which were until now not known to be involved in sleep.

Claustrum activity during REM_P sleep

Recordings were made from the claustrum (mostly laterally) (Fig. 1a) or the adjacent anterior DVR, using silicon probes (Methods). During sleep, local field potential (LFP) activity alternated regularly between SW and REM_P episodes (Fig. 1b, top), with each full sleep cycle lasting 1.5–2.5 min. SW sleep was characterized by the irregular generation of sharp-wave ripples, about once per second on average, as described previously^19,21.

**Fig. 1: REM_P activity is characterized by sharp negative field-potential events.**

REM_P activity in the claustrum consisted mainly of sharp downward (negative) extracellular potentials, occurring at irregular intervals (Fig. 1b, bottom left). Those events (which we call SNs, for sharp and negative) were quite consistent in shape but variable in both amplitude and inter-event interval (IEI; Fig. 1b, bottom). IEIs, time-to-trough and amplitude distributions over about 190,000 events are shown in Fig. 1c for one lizard (from 9 h of one night). The IEI distribution was skewed with a mode at about 40 ms (median: 60.2 ms; [25th, 75th] percentiles: [39.8, 110.5] ms). Previous results^19,21 showed that REM_P (in the DVR and claustrum) is dominated by LFP power in the 20-Hz (beta) band. The instantaneous power in the beta band (measured in a scrolling 10-s window) and the SN rate (Fig. 1d) were indeed well correlated, consistent with the mode of the IEI distribution. REM_P activity in the claustrum is thus dominated by SNs and their interval statistics (3 animals, 9 nights; Extended Data Fig. 1a–d).

Single units isolated from claustrum recordings (Fig. 1e, left) typically fired 0–2 action potentials per SN, aligned to the SN's descending phase. Given the short duration of that phase, the action potentials of different units occurred within a few milliseconds of one another at most (Fig. 1e and Extended Data Fig. 1e). SN waveforms are therefore extracellular potentials that reflect a net-depolarizing current in claustrum neurons; this, in turn, probably underlies the synchronized firing of the depolarized units. We will show that this phasic field potential must result from input to the claustrum, and not solely from intrinsic and coordinated properties of claustral units themselves.

Bilateral coordination during sleep

We next recorded simultaneously from the left and the right claustra to examine the bilateral coordination of sleep-related activities. (Whereas a corpus callosum exists only in placental mammals, reptiles have several forebrain commissures²⁴.) The regular cycling of SW and REM_P was precisely synchronized across the two sides (Fig. 2a and Extended Data Fig. 2a), but the fine coordination of the two claustra differed greatly between the two phases. During SW, the sharp-wave ripples—characteristic of this phase—were not synchronized bilaterally, and their amplitudes did not covary (Fig. 2b and Extended Data Fig. 2b,c), consistent with the independent generation of sharp-wave ripples in each claustrum²¹ and the absence of reported contralateral projections between claustra in mammals²⁵ and in Pogona²¹.

**Fig. 2: REM_P SNs are tightly coordinated between left and right claustra, but SW sharp-wave ripples are not.**

During REM_P, however, SNs in each claustrum were precisely mirrored (in timing and amplitude) on the opposite side and in sweeps from one claustrum triggered on SNs produced in the other (Fig. 2c). Over hundreds of thousands of SNs analysed, this tight correspondence between left and right sides was observed in 85 to 90% of the clearly detectable events (95–100 percentiles of SN amplitudes).

Although precisely coordinated, SNs from either side were not simultaneous, but offset by a delay of about 20 ms (median: 19.3 ms; [25th, 75th] percentiles: [15.4, 24.5] ms) with one side or the other leading (Fig. 2c). This delay was consistent throughout the night, across nights and across lizards (12 nights, 7 animals; Fig. 2d). SNs on both sides showed correlated amplitude variations (Fig. 2c), suggesting a common cause. The leading side would switch from one side to the other (Fig. 2c,d), but typically not on an event-by-event basis, as shown in a scrolling cross-correlation between left and right LFPs (Fig. 2e): switches occurred between REM_P episodes (as here between the first and second), or once (occasionally twice) within single REM_P episodes (as here during the second, fourth or sixth). The full sleep period from which this segment is shown is in Extended Data Fig. 3a.

The leading side at a given moment could usually be predicted by comparing the left- and right-claustrum LFPs: the side with dominant beta-band power tended to lead. In the second, fourth and sixth REM_P episodes (Fig. 2e), for example, the crossing of the power curves corresponds to a leading-side switch. In the first and seventh, by contrast, one side dominated and led throughout. This correspondence between leadership and LFP power dominance was corroborated by an analysis of individual SN pairs: the SN with the larger amplitude generally led its contralateral counterpart (Fig. 2f).

