Large‐scale analysis reveals populational contributions of cortical spike rate and synchrony to behavioural functions

Rie Kimura; Akiko Saiki; Yoko Fujiwara‐Tsukamoto; Yutaka Sakai; Yoshikazu Isomura

doi:10.1113/JP272794

. 2016 Nov 7;595(1):385–413. doi: 10.1113/JP272794

Large‐scale analysis reveals populational contributions of cortical spike rate and synchrony to behavioural functions

Rie Kimura ^1,^2,^3,^4,^✉, Akiko Saiki ^1,², Yoko Fujiwara‐Tsukamoto ^1,^2,^5,⁶, Yutaka Sakai ^1,², Yoshikazu Isomura ^1,²

PMCID: PMC5199735 PMID: 27488936

Abstract

Key points

There have been few systematic population‐wide analyses of relationships between spike synchrony within a period of several milliseconds and behavioural functions.
In this study, we obtained a large amount of spike data from > 23,000 neuron pairs by multiple single‐unit recording from deep layer neurons in motor cortical areas in rats performing a forelimb movement task.
The temporal changes of spike synchrony in the whole neuron pairs were statistically independent of behavioural changes during the task performance, although some neuron pairs exhibited correlated changes in spike synchrony.
Mutual information analyses revealed that spike synchrony made a smaller contribution than spike rate to behavioural functions.
The strength of spike synchrony between two neurons was statistically independent of the spike rate‐based preferences of the pair for behavioural functions.

Abstract

Spike synchrony within a period of several milliseconds in presynaptic neurons enables effective integration of functional information in the postsynaptic neuron. However, few studies have systematically analysed the population‐wide relationships between spike synchrony and behavioural functions. Here we obtained a sufficiently large amount of spike data among regular‐spiking (putatively excitatory) and fast‐spiking (putatively inhibitory) neuron subtypes (> 23,000 pairs) by multiple single‐unit recording from deep layers in motor cortical areas (caudal forelimb area, rostral forelimb area) in rats performing a forelimb movement task. After holding a lever, rats pulled the lever either in response to a cue tone (external‐trigger trials) or spontaneously without any cue (internal‐trigger trials). Many neurons exhibited functional spike activity in association with forelimb movements, and the preference of regular‐spiking neurons in the rostral forelimb area was more biased toward externally triggered movement than that in the caudal forelimb area. We found that a population of neuron pairs with spike synchrony does exist, and that some neuron pairs exhibit a dependence on movement phase during task performance. However, the population‐wide analysis revealed that spike synchrony was statistically independent of the movement phase and the spike rate‐based preferences of the pair for behavioural functions, whereas spike rates were clearly dependent on the movement phase. In fact, mutual information analyses revealed that the contribution of spike synchrony to the behavioural functions was small relative to the contribution of spike rate. Our large‐scale analysis revealed that cortical spike rate, rather than spike synchrony, contributes to population coding for movement.

Keywords: cortical spike synchronization, forelimb movement, multiple single‐unit recording

Key points

There have been few systematic population‐wide analyses of relationships between spike synchrony within a period of several milliseconds and behavioural functions.
In this study, we obtained a large amount of spike data from > 23,000 neuron pairs by multiple single‐unit recording from deep layer neurons in motor cortical areas in rats performing a forelimb movement task.
The temporal changes of spike synchrony in the whole neuron pairs were statistically independent of behavioural changes during the task performance, although some neuron pairs exhibited correlated changes in spike synchrony.
Mutual information analyses revealed that spike synchrony made a smaller contribution than spike rate to behavioural functions.
The strength of spike synchrony between two neurons was statistically independent of the spike rate‐based preferences of the pair for behavioural functions.

Abbreviations

AC: auto‐correlogram
AGm: medial agranular cortex
BDA: biotinylated dextran amine
CC: cross‐correlogram
CFA: caudal forelimb area
D: spike delays of CC peaks
DAB: diaminobenzidine
Dcs: dorsal corticospinal tract
DLS: dorsolateral striatum
ext: external
FDR: false discovery rate
FL: forelimb
FS: fast‐spiking
I: mutual information
ICMS: intracortical microstimulation
int: internal
ISI: inter‐spike interval
ITI: inter‐trial interval
Juxta: juxtacellularly recorded neurons
KS: Kolmogorov–Smirnov
M1: primary motor cortex
M2: secondary motor cortex
Multi: multiple single unit‐recorded neurons
MUS: muscimol
No‐S/no‐stim: no‐stimulation
PETH: peri‐event time histogram
PM: premotor area
R: averaged spike rate
RFA: rostral forelimb area
RS: regular‐spiking
SI: selectivity index
SMA: supplementary motor area
S/stim: stimulation
TRI: task relevance index
VTA: ventral tegmental area

Introduction

A large body of neurophysiological evidence has revealed that the spike rates of cortical neurons correlate with various behavioural functions (e.g. Evarts, 1968; Georgopoulos et al. 1986). In addition, spike synchrony, which can be statistically identified by cross‐correlograms (Perkel et al. 1967; Toyama et al. 1981; Aertsen & Gerstein, 1985), has been observed in various brain areas in the context of higher order behavioural functions (Gray et al. 1989; Vaadia et al. 1995; Sakurai, 1996; Riehle et al. 1997; Steinmetz et al. 2000; Fries et al. 2001; Narayanan & Laubach, 2006; Cohen & Newsome, 2008; Fujisawa et al. 2008; Mitchell et al. 2009; Hirabayashi et al. 2013 a, b ). Some studies have also suggested that spike synchrony may contribute to control of voluntary movements (Isomura et al. 2009; Putrino et al. 2010). In general, cortical neurons receive synaptic inputs from many other neurons, most of which are far below firing threshold, although a small population generates large depolarizations (Song et al. 2005; Lefort et al. 2009; Ikegaya et al. 2013). If presynaptic neurons synchronize their spiking within a period of several milliseconds, the postsynaptic neuron can readily discharge in response to these synchronous inputs. Therefore, spike synchrony may play a physiological role in effective integration of functional information from various neurons into the rate code of the postsynaptic neuron. The relative contributions of spike synchrony and spike rate to behavioural functions remain to be determined.

Spike synchrony is a rare event that occurs at most in a few per cent of all neuronal pairs (e.g. Fujisawa et al. 2008). In order to characterize the properties of spike synchrony, data needs to be gathered from many (more than 10,000) neuron pairs. However, most previous reports have not discussed in statistical terms how spike synchronies are related to physiological functions at the population level, although the importance of capturing the distribution of spike synchronies has recently been appreciated (Buzsáki & Mizuseki, 2014). Several reports of population spike synchrony dealt with large changes in behavioural state, e.g. running, sleeping and immobility (Mizuseki & Buzsáki, 2013, 2014), or dynamic changes associated with learning or long‐term potentiation (Baeg et al. 2007; Yun et al. 2007; Dupret et al. 2013). A few reports also examined the population‐wide relationship between spike synchronies and functional similarities between neuron pairs (Constantinidis et al. 2002; Buetfering et al. 2014). Nevertheless, it remains unknown how spike synchrony contributes to behavioural functions in association with (or independent of) spike rate on a large scale.

In this study, we focused on voluntary movements of a forelimb in rodents (Brácha et al. 1990; Iwaniuk & Whishaw 2000), and examined the population‐wide relationship between spike synchronies and behavioural functions in a motor cortical area. Specifically, we performed multiple single‐unit recording from deep layers in the caudal forelimb area (CFA) of primary motor cortex (M1) and the rostral forelimb area (RFA) of secondary motor cortex (M2) in head‐restrained rats performing tasks. These tasks required the rats to pull a lever in response to a cue tone (external) or pull it spontaneously (internal). Using our effective task‐training protocol (Kimura et al. 2012), we were able to collect an enormous amount of data regarding synchronous activities (> 23,000 neuron pairs), enabling us to examine the statistical properties of spike synchronies at the population level.

Methods

Ethical approval

All experiments were approved by the Tamagawa University Animal Care and Use Committee (H22‐32; 2010–2013), and performed in accordance with the Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions (Ministry of Education, Culture, Sports, Science and Technology, Japan), and the Guidelines for Animal Experimentation in Neuroscience (Japan Neuroscience Society). Every effort was made to minimize pain and suffering and to reduce the number of animals used.

Animal preparation

Adult male Long–Evans rats (6–7 weeks old, 244 ± 18 g (range: 200–280 g) at the time of surgery, n = 90; 77 for behavioural/recording experiments; 13 for pharmacological experiments; Institute for Animal Reproduction, Ibaraki, Japan) were kept under an inverted light schedule (lights off at 09.00 h and lights on at 21.00 h). The rats were adapted to a stainless steel cylinder (hideaway) in their home cage and each rat was briefly (10 min) handled a total of two times. We used behavioural/electrophysiological data from 73 of 77 rats (95%) for analysis; the remaining four rats were omitted because of poor task performance or bad electrophysiological recording.

Anaesthesia was induced with 5.0% isoflurane and maintained at 2.0–2.5% isoflurane (v/v; Univentor 400 Anaesthesia Unit, Univentor, Zejtun, Malta). During anaesthesia, body temperature was maintained at 37°C using an animal warmer (BWT‐100, Bio Research Center, Nagoya, Japan). For head skin incision, if necessary, we additionally used surface anaesthetic (lidocaine hydrochloride, 2% xylocaine jelly, AstraZeneca, Osaka, Japan). A sliding aluminium head attachment (custom‐made by Narishige, Tokyo, Japan) was surgically attached to the skull with tiny anchor screws (stainless steel, 1 mm in diameter, 2 or 3 mm long) and dental resin cement (Super‐Bond C&B, Sun Medical, Shiga, Japan; Panavia F2.0, Kuraray Medical, Tokyo, Japan; Unifast II, GC Corporation, Tokyo, Japan) (Kimura et al. 2012). Teflon‐coated silver wires (786000, A‐M Systems, Carlsborg, WA, USA; 200 μm in diameter) for reference and grounding were implanted above the cerebellum. In some experiments, twisted Teflon‐coated silver wire electrodes (786000, A‐M Systems) were implanted into the right upper forelimb (near the biceps brachii, trapezius, or deltoid) to measure electromyogram (EMG) activity during task performance.

After rats recovered from the surgery with satisfactory food and water intake for 2–3 days, under careful observation, the rats were deprived of drinking water in their home cages, although food was available ad libitum. Sufficient water was provided as a reward during task performance. An agar block (containing 15 ml of water) was given to the rats, when necessary, in order to maintain their body weight above 80% of its original value (cf. Schwarz et al. 2010; Kimura et al. 2012 for water control).

One day before electrophysiological recording, the rats were subjected to a second surgery under isoflurane anaesthesia, and a tiny hole (about 1 mm in diameter) was opened in the skull and dura mater above the forelimb (FL) area of either of the left motor cortices (CFA: 1.0 ± 0.5 mm anterior, 2.5 ± 0.5 mm lateral from bregma, Isomura et al. 2009; RFA: 3.5 ± 0.5 mm anterior, 2.0 ± 0.5 mm lateral, Hyland, 1998; Saiki et al. 2014). The hole was covered with silicone sealant (Dent Silicone‐V, Regular, Shofu, Kyoto, Japan) until recording started.

Behavioural tasks

For operant behavioural experiments, rats were placed in a cylinder containing a spout‐lever, which integrated the lever and reward into one device, under head‐restrained conditions in dim light, as described previously (Kimura et al. 2012; custom‐made by O'Hara & Co, Tokyo, Japan). They were initially trained (three consecutive training days, about 3 h per day) to perform an external‐trigger movement task in which they had to pull the lever using their right forelimb when a pure‐tone cue (8 kHz sound) was presented briefly (0.5 s) after holding it (keeping the lever at 0–3 mm from the front end). The hold period until the cue onset was automatically extended step‐by‐step up to 1 s (0, 0.1, 0.2, 0.3, 0.5, 0.7, 0.8 and 1.0 s), according to the total number of successful trials in each step. If the rats pulled the lever successfully (i.e. moved it over 6–9 mm) in response to cue presentation, they acquired a drop (0.01 ml) of 0.1% saccharin water as a reward after a 0.2–0.8 s delay (0.1 s steps at random, except that the first step was 0.2 s).

On the day of electrophysiological recording (basically, 2 days after the three consecutive training days), internal‐trigger trials with the cue presentation omitted were added pseudorandomly to the external‐trigger trials at a 1:1 ratio. Rats soon learned to pull the lever spontaneously after holding it for at least 1 s in the internal‐trigger trials (see Supplemental Video S1 in Kimura et al. 2012). The amount of reward for the external‐/internal‐trigger movement task was reduced to 0.005 ml (saccharin water) to allow the rats to perform this task longer without satiety. If the rats did not pull the lever correctly, they had to re‐perform the same type of trial until they succeeded (correction trials) after the inter‐trial interval (ITI, 0.2–0.8 or 0.1–0.5 s). Except in pharmacological experiments, we performed the final behavioural/electrophysiological experiment only once for each rat.

Electrophysiological recording

We performed extracellular multiple single‐unit recording and juxtacellular recording from the FL area of one of the left motor cortices (CFA or RFA) in behaving rats. In some experiments, these coordinates were confirmed by checking forelimb movements or EMG activities in the contralateral forelimb in response to intracortical microstimulation (ICMS; about −20 to −120 μA, 50 pulses at 100 Hz or sustained pulses; SEN 8203, Nihon Kohden, Tokyo, Japan) (Saiki et al. 2014). For multiple single‐unit recording, a 16‐channel silicon probe (A2×2‐tet‐3mm‐150‐150‐121/312; NeuroNexus, Ann Arbor, MI, USA; 2 shanks × 2 tetrode‐like sites, at 150 μm intervals) was placed on the brain surface in the skull hole, and then covered with 2% agarose (Agarose‐HGT; Nacalai Tesque, Kyoto, Japan) and paraffin wax (Nacalai Tesque) to prevent drying. The probe was inserted into the FL area 1.25 ± 0.2 mm (CFA) or 1.40 ± 0.2 mm (RFA) deep (putative layer 5B) vertically (0 deg) or at a 12 deg angle using manipulators (1760‐61, David Kopf Instruments, Tujunga, CA, USA; MWS‐1B, Narishige; LSS‐8000 Inchworm, Burleigh Instruments, Fishers, NY, USA) on a stereotaxic frame (SR‐8 N, Narishige). After more than 1 h of stabilization, behavioural/electrophysiological recording was started, and multichannel signals were amplified with a 16‐channel amplifier (MEG‐6116, Nihon Kohden; final gain, 2000; original band‐pass filter, 0.5–1.5 Hz to 10 kHz) and digitized with a 32‐channel hard‐disc recorder (DataMax II, R.C. Electronics, Santa Barbara, CA, USA; sampling, 20 kHz). For juxtacellular recording, a glass electrode (BF150‐75‐10, Sutter Instrument, Novato, CA, USA), prepared by a puller (PC‐10, Narishige) and filled with 2% Neurobiotin (Vector Laboratories, Burlingame, CA, USA) in 0.5 m KCl, was inserted by calculating the insertion position which can achieve the adjacent recordings within about 200 μm between the tip of the glass electrode and the recording sites of a silicon probe for multiple single‐unit recording after the insertion (R. Kimura et al., unpublished observations; see http://libds.tamagawa.ac.jp/dspace/bitstream/11078/370/3/Data2016_RieKimura_01.pdf, the same hereinafter). The juxtacellularly recorded signal was amplified with an intracellular amplifier (IR‐283, Cygnus Technologies, Delaware Water Gap, PA, USA; final gain, 1000; band‐pass filter, 300 Hz to 10 kHz) and digitized with the same recorder (sampling, 20 kHz). Single recorded neurons were juxtacellularly and specifically stimulated by a train of positive current injections (nanostimulation: Houweling et al. 2010; 2–20 nA for 500 ms, every 1 s for 5–20 min); meanwhile, Neurobiotin was administered by electroporation.

In some experiments, EMG activity for movements of the right forelimb was obtained using the amplifier and recorder (final gain, 1000 or 2000; band‐pass filter, 5–50 Hz to 10 kHz; sampling, 20 kHz). EMG power was calculated by squaring and then averaging the offset value of EMG activity.

Pharmacological inactivation

The pharmacological blockade of neuronal activity in CFA and/or RFA was achieved by unilateral microinjections of the GABA_A receptor agonist muscimol (1 μg μl⁻¹ in saline, 0.5 μl, for 5 min, Sigma‐Aldrich, St Louis, MO, USA; Smith et al. 2010) through a Hamilton microsyringe (33‐gauge, 5 μl, Hamilton, Reno, NV, USA) installed on a microinjector (IMS‐3, Narishige). For dual blockade of CFA and RFA, muscimol injections were performed serially: the microsyringe was left in place for an additional 10 min after the first injection into one area; it was then inserted into the other area, and a second injection was started after 10 min. To confirm the effective spread of drug injection, in preliminary experiments we injected dye solution (0.5 μl of 2% Chicago sky blue 6B, alias Pontamine sky blue, in saline, Sigma‐Aldrich) to separate rats using the same method as for the muscimol injection and monitored its spread microscopically.

Histological staining

Under deep anaesthesia with urethane (30% ethyl carbamate, 7–10 ml kg⁻¹, i.p., Nacalai Tesque), the rats were killed by intracardial perfusion with cold saline followed by 4% paraformaldehyde (sometimes with 0.5% glutaraldehyde) or 10% formalin in 0.1 m phosphate buffer. To check electrode positions, the brains were postfixed and sliced coronally into 50 μm‐thick serial sections using a microslicer (LinearSlicer PRO7, Dosaka EM, Kyoto, Japan). In some experiments, Neurobiotin‐labelled neurons were visualized using the avidin–biotin–horseradish peroxidase complex (ABC; 1:200 dilution; Vectastain Elite ABC, Vector Laboratories) with diaminobenzidine (DAB; Sigma‐Aldrich) and nickel (Isomura et al. 2009, 2013). The visualized neurons were reconstructed using a camera lucida following Nissl staining with 0.5% Neutral Red. We divided CFA and RFA into layers based on previous reports (Hooks et al. 2013). To confirm corticospinal projection, we injected the anterograde tracer biotinylated dextran amine (BDA; 10,000 Da molecular mass, 2% in saline; Thermo Fisher Scientific, Waltham, MA, USA) into deep layers of RFA, iontophoretically through a glass pipette (BAB‐501, Kation Scientific, Minneapolis, MN, USA; 5 μA, 7 s on/off for 20 min). Seven days after the injections, we visualized the BDA‐labelled axons by the ABC method described above (Smith et al. 2010; Isomura et al. 2013).

Data analysis

Movement–behaviour time extraction from lever trajectories

The spout‐lever was movable horizontally in both the forward (by push; front end, 0%) and backward (by pull; back end, 100%) directions (Kimura et al. 2012). The trajectory was recorded by an angle encoder (MES‐12‐1000 PC, Microtech Laboratory, Kanagawa, Japan) during the whole recording period. To reduce weight, no force transducer was attached. In the behavioural task procedure, the holding (push) position was defined as 0–20% of the working range of the lever (0–3 mm from the front end). The goal position was defined as 40–60% (centre position, 6–9 mm). Hold time was defined as the time during which the lever was kept in the hold position in each trial (see Fig. 1 B). Push‐end time was defined as the time of the nearest local minimum of the lever trajectory after the lever moved into the holding range (see Fig. 1 B). Pull‐onset time was defined as the time of the nearest local minimum of the lever trajectory before the lever moved out of the holding range. Pull‐end time was defined as the time of the nearest local maximum of the lever trajectory after the lever moved into the goal range.

A, set‐up for electrophysiological recording from the caudal and rostral forelimb areas (CFA and RFA) in head‐restrained rats during lever‐manipulation tasks. B, schematic diagram showing lever trajectories required in the tasks. Rats pushed the lever and held it for 1 s. In the externally initiated movement task, just after that, a cue sound was presented and the rats pulled the lever. In the internally initiated movement task, no cue was presented, and the rats pulled the lever in a self‐paced manner. C, timing of lever pulling. Histogram shows hold time normalized by the number of correct trials (N = 73, mean and SD). D, representative lever trajectories (means ± SD) and EMG powers for external (blue) and internal (red) trials (representative averaged EMG power area of −100 to 500 ms from pull‐onset; external: 5.63 mV²·ms, internal: 6.07 mV²·ms; pull time from pull‐onset to pull‐end: external 0.25 ± 0.06 s, internal 0.28 ± 0.06 s, N = 73) in the forelimb.

Spike extraction and refinement

Multiple single‐unit and juxtacellular recording data (acquired over an ∼3 h session including ITI) were processed to isolate spike events with the automatic spike‐sorting program EToS (Efficient Technology of Spike‐sorting; http://etos.sourceforge.net/; Takekawa et al. 2010, 2012), using wavelet feature extraction and variational Bayes inference of the likely mixture of the t‐distribution family including the Gaussian distribution. Spikes separated by more than 0.25 ms could be detected. Only spikes separated over 0.5 ms were used for the estimation of mixture distributions, and the rest were clustered on the basis of the estimated mixture distributions. The sorted spike clusters were manually combined, divided, or discarded to refine single‐neuron clusters, based on the presence/absence of refractory periods (< 2 ms) in their own auto‐correlations (related to inter‐spike interval, ISI) and cross‐correlations with other clusters, using the manual clustering tools Klusters and Neuroscope (Harris et al. 2000; Hazan et al. 2006). Furthermore, if we found a cross‐correlation of two clusters that exhibited a sharp peak just in the 0 ms bin accompanied with a refractory period up to 2 ms (putatively from the same neuron, recorded via different tetrodes), then we excluded the smaller‐size cluster. We confirmed that the final form of each cluster had a clean refractory period in its autocorrelation and no refractory periods in its cross‐correlations with other clusters.

Classification of neurons

The identified spike clusters (neurons) were classified as follows (Table 1 A). Each neuron was classified as ‘RS’ or ‘FS’ according to the spike duration between the trough and peak (R. Kimura et al., unpublished observations). Spike durations were split into two groups, which were correlated with ongoing (all averaged) spike rate. The group with longer durations was considered to represent regular‐spiking (RS) pyramidal neurons, whereas the group with shorter durations was considered to represent fast‐spiking (FS) interneurons (spike duration threshold: 0.5 ms for multiple single‐unit recording, 0.45 ms for juxtacellular recording; Barthó et al. 2004; Isomura et al. 2009, 2013; Guo et al. 2014; Saiki et al. 2014; Li et al. 2015).

Table 1.

Classification of neurons and the numbers of neurons and neuron pairs

A. Numbers of classified neurons
	CFA 1518		RFA 1533
Total recorded neurons	RS 1405	FS 113	RS 1444	FS 89
Multiunit recording	1357	105	1377	88
Juxtacellular recording	48	8	67	1
Task related	530	65	570	50
Pull onset aligned	441	54	475	41
Hold related	91	5	79	3
Pull related	177	39	188	27
Others	173	10	208	11
Push end aligned	89	11	95	9
Non‐task related	151	10	211	9
Not analysed	724	38	663	30

B. Numbers of simultaneously recorded neuron pairs
	CFA 10243				RFA 13207
Total possible neuron pairs N _basal ≥ 200	RS−RS 7985	RS−FS 1080	FS−RS 1074	FS−FS 104	RS−RS 10765	RS−FS 1185	FS−RS 1167	FS−FS 90
Significant pairs (P < 0.005)	493	136	36	18	637	159	48	27
N _basal ≥ 500	5064	855	845	92	6406	890	883	80
Task−Task	3755	665	657	78	4550	662	660	66
Hold−Hold	111	13	12	2	94	5	5	0
Hold−Pull	233	67	22	6	197	44	25	2
Pull−Hold	233	23	67	6	194	26	44	2
Pull−Pull	454	162	161	34	631	133	134	18
Others	2724	400	395	30	3434	454	452	44
Task−Non‐task	574	29	150	7	695	42	152	6
Non‐task−Task	569	152	29	7	687	153	39	6
Non‐task−Non‐task	166	9	9	0	474	33	32	2

Open in a new tab

We focused on neurons that changed their activities during task performance. In order to quantify task relevance, we evaluated the difference between the distribution of spike timings in a task sequence and the uniform distribution as –log₁₀ P (task relevance index), where P is the significance p‐value of the Kolmogorov–Smirnov (KS) test for the distribution difference (R. Kimura et al., unpublished observations). This was not a statistical test; instead, P was simply used as an index to quantify the temporal dependence of spike rate with statistical reliability. In each task sequence, two types of timings were used: push end‐aligned timings from −0.5 to 1 s, and pull onset‐aligned timings from −1 to 0.5 s. Neurons that exhibited task relevance indices greater than 6 for push end‐/pull onset‐aligned timings in the external‐/internal‐trigger trials were categorized as task‐related neurons. In particular, all neurons exhibiting task relevance indices greater than 6 for pull onset‐aligned timings were designated as pull onset‐aligned task‐related neurons, and the rest of the task‐related neurons were designated as push end‐aligned task‐related neurons. Neurons that exhibited no fewer than 4000 spikes and task relevance indices not greater than 2 for all of the alignment and motor‐initiation conditions were designated as non‐task‐related neurons.

The peri‐event time histogram (PETH) was calculated by averaging the spike timings with a Gaussian filter (resolution: 0.5 ms; σ deg = 50 ms; length: 500.5 ms). Pull onset‐aligned task‐related neurons were classified as hold related, pull related, or others according to the peak times of pull onset‐aligned PETHs with larger peak amplitude in external‐ or internal‐trigger trials. In hold‐related neurons, the peak time of the PETH was in the range −0.9 to −0.25 s, whereas in pull‐related neurons, the peak time was in the range −0.25 to 0.25 s. Other neurons had peak times outside these ranges.

Selectivity to external‐/internal‐trigger movements

In order to represent selectivity in neuronal activity, we defined the selectivity index (SI). For instance, the selectivity for the averaged spike rates of external‐ and internal‐trigger trials was described as

SI [R_{ext}, R_{int}] = \frac{R_{ext} - R_{int}}{R_{ext} + R_{int}},

(1)

where R _ext and R _int are the averaged spike rates in the range −50 to 50 ms around the peak times of the pull onset‐aligned PETH with larger peak amplitude in external‐ or internal‐trigger trials. The data analysed for the selectivity indices of the averaged spike rates were restricted by the numbers of trials and spikes. The analysed data included over 40 external‐trigger trials and 40 internal‐trigger trials (in multiple single‐unit recording; 30 trials were used for juxtacellular recording because of its high accuracy), over 2000 spikes in total, and over 200 spikes in each of the PETH time ranges (push end‐aligned PETH: −0.5 to 1 s; pull onset‐aligned PETH: −1 to 0.5 s). The significance of selectivity of each single neuron was evaluated by the Wilcoxon rank‐sum test for spike numbers in each trial for external‐ or internal‐trigger movements. The significance of selectivity of the neuronal population was evaluated by KS test to compare the distribution of SIs with that of shuffled data obtained by randomly swapping external‐ and internal‐trigger trials. Shuffling was performed 100 times for every neuron.

Evaluation of spike synchrony

The cross‐correlogram (CC) of spike timings between a pair of neurons 1 and 2, i.e. the spike rate of neuron 2 aligned with spike timings of neuron 1, was calculated with a bin width of 0.5 ms. We focused on CC peaks of spike delays in the range 0.25–5.25 ms (10 bins, window: 5 ms), regardless of the peak width, relatively stronger than the average CC on the 10‐ms scale; such CC peaks are referred to as ‘spike synchrony’ in this study. Ordered neuron pairs were considered. CC peaks in which spikes of neuron 1 preceded spikes of neuron 2 were considered only for the neuron pair of ‘neuron 1 to neuron 2’. We excluded values in the 0 ms bin (−0.25 to 0.25 ms, nearly simultaneous firings) from further analyses because spike pairs within 0.25 ms are undetectable by a single tetrode in our spike‐detection protocol, and also might include spikes from the same neuron recorded by different tetrodes.

Statistical reliability of spike synchrony: P‐value

To evaluate the statistical reliability of spike synchrony in comparison with the 10 ms scale correlation, we performed a statistical test using a jittering protocol (Fujisawa et al. 2008), a standard method for evaluating spike synchrony. We did not adopt the shift‐predictor method, the most popular technique, because it does not exclude correlation through trial‐by‐trial variability. Instead, we focused on millisecond‐order spike synchrony that was significantly stronger than that expected from the rate correlation on a 10 ms scale.

In particular, we adopted a uni‐directional jittering protocol (R. Kimura et al., unpublished observations), because a bi‐directional jittering protocol may underestimate the significance P‐value due to limited detection of spike pairs within 0.25 ms. For a CC peak in which the spikes of neuron 1 preceded those of neuron 2, the spike timings of neuron 2 were randomly jittered uni‐directionally; each jitter was sampled from 0, −0.5, −1, ..., −10 ms (21 candidates). The probability that a peak among the 0.5 to 5 ms bins of a surrogate CC calculated using a jittered spike sequence is larger than the actual peak can be described using a binominal distribution (Stark and Abeles, 2009) as P = 1 − (1 − q)^m, where: $q = \sum_{k = k_{p}}^{n}_{n} C_{k} p^{k} {(1 - p)}^{n - k}$ . The number of candidate bins for the peak was m = 10 (0.5 to 5 ms bins), the probability that a jitter would fall into a bin was p = 1/21, n was the number of candidate spike pairs, and k_p was the number of spike pairs in the actual peak. This P‐value represents the corrected probability for multiple candidates (m = 10 bins). For the sake of statistical reliability, we evaluated the P‐value only for neuron pairs with a sufficient number of spike pairs (N _basal ≥ 200) in the basal range 5.25–15.25 ms (20 bins) of the CC.

To evaluate the significance of spike synchrony at a population level, we compared the distributions of the P‐values with the uniform distribution, which is the common theoretical distribution of P‐values of jittered data. We labelled this distribution as ‘no correlation’ in figures.

Strength of spike synchrony: H‐value

To evaluate the strength of spike synchrony, we used the normalized height of a CC peak, denoted by H (R. Kimura et al., unpublished observations), because the P‐value used to determine statistical significance depends on the number of spikes and cannot be used for comparisons among data with different numbers of spikes. The H‐value was simply calculated as H = 100 × A/B (%), where A is the average of the CC values at the peak bin and adjacent bin(s) (two bins for a peak in the 0.5 ms bin, and three bins otherwise), and B is the average of the CC values in the basal range 5.25–15.25 ms (20 bins) (Sears & Stagg, 1976; Aertsen & Gerstein, 1985). The expected value of H is independent of the number of spikes, although its accuracy does depend on the number. For the sake of fair comparison with equivalent accuracy, to compare the strength of spike synchrony during the whole recording period among different neuron groups, we calculated the H‐value from a pair of restricted spike sequences with 500 fixed spike pairs in the basal range (N _basal = 500) by cutting off the first and last parts of the data.

Evaluation of functional change of spike synchrony

To evaluate the dependence of temporal changes of spike synchrony on behavioural state in a task sequence, CCs and H‐values were calculated during restricted periods in a task sequence: task/no‐task periods, hold/pull periods, and external‐/internal‐trigger trial periods. A ‘task’ period was defined as the period from the hold‐onset to the reward delivery in correct trials only. A ‘no‐task’ period was defined as the interval from 1 s after the end of a trial (correct trial: reward time; error trial: 0.2 s after performing the error) to 1 s before the hold‐onset in the next trial (if it was over 0.5 s); such intervals could include periods of rest without manipulation of the lever, sleep, etc. Because the task/no‐task periods included various lengths of time that did not correspond to each other, for the sake of fair comparison we divided each period into 500 ms periods and made pseudo‐500 ms trials. ‘Hold’ and ‘pull’ periods were defined as periods from −750 to −250 ms and from −250 to 250 ms, respectively, in pull onset‐aligned timings in correct trials. External (‘ext’) and internal (‘int’) periods were defined as the periods from 1 s after the hold‐onset to the reward delivery in correct trials. Only CCs with over 200 spike pairs in the basal range (N _basal ≥ 200) were used for analyses.

Functional change in strength of spike synchrony

Functional changes in strength of spike synchrony were evaluated by the selectivity index SI, defined in eqn (1) for the normalized peak height H _X and H _Y, SI[H _X,H _Y], where X and Y represent behavioural periods defined above, e.g. X = task and Y = no‐task. The peak heights H _X and H _Y were calculated for the common bins based on the peak spike delay of the CC peak with larger amplitude in X or Y. To evaluate the significance of the difference in H‐values between conditions X and Y, we performed a 2 × 2 chi‐square test for the number of spike pairs around the peak and adjacent bin(s) (±1 bin) and in the basal range (5.25–15.25 ms) in each pair of neurons. To determine whether the distribution of functional changes differed significantly from the distribution of changes expected by chance, we compared it with the distribution calculated from shuffled data obtained by randomly swapping trials (or pseudo‐trials) for conditions X and Y. Shuffling was performed only once for each pair of neurons.

Functional change in timing of spike synchrony

Functional changes in timing of spike synchrony were evaluated based on the difference between the spike delays of the CC peaks for conditions X and Y, D_Y – D_X. We focused on CC peaks in 0.5 to 5 ms bins, and hence the spike delay difference D_Y – D_X was in the range from –4.5 to 4.5 ms. The distribution of the spike delay difference was compared with that of shuffled data identical to those in “Functional change in strength of spike synchrony” section.

Information carried by a pair spike pattern

To quantify the information carried by a spike synchrony event, we evaluated mutual information between the condition X/Y and the spike pattern of two neurons within a pair of 1 ms windows separated by the spike delay of the CC peak with larger amplitude in X or Y. The conditional probabilities of spike patterns P(0,0|X), P(1,0|X), P(0,1|X), and P(1,1|X) were estimated by sliding the pair of 1 ms windows by 1 ms steps for the spike data in condition X, where 0 and 1 indicate the absence or presence of a spike in the 1 ms window, respectively. The probability of each condition was constrained to be equal in each experiment, i.e. P(X) = P(Y) = 1/2. Therefore, the mutual information between the condition X/Y and the pair spike pattern was calculated as:

\begin{matrix} I_{pair} & = & \frac{1}{2} \sum_{Z = X, Y} \sum_{s_{1} = 0, 1} \sum_{s_{2} = 0, 1} P (s_{1}, s_{2} | Z) \\ \times lo g_{2} \frac{2 P (s_{1}, s_{2} | Z)}{P (s_{1}, s_{2} | X) + P (s_{1}, s_{2} | Y)} . \end{matrix}

(2)

This mutual information I _pair contains information carried by the spike rate pattern of two neurons. To compare I _pair with the information of only the spike rate pattern, we introduced the rate information I _rate calculated from the spike rates (r ₁ ^Z, r ₂ ^Z) of two neurons based on the hypothesis that the two neurons are independent. If so, then the probability of spike synchrony is the product of the spike rates per 1 ms, P _rate(1,1|Z) = r ₁ ^Z r ₂ ^Z, where spike rates r ₁ ^Z = P(1,0|Z) + P(1,1|Z) and r ₂ ^Z = P(0,1|Z) + P(1,1|Z). Thus, we obtained rate information I _rate following eqn (2) for hypothesized probabilities P _rate(1,1|Z) = r ₁ ^Z r ₂ ^Z, P _rate(1,0|Z) = r ₁ ^Z (1–r ₂ ^Z), P _rate(0,1|Z) = (1 – r ₁ ^Z) r ₂ ^Z and P _rate(0,0|Z) = 1 – r ₁ ^Z – r ₂ ^Z + r ₁ ^Zr ₂ ^Z. The mutual information I _rate was compared with that of shuffled data, I _shuffle. The shuffled data were identical to those in “Functional change in strength of spike synchrony” section.