The combination of matching bilateral pairs of waveforms, fixed positive or negative lags with the same absolute value and temporal lead of the stronger side suggested the possibility of competition between the two claustra, in which the stronger side at a given moment imposes its output on the weaker one, with a delay. By this hypothesis, leadership would depend on instantaneous variations of the relative 'strengths' of activity on the two sides, and the delay could be due to signal propagation and synaptic transfer in mirror-symmetric circuits. The suppression of the weaker side's own output—and its replacement by the stronger side's—suggested a winner-take-all type of bilateral competition.

Interhemispheric competition during REM

To understand the nature of this hypothetical competition, we examined the transitions between leading sides. Two successive REM_P episodes are shown in Fig. 3a, together with three short segments of the left and right LFP traces from the second episode: in the blue frame, left power dominates and the SNs on the left lead; in the red frame, the converse happens; in the yellow frame, leadership is not settled. The yellow segment corresponds precisely to the time when the left and right beta-band powers are balanced (top traces).

**Fig. 3: Dominance switching obeys slow dynamics.**

By scoring an instantaneous-leadership tendency of one side versus the other (Fig. 3b and Methods) and applying it to nine hours of sleep (around 208,000 SNs), we reveal high-density domains that correspond to identifiable states: in grey is SW, when the two sides are not correlated (see Fig. 2b); in red and blue are states in which one side leads the other (right or left, respectively); and in yellow is a state in which dominance switches ('competition'). The trajectory shown in black (Fig. 3b) corresponds to the first REM_P episode in Fig. 3a; it remains on the red side from one SW state to the next. The cyan trajectory corresponds to the second REM_P episode, in which dominance starts on the left side, enters the unsettled state (yellow) and finishes on the right side before moving back to SW. Over all recordings, the competitive (yellow) state was the least common of the three REM_P states—for example, over one night's 210 REM_P episodes, plotted in Fig. 3c, the average REM_P episode consisted of 22 s with left dominating, 26 s with right dominating and 4 s as unsettled. When one side dominated, it tended to do so over many successive SNs. Thus, if the left and right sides do compete for dominance, they must do so with dynamics slower than individual SNs, suggesting that the drive causing left–right competition has different mechanisms than those that underlie the production of SNs.

The fraction of REM_P time spent in each one of these three states (left dominating, right dominating or competition) is plotted for one night (Fig. 3c). This plot suggests that nearby REM_P episodes are not entirely independent of one another; a temporal correlation exists such that the fraction of time that either side dominated waxed and waned throughout the night (Extended Data Fig. 4). Moreover, if one side led for the longest in a REM_P cycle, it tended to do so over many successive sleep cycles (sometimes more than ten) in the last two to three hours of the night (Fig. 3d). This long-lasting side dominance at the end of the night thus indicates the existence of a slow competitive process between the two sides of the brain, on a timescale longer than that of the sleep cycle, and increasingly long as the night proceeds.

The great majority of REM_P episodes contained either 0 or 1 dominance switch (Fig. 3e). REM_P episodes with no switch lasted about 60 s (mean, 57 s; s.d., 16.6 s; n = 59), against 90 s (mean, 88.5 s; s.d., 19.6 s; n = 103) for those with one switch. In REM_P episodes with one switch, the durations of each sub-period of dominance (left-dominant and right-dominant) self-adjusted such that the duration of the REM_P period remained more or less constant (diagonal trend for black points, Fig. 3f).

A midbrain nucleus active during REM_P

Guided by retrograde tracing from the claustrum, we searched for circuits that might underlie the left–right claustrum competition during REM_P. We found no evidence for direct connections between the two claustra, and despite anatomical evidence for ipsi- and contralateral projections from the amygdala and cortex to the claustrum (Extended Data Fig. 5a,b), lesions in or ablation of either had no effect on the bilateral switching of REM_P claustrum activities (Extended Data Fig. 5c–f). We next examined areas at the mid-to-hindbrain junction, given their known involvement in REM_P control in mammals^1,3,17. Recordings were obtained after traversing part of the optic tectum (green; Fig. 4a). We detected, from mesencephalic recording sites ventral to the caudal end of the optic tectum (Fig. 4b–e), REM_P sleep activity that matched precisely activity recorded in the ipsilateral claustrum. These recordings were from a nucleus that contained large, sparsely distributed GABAergic (Vgat⁺) and parvalbumin-positive (Pvalb⁺) somata at the edge of the tegmentum, next to a tecto-thalamic fibre tract (Fig. 4b–d and Extended Data Fig. 6). We suggest that this is likely to be the homologue of the nucleus isthmi pars magnocellularis (Imc) in avians, in which it has been intensively studied^26,27,28,29.

**Fig. 4: A mesencephalic nucleus coactive with the claustrum during REM_P.**

Simultaneous recordings were made from both claustra (only the left shown) and the left Imc (Fig. 4e and Extended Data Fig. 7a–c). We noted a marked correspondence between the two (Fig. 4e–g); SNs characteristic of REM_P in the claustrum aligned precisely with the Imc waveforms, which themselves corresponded to brisk firing bursts of Imc neurons (Fig. 4h). Imc field potentials led those in the claustrum by 30 ms ipsilaterally (median: −30 ms; [25th, 75th] percentiles: [−33.0, −27.0] ms) or 50 ms contralaterally (median: −45 ms; [25th, 75th] percentiles: [−51, −26] ms) (Fig. 4g,i and Extended Data Fig. 7b,c), a difference consistent with the 20 ms delay between the left and the right claustra (Fig. 2). This suggested that each claustrum receives excitatory synaptic drive coupled to Imc activity in the midbrain, and that the apparent competition between claustra (Figs. 2 and 3) might instead be the expression of bilateral competition in the midbrain.

Indeed, although the Imc was clearly active during REM_P, its activity on a given side was strongest when the ipsilateral claustrum was also dominant (that is, phase-leading) (Fig. 4j). This suggested that the two Imcs compete with one another, and that claustrum dominance patterns reflect the results of Imc competition.

We recorded simultaneously from both Imcs (Fig. 4k and Extended Data Fig. 7d–j); sorted units from each side formed alternating bursts (two transitions in this REM_P cycle; Fig. 4k; see also Extended Data Fig. 7g), the timing of which matched power changes in the claustrum. As predicted, the instantaneous power in the claustrum increased when the ipsilateral Imc was dominant. These results suggest that inter-claustral competition in fact results from bilateral competition within the midbrain, in which the Imc has a key role.

The Imc is required for bilateral competition

To test this hypothesis, we lesioned the Imc unilaterally using ibotenic acid (IBA) (Fig. 5a,b and Extended Data Fig. 8b) (n = 3 animals). The effects of these lesions were clear: the claustrum on the side of the lesioned Imc lost its ability to lead or dominate (Fig. 5c), and all REM_P episodes were now dominated by the intact side, contralateral to the lesion, as seen in two hours of one night (Fig. 5c; see full 9 h in Extended Data Fig. 8a) and in the now unimodal claustrum phase-lag distributions (Fig. 5d) (3 animals, 8 nights). Correspondingly, the phase-leading claustrum (contralateral to the Imc lesions) was also the one with dominating beta power (blue; Extended Data Fig. 8a). Notably, lizards with unilateral Imc lesions also exhibited shorter REM_P episodes (mean duration in control animals: 65 s, n = 12, versus 33 s in lesioned animals, n = 8; P = 0.000028, t = −6.992; Welch's two-sided t-test) and an excess of SW, resulting in less time spent in REM_P sleep overall (mean REM_P duration: 4:40 h over 9 h of sleep in control animals, versus 1:37 h over the same duration in lesioned animals; P = 0.000062, t = −7.204), suggesting that the Imc also has a role in the SW–REM_P transition.

**Fig. 5: The Imc is required for side dominance during REM_P.**

Discussion

We have described competitive dynamics between the two sides of the brain, which are detectable in the claustrum during REM_P but not during SW sleep. Claustrum activity during REM_P sleep consisted of very brief synchronized firing bursts coincident with sharp negative LFP events (SNs). Contrary to sharp-wave ripples during SW sleep, SNs were tightly correlated across the two claustra, with precise time delays and rapid switches of leadership correlated with reversals of amplitude dominance, suggesting a winner-take-all type of competition between the two claustra. This competition, however, did not have a telencephalic origin: the dominance of either claustrum required the integrity of the ipsilateral member of a bilateral pair of GABAergic mesencephalic nuclei (Imc). Hence, the apparent claustral competition during REM_P sleep is the expression of competition in the midbrain, transferred bilaterally to the two claustra. The proposed functional circuits that underlie this relationship are schematized in Fig. 5e: in this model, the two sides of the midbrain compete during REM_P, and side dominance depends on Imc integrity (Figs. 4 and 5). Each side of the midbrain projects to both claustra (possibly, but not necessarily or exclusively, via the thalamic nucleus rotundus^30,31) such that, as one side dominates, left and right claustrum LFPs covary (owing to their common source) but with a time lag (owing to a longer decussating path). The smaller SN amplitude on the contralateral (lagging) side also suggests a lesser gain or reliability of the midline-crossing pathway. During the brief transitions when no midbrain side dominates, the left and right claustra are coactive, which results in disordered SN patterns in both claustra (Fig. 3a).