Effect of juxtacellular nanostimulation on spike synchrony

To evaluate the effect of nanostimulation on spike synchrony, we used only data in which juxtacellular nanostimulation induced a greater than 50% increase in spike rate relative to that during no‐stimulation periods. We extracted spike sequences starting 30 s after the initiation of iterative stimulations, and then calculated the selectivity index defined in eqn (1) for H‐values in stimulation (‘stim’) periods (repeating 0.5 s stimulation at 1 Hz) and no‐stimulation (‘no‐stim’) periods (see Fig. 4), SI[H _stim,H _no‐stim]. Only CC data with over 150 spike pairs in the basal range (N _basal ≥ 150) were adopted for this analysis. The distribution of the selectivity index was compared with that of shuffled data calculated by randomly swapping stimulation and no‐stimulation periods. Because there were a small number of such neuron pairs, in this case we analysed 100 shuffled data for every neuron pair.

A, representative juxtacellularly recorded trace, collected in order to reveal the effects of artificial activation of a single neuron on spike synchrony. Nanostimulation was repeatedly applied in current pulses (S: 500 ms on; No‐S: 500 ms off, < 10 nA). Mean spike rate of the juxtacellularly recorded neuron during stimulation was 166% of the rate in the absence of stimulation. The downward‐pointing triangle (▼) represents a stimulation artifact. B, changes in mean spike rates of juxtacellularly recorded neurons (Juxta) and the others (multiple single unit‐recorded neurons; Multi) by nanostimulation (Table 3; ^** and n.s., as determined by paired t test). Mean spike rates of the same neuron during no stimulation (No‐S) and stimulation (S) are linked by lines. C, representative CCs during stimulation and no stimulation in the experiment shown in A in CFA RS–FS. D, top, histograms of SI[H _stim,H _no‐stim] in RS (Juxta)‐RS/FS (Multi) shown in CFA/RFA (n.s., as determined by one‐sample t test against 0). Black parts of histograms represent neuron pairs exhibiting significant changes at the single‐pair level (p < 0.05, chi‐square test). Bottom, cumulative distributions of SI[H _stim,H _no‐stim] obtained from real data (black) and shuffled data (cyan) in which the stimulation and no‐stimulation periods were shuffled (Table 3; n.s., as determined by KS test). E, the same as D, but the order of neuron pairs was reversed [RS/FS (Multi)–RS (Juxta)].

Network characteristics of significant spike synchrony

We characterized network structures linked with neuron pairs exhibiting significant spike synchrony (P < 0.005, N _basal ≥ 200). The P‐value was used as the significance level, although it might be affected by the firing rates. The histogram of the link number (network degree) per neuron (network node) was calculated for different types of links. We analysed only three types of links: divergent links from an RS neuron to other RS neurons, convergent links from other RS neurons to an RS neuron, and convergent links from other RS neurons to an FS neuron, all of which were present in sufficiently large numbers. We tested whether the link number histogram was significantly different from that of a random network obtained by shuffling links among simultaneously recorded neuron pairs with over 200 spike pairs in the basal range (N _basal ≥ 200). By iterating the shuffling 200,000 times, we numerically calculated the probability distribution of the link number histogram of a shuffled network, and evaluated the statistical significance with the probability distribution. To depict a representative example of network structure, we used Pajek (http://pajek.imfm.si), a free program for analysis of large networks.

Statistics

We primarily adopted two significance levels, ^** (p < 0.01, P < 0.005) and ^* (p < 0.05, P < 0.025) where p is the two‐sided probability and P is the one‐sided probability of the null‐hypothesis distribution. Non‐significant results (p ≥ 0.05, P ≥ 0.025) were marked as n.s. We used the one‐sided probability only in the exact test for CC peaks (P‐value), which is tested in the direction toward the CC peak but not the trough. Most of the p‐values of the results are summarized in Tables 2 and 3, and some are described in the text, rounded down to one effective digit. For Welch's t test, we also described the degree as a subscript, t _degree.

Table 2.

Statistical tests and analyses of false discovery rate (FDR)

A. Statistical tests for spike rate (PETH)
			RS								FS
Statistics	Comparison		n vs. n		KS	p	FDR	t or F		p	n vs. n		KS	p	FDR	t or F		p	Figure
Activity peak time	CFA	External vs. Internal	140	140		0.9					34	34		0.9					Fig. 2 C
	RFA	External vs. Internal	157	157		0.8					25	25		0.9
	External	CFA vs. RFA	140	157		0.7					34	25		0.8
	Internal	CFA vs. RFA	140	157		0.8					34	25		0.9
SI[R _ext,R _int]	CFA	Real vs. Shuffle	140	14000	**	1 × 10⁻⁶	††	t ₁₄₀ F	n.s. > **	0.1 1 × 10⁻⁴¹	34	3400	**	2 × 10⁻³	††	t ₃₃ F	> >	2 × 10⁻³ 3 × 10⁻³	Fig. 2 D
	RFA	Real vs. Shuffle	157	15700	**	3 × 10⁻¹⁸	††	t ₁₅₇ F	> >	3 × 10⁻⁵ 5 × 10⁻⁸⁶	25	2500	**	1 × 10⁻⁴	††	t ₂₄ F	> >	2 × 10⁻³ 7 × 10⁻⁵
		CFA vs. RFA	140	157	**	2 × 10⁻³	††	t ₂₉₀	<*	0.01	34	25		0.6

B. Statistical tests for spike cross‐correlogram (CC)
Statistics		CFA								RFA


(restriction)	Comparison	n vs. n		KS	p	FDR	t or F		p	n vs. n		KS	p	FDR	t or χ²		p	Figure
CC peak significance P	RS–RS vs. No correlation	7985	—	**	1 × 10⁻¹¹⁰	††	χ²	> **	0	10765	—	**	1 × 10⁻¹⁴⁰	††	χ²	> **	0	Fig. 3 B
	RS–FS vs. No correlation	1080	—	**	4 × 10⁻³⁵	††	χ²	> **	0	1185	—	**	6 × 10⁻⁵⁵	††	χ²	> **	0
(N _basal ≥ 200)	FS–RS vs. No correlation	1074	—	**	1 × 10⁻⁴	††	χ²	> **	4 × 10⁻⁴⁰	1167	—	**	4 × 10⁻⁶	††	χ²	> **	1 × 10⁻⁶⁸
	FS–FS vs. No correlation	104	—	**	8 × 10⁻⁹	††	χ²	> **	1 × 10⁻¹³⁰	90	—	**	1 × 10⁻¹⁵	††	χ²	> **	0
CC peak height H (N _basal = 500)	RS–RS vs. RS–FS	5064	855	*	0.02		t ₁₀₀₉	< **	7 × 10⁻⁴	6406	890	**	4 × 10⁻³	†	t ₁₄₉₀	< **	1 × 10⁻⁵	Fig. 3 C
	RS–RS vs. FS–RS	5064	845	**	1 × 10⁻⁸	††	t ₂₄₀₃	> **	2 × 10⁻¹⁴	6406	883	**	7 × 10⁻²³	††	t ₅₄₁₉	> **	1 × 10⁻²⁰
	RS–RS vs. FS–FS	5064	92		0.1					6406	80		0.05
	RS–FS vs. FS–RS	855	845	**	6 × 10⁻¹⁰	††	t ₁₀₄₆	> **	6 × 10⁻¹¹	890	883	**	4 × 10⁻²²	††	t ₁₁₂₉	> **	3 × 10⁻²⁴
	RS–FS vs. FS–FS	855	92		0.4					890	80		0.3
	FS–RS vs. FS–FS	845	92	**	2 × 10⁻³	††	t ₁₁₀	< **	8 × 10⁻⁵	883	80	**	1 × 10⁻⁷	††	t ₈₄	< **	2 × 10⁻⁵
SI[H _task,H _no‐task] (N _basal ≥ 200)	RS–RS Real vs. Shuffle	1528	1528	**	8 × 10⁻¹⁵	††	t ₃₀₁₈	< **	1 × 10⁻¹³	2145	2145	*	0.01		t ₄₂₅₇	< *	0.01	Fig. 6 A
	RS–FS Real vs. Shuffle	363	363	*	0.01	†	t ₇₀₀	> *	0.02	416	416		0.1
SI[H _hold,H _pull] (N _basal ≥ 200)	RS–RS Real vs. Shuffle	255	255		0.1					167	167		0.4					Fig. 6 B
	RS–FS Real vs. Shuffle	139	139		0.2					99	99		0.7
SI[H _ext,H _int] (N _basal ≥ 200)	RS–RS Real vs. Shuffle	769	769		0.5					628	628		0.7					Fig. 6 C
	RS–FS Real vs. Shuffle	275	275		0.2					237	237		0.8
D _hold – D _pull (N _basal ≥ 200)	RS–RS Real vs. Shuffle	255	255		0.7					167	167		0.4					Fig. 7 A
	RS–FS Real vs. Shuffle	139	139		0.9					99	99		0.9
D _ext – D _int (N _basal ≥ 200)	RS–RS Real vs. Shuffle	769	769		0.9					628	628		0.8					Fig. 8 B
	RS–FS Real vs. Shuffle	275	275		0.05					237	237		0.7
Pair‐spike information	RS–RS I _rate vs. I _shuffle	255	255	**	3 × 10⁻⁹⁰	††	t ₂₅₄	> **	2 × 10⁻¹⁰	167	167	**	2 × 10⁻⁵⁸	††	t ₁₆₆	> **	7 × 10⁻¹²	Fig. 7 C
I _pair (hold/pull)	RS–RS I _pair vs. I _rate	255	255		0.9					167	167		0.9
(N _basal ≥ 200)	RS–FS I _rate vs. I _shuffle	139	139	**	3 × 10⁻⁴⁰	††	t ₁₃₈	> **	1 × 10⁻⁵	99	99	**	7 × 10⁻⁴⁰	††	t ₉₈	> **	9 × 10⁻¹⁵
	RS–FS I _pair vs. I _rate	139	139		0.9					99	99		0.9
Pair‐spike information	RS–RS I _rate vs. I _shuffle	769	769	**	3 × 10⁻¹⁰⁹	††	t ₇₈₈	> **	1 × 10⁻⁴⁷	628	628	**	2 × 10⁻⁹¹	††	t ₆₂₈	> **	1 × 10⁻²³	Fig. 7 D
I _pair (external/internal)	RS–RS I _pair vs. I _rate	769	769		0.8					628	628		0.9
(N _basal ≥ 200)	RS–FS I _rate vs. I _shuffle	275	275	**	2 × 10⁻²¹	††	t ₃₀₃	> **	6 × 10⁻¹⁰	237	237	**	2 × 10⁻²⁸	††	t ₂₃₉	> **	1 × 10⁻¹³
	RS–FS I _pair vs. I _rate	275	275		0.9					237	237		0.9
CC peak height H (N _basal = 500)	RS_task–RS_task vs. RS_task–RS_non	3755	574	*	0.04		t ₅₉₂	n.s.	0.06	4550	695		0.4					Fig. 10 A
	RS_task–RS_task vs. RS_non–RS_task	3755	569		0.9					4550	687		0.2
	RS_task–RS_task vs. RS_non–RS_non	3755	166		0.1					4550	474		0.9
	RS_task–RS_non vs. RS_non–RS_task	574	569		0.1					695	687	*	0.03		t ₁₂₆₃	> *	0.01
	RS_task–RS_non vs. RS_non–RS_non	574	166	*	0.03		t ₆₉₉	> *	0.01	695	474		0.5
	RS_non–RS_task vs. RS_non–RS_non	569	166		0.3					687	474		0.3
	RS_task–FS_task vs. RS_non–FS_task	665	152		0.9					662	153	*	0.01	†	t ₁₉₃	n.s.	0.06
CC peak height H (N _basal = 500)	RS_hold–RS_hold vs. RS_hold–RS_pull	111	233		0.2					94	197	*	0.01	†	t ₉₃	n.s.	0.2	Fig. 10 B
	RS_hold–RS_hold vs. RS_pull–RS_hold	111	233		0.6					94	194	*	0.01		t ₉₃	n.s.	0.2
	RS_hold–RS_hold vs. RS_pull–RS_pull	111	454		0.8					94	631	*	0.02		t ₉₃	n.s.	0.2
	RS_hold–RS_pull vs. RS_pull–RS_hold	233	233		0.7					197	194		0.9
	RS_hold–RS_pull vs. RS_pull–RS_pull	233	454		0.5					197	631		0.8
	RS_pull–RS_hold vs. RS_pull–RS_pull	233	454		0.6					194	631		0.9
	RS_hold–FS_pull vs. RS_pull–FS_pull	67	162		0.4					44	133		0.3
CC peak height H (N _basal = 500, RS_pull–RS_pull)	Δt in Range 1 vs. Range 2	112	116		0.9					144	171		0.9					Fig. 10 C
	Δt in Range 1 vs. Range 3	112	116		0.3					144	171		0.2
	Δt in Range 1 vs. Range 4	112	110		0.5					144	145		0.5
	Δt in Range 2 vs. Range 3	116	116		0.1					171	171		0.2
	Δt in Range 2 vs. Range 4	116	110		0.3					171	145		0.8
	Δt in Range 3 vs. Range 4	116	110		0.9					171	145		0.6
CC peak height H (N_basal=500, RS_pull–RS_pull)	ΔSI in Range 1 vs. Range 2	87	112		0.9					161	123		0.1					Fig. 10 D
	ΔSI in Range 1 vs. Range 3	87	112		0.7					161	121		0.4
	ΔSI in Range 1 vs. Range 4	87	85		0.9					161	160		0.7
	ΔSI in Range 2 vs. Range 3	112	112		0.4					123	121	*	0.01		t ₂₄₂	n.s.	0.06
	ΔSI in Range 2 vs. Range 4	112	85		0.9					123	160		0.2
	ΔSI in Range 3 vs. Range 4	112	85		0.7					121	160		0.3

Open in a new tab

A, KS test, Welch's t test, F test: ^** p < 0.01, ^* p < 0.05, p‐values rounded down to one effective digit. False discovery rate (FDR) for 14 KS tests: ^††FDR < 0.01, ^†FDR < 0.05. B, RS_non: non‐task‐related RS neuron; KS test, Welch's t test, χ² test: ^** p < 0.01, ^* p < 0.05, p‐values rounded down to one effective digit. FDR for 108 KS tests: ^††FDR < 0.01, ^†FDR < 0.05.

Table 3.

Additional statistical tests for validation

			CFA							RFA

Statistics (restriction)	Comparison		n vs. n		KS	p	t test		p	n vs. n		KS	p	t test		p	Figure
For juxtacellular nanostimulation
Spike rate	Juxta RS	Stim vs. No‐stim	10	10			paired t	>**	1 × 10⁻³	8	8			paired t	>**	2 × 10⁻³	Fig. 4 B
	Multi RS/FS	Stim vs. No‐stim	63	63			paired t	n.s.	0.9	29	29			paired t	n.s.	0.9
SI[H _stim,H _no‐stim]	Juxta RS– Multi RS/FS	Real vs. Shuffle	61	6100		0.05				28	2800		0.5				Fig. 4 D
(N _basal ≥ 150)
	Multi RS/FS– Juxta RS	Real vs. Shuffle	60	6000		0.1				§	§		§				Fig. 4 E
										(§ Inappropriate for the analysis because the number of pairs is small)
For different thresholds of task relevance index (TRI)
SI[H _task, H _no‐task]	RS–RS Real vs. Shuffle		773	773	**	3 × 10⁻⁶	t ₁₅₃₁	<**	1 × 10⁻⁴	874	874		0.3
(TRI > 12, N _basal ≥ 200)	RS–FS Real vs. Shuffle		195	195		0.1				182	182		0.7
SI[H _task, H _no‐task]	RS–RS Real vs. Shuffle		412	412	**	2 × 10⁻⁴	t ₈₂₀	<**	4 × 10⁻⁴	440	440		0.7
(TRI > 24, N _basal ≥ 200)	RS–FS Real vs. Shuffle		113	113		0.07				92	92		0.1
For the subgroup of rats who were likely to discriminate external/internal movements
SI[H _ext,H _int]	RS–RS Real vs. Shuffle		230	230		0.5				145	145		0.2				Fig. 8 B
(N _basal ≥ 200)	RS–FS Real vs. Shuffle		93	93		0.3				42	42		0.9
D _ext – D _int	RS–RS Real vs. Shuffle		230	230		0.8				145	145		0.9				Fig. 8 C
(N _basal ≥ 200)	RS–FS Real vs. Shuffle		93	93		0.05				42	42		0.9
Pair‐spike information	RS–RS I _rate vs. I _shuffle		230	230	**	5 × 10⁻⁵⁷	t ₂₃₂	>**	5 × 10⁻²¹	145	145	**	1 × 10⁻⁴⁰	t ₁₄₄	>**	8 × 10⁻⁹
I _pair (external/ internal)	RS–RS I _pair vs. I _rate		230	230		0.9				145	145		0.9				Fig. 8 D
(N _basal ≥ 200)	RS–FS I _rate vs. I _shuffle		93	93	**	3 × 10⁻¹¹	t ₁₀₁	>**	5 × 10⁻⁹	42	42	**	2 × 10⁻¹⁶	t ₄₁	>**	9 × 10⁻⁷
	RS–FS I _pair vs. I _rate		93	93		0.9				42	42		0.9
For the subgroup of the neuronal pairs which exhibited significant synchronous activities (P < 0.005)
SI[H _task,H _no‐task]	RS–RS Real vs. Shuffle		227	22700	**	2 × 10⁻¹¹	t ₂₂₉	<**	9 × 10⁻⁸	348	34800	*	0.02	t ₃₅₂	n.s	0.08	Fig. 9 A
(N _basal ≥ 200)	RS–FS Real vs. Shuffle		65	6500	*	0.03	t ₆₅	n.s.	0.6	91	9100		0.2
SI[H _hold,H _pull]	RS–RS Real vs. Shuffle		69	6900	*	0.02	t ₆₉	>*	0.04	52	5200		0.9				Fig. 9 B
(N _basal ≥ 200)	RS–FS Real vs. Shuffle		33	3300		0.8				44	4400		0.9
SI[H _ext,H _int]	RS–RS Real vs. Shuffle		150	15000		0.3				163	16300		0.1				Fig. 9 C
(N _basal ≥ 200)	RS–FS Real vs. Shuffle		63	6300		0.2				69	6900		0.9
Temporal consistency of significant selectivity of spike synchrony (significant neuron pairs with p < 0.05 in χ2 test of RS–RS, RS–FS pairs in CFA and RFA)
SI[H _hold,H _pull]	RS–RS/FS	1st half vs. 2nd half	38	38	Pearson's correlation test			r = 0.664		** p = 5 × 10⁻⁶		2 × 2 χ² test			** p = 3 × 10⁻⁵		Fig. 7 E
SI[H _ext,H _inh]	RS–RS/FS	1st half vs. 2nd half	27	27	Pearson's correlation test			r = 0.161		p = 0.4							Fig. 7 F

Open in a new tab

KS test, Welch's t test, paired t test, χ² test: ^** p < 0.01, ^* p < 0.05, p‐values rounded down to one effective digit.

Our main statistical conclusions were based on differences at the population level. First, we performed the Kolmogorov–Smirnov (KS) test to evaluate differences between distributions of certain features in spike rate or synchrony of various neuronal attributes. If a difference was significant, we performed additional tests. Welch's t test or the chi‐square test was performed in order to determine whether one group had a significantly larger mean, and the F test was performed in order to determine whether one group had a significantly larger variance. For combinatorial comparisons, we confirmed the main effects by ANOVA.

We performed multiple comparisons of many features in spike rate and synchrony for many neuronal attributes and the combinations on many conditions in task sequences. Therefore, we evaluated the false discovery rate (FDR) to determine whether the statistical significance of each KS test occurred by chance. Because the numbers of data sources were strikingly different, the FDR analyses for spike rate and spike synchrony were performed separately, and the tests for behavioural and juxtacellular data were not included in the FDR analyses. The additional tests for the same statistical features (Table 3) were not included in the FDR analyses. A list of tests subjected to FDR analyses is provided in Table 2. Our FDR analysis followed the Benjamini–Hochberg method (Benjamini & Hochberg, 1995). The p‐values of respective KS tests were sorted as p ₁ ≤ p ₂ ≤ … p_i ≤ … p_M, where M is the number of KS tests. Tests from 1 to the maximum order i, such that p_iM⁄i < q, were accepted at a significance level q. We used two significance levels for FDR, ††: q = 0.01, †: q = 0.05.

All error bars and centre circles in figures and data in the text indicate means ± standard deviations (SD).

Results

Contribution of CFA and RFA to behavioural task performance

To investigate spike rate and synchrony in motor cortex neurons, we trained adult rats to perform an external‐ and internal‐trigger forelimb movement task under head fixation (Fig. 1 A and B; see Supplemental Video S1 is shown in Kimura et al. 2012; Isomura et al. 2009. In the external‐trigger trials, rats had to push a lever, hold it for 1 s, and then pull it in response to auditory cue presentation in order to receive reward water; in the internal‐trigger trials, they had to do so spontaneously without cue presentation. The rats successfully performed 1230 ± 460 trials, including both trial types, for 3 h (N = 77 rats; 73 of the 77 rats for the subsequent analysis, see Methods). The rats exhibited a transient peak hold time of 1.15–1.20 s (0.15–0.20 s after the cue onset) in the external‐trigger trials, and a broad peak around 1.10–1.15 s in the internal‐trigger trials (Fig. 1 C; N = 73). Although there was a significant difference in the hold time for correct trials between external‐ and internal‐trigger trials (external: 1.14 ± 0.07 s; internal: 1.34 ± 0.14 s; p = 3×10⁻²¹, paired t test), we did not observe such a large difference between the two trial types in either lever trajectory or EMG activity in the right forelimb muscles during pull movements (Fig. 1 D). Thus, it was possible to directly compare neuronal activities between motor initiations.

To determine whether the external‐ and internal‐trigger forelimb movements were dependent on activation of the two motor cortices, CFA and RFA, we performed pharmacological inactivation experiments with muscimol, a GABA_A receptor agonist (R. Kimura et al., unpublished observations). There was a significant effect of the muscimol on task performance with the right forelimb in both external‐ and internal‐trigger trials (p < 0.01, one‐way ANOVA). Unilateral (left‐side) inactivation of both CFA and RFA, or only RFA, by muscimol (1 μg μl⁻¹, 0.5 μl) strongly inhibited task performance (relative changes in correct trials: CFA and RFA 31.4 ± 42.6%, N = 4, p = 0.04, one‐sample t test compared with 100%; RFA 44.2 ± 28.5%, N = 4, p = 0.02), whereas unilateral inactivation of only CFA did not significantly alter performance (129 ± 32 %, N = 6, n.s.), even at higher doses (5 or 10 μg μl⁻¹, 0.5 μl; data not shown). In the reference group that received saline injections in both CFA and RFA, there was little change in performance (93.6%, N = 2). Furthermore, the intact group exhibited no significant changes between the early phase (0–10 min) and late phase (70–85 min) after task initiation, which were separated by the same elapsed time as in the pharmacological experiments (99.7 ± 60.4%, N = 77, n.s., one‐sample t test). Thus, along with CFA, which is believed to play crucial roles in motor execution, RFA also plays a role in external and internal initiation of forelimb movements. Injections of biotinylated dextran amine (BDA), an anterograde tracer, into unilateral RFA (R. Kimura et al., unpublished observations) revealed axonal projections from there to ipsilateral CFA, contralateral RFA, ipsilateral dorsolateral striatum (DLS), and contralateral spinal cord, as previously reported (Rouiller et al. 1993; Hira et al. 2013 b). Also, intracortical microstimulation to CFA or RFA evoked an EMG response with short latency (about 10 ms; data not shown; Neafsey et al. 1986; Liang et al. 1993; Isomura et al. 2009; Hira et al. 2013 a, b ).

Functional spike rate in motor cortices during external‐ and internal‐trigger movements

We examined neuronal activity by multiple single‐unit recording and/or juxtacellular recording from deep layers (putative layer 5B) in CFA or RFA during task performance. All isolated units from both recordings were classified by spike duration as regular‐spiking (RS) or fast‐spiking (FS) neurons (Table 1 A; Isomura et al. 2009; see Methods). Most RS neurons are excitatory pyramidal neurons, whereas most FS neurons are inhibitory interneurons (Barthó et al. 2004; Isomura et al. 2009). Averaged ongoing activity was higher in FS neurons than in RS neurons (CFA: RS, n = 1,405, 3.12 ± 4.71 Hz; FS, n = 113, 15.0 ± 22.0 Hz; p = 7 × 10⁻⁸ in Welch's t test; RFA: RS, n = 1,444, 2.82 ± 3.87 Hz; FS, n = 89, 14.0 ± 14.2 Hz; p = 6×10⁻¹¹). RS and FS neurons exhibited various functional activities related to forelimb movements (Isomura et al. 2009; Saiki et al. 2014). We defined task‐related and non‐task‐related neurons using our task relevance index and the total number of spikes (Table 1 A; see Methods). The task‐related neurons were further classified as pull onset‐/push end‐aligned neurons or hold‐/pull‐related neurons (see Methods; Table 1 A).

We next investigated how the peak activity of the pull‐related neurons was affected by the difference in motor initiation (external‐ and internal‐trigger; Fig. 2). We confirmed that activity peak time in pull‐related neurons differed neither between external and internal motor initiations nor between CFA and RFA (Fig. 2 C; KS test, see Table 2 A; e.g. CFA RS, external, 0.052 ± 0.269 s; internal, 0.042 ± 0.280 s from pull onset). Some RS and FS neurons exhibited significantly different activities between the external‐ and internal‐trigger movements (Fig. 2 A). Therefore, we examined the distribution of the selectivity index (SI[R _ext,R _int]), which represents the selectivity of individual peak activity to external‐ or internal‐trigger movements (Fig. 2 D). Many neurons exhibited significant selectivity (p < 0.05; Fig. 2 D, dark‐colour bars in histograms; RS in CFA, external 16.4%, internal 14.3%; RS in RFA, external 31.2%, internal 12.7%; FS in CFA, external 20.6%, internal 0%; FS in RFA, external 32.0%, internal 8.0%) in the Wilcoxon rank‐sum test for spike numbers in each trial (external or internal). The distribution of SI[R _ext,R _int] for RS/FS neurons in CFA/RFA differed significantly from surrogate data in which external‐ and internal‐trigger trials were shuffled (KS test, see Table 2 A). The variance of SI[R _ext,R _int] was significantly larger than in trial‐shuffled data (F test, see Table 2 A). RS neurons in RFA, as well as FS neurons in CFA and RFA, had a significant tendency to discharge more frequently during the external‐trigger movements (Welch's t test compared with shuffled data, see Table 2 A). Their distribution for RS neurons differed significantly between CFA and RFA (KS test, see Table 2 A). The tendency of RS neurons toward external‐trigger movements was larger in RFA than in CFA (Welch's t test, see Table 2 A).

A, representative pull‐related neuron exhibiting distinct patterns of spike activities in relation to pull movement for external‐trigger (left) and internal‐trigger (right) trials in the neuron (top, mean and SD of lever trajectories; middle, raster plots; bottom, peri‐event time histogram (PETH) of spike activities aligned to pull onset). B, PETH normalized to the respective maximal activity, aligned to push end/pull onset. Blocks (ordered sequentially from the top) correspond to push end‐aligned task‐related RS and FS neurons sorted by peak time for the task type with larger peak amplitude (external or internal), pull onset‐aligned task‐related RS and FS neurons sorted by peak time for the task type with larger peak amplitude, and non‐task‐related RS and FS neurons sorted by decreasing order of the task relevance indices (Table 1 A). Bottom, averaged lever trajectories (N = 73). C, cumulative histograms of peak time of pull onset‐aligned activities in pull‐related neurons (n.s., as determined by KS test; Table 2 A). D, neuronal selectivity for external and internal movements. Top and middle, histograms of selectivity index (SI[R _ext,R _int], 100 ms window) for external or internal movements of pull‐related neurons (n.s. and ^**, as determined by one‐sample t test against 0). Darker‐coloured bars represent neurons with significant differences at the single‐neuron level (p < 0.05, Wilcoxon rank‐sum test). Bottom, cumulative histograms of SI[R _ext,R _int] of real and trial‐shuffled data (Table 2 A; ^**, as determined by KS test). [Correction made on 1 January 2017 after first online publication: a label in figure 2B was corrected.]

Significant emergence of spike synchrony in motor cortices

Neurons can exhibit synchronous activities on various time scales. Here, we focused on millisecond‐order spike synchrony independent of slower spike rate changes (Fig. 2). The spike synchrony of a neuron pair tended to be stable during the recording period (R. Kimura et al., unpublished observations). We evaluated the peak within a 5 ms delay in a spike cross‐correlogram (CC) of a pair of simultaneously recorded neurons, in comparison with the 10 ms scale correlation. For this purpose, we introduced two indicators: the statistical significance (P‐value) and the relative height (H‐value) of the CC peak. The P‐value refers to the statistical reliability of the CC peak, calculated by a uni‐directional random jittering protocol in the 10 ms range. The H‐value refers to the strength of the CC peak, calculated as the ratio of the peak height to the mean height in the basal range (10 ms length).

Figure 3 A illustrates a pair consisting of a hold‐related RS neuron and a pull‐related FS neuron, which exhibited a single peak with a several‐millisecond delay in their CC (RS–FS; RS prior to FS). The distributions of the statistical significance, P‐values, for all combinations were significantly different from the theoretical distribution for no correlation (Fig. 3 B; KS test, see Table 2 B). The ratios of neuron pairs exhibiting significant spike synchrony (P < 0.005; CFA RS–RS: 6.17%, RS–FS: 12.6%, FS–RS: 3.35%, FS–FS: 17.3%, RFA RS–RS: 5.92%, RS–FS: 13.4%, FS–RS: 4.11%, FS–FS: 30.0%) were significantly larger than predicted to occur by chance (chi‐square goodness‐of‐fit test, see Table 2 B). These distributions of P‐value clearly indicate that spike synchrony existed in all combinations of RS/FS subtypes.

A, an example cross‐correlogram (CC) between a hold‐related RS neuron and a pull‐related FS neuron in CFA. PETHs of the two neurons in external movements are shown on both sides. The reliability (P‐value; red, probability distribution for significance threshold) and strength (H‐value; orange and green, relative peak height to baseline) of the CC peak were evaluated as described in Methods. B, cumulative distribution of P‐value among RS/FS neuron subtypes, which had over 200 spike pairs within the basal range, compared with the theoretically expected uniform distribution of jittered data with no spike correlation (dashed curve, ‘no correlation’; common to all pairs of neuron subtypes). In all combinations, the distributions were significantly different from that of no correlation (Table 2 B; ^**, as determined by KS test). C, cumulative distributions of H‐value for combinations of RS/FS neuron subtypes. ^** and ^* indicate statistical significances of distribution differences, as determined by KS test (Table 2 B).

Although we observed differences in the significance of spike synchrony among pairs of RS/FS‐subtype groups (P‐value, Fig. 3 B), this might have depended on the number of recorded spikes. Closer inspection, using the strength of spike synchrony (H‐value) calculated from cutoff spike sequence data with a fixed number of spike pairs within the basal range (N _basal = 500), revealed significant differences in the H‐value distributions among the pairs of RS/FS‐subtype groups (Fig. 3 C; KS test, see Table 2 B). The FS–RS pair group had smaller H‐values than others, and the RS–FS pair group had larger H‐values than the RS–RS pair group (Welch's t test, see Table 2 B). Thus, we found a difference in spike synchrony among the pairs of RS/FS‐subtype groups.

Validity for the evaluation methods of spike synchrony

We confirmed that the significance of spike synchrony was not sensitive to various parameters in the CC analysis (R. Kimura et al., unpublished observations), including jittering widths (2, 5, 10, 20 ms) and bin widths (0.5, 1.0, 2.0 ms); therefore, we adopted 10 ms for the jittering width and 0.5 ms for bin width. We observed unbiased distributions of spike synchrony across different rats and between the same and different tetrodes, although the number of possible pairs varied depending on the recording conditions. Because most significant spike synchrony emerged within a period of several milliseconds (CFA: RS–RS 1.28 ± 1.07 ms, RS–FS 1.50 ± 0.67; RFA: RS–RS 1.38 ± 1.06, RS–FS 1.84 ± 0.78), we focused on spike synchrony within a 5 ms delay.

In order to determine the robustness of our spike synchrony analysis using the H‐value against alteration in spike rate, we investigated whether spike synchrony was affected by electrical stimulation of a single neuron through a juxtacellular recording electrode (nanostimulation; Houweling et al. 2010). Nanostimulation evokes additional spikes in a single juxtacellularly recorded neuron by repeatedly (1210 ± 347 times, n = 18) injecting positive current in the nano‐ampere range, thereby mimicking temporary changes in single‐cell excitability during ongoing cortical activity. Indeed, we were able to significantly increase the spike rate in juxtacellularly recorded RS neurons (Fig. 4 A and B; CFA, 22.99 ± 15.54 Hz during stimulation, 9.88 ± 7.24 Hz during no‐stimulation, p = 1 × 10⁻³ in paired t test; RFA, 16.16 ± 7.57 Hz, 4.44 ± 1.52 Hz, p = 2 × 10⁻³ in paired t test; see Table 3), but not in their neighbours simultaneously captured by multiple single‐unit recording (RS and FS in CFA, 9.10 ± 8.31 Hz during stimulation, 9.10 ± 8.35 Hz during no‐stimulation; RS and FS in RFA, 10.9 ± 8.37 Hz, 10.9 ± 8.38 Hz; n.s. in paired t test). The distribution of H‐value changes in response to nanostimulation, reflected by the selectivity index (SI[H _stim,H _no‐stim]) in RS–RS/FS pairs that these juxtacellularly stimulated RS neurons formed with the other multiple single unit‐recorded RS/FS neurons (Fig. 4 C and D) or RS/FS–RS pairs that the other multiple single unit‐recorded RS/FS neurons formed with juxtacellularly stimulated RS neurons (Fig. 4 E), did not significantly differ from that of surrogate data in which the stimulation and no‐stimulation periods were shuffled (KS test, see Table 3). Thus, an artificial change in single‐cell excitability did not induce a significant change in spike synchrony.

We reconstructed networks by linking RS–RS and RS–FS neuron pairs exhibiting significant spike synchrony (P < 0.005), and characterized the network architecture in regard to two circuit components, convergence and divergence, since each link has directionality corresponding to the temporal order of spikes in the significant spike synchrony (Buzsáki 2010; Harris & Mrsic‐Flogel, 2013). The degree, i.e. the number of directional links per one neuron, did not exhibit a biased distribution across different rats, and the number of neurons with higher degrees (i.e. hub‐like neurons), was significantly larger than expected (Fig. 5).

A, representative network architecture reconstructed by linking neuron pairs exhibiting significant spike synchrony (P < 0.005) with pseudo‐colored lines. This network consisted of 28 RFA neurons. Neuron 1 had 10 convergent links, and neuron 2 had 8 divergent links. B, degree distribution for convergent and divergent circuit components among RS neurons; here “degree” is used to refer to the concept in network analysis. Statistical significance was calculated by 200,000 samplings of the degree distribution in shuffled data (gray lines: statistical significance thresholds; ^***, p < 0.001; ^**, p < 0.01; ^*, p < 0.05; red, increase; blue, decrease).