The Imc is part of a complex of cholinergic, glutamatergic and GABAergic isthmic nuclei that has been studied mainly in birds^26,28,29,32 and is implicated in bottom-up attention and gaze control. The Imc, which is embryologically a pre-isthmic nucleus³³, mediates a type of broad lateral inhibition of excitatory neurons in both the ipsilateral optic tectum and the nucleus isthmi pars parvocellularis (Ipc)—a companion cholinergic–glutamatergic nucleus that forms a positive feedback loop with the ipsilateral tectum^{27,29,31,32,34}. When two stimuli fall onto one retina, these circuits undergo a winner-take-all-type competitive interaction such that responses to the stronger stimulus are selected, and attention and gaze are directed towards it, rather than to a weighted mean of the two stimuli. The Imc, by virtue of its heterotypic connectivity with the optic tectum and the Ipc, underlies this competition^27,28. This work was at first concerned with the selection of competing visual stimuli falling on the same retina, but experiments have now established that competition also takes place between contralateral stimuli, and even between sensory modalities³⁵, although the underlying circuits are not well understood. Likely homologues of these avian (reptilian) isthmic nuclei are found in fish^36,37,38, amphibians³⁹ and non-avian reptiles^40,41, as well as in mammals (parabigeminal nucleus), in which a role in visual attention has been hypothesized⁴².

The isthmo–claustral connection uncovered here is notable in part because some of the claustrum's hypothesized functions are attentional^43,44. But our results raise many new questions. First, what pathways underlie the bilateral mesencephalic competition (Fig. 5e)? Although the Imc is inhibitory (Fig. 4c) and necessary for this competition (Fig. 5), we are not aware of contralateral Imc projections. Second, what pathways link Imc, a GABAergic nucleus, to the activation of both claustra? The high correlation between isthmic and claustral LFPs during REM_P suggests high-gain excitatory pathways between those areas, or drive from a common source, the dominance of which depends on the integrity of the ipsilateral Imc. Third, what is the functional importance of the tight correlation and ±20-ms delays between the claustra during REM_P? Although there is no evidence for direct projections between the claustra, bilateral claustral projections to the telencephalon exist^25,44, such that convergence on common targets is plausible. If so, the delays might, in conjunction with the activation of spike-timing-dependent plasticity rules, have a role in synaptic homeostasis^16,45,46 or plasticity during REM_P. These hypotheses are especially intriguing when considering that the temporal profiles of dominance are different early and late in the night (Fig. 3), and that sharp-wave ripples are uncorrelated bilaterally during SW (Fig. 2 and Extended Data Fig. 2). Fourth, several species of birds and marine mammals show unilateral sleep^47,48. Could the same mesencephalic circuits underlie this (slower) alternation? Fifth, side dominance is temporary and balanced between the two sides over each night. What is the cause of the recurrent transfer of side dominance (Fig. 3), presumably located upstream of the Imc? Because the duration of one-sided dominance varies between a fraction of one sleep cycle and many cycles, it probably does not originate in the circuits that cause SW–REM_P alternation. Sixth, what role does the isthmic complex have in SW–REM_P transitions? Imc lesions revealed a decrease in REM_P and a concurrent increase in SW sleep, indicating an interaction with circuits that control the ultradian sleep cycle. Finally, how are activities between the isthmi and the claustra coordinated in the awake state, given that brain activity during REM_P sleep most resembles that in the awake state^9,17?

Hence, the claustrum—a well-described^44,49 but functionally poorly understood area of the telencephalon—seems to have a role in SW sleep in reptiles and mammals^21,50, and reflects competitive brain dynamics during REM_P sleep (this paper), with origins in the midbrain. Antiphase activity between the two hemispheres has been identified in running rodents and during REM with fast rhythms (140-Hz 'splines') in superficial retrosplenial cortex⁶, but the mechanisms that underlie this relationship are unknown. The relative simplicity of the reptilian model may thus help to reveal not only some mechanisms of sleep, but also some of the functions of the claustrum and of the midbrain in sleep.

Finally, our data emphasize the value of investigating non-mammalian sleep. Comparing stages of sleep (for example, SW and REM), electrophysiological waveforms (for example, SWRs) or even brain areas (for example, the claustrum) across distant species can be difficult; indeed, the terminology of sleep originated with data from humans and cats, the phenomena to be compared are usually multiparametric and sleep is polymorphic even among mammals. Matches across species are thus often only partial. But by comparing data using exhaustive and increasingly precise approaches, hypotheses about structural homologies can now be tested more accurately, at least among vertebrates. This should in turn help us to determine which phenotypic and mechanistic attributes of sleep are common across species, how they came to diverge or converge, and eventually whether they subserve similar functions.

Methods

Animals

Lizards (P. vitticeps) of either sex, weighing 150–250 g, were obtained from our institute colony and selected for size, weight and health status. The lizards were housed in our state-of-the-art animal facility. All experimental procedures were approved by the relevant animal welfare authority (Regierungspräsidium Darmstadt, Germany) and conducted following the strict federal guidelines for the use and care of laboratory animals (permit numbers V54-19c20/15-F126/1005_1011 and 2006).

Lizard surgery for chronic recordings

Twenty-four hours before surgery, lizards were administered analgesics (butorphanol, 0.5 mg kg⁻¹ subcutaneously; meloxicam, 0.2 mg kg⁻¹ subcutaneously) and antibiotics (marbofloxacin, marbocyl, 2 mg kg⁻¹). On the day of surgery, anaesthesia was initiated with 5% isoflurane, and maintained with isoflurane (1–4 vol%) after intubation. The lizards were placed in a stereotactic apparatus after ensuring deep anaesthesia (absence of corneal reflex). Body temperature during surgery was maintained at 30 °C using a heating pad and an oesophageal temperature probe. Heart rate was continuously monitored using a Doppler flow detector. The skin covering the skull was disinfected using 10% povidone-iodine solution (Betai-NE) before removal with a scalpel. A cranial window was made to reach the various regions of interest; the dura and arachnoid were then removed with fine forceps and scissors. The pia was carefully removed over the area of electrode insertion (dorsal cortex for claustrum and optic tectum for Imc recordings). The exposed skull was covered with a layer of ultraviolet (UV)-hardening glue, and the stripped ends of insulated stainless steel wires were secured in place subdurally with UV-hardening glue, to serve as reference and ground.

Silicon probes were mounted on a Nanodrive (Cambridge Neurotech) and secured to a stereotactic adaptor for insertion. On the day after the surgery, probes were slowly lowered into the tissue (claustrum: 0.8–1.2 mm; Imc: 2.5–3.0 mm). For Imc recordings, the probe was advanced in small steps over two to three days until the signal began to show the area-typical signature.

The brain was covered with Duragel (Cambridge Neurotech) followed by Vaseline, and the probes were secured with UV-hardening glue. After surgery, lizards were released from the stereotactic apparatus and left on a heating pad set to 30 °C until full recovery from anaesthesia.

In vivo electrophysiology

Two to three days before surgery, lizards were habituated to a sleep arena, which was itself placed in a 3 × 3 × 3-m electromagnetic-shielded room. One hour before lights off (t_off = 18:00/19:00, winter/summer), lizards were placed into the arena and left to sleep and behave naturally overnight. They were returned to their home terrarium 3 to 4 hours after lights on (t_on = 06:00/07:00, winter/summer). The lizards then received food and water. Experiments were performed at a constant room temperature of around 21.5 °C.

Electrodes were either 32-channel silicon probes (NeuroNexus; 50-μm pitch, 177-μm² surface area for each site; in 2 rows of 16 contacts), or Neuropixels 1.0 probes⁵¹. Recordings using 32-channel probes were performed with a Cheetah Digital Lynx SX system and HS-36 headstages as previously described²¹. Signals were sampled at 32 kHz, and IronClust with manual curation was used for spike sorting (https://github.com/flatironinstitute/ironclust#readme). Neuropixels data were acquired nominally at 30 kHz using SpikeGLX software (http://billkarsh.github.io/SpikeGLX/). The true sampling frequency varied slightly with different headstages and could drift throughout a recording. Data acquired from multiple probes were synchronized by linear interpolation, using as a reference a common 1-Hz squarewave signal recorded with all probes.

Action potentials were sorted using Kilosort2 (ref. ⁵²; https://github.com/MouseLand/Kilosort) and the ecephys pipeline (https://github.com/jenniferColonell/ecephys_spike_sorting); clusters were curated manually in Phy (https://github.com/cortex-lab/phy).

Lesion experiments

In preparation for Imc lesion experiments, we carefully removed the pia overlaying the optic tectum and inserted a bevelled quartz micropipette at an angle of 6–8° from the vertical axis to depths of 2,700–2,900 μm. IBA (200–350 nl; 10 μg μl^...

Search This Blog

Reptiles Definition