Independence of temporal changes in spike synchrony and movement state

Spike synchrony can change dynamically depending on behavioural phase or top‐down control, as reported previously (Vaadia et al. 1995; Riehle et al. 1997; Hatsopoulos et al. 1998; Steinmetz et al. 2000; Narayanan & Laubach, 2006; Cohen & Newsome, 2008; Fujisawa et al. 2008; Mitchell et al. 2009; Hirabayashi et al. 2013 b). Hence, we examined the changeability of spike synchrony during task performance.

First, we examined differences between the strength of spike synchrony during task (H _task) and no‐task (resting) periods (H _no‐task) (Fig. 6 A). As reported previously, many neuron pairs exhibited significant differences (neuron pairs with p < 0.05 by 2 × 2 chi‐square test, shown as black bars in histograms in the right column of Fig. 6 A; RS–RS in CFA, 14.4%; RS–RS in RFA, 11.0%; RS–FS in CFA, 13.5%; RS–FS in RFA, 15.4%). To determine whether the changes in spike synchrony depended on task/no‐task periods, we compared the distribution of H‐value changes, indicated as the selectivity index (SI[H _task,H _no‐task]), with that of surrogate data in which task and no‐task periods were shuffled (pseudo‐500 ms periods). In the RS–RS pair group in CFA, the distribution of the selectivity index (SI[H _task,H _no‐task]) was shifted significantly leftward (i.e. biased to the no‐task period) relative to that obtained from the shuffled data (KS test and Welch's t test, see Table 2 B); a similar trend was observed in RFA, although the significance was weak (KS test and Welch's t test, see Table 2 B). We also obtained a similar result for SI[H _task,H _no‐task] in the RS–RS pair group in CFA, even when we extracted only the task‐related neurons, whose definition was changed by shifting the threshold of the task relevance index (from 6 to 12 or 24; significant in KS test and Welch's t test, see Table 3).

A, difference between spike synchrony strengths during task periods (H _task) and no‐task periods (H _no‐task). Left, three examples of CCs during task and no‐task periods exhibiting apparent differences (top, middle) or no difference (bottom) (^** and n.s. on CCs indicate significances of spike synchrony, as determined by P‐value; ^**, P < 0.005; n.s., P ≥ 0.025). Right, histograms and cumulative distributions of spike synchrony change, represented by selectivity index SI[H _task,H _no‐task]. Filled circles and error bars represent means and SD (^**, ^* and n.s., as determined by one‐sample t test against 0). Black bars of histograms represent neuron pairs exhibiting significant changes at the single‐pair level (p < 0.05, 2 × 2 chi‐square test). Cumulative distributions (black curves) were compared with those of surrogate data (cyan curves) in which task and no‐task periods (pseudo–500 ms periods; see Methods) were shuffled (n.s., ^* and ^**, as determined by KS test; Table 2 B). B, difference between spike synchrony strengths during hold periods (H _hold) and pull periods (H _pull). An example of CCs, histograms, and cumulative distributions of spike synchrony change (SI[H _hold,H _pull]) as shown in A. Shuffled data were obtained by randomly swapping hold and pull periods. C, difference between spike synchrony strengths during external‐trigger (H _ext) and internal‐trigger movement periods (H _int). An example of CCs, histograms and cumulative distributions of spike synchrony change (SI[H _ext,H _int]) as shown in A. Shuffled data were obtained by randomly swapping external‐ and internal‐trigger trials.

Next, we examined differences between the strength of spike synchrony during hold (H _hold) and pull periods (H _pull) (Fig. 6 B). Many neuron pairs exhibited significant differences (neuron pairs with p < 0.05 by 2 × 2 chi‐square test, shown as black bars in histograms of Fig. 6 B; RS–RS in CFA, 11.0%; RS–RS in RFA, 8.98%; RS–FS in CFA, 10.1%; RS–FS in RFA, 7.07%). However, in all RS–RS and RS–FS pair groups, the SI[H _hold,H _pull] distribution did not differ significantly from that of surrogate data in which the hold and pull periods were shuffled (KS test, see Table 2 B). At the population level, temporal changes in strength of spike synchrony were statistically independent of hold and pull movements.

Similarly, we examined possible differences between the strength of spike synchrony during external‐ (H _ext) and internal‐trigger trials (H _int) (Fig. 6 C). Many neuron pairs exhibited significant differences (neuron pairs with p < 0.05 by 2 × 2 chi‐square test, shown as black bars in histograms of Fig. 6 C; RS–RS in CFA, 9.49%; RS–RS in RFA, 8.92%; RS–FS in CFA, 12.0%; RS–FS in RFA, 10.5%). However, in all RS–RS and RS–FS pair groups, the SI[H _ext,H _int] distribution did not differ significantly from that of surrogate data in which the external‐ and internal‐trigger trials were shuffled (KS test, see Table 2 B), namely, temporal changes in strength of spike synchrony were statistically independent of externally and internally initiated movements.

Furthermore, we found that temporal changes in timing of spike synchrony were also statistically independent of behavioural state (hold/pull or external/internal). The distribution of the difference between spike delays of CC peaks during hold and pull periods (D _hold – D _pull) or external‐ and internal‐trigger movement periods (D _ext – D _int) did not differ significantly from that of the shuffled data (Fig. 7 A and B; KS test, see Table 2 B).

A, differences between the spike delays of CC peaks during hold periods (D _hold) and pull periods (D _pull). Cumulative distributions (black curves) of the spike delay difference D _hold – D _pull were compared with those of shuffled data (cyan curves) as shown in Fig. 6 B (n.s. as determined by KS test; Table 2 B)*. B*, difference between spike delays of CC peaks during external‐trigger (D _ext) and internal‐trigger movement periods (D _int) as shown in A. Shuffled data were identical to those in Fig. 6 C. C, mutual information between the motor phase (hold/pull) and spike patterns of each neuron pair within 1 ms windows separated by the spike delay of the CC peak (see Methods). Cumulative distributions of the information carried by the real pair‐spike pattern (I _pair; black curves), by the pair‐spike pattern of shuffled data identical to those in Fig. 6 B (I _shuffle; cyan curves), and by spike rate pattern (I _rate; pink curves) (^** and n.s., as determined by KS test; Table 2 B). The distributions of I _pair almost coincided with those of I _rate. D, mutual information between motor initiation (external/internal) and pair‐spike patterns as shown in C. Shuffled data were identical to those in Fig. 6 C. E, temporal stability of significant motor‐phase specificity (hold/pull; n = 38 pairs with p < 0.05 in 2 × 2 chi‐square test; black bars of histograms in Fig. 6 B) of spike synchrony (SI[H _hold,H _pull], first half vs. second half of the recording session, r = 0.664, p = 5 × 10⁻⁶, as determined by Pearson's correlation coefficient test; p = 3 × 10⁻⁵, among four quadrants, 2 × 2 chi‐square test; Table 3). F, no temporal stability of significant motor‐initiation preference (external/internal; n = 27 pairs with p < 0.05 in 2 × 2 chi‐square test; black bars of histograms in Fig. 6 C) of spike synchrony (SI[H _ext,H _int], first half vs. second half, r = 0.161, n.s., as determined by Pearson's correlation coefficient test).

In order to determine how much information is carried by spike synchrony, we evaluated mutual information I _pair between the behavioural state and spike pattern of a neuron pair within 1 ms windows separated by the spike delay of the CC peak (see Methods). This I _pair also contains information for the spike rate. Therefore, we also evaluated mutual information I _rate calculated from the respective spike rates of the neuron pair, based on the hypothesis that the two neurons were independent. The distribution of I _pair almost coincided with the distribution of I _rate (Fig. 7 C and D; n.s. in KS test, see Table 2 B). However, the I _rate was significantly larger than the I _shuffle of shuffled data (KS test and Welch's t test, see Table 2 B). These results showed that information about behavioural state was primarily carried by the spike rates alone, and that the detailed spike pair pattern carried little information.

We also confirmed this conclusion for only a subgroup of rats that were more likely to discriminate the external‐ and internal‐trigger trial types (Fig. 8). This subgroup consisted of 15 rats (6 for CFA, 9 for RFA) in which more than 50% of pull‐related neurons exhibited significantly different spike activity between external‐ and internal‐trigger trials (p < 0.05 by Wilcoxon rank test). For neurons of the effectively discriminating subgroup (Fig. 8 A; p < 0.05 by Wilcoxon rank test; CFA RS, n = 20, 62.5% of analysed pull‐related neurons, FS, n = 4, 66.7%; RFA RS, n = 31, 63.3%, FS, n = 3, 100%), we reproduced the results shown in Figs 6 C and 7 B and D (Fig. 8 B, C, and D; Table 3).

A, histograms and cumulative distributions of the selectivity index SI[R _ext,R _int] (see Figure 2 D) for a subpopulation of rats who were more likely to discriminate between external‐ and internal‐trigger movements. In these rats, over 50% of pull‐related neurons exhibited significant selectivity of spike rates (p < 0.05, Wilcoxon rank‐sum test) to external or internal movement (CFA: N = 6; RFA: N = 9). B, cumulative distributions of the spike synchrony change SI[H _ext,H _int] (see Figure 6 C) for the subpopulation of rats defined in A (see Table 3). C, cumulative distributions of the difference in timing of spike synchrony, D _ext ‐ D _int, (see Figure 7 B) for the subpopulation of efficiently task‐discriminating rats. D, cumulative distributions of the mutual information between motor initiations and pair‐spike patterns (see Figure 7 D) for these rats.

In addition, we extracted only the neuron pairs that exhibited significant synchronous activities during the whole recording period (P < 0.005) and examined the relation between the temporal change of their spike synchrony and behavioural states (Fig. 9). We obtained results similar to those shown in Fig. 6 in almost all cases, with the exception of one comparison (Fig. 9 B; SI[H _hold,H _pull] of RS–RS pair group in CFA; weakly significant in KS test and Welch's t test, see Table 3).

A, difference between spike synchrony strengths during task periods (H _task) and no‐task periods (H _no‐task) in a subgroup of neuronal pairs with significant synchronous activities (P < 0.005) in a whole recording period. Cumulative distributions of spike synchrony change, represented by selectivity index SI[H _task,H _no‐task]. Cumulative distributions (black curves) were compared with those of surrogate data (cyan curves; 100 shuffled data for every neuron pair because of the small number of neurons in this subgroup) by KS test. B, difference between spike synchrony strengths during hold periods (H_hold) and pull periods (H_pull). C, difference between spike synchrony strengths during external‐trigger movement periods (H _ext) and internal‐trigger movement periods (H_int).

We also found that neuron pairs exhibiting spike synchrony that was significantly selective for either the hold or pull period (Fig. 6 B) maintained stable selectivity (SI[H _hold,H _pull]) between the early and late halves of the recording session (Fig. 7 E; significant in Pearson's correlation coefficient test and 2 × 2 chi‐square test among four quadrants, see Table 3). By contrast, neuron pairs selective to external‐ or internal‐trigger movement (Fig. 6 C) did not exhibit such stability (Fig. 7 F and Table 3).

Independence between spike synchrony and rate‐based neuronal functions

Next, we asked whether a pair of the same functional type of neurons would exhibit stronger spike synchrony. We colour‐coded neuron pairs that exhibited significant spike synchrony (P < 0.005) according to their respective PETH peak times (R. Kimura et al., unpublished observations), but found no clear structure in the dependence between spike synchrony and the task‐related activities. In order to examine this relationship in detail, we compared the distribution of H‐values among pairs of subgroups determined by rate‐based neuronal functions (i.e. task relevance, motor‐phase specificity, motor‐initiation preference).

First, we investigated whether the H‐value distribution depended on combinations of the task‐related and non‐task‐related neurons (Fig. 10A). We observed weak significance in four comparisons (out of a total of 14) of different combinations of neuronal attributes (KS test, see Table 2 B), and only two of the four comparisons had weakly significant bias (Welch's t test, see Table 2 B). Next, we investigated whether the H‐value distribution depended on combinations of the hold‐related and pull‐related neurons (Fig. 10B). We found weakly significant differences between pairs of hold‐related RS neurons (RS_hold–RS_hold) and other pairs in RFA (KS tests for 3 of 14 comparisons, see Table 2 B), which had no significant bias (Welch's t test, see Table 2 B).

We performed multiple combinatorial comparisons for hierarchical classification of neuronal attributes (Table 1 B). To validate these confirmed statistical significances, we performed five‐way ANOVA for all H‐values (n = 15,115 pairs) considering five factors: areas (CFA/RFA), neuron subtype (RS/FS) of one neuron of the pair, neuron subtype of the other neuron, rate‐based functions (non‐task‐related/pull‐related/hold‐related/other‐task‐related) for one neuron, and rate‐based functions for the other neuron. We found significant main effects only for two factors: the subtype (RS/FS) of the former neuron of the pair (p = 0.01) and the subtype of the latter (p = 0.007).

Pull‐related neurons could contribute various timings to a pull‐movement sequence. Therefore, we examined the relationship between spike synchrony in pairs of pull‐related RS neurons and the time difference Δt between their peaks in pull onset‐aligned PETH (Fig. 10 C). In this analysis, a positive value (+) for ∆t means that the order of spiking in spike synchrony is coincident with the order of peaks of pull‐related activity in the pair. The smaller the absolute ∆t value is, the more similar the peak time positions of the two neurons are. All possible pairs were sorted according to ∆t, and then classified into four groups (in order, Range 1, 2, 3, 4) by three ∆t thresholds (−0.1, 0, 0.1 s). In both CFA and RFA, the H‐value distribution did not significantly depend on the similarity in peak time of pull‐related activities (KS test, see Table 2 B).

A, cumulative distributions of H‐value for combinations of task‐related and non‐task‐related neurons. ^* indicates statistical significance of distribution differences, as determined by KS test (Table 2 B). B, cumulative distributions of H‐value for combinations of hold‐related and pull‐related neurons, as shown in A. C, strength of spike synchrony and similarity in peak time positions in motor phase among pairs of pull‐related RS neurons. H‐values of neuron pairs are plotted as red dots for neuron pairs with a significant CC peak (P < 0.005), or black dots for neuron pairs with a non‐significant CC peak (P ≥ 0.005), sorted by the difference in peak time positions of pull onset‐aligned PETHs (∆t = t _n2 – t _n1, where t _n1 and t _n2 are the PETH peak times for the neuron pair respectively). The peak time differences were divided into four groups using three thresholds (−0.1, 0, 0.1 s), indicated by the four‐colour bars (in order: Range 1, 2, 3, 4). Inset, cumulative distributions of H‐value for the four groups (Table 2 B; n.s. by KS test). The four line colours in the inset correspond to the bar colours in the left plot. D, strength of spike synchrony and similarity in rate‐based selectivity to externally or internally initiated movements among pairs of pull‐related RS neurons, as shown in C. Neuron pairs (red: P < 0.005; black: P ≥ 0.005) were sorted by the difference in the selectivity index SI[R _ext,R _int] (∆SI[R _ext,R _int] = SI_n2 – SI_n1, where SI_n1 and SI_n2 are the SI for the respective neuron pairs) and classified into four groups by thresholds (−0.1, 0, 0.1). There was no significant difference among the distributions of the four groups, except for one pair with mutually similar SIs in RFA (cyan, Range 2, vs. light green, Range 3, Table 2 B).

Next, we examined the relationship of spike synchrony (neuron 1 to neuron 2) in pairs of pull‐related RS neurons with the difference in their rate‐based selectivity (∆SI[R _ext,R _int]) for external‐ and internal‐trigger movements (Fig. 10 D). Here, the smaller the absolute ∆SI[R _ext,R _int] is, the more similar the selectivities of the two neurons are. A positive value (+) for ∆SI[R _ext,R _int] means that the activity of neuron 2 is more biased to external movements than that of neuron 1. All possible pairs were sorted according to ∆SI[R _ext,R _int], and classified into four groups according to three ∆SI[R _ext,R _int] thresholds (−0.1, 0, 0.1; Fig. 10 D). In CFA, the H‐value distribution did not depend significantly on the similarity in ∆SI[R _ext,R _int] (KS test, see Table 2 B). In RFA, although we observed a weakly significant difference between Range 2 and 3 (KS test, see Table 2 B), the bias was not significant (Welch's t test, see Table 2 B). Moreover, the significance disappeared when the thresholds were changed slightly (−0.08, 0, 0.08) (n.s. p = 0.08 in KS test). Furthermore, we observed no significant effect of either of two factors for the H‐values in RFA (similar/dissimilar, [Range 2 + Range 3]/[Range 1 + Range 4], n.s. p = 0.9; direction of difference, [Range 1 + Range 2]/[Range 3 + Range 4], n.s. p = 0.5; interaction, n.s. p = 0.06 in two‐way ANOVA).

Discussion

In this study, we quantitatively examined behaviour‐related spike rate and spike synchrony within a period of several milliseconds in deep layers of CFA and RFA in rats. Although many RS and FS neurons participated in external and internal initiation of voluntary movements in both regions, spike rate in RS neurons in RFA was more biased to external movements than that in CFA. Spike synchrony was clearly observed in a significantly large fraction of neuron pairs, and differed among RS/FS pair types. We found many neuron pairs that exhibited spike synchrony depending on behavioural state. The distribution of the dependence on task/no‐task state was significantly different from the distribution of the shuffled surrogate data. However, during task performance, the distribution of spike synchrony dependence on movement phase (hold/pull, external/internal) did not significantly differ from the distribution of the shuffled surrogate data. This suggests that temporal changes in spike synchrony were statistically independent of behavioural changes during task performance, whereas spike rate clearly depended on movement. Mutual information analyses revealed that the populational contribution of spike synchrony to the behavioural functions was less than the contribution of spike rate. Furthermore, the strength of spike synchrony was independent of rate‐based neuronal preferences of the pair, such as task relevance, motor‐phase specificity and motor‐initiation preference.

Spike synchrony independent of rate‐based behavioural functions

In contrast to our conclusions, many previous reports (e.g. Riehle et al. 1997; Hatsopoulos et al. 1998; Putrino et al. 2010) have suggested that spike synchrony contributes dynamically to functional control of behaviours. However, most previous studies were less quantitative than our large‐scale analysis. Indeed, we also found many neuron pairs that exhibited spike synchrony dependent on behavioural functions or rate‐based neuronal preferences. In this sense, our results are consistent with the previous studies. In contrast, our population analyses did not show significant dependence between spike synchrony and rate‐based behavioural functions. The only notable exception was dependence on significantly different behavioural states, task/no‐task, which is consistent with previous large‐scale analyses (Mizuseki & Buzsáki, 2013, 2014).

Several quantitative studies have suggested that spike synchrony is independent of behavioural/cortical functions/states. Both elevation and reduction in spike synchrony are observed during learning, but their net change is near 0 in prefrontal cortex (Baeg et al. 2007). Moreover, population functional connectivity or spike synchrony is stable even when long‐term potentiation is induced or brain state is changed in hippocampal CA3 (Yun et al. 2007; Mizuseki & Buzsáki, 2013). The spatial similarity of the firing patterns of entorhinal grid cells is uniform irrespective of spike synchrony (Buetfering et al. 2014).

Nevertheless, a few neuron pairs exhibited differential spike synchrony that depended upon motor functions/states. Even if only a few pairs exhibited such differential spike synchrony, it might be sufficient to accomplish behavioural functions. Indeed, our observation of the stable motor‐phase specificity of spike synchrony (Fig. 7 E) implies the possible functional importance of spike synchrony in only a few pairs. Although the fraction of neuron pairs with functional spike synchrony was small in recording sites and our task, it is possible that a significant amount of functional spike synchrony might occur in a different brain area in a different behavioural task.

Motor representation in rodent motor cortices

Our observations suggest that different modes of motor initiation, external or internal, are represented by spike rate (Baker & Lemon, 2000; London et al. 2010; but see Laubach et al. 2000) rather than spike synchrony, especially in putative excitatory neurons in RFA and putative inhibitory neurons in both CFA and RFA.

Here, however, the difference in spike rate between the two types of motor initiation was comparatively small, probably because our behavioural task was rather simple (Elsinger et al. 2006). It remains possible that motor initiation could also be represented by spike synchrony in rats performing more difficult behaviours with a cognitive demand or a complex movement, as observed in monkeys (Riehle et al. 1997; Hatsopoulos et al. 1998).

Different types of motor initiation are applicable to daily locomotion. For example, stair climbing is an externally initiated movement using visual cues, i.e. step lines and intervals, whereas level‐ground walking is an internally initiated movement using internal memory. In Parkinson's disease patients, level‐ground walking is more difficult than stair climbing (Morris et al. 1994; Rochester et al. 2011). This clinical observation suggests that different circuits are involved in these types of motor initiation. In primates, externally triggered movements are driven by signalling from the cerebellum and premotor area (PM) to M1, whereas internally triggered movements are driven by signalling from the basal ganglia (e.g. ventral tegmental area, VTA) and supplementary motor area (SMA) to M1 (Okano & Tanji, 1987; Romo & Schultz, 1992; Hosp et al. 2011, 2015). It remains unclear how cortical neurons encode different motor initiations in rodents.

Physiological functions of primary and secondary motor cortices

In contrast to M1, M2 has both motor and higher order functions (Ostlund et al. 2009; Smith et al. 2010; Brecht, 2011; Erlich et al. 2011; Sul et al. 2011; Murakami et al. 2014). As we observed, M2 neurons project to the spinal cord. They receive inputs from multiple sensory cortices, and have reciprocal connections with the prefrontal and parietal cortices in addition to CFA (Reep et al. 1990; Rouiller et al. 1993; Uylings et al. 2003). RFA and the rostral medial agranular cortex (AGm) in rat M2 may be homologous to PM and SMA in primates, respectively (Donoghue & Wise, 1982; Rouiller et al. 1993; Sul et al. 2011; Brown & Teskey, 2014). Indeed, our results showed that the spike rate in RFA is larger for externally triggered movements than internally triggered movements, as reported in primate PM (Okano & Tanji, 1987).

We defined RFA as the area in which ICMS evokes forelimb movements or EMG activities. Our muscimol experiments revealed that unilateral inactivation of only RFA or of both CFA and RFA degraded task performance, whereas inactivation of only CFA did not. This finding is partly consistent with a previous observation (Erlich et al. 2011; comparisons of the muscimol effects between the M1 and frontal orienting field, as a part of M2), but different from other reports (cf. Hira et al. 2013 a; Brown & Teskey, 2014; Peters et al. 2014; Otchy et al. 2015). The low susceptibility of behavioural performance to CFA inactivation may be due to incomplete inactivation of CFA. Alternatively, the physical load of lever manipulation might be too light to affect task performance; this is likely given the fact that EMG activities tended to be decreased by CFA inactivation (data not shown).

Interlaminar/intralaminar signalling in motor cortex

To initiate voluntary movements internally, frontal cortical areas may directly activate layer 5B of CFA, which sends projections to the spinal cord. To initiate them externally, sensory stimuli may activate superficial and middle layers (layers 2/3 and 5A) of CFA, and these may in turn also activate layer 5B (Hooks et al. 2013). In other words, CFA may drive differently initiated movements by switching these pathways to sequential (layer‐by‐layer) transmission of motor information (Sakata & Harris, 2009; Takeuchi et al. 2011). In fact, individual neurons across distinct layers of CFA are activated at various phases of movements (cooperative multilayer activation; Isomura et al. 2009); thus, motor information may be processed through parallel interlaminar transmissions.

Previous reports (Morishima & Kawaguchi, 2006; Morishima et al. 2011; Kiritani et al. 2012) suggested that spike synchrony may depend on different RS neuron subclasses with specific long‐distance targets. Because each subclass of RS neurons is distributed predominantly in a specific layer, it is possible that we did not simultaneously record multiple subclasses of RS neurons, potentially explaining our observation that spike synchrony among RS neurons was independent of rate‐based functions.

In the intralaminar network, the degree distribution of neurons that fired synchronously was long‐tailed, suggesting the existence of hub‐like neurons (Song et al. 2005; Ikegaya et al. 2013) or small‐world networks (Yu et al. 2008).

Circuit mechanism of spike rate and synchrony

Spike synchrony between two neurons can be elicited by direct excitatory synaptic input from one of them or a common input from a different neuron. Our experiments showed that spike synchrony that juxtacellularly stimulated RS neurons formed with the other multiple single unit‐recorded neurons was not diminished by artificial nanostimulation (Fig. 4), suggesting that it might result from a direct synaptic connection. In that case, the synaptic input should be large enough to evoke spikes. In fact, FS neurons exhibit excitatory postsynaptic potentials (EPSPs) of larger amplitudes than RS neurons, with small trial‐to‐trial fluctuations (Galarreta & Hestrin, 2001; Beierlein et al. 2003). Such large‐amplitude synaptic connections have been observed even in RS neurons (Song et al. 2005; Lefort et al. 2009; Morishima et al. 2011; Ikegaya et al. 2013; Mizuseki & Buzsáki, 2013). Moreover, several studies have shown that the impacts of single neurons can be strong (Miles & Wong, 1983; Brecht et al. 2004; Fujisawa et al. 2006; Houweling & Brecht, 2008; Li et al. 2009; cf. London et al. 2010). Synchronous inputs do not inevitably change the spike rate of the postsynaptic neuron. Our artificial nanostimulation changed the spike rate of stimulated neurons while preserving spike synchrony with the other non‐stimulated neurons (Fig. 4). These observations suggest that processing of spike rate code can be performed independently of spike synchrony. Population coding using spike rate might have more powerful information related to movement with robustness to noise (including both intrinsic noise, such as stochastic synaptic release, and extrinsic noise such as activity from other brain areas depending on the degree of arousal; London et al. 2010; Brette, 2015). Recently, artificial neural network models based on rate coding led to a breakthrough in the field of artificial intelligence (LeCun et al. 2015). Thus, rate coding may be sufficient to perform functional information processing.

However, spike synchrony does exist in cortical neurons. Our results and some previous studies (Baeg et al. 2007; Yun et al. 2007; Dupret et al. 2013; Mizuseki & Buzsáki, 2013, 2014) revealed functional changes in spike synchrony associated with large changes of behavioural states. Spikes are often correlated with network activity (Steriade et al. 1993; Léger et al. 2005; Kwan & Dan, 2012), which is so strong that the influence of network activity cannot be eliminated by artificial stimulation. Some studies revealed that functionally similar neurons are well organized (Komiyama et al. 2010; Yassin et al. 2010; Ko et al. 2011). Reliable spike synchrony might reflect an effective neural circuit, which might sometimes switch modes as a function of behavioural state. Large‐scale analyses of spike synchrony remain important for elucidation of unknown mechanisms of neural circuits involved in information processing.

Additional information

Competing interests

None declared.

Author contributions

R.K., Y.S. and Y.I. designed the research; R.K. performed the experiments; A.S. and Y.F.T. assisted with performing the experiments; R.K. and Y.S. analysed the data; and R.K., Y.S. and Y.I. interpreted the data and wrote the manuscript. All authors revised the manuscript. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was funded by JST CREST (Y.I.); Grants‐in‐Aid for Scientific Research on Innovative Areas (Y.I., JP 22120520, JP 24120715, JP 26112005; R.K., JP 26115521, ‘Memory dynamism’), for Scientific Research (B) (Y.I., JP 24300143), and for Young Scientists (B) (R.K., JP 24700399, JP 26830019) from MEXT; and research grants from the Research Foundation for Pharmaceutical Sciences (Group A) (R.K.) and the Narishige Neuroscience Research Foundation (R.K.).

Acknowledgements

We thank Profs K. Kitamura, M. Matsuzaki and Y. Yoshimura, and the members of CREST for helpful discussions; C. Chang, S. Ohno and C. Soai for technical assistance; and K. O'Hara (O'Hara & Co, Tokyo, Japan) and N. Kurihara (PRIMELAB, Tokyo, Japan) for developing the devices for behavioural experiments.

References

Aertsen AM & Gerstein GL (1985). Evaluation of neuronal connectivity: sensitivity of cross‐correlation. Brain Res 340, 341–354. [DOI] [PubMed] [Google Scholar]
Baeg EH, Kim YB, Kim J, Ghim JW, Kim JJ & Jung MW (2007). Learning‐induced enduring changes in functional connectivity among prefrontal cortical neurons. J Neurosci 27, 909–918. [DOI] [PMC free article] [PubMed] [Google Scholar]
Baker SN & Lemon RN (2000). Precise spatiotemporal repeating patterns in monkey primary and supplementary motor areas occur at chance levels. J Neurophysiol 84, 1770–1780. [DOI] [PubMed] [Google Scholar]
Barthó P, Hirase H, Monconduit L, Zugaro M, Harris KD & Buzsáki G (2004). Characterization of neocortical principal cells and interneurons by network interactions and extracellular features. J Neurophysiol 92, 600–608. [DOI] [PubMed] [Google Scholar]
Beierlein M, Gibson JR & Connors BW (2003). Two dynamically distinct inhibitory networks in layer 4 of the neocortex. J Neurophysiol 90, 2987–3000. [DOI] [PubMed] [Google Scholar]
Benjamini Y & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Statist Soc Ser B 57, 289–300. [Google Scholar]
Brácha V, Zhuravin IA & Bureš J (1990). The reaching reaction in the rat: a part of the digging pattern? Behav Brain Res 36, 53–64. [DOI] [PubMed] [Google Scholar]
Brecht M (2011). Movement, confusion, and orienting in frontal cortices. Neuron 72, 193–196. [DOI] [PubMed] [Google Scholar]
Brecht M, Schneider M, Sakmann B & Margrie TW (2004). Whisker movements evoked by stimulation of single pyramidal cells in rat motor cortex. Nature 427, 704–710. [DOI] [PubMed] [Google Scholar]
Brette R (2015). Philosophy of the spike: rate‐based vs. spike‐based theories of the brain. Front Syst Neurosci 9, 151. [DOI] [PMC free article] [PubMed] [Google Scholar]
Brown AR & Teskey GC (2014). Motor cortex is functionally organized as a set of spatially distinct representations for complex movements. J Neurosci 34, 13574–13585. [DOI] [PMC free article] [PubMed] [Google Scholar]
Buetfering C, Allen K & Monyer H (2014). Parvalbumin interneurons provide grid cell‐driven recurrent inhibition in the medial entorhinal cortex. Nat Neurosci 17, 710–718. [DOI] [PubMed] [Google Scholar]
Buzsáki G (2010). Neural syntax: cell assemblies, synapsembles, and readers. Neuron 68, 362–385. [DOI] [PMC free article] [PubMed] [Google Scholar]
Buzsáki G & Mizuseki K (2014). The log‐dynamic brain: how skewed distributions affect network operations. Nat Rev Neurosci 15, 264–278. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cohen MR & Newsome WT (2008). Context‐dependent changes in functional circuitry in visual area MT. Neuron 60, 162–173. [DOI] [PMC free article] [PubMed] [Google Scholar]
Constantinidis C, Williams GV & Goldman‐Rakic PS (2002). A role for inhibition in shaping the temporal flow of information in prefrontal cortex. Nat Neurosci 5, 175–180. [DOI] [PubMed] [Google Scholar]
Donoghue JP & Wise SP (1982). The motor cortex of the rat: cytoarchitecture and microstimulation mapping. J Comp Neurol 212, 76–88. [DOI] [PubMed] [Google Scholar]
Dupret D, O'Neill J & Csicsvari J (2013). Dynamic reconfiguration of hippocampal interneuron circuits during spatial learning. Neuron 78, 166–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
Elsinger CL, Harrington DL & Rao SM (2006). From preparation to online control: reappraisal of neural circuitry mediating internally generated and externally guided actions. Neuroimage 31, 1177–1187. [DOI] [PubMed] [Google Scholar]
Erlich JC, Bialek M & Brody CD (2011). A cortical substrate for memory‐guided orienting in the rat. Neuron 72, 330–343. [DOI] [PMC free article] [PubMed] [Google Scholar]
Evarts EV (1968). Relation of pyramidal tract activity to force exerted during voluntary movement. J Neurophysiol 31, 14–27. [DOI] [PubMed] [Google Scholar]
Fries P, Reynolds JH, Rorie AE & Desimone R (2001). Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291, 1560–1563. [DOI] [PubMed] [Google Scholar]
Fujisawa S, Amarasingham A, Harrison MT & Buzsáki G (2008). Behavior‐dependent short‐term assembly dynamics in the medial prefrontal cortex. Nat Neurosci 11, 823–833. [DOI] [PMC free article] [PubMed] [Google Scholar]
Fujisawa S, Matsuki N & Ikegaya Y (2006). Single neurons can induce phase transitions of cortical recurrent networks with multiple internal states. Cereb Cortex 16, 639–654. [DOI] [PubMed] [Google Scholar]
Galarreta M & Hestrin S (2001). Spike transmission and synchrony detection in networks of GABAergic interneurons. Science 292, 2295–2299. [DOI] [PubMed] [Google Scholar]
Georgopoulos AP, Schwartz AB & Kettner RE (1986). Neuronal population coding of movement direction. Science 233, 1416–1419. [DOI] [PubMed] [Google Scholar]
Gray CM, König P, Engel AK & Singer W (1989). Oscillatory responses in cat visual cortex exhibit inter‐columnar synchronization which reflects global stimulus properties. Nature 338, 334–337. [DOI] [PubMed] [Google Scholar]
Guo ZV, Li N, Huber D, Ophir E, Gutnisky D, Ting JT, Feng G & Svoboda K (2014). Flow of cortical activity underlying a tactile decision in mice. Neuron 81, 179–194. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harris KD, Henze DA, Csicsvari J, Hirase H & Buzsáki G (2000). Accuracy of tetrode spike separation as determined by simultaneous intracellular and extracellular measurements. J Neurophysiol 84, 401–414. [DOI] [PubMed] [Google Scholar]
Harris KD & Mrsic‐Flogel TD (2013). Cortical connectivity and sensory coding. Nature 503, 51–58. [DOI] [PubMed] [Google Scholar]
Hatsopoulos NG, Ojakangas CL, Paninski L & Donoghue JP (1998). Information about movement direction obtained from synchronous activity of motor cortical neurons. Proc Natl Acad Sci USA 95, 15706–15711. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hazan L, Zugaro M & Buzsáki G (2006). Klusters, NeuroScope, NDManager: a free software suite for neurophysiological data processing and visualization. J Neurosci Methods 155, 207–216. [DOI] [PubMed] [Google Scholar]
Hira R, Ohkubo F, Ozawa K, Isomura Y, Kitamura K, Kano M, Kasai H & Matsuzaki M (2013. a). Spatiotemporal dynamics of functional clusters of neurons in the mouse motor cortex during a voluntary movement. J Neurosci 33, 1377–1390. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hira R, Ohkubo F, Tanaka YR, Masamizu Y, Augustine GJ, Kasai H & Matsuzaki M (2013. b). In vivo optogenetic tracing of functional corticocortical connections between motor forelimb areas. Front Neural Circuits 7, 55. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hirabayashi T, Takeuchi D, Tamura K & Miyashita Y (2013. a). Microcircuits for hierarchical elaboration of object coding across primate temporal areas. Science 341, 191–195. [DOI] [PubMed] [Google Scholar]
Hirabayashi T, Takeuchi D, Tamura K & Miyashita Y (2013. b). Functional microcircuit recruited during retrieval of object association memory in monkey perirhinal cortex. Neuron 77, 192–203. [DOI] [PubMed] [Google Scholar]
Hooks BM, Mao T, Gutnisky DA, Yamawaki N, Svoboda K & Shepherd GM (2013). Organization of cortical and thalamic input to pyramidal neurons in mouse motor cortex. J Neurosci 33, 748–760. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hosp JA, Nolan HE & Luft AR (2015). Topography and collateralization of dopaminergic projections to primary motor cortex in rats. Exp Brain Res 233, 1365–1375. [DOI] [PubMed] [Google Scholar]
Hosp JA, Pekanovic A, Rioult‐Pedotti MS & Luft AR (2011). Dopaminergic projections from midbrain to primary motor cortex mediate motor skill learning. J Neurosci 31, 2481–2487. [DOI] [PMC free article] [PubMed] [Google Scholar]
Houweling AR & Brecht M (2008). Behavioural report of single neuron stimulation in somatosensory cortex. Nature 451, 65–68. [DOI] [PubMed] [Google Scholar]
Houweling AR, Doron G, Voigt BC, Herfst LJ & Brecht M (2010). Nanostimulation: manipulation of single neuron activity by juxtacellular current injection. J Neurophysiol 103, 1696–1704. [DOI] [PubMed] [Google Scholar]
Hyland B (1998). Neural activity related to reaching and grasping in rostral and caudal regions of rat motor cortex. Behav Brain Res 94, 255–269. [DOI] [PubMed] [Google Scholar]
Ikegaya Y, Sasaki T, Ishikawa D, Honma N, Tao K, Takahashi N, Minamisawa G, Ujita S & Matsuki N (2013). Interpyramid spike transmission stabilizes the sparseness of recurrent network activity. Cereb Cortex 23, 293–304. [DOI] [PubMed] [Google Scholar]
Isomura Y, Harukuni R, Takekawa T, Aizawa H & Fukai T (2009). Microcircuitry coordination of cortical motor information in self‐initiation of voluntary movements. Nat Neurosci 12, 1586–1593. [DOI] [PubMed] [Google Scholar]
Isomura Y, Takekawa T, Harukuni R, Handa T, Aizawa H, Takada M & Fukai T (2013). Reward‐modulated motor information in identified striatum neurons. J Neurosci 33, 10209–10220. [DOI] [PMC free article] [PubMed] [Google Scholar]
Iwaniuk AN & Whishaw IQ (2000). On the origin of skilled forelimb movements. Trends Neurosci 23, 372–376. [DOI] [PubMed] [Google Scholar]
Kimura R, Saiki A, Fujiwara‐Tsukamoto Y, Ohkubo F, Kitamura K, Matsuzaki M, Sakai Y & Isomura Y (2012). Reinforcing operandum: rapid and reliable learning of skilled forelimb movements by head‐fixed rodents. J Neurophysiol 108, 1781–1792. [DOI] [PubMed] [Google Scholar]
Kiritani T, Wickersham IR, Seung HS & Shepherd GM (2012). Hierarchical connectivity and connection‐specific dynamics in the corticospinal‐corticostriatal microcircuit in mouse motor cortex. J Neurosci 32, 4992–5001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ko H, Hofer SB, Pichler B, Buchanan KA, Sjöström PJ & Mrsic‐Flogel TD (2011). Functional specificity of local synaptic connections in neocortical networks. Nature 473, 87–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
Komiyama T, Sato TR, O'Connor DH, Zhang YX, Huber D, Hooks BM, Gabitto M & Svoboda K (2010). Learning‐related fine‐scale specificity imaged in motor cortex circuits of behaving mice. Nature 464, 1182–1186. [DOI] [PubMed] [Google Scholar]
Kwan AC & Dan Y (2012). Dissection of cortical microcircuits by single‐neuron stimulation in vivo. Curr Biol 22, 1459–1467. [DOI] [PMC free article] [PubMed] [Google Scholar]
LeCun Y, Bengio Y & Hinton G (2015). Deep learning. Nature 521, 436–444. [DOI] [PubMed] [Google Scholar]
Laubach M, Wessberg J & Nicolelis MA (2000). Cortical ensemble activity increasingly predicts behaviour outcomes during learning of a motor task. Nature 405, 567–571. [DOI] [PubMed] [Google Scholar]
Lefort S, Tomm C, Floyd Sarria JC & Petersen CC (2009). The excitatory neuronal network of the C2 barrel column in mouse primary somatosensory cortex. Neuron 61, 301–316. [DOI] [PubMed] [Google Scholar]
Léger JF, Stern EA, Aertsen A & Heck D (2005). Synaptic integration in rat frontal cortex shaped by network activity. J Neurophysiol 93, 281–293. [DOI] [PubMed] [Google Scholar]
Li CY, Poo MM & Dan Y (2009). Burst spiking of a single cortical neuron modifies global brain state. Science 324, 643–646. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li N, Chen TW, Guo ZV, Gerfen CR & Svoboda K (2015). A motor cortex circuit for motor planning and movement. Nature 519, 51–56. [DOI] [PubMed] [Google Scholar]
Liang F, Rouiller EM & Wiesendanger M (1993). Modulation of sustained electromyographic activity by single intracortical microstimuli: comparison of two forelimb motor cortical areas of the rat. Somatosens Mot Res 10, 51–61. [DOI] [PubMed] [Google Scholar]
London M, Roth A, Beeren L, Häusser M & Latham PE (2010). Sensitivity to perturbations in vivo implies high noise and suggests rate coding in cortex. Nature 466, 123–127. [DOI] [PMC free article] [PubMed] [Google Scholar]
Miles R & Wong RK (1983). Single neurones can initiate synchronized population discharge in the hippocampus. Nature 306, 371–373. [DOI] [PubMed] [Google Scholar]
Mitchell JF, Sundberg KA & Reynolds JH (2009). Spatial attention decorrelates intrinsic activity fluctuations in macaque area V4. Neuron 63, 879–888. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mizuseki K & Buzsáki G (2013). Preconfigured, skewed distribution of firing rates in the hippocampus and entorhinal cortex. Cell Rep 4, 1010–1021. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mizuseki K & Buzsáki G (2014). Theta oscillations decrease spike synchrony in the hippocampus and entorhinal cortex. Philos Trans R Soc Lond B Biol Sci 369, 20120530. [DOI] [PMC free article] [PubMed] [Google Scholar]
Morishima M & Kawaguchi Y (2006). Recurrent connection patterns of corticostriatal pyramidal cells in frontal cortex. J Neurosci 26, 4394–4405. [DOI] [PMC free article] [PubMed] [Google Scholar]
Morishima M, Morita K, Kubota Y & Kawaguchi Y (2011). Highly differentiated projection‐specific cortical subnetworks. J Neurosci 31, 10380–10391. [DOI] [PMC free article] [PubMed] [Google Scholar]
Morris ME, Iansek R, Matyas TA & Summers JJ (1994). The pathogenesis of gait hypokinesia in Parkinson's disease. Brain 117, 1169–1181. [DOI] [PubMed] [Google Scholar]
Murakami M, Vicente MI, Costa GM & Mainen ZF (2014). Neural antecedents of self‐initiated actions in secondary motor cortex. Nat Neurosci 17, 1574–1582. [DOI] [PubMed] [Google Scholar]
Narayanan NS & Laubach M (2006). Top‐down control of motor cortex ensembles by dorsomedial prefrontal cortex. Neuron 52, 921–931. [DOI] [PMC free article] [PubMed] [Google Scholar]
Neafsey EJ, Bold EL, Haas G, Hurley‐Gius KM, Quirk G, Sievert CF & Terreberry RR (1986). The organization of the rat motor cortex: a microstimulation mapping study. Brain Res Rev 396, 77–96. [DOI] [PubMed] [Google Scholar]
Okano K & Tanji J (1987). Neuronal activities in the primate motor fields of the agranular frontal cortex preceding visually triggered and self‐paced movement. Exp Brain Res 66, 155–166. [DOI] [PubMed] [Google Scholar]
Ostlund SB, Winterbauer NE & Balleine BW (2009). Evidence of action sequence chunking in goal‐directed instrumental conditioning and its dependence on the dorsomedial prefrontal cortex. J Neurosci 29, 8280–8287. [DOI] [PMC free article] [PubMed] [Google Scholar]
Otchy TM, Wolff SB, Rhee JY, Pehlevan C, Kawai R, Kempf A, Gobes SM & Ölveczky BP (2015). Acute off‐target effects of neural circuit manipulations. Nature 528, 358–363. [DOI] [PubMed] [Google Scholar]
Perkel DH, Gerstein GL & Moore GP (1967). Neuronal spike trains and stochastic point processes. II. Simultaneous spike trains. Biophys J 7, 419–440. [DOI] [PMC free article] [PubMed] [Google Scholar]
Peters AJ, Chen SX & Komiyama T (2014). Emergence of reproducible spatiotemporal activity during motor learning. Nature 510, 263–267. [DOI] [PubMed] [Google Scholar]
Putrino D, Brown EN, Mastaglia FL & Ghosh S (2010). Differential involvement of excitatory and inhibitory neurons of cat motor cortex in coincident spike activity related to behavioral context. J Neurosci 30, 8048–8056. [DOI] [PMC free article] [PubMed] [Google Scholar]
Reep RL, Goodwin GS & Corwin JV (1990). Topographic organization in the corticocortical connections of medial agranular cortex in rats. J Comp Neurol 294, 262–280. [DOI] [PubMed] [Google Scholar]
Riehle A, Grün S, Diesmann M & Aertsen A (1997). Spike synchronization and rate modulation differentially involved in motor cortical function. Science 278, 1950–1953. [DOI] [PubMed] [Google Scholar]
Rochester L, Baker K, Nieuwboer A & Burn D (2011). Targeting dopa‐sensitive and dopa‐resistant gait dysfunction in Parkinson's disease: selective responses to internal and external cues. Mov Disord 26, 430–435. [DOI] [PubMed] [Google Scholar]
Romo R & Schultz W (1992). Role of primate basal ganglia and frontal cortex in the internal generation of movements. III. Neuronal activity in the supplementary motor area. Exp Brain Res 91, 396–407. [DOI] [PubMed] [Google Scholar]
Rouiller EM, Moret V & Liang F (1993). Comparison of the connectional properties of the two forelimb areas of the rat sensorimotor cortex: support for the presence of a premotor or supplementary motor cortical area. Somatosens Mot Res 10, 269–289. [DOI] [PubMed] [Google Scholar]
Saiki A, Kimura R, Samura T, Fujiwara‐Tsukamoto Y, Sakai Y & Isomura Y (2014). Different modulation of common motor information in rat primary and secondary motor cortices. PLoS One 9, e98662. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sakata S & Harris KD (2009). Laminar structure of spontaneous and sensory‐evoked population activity in auditory cortex. Neuron 64, 404–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sakurai Y (1996). Hippocampal and neocortical cell assemblies encode memory processes for different types of stimuli in the rat. J Neurosci 16, 2809–2819. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schwarz C, Hentschke H, Butovas S, Haiss F, Stüttgen MC, Gerdjikov TV, Bergner CG & Waiblinger C (2010). The head‐fixed behaving rat—Procedures and pitfalls. Somatosens Mot Res 27, 131–148. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sears TA & Stagg D (1976). Short‐term synchronization of intercostal motoneurone activity. J Physiol 263, 357–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
Smith NJ, Horst NK, Liu B, Caetano MS & Laubach M (2010). Reversible inactivation of rat premotor cortex impairs temporal preparation, but not inhibitory control, during simple reaction‐time performance. Front Integr Neurosci 4, 124. [DOI] [PMC free article] [PubMed] [Google Scholar]
Song S, Sjöström PJ, Reigl M, Nelson S & Chklovskii DB (2005). Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS Biol 3, e68. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stark E & Abeles M (2009). Unbiased estimation of precise temporal correlations between spike trains. J Neurosci Methods 179, 90–100. [DOI] [PubMed] [Google Scholar]
Steinmetz PN, Roy A, Fitzgerald PJ, Hsiao SS, Johnson KO & Niebur E (2000). Attention modulates synchronized neuronal firing in primate somatosensory cortex. Nature 404, 187–190. [DOI] [PubMed] [Google Scholar]
Steriade M, Contreras D, Curró DR & Nuñez A (1993). The slow (< 1 Hz) oscillation in reticular thalamic and thalamocortical neurons: scenario of sleep rhythm generation in interacting thalamic and neocortical networks. J Neurosci 13, 3284–3299. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sul JH, Jo S, Lee D & Jung MW (2011). Role of rodent secondary motor cortex in value‐based action selection. Nat Neurosci 14, 1202–1208. [DOI] [PMC free article] [PubMed] [Google Scholar]
Takekawa T, Isomura Y & Fukai T (2010). Accurate spike sorting for multi‐unit recordings. Eur J Neurosci 31, 263–272. [DOI] [PubMed] [Google Scholar]
Takekawa T, Isomura Y & Fukai T (2012). Spike sorting of heterogeneous neuron types by multimodality‐weighted PCA and explicit robust variational Bayes. Front Neuroinform 6, 5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Takeuchi D, Hirabayashi T, Tamura K & Miyashita Y (2011). Reversal of interlaminar signal between sensory and memory processing in monkey temporal cortex. Science 331, 1443–1447. [DOI] [PubMed] [Google Scholar]
Toyama K, Kimura M & Tanaka K (1981). Cross‐correlation analysis of interneuronal connectivity in cat visual cortex. J Neurophysiol 46, 191–201. [DOI] [PubMed] [Google Scholar]
Uylings HB, Groenewegen HJ & Kolb B (2003). Do rats have a prefrontal cortex? Behav Brain Res 146, 3–17. [DOI] [PubMed] [Google Scholar]
Vaadia E, Haalman I, Abeles M, Bergman H, Prut Y, Slovin H & Aertsen A (1995). Dynamics of neuronal interactions in monkey cortex in relation to behavioural events. Nature 373, 515–518. [DOI] [PubMed] [Google Scholar]
Yassin L, Benedetti BL, Jouhanneau JS, Wen JA, Poulet JF & Barth AL (2010). An embedded subnetwork of highly active neurons in the neocortex. Neuron 68, 1043–1050. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yu S, Huang D, Singer W & Nikolic D (2008). A small world of neuronal synchrony. Cereb Cortex 18, 2891–2901. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yun SH, Lee DS, Lee H, Baeg EH, Kim YB & Jung MW (2007). LTP induction modifies functional relationship among hippocampal neurons. Learn Mem 14, 190–194. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0001] Aertsen AM & Gerstein GL (1985). Evaluation of neuronal connectivity: sensitivity of cross‐correlation. Brain Res 340, 341–354. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0002] Baeg EH, Kim YB, Kim J, Ghim JW, Kim JJ & Jung MW (2007). Learning‐induced enduring changes in functional connectivity among prefrontal cortical neurons. J Neurosci 27, 909–918. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0003] Baker SN & Lemon RN (2000). Precise spatiotemporal repeating patterns in monkey primary and supplementary motor areas occur at chance levels. J Neurophysiol 84, 1770–1780. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0004] Barthó P, Hirase H, Monconduit L, Zugaro M, Harris KD & Buzsáki G (2004). Characterization of neocortical principal cells and interneurons by network interactions and extracellular features. J Neurophysiol 92, 600–608. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0005] Beierlein M, Gibson JR & Connors BW (2003). Two dynamically distinct inhibitory networks in layer 4 of the neocortex. J Neurophysiol 90, 2987–3000. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0006] Benjamini Y & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Statist Soc Ser B 57, 289–300. [Google Scholar]

[tjp7448-bib-0007] Brácha V, Zhuravin IA & Bureš J (1990). The reaching reaction in the rat: a part of the digging pattern? Behav Brain Res 36, 53–64. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0008] Brecht M (2011). Movement, confusion, and orienting in frontal cortices. Neuron 72, 193–196. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0009] Brecht M, Schneider M, Sakmann B & Margrie TW (2004). Whisker movements evoked by stimulation of single pyramidal cells in rat motor cortex. Nature 427, 704–710. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0010] Brette R (2015). Philosophy of the spike: rate‐based vs. spike‐based theories of the brain. Front Syst Neurosci 9, 151. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0011] Brown AR & Teskey GC (2014). Motor cortex is functionally organized as a set of spatially distinct representations for complex movements. J Neurosci 34, 13574–13585. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0012] Buetfering C, Allen K & Monyer H (2014). Parvalbumin interneurons provide grid cell‐driven recurrent inhibition in the medial entorhinal cortex. Nat Neurosci 17, 710–718. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0013] Buzsáki G (2010). Neural syntax: cell assemblies, synapsembles, and readers. Neuron 68, 362–385. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0014] Buzsáki G & Mizuseki K (2014). The log‐dynamic brain: how skewed distributions affect network operations. Nat Rev Neurosci 15, 264–278. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0015] Cohen MR & Newsome WT (2008). Context‐dependent changes in functional circuitry in visual area MT. Neuron 60, 162–173. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0016] Constantinidis C, Williams GV & Goldman‐Rakic PS (2002). A role for inhibition in shaping the temporal flow of information in prefrontal cortex. Nat Neurosci 5, 175–180. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0017] Donoghue JP & Wise SP (1982). The motor cortex of the rat: cytoarchitecture and microstimulation mapping. J Comp Neurol 212, 76–88. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0018] Dupret D, O'Neill J & Csicsvari J (2013). Dynamic reconfiguration of hippocampal interneuron circuits during spatial learning. Neuron 78, 166–180. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0019] Elsinger CL, Harrington DL & Rao SM (2006). From preparation to online control: reappraisal of neural circuitry mediating internally generated and externally guided actions. Neuroimage 31, 1177–1187. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0020] Erlich JC, Bialek M & Brody CD (2011). A cortical substrate for memory‐guided orienting in the rat. Neuron 72, 330–343. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0021] Evarts EV (1968). Relation of pyramidal tract activity to force exerted during voluntary movement. J Neurophysiol 31, 14–27. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0022] Fries P, Reynolds JH, Rorie AE & Desimone R (2001). Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291, 1560–1563. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0023] Fujisawa S, Amarasingham A, Harrison MT & Buzsáki G (2008). Behavior‐dependent short‐term assembly dynamics in the medial prefrontal cortex. Nat Neurosci 11, 823–833. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0024] Fujisawa S, Matsuki N & Ikegaya Y (2006). Single neurons can induce phase transitions of cortical recurrent networks with multiple internal states. Cereb Cortex 16, 639–654. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0025] Galarreta M & Hestrin S (2001). Spike transmission and synchrony detection in networks of GABAergic interneurons. Science 292, 2295–2299. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0026] Georgopoulos AP, Schwartz AB & Kettner RE (1986). Neuronal population coding of movement direction. Science 233, 1416–1419. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0027] Gray CM, König P, Engel AK & Singer W (1989). Oscillatory responses in cat visual cortex exhibit inter‐columnar synchronization which reflects global stimulus properties. Nature 338, 334–337. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0028] Guo ZV, Li N, Huber D, Ophir E, Gutnisky D, Ting JT, Feng G & Svoboda K (2014). Flow of cortical activity underlying a tactile decision in mice. Neuron 81, 179–194. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0029] Harris KD, Henze DA, Csicsvari J, Hirase H & Buzsáki G (2000). Accuracy of tetrode spike separation as determined by simultaneous intracellular and extracellular measurements. J Neurophysiol 84, 401–414. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0030] Harris KD & Mrsic‐Flogel TD (2013). Cortical connectivity and sensory coding. Nature 503, 51–58. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0031] Hatsopoulos NG, Ojakangas CL, Paninski L & Donoghue JP (1998). Information about movement direction obtained from synchronous activity of motor cortical neurons. Proc Natl Acad Sci USA 95, 15706–15711. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0032] Hazan L, Zugaro M & Buzsáki G (2006). Klusters, NeuroScope, NDManager: a free software suite for neurophysiological data processing and visualization. J Neurosci Methods 155, 207–216. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0033] Hira R, Ohkubo F, Ozawa K, Isomura Y, Kitamura K, Kano M, Kasai H & Matsuzaki M (2013. a). Spatiotemporal dynamics of functional clusters of neurons in the mouse motor cortex during a voluntary movement. J Neurosci 33, 1377–1390. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0034] Hira R, Ohkubo F, Tanaka YR, Masamizu Y, Augustine GJ, Kasai H & Matsuzaki M (2013. b). In vivo optogenetic tracing of functional corticocortical connections between motor forelimb areas. Front Neural Circuits 7, 55. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0035] Hirabayashi T, Takeuchi D, Tamura K & Miyashita Y (2013. a). Microcircuits for hierarchical elaboration of object coding across primate temporal areas. Science 341, 191–195. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0036] Hirabayashi T, Takeuchi D, Tamura K & Miyashita Y (2013. b). Functional microcircuit recruited during retrieval of object association memory in monkey perirhinal cortex. Neuron 77, 192–203. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0037] Hooks BM, Mao T, Gutnisky DA, Yamawaki N, Svoboda K & Shepherd GM (2013). Organization of cortical and thalamic input to pyramidal neurons in mouse motor cortex. J Neurosci 33, 748–760. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0038] Hosp JA, Nolan HE & Luft AR (2015). Topography and collateralization of dopaminergic projections to primary motor cortex in rats. Exp Brain Res 233, 1365–1375. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0039] Hosp JA, Pekanovic A, Rioult‐Pedotti MS & Luft AR (2011). Dopaminergic projections from midbrain to primary motor cortex mediate motor skill learning. J Neurosci 31, 2481–2487. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0040] Houweling AR & Brecht M (2008). Behavioural report of single neuron stimulation in somatosensory cortex. Nature 451, 65–68. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0041] Houweling AR, Doron G, Voigt BC, Herfst LJ & Brecht M (2010). Nanostimulation: manipulation of single neuron activity by juxtacellular current injection. J Neurophysiol 103, 1696–1704. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0042] Hyland B (1998). Neural activity related to reaching and grasping in rostral and caudal regions of rat motor cortex. Behav Brain Res 94, 255–269. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0043] Ikegaya Y, Sasaki T, Ishikawa D, Honma N, Tao K, Takahashi N, Minamisawa G, Ujita S & Matsuki N (2013). Interpyramid spike transmission stabilizes the sparseness of recurrent network activity. Cereb Cortex 23, 293–304. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0044] Isomura Y, Harukuni R, Takekawa T, Aizawa H & Fukai T (2009). Microcircuitry coordination of cortical motor information in self‐initiation of voluntary movements. Nat Neurosci 12, 1586–1593. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0045] Isomura Y, Takekawa T, Harukuni R, Handa T, Aizawa H, Takada M & Fukai T (2013). Reward‐modulated motor information in identified striatum neurons. J Neurosci 33, 10209–10220. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0046] Iwaniuk AN & Whishaw IQ (2000). On the origin of skilled forelimb movements. Trends Neurosci 23, 372–376. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0047] Kimura R, Saiki A, Fujiwara‐Tsukamoto Y, Ohkubo F, Kitamura K, Matsuzaki M, Sakai Y & Isomura Y (2012). Reinforcing operandum: rapid and reliable learning of skilled forelimb movements by head‐fixed rodents. J Neurophysiol 108, 1781–1792. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0048] Kiritani T, Wickersham IR, Seung HS & Shepherd GM (2012). Hierarchical connectivity and connection‐specific dynamics in the corticospinal‐corticostriatal microcircuit in mouse motor cortex. J Neurosci 32, 4992–5001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0049] Ko H, Hofer SB, Pichler B, Buchanan KA, Sjöström PJ & Mrsic‐Flogel TD (2011). Functional specificity of local synaptic connections in neocortical networks. Nature 473, 87–91. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0050] Komiyama T, Sato TR, O'Connor DH, Zhang YX, Huber D, Hooks BM, Gabitto M & Svoboda K (2010). Learning‐related fine‐scale specificity imaged in motor cortex circuits of behaving mice. Nature 464, 1182–1186. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0051] Kwan AC & Dan Y (2012). Dissection of cortical microcircuits by single‐neuron stimulation in vivo. Curr Biol 22, 1459–1467. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0052] LeCun Y, Bengio Y & Hinton G (2015). Deep learning. Nature 521, 436–444. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0053] Laubach M, Wessberg J & Nicolelis MA (2000). Cortical ensemble activity increasingly predicts behaviour outcomes during learning of a motor task. Nature 405, 567–571. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0054] Lefort S, Tomm C, Floyd Sarria JC & Petersen CC (2009). The excitatory neuronal network of the C2 barrel column in mouse primary somatosensory cortex. Neuron 61, 301–316. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0055] Léger JF, Stern EA, Aertsen A & Heck D (2005). Synaptic integration in rat frontal cortex shaped by network activity. J Neurophysiol 93, 281–293. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0056] Li CY, Poo MM & Dan Y (2009). Burst spiking of a single cortical neuron modifies global brain state. Science 324, 643–646. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0057] Li N, Chen TW, Guo ZV, Gerfen CR & Svoboda K (2015). A motor cortex circuit for motor planning and movement. Nature 519, 51–56. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0058] Liang F, Rouiller EM & Wiesendanger M (1993). Modulation of sustained electromyographic activity by single intracortical microstimuli: comparison of two forelimb motor cortical areas of the rat. Somatosens Mot Res 10, 51–61. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0059] London M, Roth A, Beeren L, Häusser M & Latham PE (2010). Sensitivity to perturbations in vivo implies high noise and suggests rate coding in cortex. Nature 466, 123–127. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0060] Miles R & Wong RK (1983). Single neurones can initiate synchronized population discharge in the hippocampus. Nature 306, 371–373. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0061] Mitchell JF, Sundberg KA & Reynolds JH (2009). Spatial attention decorrelates intrinsic activity fluctuations in macaque area V4. Neuron 63, 879–888. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0062] Mizuseki K & Buzsáki G (2013). Preconfigured, skewed distribution of firing rates in the hippocampus and entorhinal cortex. Cell Rep 4, 1010–1021. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0063] Mizuseki K & Buzsáki G (2014). Theta oscillations decrease spike synchrony in the hippocampus and entorhinal cortex. Philos Trans R Soc Lond B Biol Sci 369, 20120530. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0064] Morishima M & Kawaguchi Y (2006). Recurrent connection patterns of corticostriatal pyramidal cells in frontal cortex. J Neurosci 26, 4394–4405. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0065] Morishima M, Morita K, Kubota Y & Kawaguchi Y (2011). Highly differentiated projection‐specific cortical subnetworks. J Neurosci 31, 10380–10391. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0066] Morris ME, Iansek R, Matyas TA & Summers JJ (1994). The pathogenesis of gait hypokinesia in Parkinson's disease. Brain 117, 1169–1181. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0067] Murakami M, Vicente MI, Costa GM & Mainen ZF (2014). Neural antecedents of self‐initiated actions in secondary motor cortex. Nat Neurosci 17, 1574–1582. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0068] Narayanan NS & Laubach M (2006). Top‐down control of motor cortex ensembles by dorsomedial prefrontal cortex. Neuron 52, 921–931. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0069] Neafsey EJ, Bold EL, Haas G, Hurley‐Gius KM, Quirk G, Sievert CF & Terreberry RR (1986). The organization of the rat motor cortex: a microstimulation mapping study. Brain Res Rev 396, 77–96. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0070] Okano K & Tanji J (1987). Neuronal activities in the primate motor fields of the agranular frontal cortex preceding visually triggered and self‐paced movement. Exp Brain Res 66, 155–166. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0071] Ostlund SB, Winterbauer NE & Balleine BW (2009). Evidence of action sequence chunking in goal‐directed instrumental conditioning and its dependence on the dorsomedial prefrontal cortex. J Neurosci 29, 8280–8287. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0072] Otchy TM, Wolff SB, Rhee JY, Pehlevan C, Kawai R, Kempf A, Gobes SM & Ölveczky BP (2015). Acute off‐target effects of neural circuit manipulations. Nature 528, 358–363. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0073] Perkel DH, Gerstein GL & Moore GP (1967). Neuronal spike trains and stochastic point processes. II. Simultaneous spike trains. Biophys J 7, 419–440. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0074] Peters AJ, Chen SX & Komiyama T (2014). Emergence of reproducible spatiotemporal activity during motor learning. Nature 510, 263–267. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0075] Putrino D, Brown EN, Mastaglia FL & Ghosh S (2010). Differential involvement of excitatory and inhibitory neurons of cat motor cortex in coincident spike activity related to behavioral context. J Neurosci 30, 8048–8056. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0076] Reep RL, Goodwin GS & Corwin JV (1990). Topographic organization in the corticocortical connections of medial agranular cortex in rats. J Comp Neurol 294, 262–280. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0077] Riehle A, Grün S, Diesmann M & Aertsen A (1997). Spike synchronization and rate modulation differentially involved in motor cortical function. Science 278, 1950–1953. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0078] Rochester L, Baker K, Nieuwboer A & Burn D (2011). Targeting dopa‐sensitive and dopa‐resistant gait dysfunction in Parkinson's disease: selective responses to internal and external cues. Mov Disord 26, 430–435. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0079] Romo R & Schultz W (1992). Role of primate basal ganglia and frontal cortex in the internal generation of movements. III. Neuronal activity in the supplementary motor area. Exp Brain Res 91, 396–407. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0080] Rouiller EM, Moret V & Liang F (1993). Comparison of the connectional properties of the two forelimb areas of the rat sensorimotor cortex: support for the presence of a premotor or supplementary motor cortical area. Somatosens Mot Res 10, 269–289. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0081] Saiki A, Kimura R, Samura T, Fujiwara‐Tsukamoto Y, Sakai Y & Isomura Y (2014). Different modulation of common motor information in rat primary and secondary motor cortices. PLoS One 9, e98662. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0082] Sakata S & Harris KD (2009). Laminar structure of spontaneous and sensory‐evoked population activity in auditory cortex. Neuron 64, 404–418. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0083] Sakurai Y (1996). Hippocampal and neocortical cell assemblies encode memory processes for different types of stimuli in the rat. J Neurosci 16, 2809–2819. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0084] Schwarz C, Hentschke H, Butovas S, Haiss F, Stüttgen MC, Gerdjikov TV, Bergner CG & Waiblinger C (2010). The head‐fixed behaving rat—Procedures and pitfalls. Somatosens Mot Res 27, 131–148. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0085] Sears TA & Stagg D (1976). Short‐term synchronization of intercostal motoneurone activity. J Physiol 263, 357–381. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0086] Smith NJ, Horst NK, Liu B, Caetano MS & Laubach M (2010). Reversible inactivation of rat premotor cortex impairs temporal preparation, but not inhibitory control, during simple reaction‐time performance. Front Integr Neurosci 4, 124. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0087] Song S, Sjöström PJ, Reigl M, Nelson S & Chklovskii DB (2005). Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS Biol 3, e68. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0088] Stark E & Abeles M (2009). Unbiased estimation of precise temporal correlations between spike trains. J Neurosci Methods 179, 90–100. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0089] Steinmetz PN, Roy A, Fitzgerald PJ, Hsiao SS, Johnson KO & Niebur E (2000). Attention modulates synchronized neuronal firing in primate somatosensory cortex. Nature 404, 187–190. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0090] Steriade M, Contreras D, Curró DR & Nuñez A (1993). The slow (< 1 Hz) oscillation in reticular thalamic and thalamocortical neurons: scenario of sleep rhythm generation in interacting thalamic and neocortical networks. J Neurosci 13, 3284–3299. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0091] Sul JH, Jo S, Lee D & Jung MW (2011). Role of rodent secondary motor cortex in value‐based action selection. Nat Neurosci 14, 1202–1208. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0092] Takekawa T, Isomura Y & Fukai T (2010). Accurate spike sorting for multi‐unit recordings. Eur J Neurosci 31, 263–272. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0093] Takekawa T, Isomura Y & Fukai T (2012). Spike sorting of heterogeneous neuron types by multimodality‐weighted PCA and explicit robust variational Bayes. Front Neuroinform 6, 5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0094] Takeuchi D, Hirabayashi T, Tamura K & Miyashita Y (2011). Reversal of interlaminar signal between sensory and memory processing in monkey temporal cortex. Science 331, 1443–1447. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0095] Toyama K, Kimura M & Tanaka K (1981). Cross‐correlation analysis of interneuronal connectivity in cat visual cortex. J Neurophysiol 46, 191–201. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0096] Uylings HB, Groenewegen HJ & Kolb B (2003). Do rats have a prefrontal cortex? Behav Brain Res 146, 3–17. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0097] Vaadia E, Haalman I, Abeles M, Bergman H, Prut Y, Slovin H & Aertsen A (1995). Dynamics of neuronal interactions in monkey cortex in relation to behavioural events. Nature 373, 515–518. [DOI] [PubMed] [Google Scholar]

[tjp7448-bib-0098] Yassin L, Benedetti BL, Jouhanneau JS, Wen JA, Poulet JF & Barth AL (2010). An embedded subnetwork of highly active neurons in the neocortex. Neuron 68, 1043–1050. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0099] Yu S, Huang D, Singer W & Nikolic D (2008). A small world of neuronal synchrony. Cereb Cortex 18, 2891–2901. [DOI] [PMC free article] [PubMed] [Google Scholar]

[tjp7448-bib-0100] Yun SH, Lee DS, Lee H, Baeg EH, Kim YB & Jung MW (2007). LTP induction modifies functional relationship among hippocampal neurons. Learn Mem 14, 190–194. [DOI] [PubMed] [Google Scholar]

PERMALINK

Large‐scale analysis reveals populational contributions of cortical spike rate and synchrony to behavioural functions

Rie Kimura

Akiko Saiki

Yoko Fujiwara‐Tsukamoto

Yutaka Sakai

Yoshikazu Isomura

Abstract

Key points

Abstract

Key points

Abbreviations

Introduction

Methods

Ethical approval

Animal preparation

Behavioural tasks

Electrophysiological recording

Pharmacological inactivation

Histological staining

Data analysis

Movement–behaviour time extraction from lever trajectories

Figure 1. Behavioural tasks.

Spike extraction and refinement

Classification of neurons

Table 1.

Selectivity to external‐/internal‐trigger movements

Evaluation of spike synchrony

Statistical reliability of spike synchrony: P‐value

Strength of spike synchrony: H‐value

Evaluation of functional change of spike synchrony

Functional change in strength of spike synchrony

Functional change in timing of spike synchrony

Information carried by a pair spike pattern

Effect of juxtacellular nanostimulation on spike synchrony

Figure 4. Robustness of our spike synchrony analysis against artificial alteration in spike rate.

Network characteristics of significant spike synchrony

Statistics

Table 2.

Table 3.

Results

Contribution of CFA and RFA to behavioural task performance

Functional spike rate in motor cortices during external‐ and internal‐trigger movements

Figure 2. Spike activities depending on type of motor initiation in CFA and RFA.

Significant emergence of spike synchrony in motor cortices

Figure 3. Reliability and strength of spike synchrony in combinations of neuron subtypes.

Validity for the evaluation methods of spike synchrony

Figure 5. Network analysis of neuron pairs with spike synchrony.

Independence of temporal changes in spike synchrony and movement state

Figure 6. Dependence of spike synchrony on behavioural state.

Figure 7. Other evaluations of functional changes in spike synchrony.

Figure 8. Confirmation in efficiently task‐discriminating rats.

Figure 9. Temporal change of spike synchrony and behavioural state in neuronal pairs with significant spike synchrony.

Independence between spike synchrony and rate‐based neuronal functions

Figure 10. Dependence of spike synchrony on rate‐based neuronal preferences.

Discussion

Spike synchrony independent of rate‐based behavioural functions

Motor representation in rodent motor cortices

Physiological functions of primary and secondary motor cortices

Interlaminar/intralaminar signalling in motor cortex

Circuit mechanism of spike rate and synchrony

Additional information

Competing interests

Author contributions

Funding

Acknowledgements

References

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases