Are there neuronal firing artifacts produced by head movement?

I'm experimenting with a consumer-grade ElectroEncephaloGram (EEG) sensor and have created the image below using the device. Because the sensor on the device does not use a suction cup, there are a lot of motion artifacts when the headband contact is poor (during motion).

This got me thinking - neurons have mass and thus inertia, and they are not glued to each other. When the person moves (actually rotates) the head around, forces of inertia are applied to neurons/axons. I'm interested if such motion produces any kind of firing artifacts within the brain or how the brain filters out such firing?

This reminded me of this article: (5-HT and motor control: a hypothesis). Could it be that some neurons stop firing in response to movement?

There are certainly head-orientation cells (e.g. in the hippocampus). But neurons are reasonably immune to the kind of mild physical stresses that come from turning the head around; computing head orientation requires complex analysis of input from e.g. the visual system (optic flow) and vestibular system.

However, the electrical activity of muscles tends to swamp that of (nearby) neurons, so various muscle-related artifacts are often visible in an EEG if not carefully filtered out. And, of course, if the contact is poor you'll get artifacts from that: you're just measuring the variability in resistance between the sensor and your skin, not anything interesting about what faint potential changes are visible at your skin as a result of neuronal activity.

Are there neuronal firing artifacts produced by head movement? - Biology

To move an object, referred to as load, the sarcomeres in the muscle fibers of the skeletal muscle must shorten. The force generated by the contraction of the muscle (or shortening of the sarcomeres) is called muscle tension. However, muscle tension also is generated when the muscle is contracting against a load that does not move, resulting in two main types of skeletal muscle contractions: isotonic contractions and isometric contractions.

In isotonic contractions, where the tension in the muscle stays constant, a load is moved as the length of the muscle changes (shortens). There are two types of isotonic contractions: concentric and eccentric. A concentric contraction involves the muscle shortening to move a load. An example of this is the biceps brachii muscle contracting when a hand weight is brought upward with increasing muscle tension. As the biceps brachii contract, the angle of the elbow joint decreases as the forearm is brought toward the body. Here, the biceps brachii contracts as sarcomeres in its muscle fibers are shortening and cross-bridges form the myosin heads pull the actin. An eccentric contraction occurs as the muscle tension diminishes and the muscle lengthens. In this case, the hand weight is lowered in a slow and controlled manner as the amount of cross-bridges being activated by nervous system stimulation decreases. In this case, as tension is released from the biceps brachii, the angle of the elbow joint increases. Eccentric contractions are also used for movement and balance of the body.

An isometric contraction occurs as the muscle produces tension without changing the angle of a skeletal joint. Isometric contractions involve sarcomere shortening and increasing muscle tension, but do not move a load, as the force produced cannot overcome the resistance provided by the load. For example, if one attempts to lift a hand weight that is too heavy, there will be sarcomere activation and shortening to a point, and ever-increasing muscle tension, but no change in the angle of the elbow joint. In everyday living, isometric contractions are active in maintaining posture and maintaining bone and joint stability. However, holding your head in an upright position occurs not because the muscles cannot move the head, but because the goal is to remain stationary and not produce movement. Most actions of the body are the result of a combination of isotonic and isometric contractions working together to produce a wide range of outcomes (Figure 1).

Figure 1. Types of Muscle Contractions. During isotonic contractions, muscle length changes to move a load. During isometric contractions, muscle length does not change because the load exceeds the tension the muscle can generate.

All of these muscle activities are under the exquisite control of the nervous system. Neural control regulates concentric, eccentric and isometric contractions, muscle fiber recruitment, and muscle tone. A crucial aspect of nervous system control of skeletal muscles is the role of motor units.


Chronic high-frequency (>100 Hz) deep brain stimulation (DBS) is an established medical treatment for movement disorders such as Parkinson’s disease (PD) and is being explored for the treatment of many other neurological and psychiatric indications 1,2,3 . Yet, despite decades of clinical use, its underlying therapeutic mechanism is still unclear 1,2 . In particular, there is limited knowledge regarding the neural oscillatory modulations induced with therapeutic high-frequency stimulation (HFS). If robust and target-specific neural signatures associated with HFS can be discovered, they can both assist to uncover the mechanism of DBS therapy and open the path for the construction of adaptive therapies that can tune the stimulation parameters for individual PD patients.

The studies seeking an electrophysiological basis for the mechanisms of DBS have focused on the investigation of neuronal spiking and oscillatory activity from the basal ganglia. Early hypotheses suggested that high-frequency DBS mimics lesioning by inhibiting neuronal firing from the stimulated structure 4,5,6,7,8 . Others proposed that DBS therapy overrides the pathological burst-type firing with its stimulus-induced regular (tonic) pattern and thereby ameliorated parkinsonian symptoms 9,10,11 . This effect is not only in the stimulated structure but also travels downstream to the basal ganglia-thalamo-cortical circuit 10,12 and creates an “informational lesion” preventing the relay of pathological firing and oscillations 13 . However, other studies suggest that DBS, by regularizing basal ganglia spiking activity, enhances the information processing and restores responsiveness of the thalamocortical cells to the incoming sensorimotor information 14,15,16 , indicating that rather than causing “lesioning”, DBS might exert its therapeutic effect through promotion of neural activity similar to the “healthy” state 17,18 .

Local field potentials (LFP) of the basal ganglia have long attracted interest due to their utility as a feedback modality for closed-loop DBS. Particularly in the subthalamic nucleus (STN), one of the frequently targeted structures in PD patients 19,20 , excessive beta (12–30 Hz) band oscillations are considered as the hallmark 21,22 and have shown to diminish with DBS 12,23,24 and dopaminergic medication 23,25,26 . More recently, the broadband high-frequency oscillations (200–450 Hz, HFO) of LFPs and cross-frequency coupling between beta and HFO bands 27,28,29,30 have been identified as important markers in PD electrophysiology. Although the pharmaceutical modulations of the LFP bands (e.g., suppression of beta and enhancement in the HFO bands) have been well-documented 29,30,31,32,33 , the large stimulus artifact observed during DBS have hindered further investigation of these biomarkers, especially in the HFO range, for closed-loop neuromodulation applications 34 . Consequently, the contribution of LFPs in uncovering the mechanisms of DBS have been limited due to the inability to record these oscillations during stimulation.

With these motivations, we established an intraoperative system to record LFPs during acute stimulation of STN in PD patients undergoing DBS surgery. We hypothesized that HFS exerts its therapeutic effect by modulating oscillatory neural activity in the STN, similar to the effect of pharmaceutical treatment. To test this hypothesis, we recorded LFPs from microelectrodes intraoperatively and studied their modulation during multiple low- and HFS paradigms, both outside and within the STN. We observed that high-frequency therapeutic DBS (>100 Hz) induced HFO activity similar to the reports in the pharmacologically treated patients and healthy animals. In conjunction, we noted an evoked activity after each stimulus pulse, which was more resonant with the HFS. More interestingly, the strength of induced HFO was related to the interaction of stimulation pulses with the phase of the evoked waveform, indicating that both measures and their characteristics can be used functionally to optimize electroceutical therapy.

Materials and Methods

The experiments were conducted in two rhesus monkeys (Macaca mulatta R7160 and R370 weighing 5.2 and 6.9 kg, respectively). The studies were performed in compliance with The National Institutes of Health Guide for Care and Use of Laboratory Animals (1996) and with the Emory University guidelines for the use and care of laboratory animals in research.

MPTP treatment. The monkeys were treated with MPTP via a single injection through the internal carotid artery (left side in R7160, right side in R370). The total amounts of MPTP were 3.2 mg (0.6 mg/kg) and 4.1 mg (0.6 mg/kg), respectively. Both monkeys developed a stable parkinsonian state characterized by contralateral rigidity and bradykinesia. Tremor was not present at rest or with action in either monkey.

Surgical procedure. A metal chamber was anchored over the left cerebral hemisphere in monkey R7160 and the right cerebral hemisphere in monkey R370. The chamber was placed aseptically under isofluorane anesthesia. A chronic stimulating electrode was implanted through a recording chamber targeting the STN (Fig.1), previously identified by microelectrode mapping. The tips of the chronic stimulating electrode were connected to a programmable pulse generator (Itrel II, Medtronic Inc.) implanted subcutaneously in the monkey's back. The stimulating lead was a scaled-down version of the chronic stimulation electrode used in humans (Model 3387, Medtronic Inc.) and consisted of four metal contacts (impedances of 100–150 MΩ) each with a diameter of 0.76 mm, thickness of 0.50 mm, and separation between contacts of 0.50 mm.

Location of the electrode contacts and neurons recorded during 136 Hz at 1.8 V in monkey R7160 and at 3.0 V in monkey R370, and changes in the firing rate. The cathode of the stimulation electrode was located in the posteromedial portion of the STN in monkey R7160 and on the posterior portion of the STN in monkey R370, 1 mm lateral to that of monkey R7160. Scale bars, 5 mm. OT, Optic tract Ret, thalamic reticular nucleusSN, substantia nigra STR, striatumTH, thalamus.

Behavioral assessment. The amount of spontaneous movement was assessed using a computer-assisted method of behavioral assessment to quantify the amount of movement (Bergman et al., 1990) while the monkey was in a Plexiglas cage. Each session was videotaped for subsequent rating by examiners blinded to the experimental condition. During the videotape ratings the order of stimulation conditions was randomized, and two scorers (blinded to the experimental condition) counted the total movement time per 10 min for the arm and leg on the right and left sides of the body from the video. A post hocanalysis (Tukey's honestly significant difference) was used to determine the significance of the difference in the amount of time of limb movement. Muscle tone of the biceps brachii muscles evoked by manual elbow extension contralateral to the HFS was assessed using electromyography (EMG). The most effective pair of stimulating electrode contacts was chosen for bipolar stimulation in each animal after evaluation of the clinical effect and the adverse effects. The threshold for adverse effects was determined by inspection of the animal for capsular responses with the onset of stimulation. The effect of STN stimulation on spontaneous movement and muscle tone was compared under the different experimental conditions. Stimulation conditions were 210 μsec pulse width, 20 and 136 Hz, and 1.4, 2.4, and 3.0 V in R7160, and 90 μsec pulse width, 2, 136, and 185 Hz, and 2.0 and 3.5 V in R370. Maximal voltage for the behavioral assessment was set to just below the threshold for corticospinal contraction at 136 or 185 Hz in each monkey.

Recording procedure and data collection. Neuronal activity was recorded extracellularly from the external globus pallidus (GPe) and GPi. A glass-coated platinum–iridium microelectrode (impedances of 0.4–0.8 MΩ at 2 kHz) was positioned within the chamber with the use of an xy coordinate microdrive (MO-95-lp, Narishige Scientific Instruments). Recording penetrations were made in parasagittal planes moving in a rostral to caudal manner at an angle of 70° to the orbitomeatal line. Neurons in GPi were examined for their response to passive manipulations of the limbs and orofacial structures. Spontaneous neuronal activity (with the animal sitting still with head fixed) was recorded under the following conditions: prestimulation, on-stimulation, and poststimulation. The duration of the prestimulation and poststimulation periods was set at 15–25 sec, and the duration of the on-stimulation period was set at 25–35 sec for 136 Hz and at 100–110 sec for 2 Hz stimulation. The change in neuronal activity was evaluated at 2 Hz stimulation with 2.4 V in R7160 and 3.0 V in R370, and at 136 Hz with 1.4 and 1.8 V in R7160, and with 2.0 and 3.0 V in R370. The 136 Hz stimulation of 1.4 V in R7160 and 2.0 V in R370 produced no apparent effect, but that of 1.8 V in R7160 and 3.0 V in R370 produced a consistent improvement in rigidity and bradykinesia, on the basis of clinical examination of animals in the primate chair. The effect of 5 min of extended 136 Hz, 3.0 V stimulation was studied in nine GP neurons in R370. The analog neuronal signal was amplified, bandpass filtered at 300–10,000 Hz, digitized, and sampled at 50 kHz with 4096-point vertical resolution for off-line analysis.

Data analysis. The software for analysis of neural signals during stimulation was developed using a C compiler running on DOS for off-line analysis (Hashimoto et al., 2002). The template of the stimulus artifact is constructed by averaging across all peristimulus segments. During stimulation, the stimulus artifact template was subtracted from the individual traces, and neuronal spikes were detected. A peristimulus time histogram (PSTH) was constructed, and mean discharge rates were determined. For comparison of the mean frequency in the prestimulation, on-stimulation, and poststimulation periods, Student's t test (two-tailed p < 0.05) was used. A significant increase or decrease in firing probability was accepted if a single bin in the PSTH was higher or lower than the mean prestimulation firing probability ± 3.3 SDs (p = 0.001), or when the p value of two to four consecutive bins by the Wilcoxon signed rank test was <0.01.

Histological analysis. After completion of the study the monkeys were killed with an overdose of pentobarbital (100 mg/kg), and the brains were processed histologically. The brain was sectioned in the frontal plane in R7160 and in the sagittal plane in R370. Recording sites in the GPe and GPi were reconstructed by identification of gliosis along the microelectrode and HFS electrode tracks and electrophysiological landmarks (DeLong, 1971). The stimulating lead was positioned in the STN 6 mm from the midline in monkey R7160 and 7 mm from the midline in monkey R370 (Fig. 1). In both monkeys, tyrosine hydroxylase staining revealed a nearly complete loss of dopaminergic cells in the substantia nigra pars compacta.

Drowsiness and Sleep

During drowsiness, the first discernible change is gradual loss of the frequent muscle and movement artifacts and reduction of blinks and rapid lateral eye movements. Instead, a very slow frequency of 0.25 to 1.0 Hz in the frontal and lateral frontal channels emerges. These are slow rolling eye movements, or SEMS (slow eye movements of sleep), which begin in drowsiness and persist through stage 1 sleep, until gradually being lost with deeper stages of non-rapid eye movement (NREM) sleep. The EEG during drowsiness contains slower, synchronous frequencies of theta and delta throughout the background (see Figure 14).

Figure 14.

Example of drowsiness from a normal adult EEG recording. Note the prominent theta and delta activity, lack of eye movements or blinks, lack of muscle or movement artifact, and early suggestion of slow lateral rolling eye movements best seen in the F7 (more. )

Defining features of sleep stages are listed in Table 1. NREM sleep is classified as light NREM (stages 1𠄲 now termed N1𠄲) and deeper slow wave sleep (SWS, formerly known as stages 3𠄴 N3𠄴), as well as REM sleep. Typically, approximately 75% of the night is spent in NREM sleep and up to 25% in REM sleep. Stage 1 (N1) sleep is contiguous with drowsiness and is characterized by SEMS and slower theta and delta EEG frequencies of 1 to 7 Hz, with less than 50% alpha frequency activity in a 30-second epoch. It is easily marked by the appearance of vertex waves (V-waves) sharply contoured, fronto-centrally predominant waves (Figure 15). During stage 2 (N2) sleep, more delta frequency background begins to emerge, and the defining features of sleep spindles, K-complexes, and posterior occipital sharp transients of sleep (POSTS) are seen (Figures 16, 17). Sleep spindles are thought to reflect the synchronous activity mediated by thalamo-cortical neuronal networks. SWS (N3) has similar features, but less spindles, K-complexes, and POSTS are seen and even more delta frequency activity emerges (Figure 18).


Defining Features of Sleep Stages on the EEG

Figure 15.

Stage 1 (N1) sleep. Characterized by slow rolling eye movement artifacts, and slower theta and some delta frequencies in the EEG background. V-waves (V) also typically occur. Copyright 2013. Mayo Foundation for Medical Education and Research. All rights (more. )

Figure 16.

Stage 2 (N2) sleep. Slower theta and some delta (by definition, less than 20% of background of delta range slowing) frequencies in the EEG background. K-complexes and sleep spindles are the hallmarks of N2 architecture. Copyright 2013. Mayo Foundation (more. )

Figure 17.

Slow wave sleep (N3) contains greater than 20% high-voltage (㹵 μV crest to crest) delta frequencies and fewer K-complexes and spindles. The figure below was taken from a full EEG recording during a polysomnogram, since N3 sleep is typically (more. )

Figure 18.

REM sleep is characterized by a more typically wake-appearing, desynchronized, mixed-frequency background, which may contain alpha frequencies, characteristic centrally dominant sharply contoured sawtooth waves, and rapid eye movement artifacts in lateral (more. )

REM sleep was previously known as paradoxical sleep, because REM actually resembles the waking EEG more closely than NREM sleep, having a desynchronized, low-voltage background. There are also fronto-central, sharply contoured theta frequencies called sawtooth waves, as well as REM artifacts seen in lateral frontal sites (Figure 18). Proper sleep-staging criteria also require features of very low-voltage chin electromyography (EMG) and eye movements recorded by electrooculogram (EOG) channels, but these polysomnographic channels are not routinely recorded during outpatient EEGs.

Except where indicated, this publication is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License (BY-NC-SA:

No permission is required from the editors, authors, or publisher to reuse or repurpose content, provided the original work is properly cited. Figures, tables, and images included in this work are also published under the CC BY-NC-SA license and should be properly cited when reused or repurposed.

For any reuse or redistribution of a work, you must also make clear the license terms under which the work was published. You may not apply legal or technological measures that legally restrict others from doing anything the license permits.

Materials and Methods


Physiological experiments were carried out at the University of Modena and Reggio Emilia on two (MK1 and MK2) adult female macaque monkeys (Macaca fascicularis, 3–4 kg, 4–5 years old). Subjects’ water intake was controlled daily to enhance motivation, and they were trained to sit in a primate chair and to interact with the experimenters. Two (MK3 and MK4) naïve monkeys (Macaca fascicularis and Macaca mulatta, 3–4 kg, 4–5 years old) were used for injection of neural tracers. The anatomical experiments were carried out at the University of Parma. All experimental protocols complied with the European law on the humane care and use of laboratory animals (directives 86/609/EEC, 2003/65/CE, and 2010/63/EU), were authorized by the Italian Ministry of Health (n. 65/2010-B e n. 66/2010-C released on 03/29/2010 n. 155/2013-C released on 06/25/2013 n. 294/2012-C released on 12/11/2012 n. 48/2016-PR released on 01/20/2016), and were approved by the local Ethics Committee of the University of Modena and Reggio Emilia and University of Parma.


Monkeys (MK1 and MK2) sat in the primate chair in front of a panel at a distance of 114 cm, on which 49 bicolored light-emitting diodes (LEDs) with a diameter of 0.05° were placed. Eye position was monitored by using a magnetic search coil surgically implanted beneath the conjunctiva of one eye and sampled at 1 KHz 75 . Head position was maintained with a surgically implanted stainless steel prosthesis, and monkeys’ heads were painlessly restricted by MUPRO 23 , a homemade multipurpose neck robot. MUPRO was designed to record both the isometric neck forces and to enable head rotation on the horizontal plane. It consists of a mechanical device, comprising a cardan joint, a potentiometer, an electromagnetic brake, and four flexion load cells (FLCs), which identify the isometric forces applied in four directions of the space (i.e., forward, backward, rightward, and leftward), plus an oleodynamic system allowing head rotation on the horizontal plane between +/− 20°. These components are assembled on a column bolted to the primate’s chair. An electrical device provides DC power for the potentiometer and the brake. To allow electrophysiological recordings, macaques were additionally implanted with a stainless-steel recording chamber (Thomas Recording) using stereotaxic coordinates. The inner diameter of the recording chamber was 19 mm, and it was vertically oriented to enable a perpendicular approach to the region of interest. All surgical procedures were conducted using an aseptic technique and under general anesthesia (Zoletil 10 mg/kg i.m.), and after each surgical intervention, treatment with antibiotics, cortisone, and analgesics was administered for up to one week. Before each session, the chamber was aseptically opened and rinsed thoroughly with sterile saline, and through a microdrive system (Mini Matrix Thomas Recording) a quartz-glass electrode (0.5–1.0 MΩ) passed through the dura. The biological signal was preamplified (PreAmplifier DPA-4), amplified, and filtered (Main Amplifier/Filter System MAF-05) to eliminate artifacts from 5 to 75 KHz and coil drivers. A SPS 8701E Waveform Discriminator System selected the amplified unit activity, which was monitored using an oscilloscope and also audio-monitored. Eye position, LED levels, unit activity, auditory markers, head forces, and head-rotation signals were sampled continuously during the experimental session at 1 KHz, stored by SuperScope II (GWI) software, and imported into Matlab for further analysis by custom scripts.

Behavioral Tasks

Once a neuron was isolated, macaque subjects performed a head-rotation task (HRe) (Fig. 1a). The monkey’s head was partially unrestrained (i.e., forces were measured through the MUPRO system, and head rotation of +/− 20° was allowed), and subjects received a piece of fruit (1 × 1 cm), positioned to their right or left, as a reward. In order to reach the food, they had to turn their heads and reach 20° rightward or leftward. Head forces and position were signaled by analog signals expressed in volts. Once subjects performed the HRe, neurons were also tested during a visually guided saccade task (ST) (Supplementary Fig. S1a) in which they first fixated, for a fixed period of 1 s, a central red LED (red period) within an electronic window ranging from 3 to 5 deg. When the central red target was abruptly switched off, a peripheral red target (20° up, down, left, or right) simultaneously appeared. Subjects shifted their gaze from fixation to this peripheral target as quickly as possible (within 0.7 s), maintaining a new fixation for 1 s to receive a juice reward. A 2-s intertrial period followed each trial. The visual stimuli were presented by homemade software running on a personal computer, and an acoustic cue, with an intensity of 40–50 dB, was switched on at the beginning of each trial session and switched off at the end, thus signaling to the monkey the beginning and the end of the working period. The onset and offset of each epoch in ST was signaled by analog LED levels. Finally, a subset of neurons recorded in BA 9/46dr was also tested during the head-rotation observation task (HRo Fig. 1b). The monkey’s head was restricted in the central position (i.e., forces were measured through the MUPRO system, but head rotation was not allowed), and the experimenter, facing the monkey at a distance of 60 cm, turned his head with two fixed sequences: first, rightward (with respect to the monkey), from 0° to 90° away from the monkey’s face (Averted Right epoch) and then, with no relevant delay, leftward, from 90° to 0° toward the monkey’s face (Directed Left epoch) second, leftward, from 0° to 90° away from the monkey’s face (Averted Left epoch) and then, with no relevant delay, rightward, from 90° to 0° toward the monkey’s face (Directed Right epoch). The intertrial period between two sequences during HRo was at least of 2 s and it was used as baseline. The onset and offset of HRo epochs were signaled by analog auditory markers produced by experimenters. During the HRo, monkeys were randomly rewarded, with no temporal relation to the end of trials. Eye and head position as well as neck forces were continuously recorded during the HRe, ST, and HRo. The activity of each neuron was recorded in at least 10 trials for each basic condition. Neurons were classified as task related if they had a significant response (Bonferroni post hoc test, p < 0.05) to the HRe and/or the HRo.


Single-neuron activity was analyzed in relation to the analog signals related to the main behavioral events. Spikes were recorded continuously and were convolved with a 20-ms Gaussian smoothing window. Self-initiated head rotation was analyzed, synchronizing neural responses to the onset of head-rotation movements considering the following epochs: (1) baseline, from 1.5 to 1 s before head-rotation onset, during the intertrial period (2) premovement, 0.5 s before head-rotation onset (3) movement, 1.5 s after head-rotation onset. Possible responses to self head rotation relative to baseline, expressed as mean firing rate (spikes/s), were assessed considering both directions (rightward and leftward) by means of a 2 × 3 repeated measures ANOVA (factors: Direction, Epoch) with a significance criterion of p < 0.05. Only neurons exhibiting at least a significant effect of the factor Epoch, alone or in interaction with the other factors, were classified as head-rotation neurons (Bonferroni post hoc test, p < 0.05).

A subset of single neurons was tested during the HRo. The neuronal responses were analyzed, synchronizing the neuronal activity to the onset of experimenter’s head rotation considering the following epochs: (1) baseline, 0.5 s before movement onset, during the intertrial period (2) averted epochs (Averted Right, Averted Left), from movement onset to movement offset (of variable duration, calculated on a trial-by-trial basis) (3) directed epochs (Directed Left, Directed Right), 1.5 s after offset of averted epochs. Possible responses to the experimenter’s head rotations relative to baseline, expressed as mean firing rate (spikes/s), were assessed considering both hemifields (right hemifield and left hemifield) by means of a 2 × 3 repeated measures ANOVA (factors: Hemifield, Epoch) with a significance criterion of p < 0.05. Only neurons exhibiting at least a significant effect of the factor Epoch, alone or in interaction with the other factors, were classified as visually triggered (Bonferroni post hoc test, p < 0.05).

Finally, saccadic eye movements were analyzed, synchronizing neural responses to the onset and offset of saccades produced during ST considering the following epochs of interest: (1) baseline, from 0.7 to 0.9 s after the red central target onset, while monkeys were fixating in central position (2) premovement, 0.2 s before the saccade onset (3) movement, from saccade onset to saccade offset (of variable duration, calculated on a trial-by-trial basis) (4) reaching position, 0.2 s after saccade offset. Eye onset and offset were defined as the last points on either side of the peak velocity before which the tangential velocity fell below 30°/s 17, 30 . Possible responses to the saccadic eye movement relative to baseline, expressed as mean firing rate (spikes/s), were assessed considering all directions (20° up, down, left, right) by means of a 4 × 4 repeated measures ANOVA (factors: Direction, Epoch) with a significance criterion of p < 0.05. Only neurons exhibiting at least a significant effect of the factor Epoch, alone or in interaction with the other factors, were classified as eye-motor neurons (Bonferroni post hoc test, p < 0.05).

Population analyses were carried out, taking into account single-neuron responses expressed in terms of normalized mean activity using a moving window of 200-ms slit forward in steps of 20 ms and analyzed with different repeated measures ANOVAs depending on the conditions to be compared, as previously described for single neurons. Population analyses for Fig. 4b and c were carried out by means of 2 × 2 repeated measures ANOVA (factors: Direction, Epoch) with a significance criterion of p < 0.05, followed by a Bonferroni post hoc test correction (p < 0.05). Finally, single-neuron responses in Figs 3c and 4c expressed as normalized activity (color-map plots) were carried out using a moving window of 100-ms slit forward in steps of 20 ms. Furthermore, relative to each neuron peak of activity (equal to 1), we also calculated the burst duration (Figs 3d and 4d) as the time interval between the first bin before and after the peak of activity whose value was higher and lower than (1 − B) × 0.25 + B, respectively, where B is the mean baseline activity. Differences between the burst durations were highlighted by means of a t-test with a significance criterion of p < 0.05. Once we obtained the first bin before the peak of activity whose value was higher than formula previously mentioned, we calculated the neuronal latency as the time interval between the first bin and the onset of the behavioral events of interest. Then, we plotted cumulative distributions of latencies relative to the behavioral events of interest (Figs 3d and 4d). By means of an χ 2 test performed bin per bin (bin = 100 ms) between cumulative distributions, we established the time intervals in which significant differences in the proportion of neurons were found (χ 2 , p < 0.05).

Population analyses in Supplementary Fig. S2c were carried out comparing trials in which monkeys generated smooth-pursuit eye movements and those in which they generated saccadic eye movements during HRo. Saccades and smooth-pursuit were grouped taking into account for each trial the ocular velocity pattern as shown in Supplementary Fig. S2a.


Internal Edit

Homeostatic imbalances Edit

Homeostatic outbalances are the main driving force for changes of the body. These stimuli are monitored closely by receptors and sensors in different parts of the body. These sensors are mechanoreceptors, chemoreceptors and thermoreceptors that, respectively, respond to pressure or stretching, chemical changes, or temperature changes. Examples of mechanoreceptors include baroreceptors which detect changes in blood pressure, Merkel's discs which can detect sustained touch and pressure, and hair cells which detect sound stimuli. Homeostatic imbalances that can serve as internal stimuli include nutrient and ion levels in the blood, oxygen levels, and water levels. Deviations from the homeostatic ideal may generate a homeostatic emotion, such as pain, thirst or fatigue, that motivates behavior that will restore the body to stasis (such as withdrawal, drinking or resting). [2]

Blood pressure Edit

Blood pressure, heart rate, and cardiac output are measured by stretch receptors found in the carotid arteries. Nerves embed themselves within these receptors and when they detect stretching, they are stimulated and fire action potentials to the central nervous system. These impulses inhibit the constriction of blood vessels and lower the heart rate. If these nerves do not detect stretching, the body determines perceives low blood pressure as a dangerous stimulus and signals are not sent, preventing the inhibition CNS action blood vessels constrict and the heart rate increases, causing an increase in blood pressure in the body. [3]

External Edit

Touch and pain Edit

Sensory feelings, especially pain, are stimuli that can elicit a large response and cause neurological changes in the body. Pain also causes a behavioral change in the body, which is proportional to the intensity of the pain. The feeling is recorded by sensory receptors on the skin and travels to the central nervous system, where it is integrated and a decision on how to respond is made if it is decided that a response must be made, a signal is sent back down to a muscle, which behaves appropriately according to the stimulus. [2] The postcentral gyrus is the location of the primary somatosensory area, the main sensory receptive area for the sense of touch. [4]

Pain receptors are known as nociceptors. Two main types of nociceptors exist, A-fiber nociceptors and C-fiber nociceptors. A-fiber receptors are myelinated and conduct currents rapidly. They are mainly used to conduct fast and sharp types of pain. Conversely, C-fiber receptors are unmyelinated and slowly transmit. These receptors conduct slow, burning, diffuse pain. [5]

The absolute threshold for touch is the minimum amount of sensation needed to elicit a response from touch receptors. This amount of sensation has a definable value and is often considered to be the force exerted by dropping the wing of a bee onto a person's cheek from a distance of one centimeter. This value will change based on the body part being touched. [6]

Vision Edit

Vision provides opportunity for the brain to perceive and respond to changes occurring around the body. Information, or stimuli, in the form of light enters the retina, where it excites a special type of neuron called a photoreceptor cell. A local graded potential begins in the photoreceptor, where it excites the cell enough for the impulse to be passed along through a track of neurons to the central nervous system. As the signal travels from photoreceptors to larger neurons, action potentials must be created for the signal to have enough strength to reach the CNS. [3] If the stimulus does not warrant a strong enough response, it is said to not reach absolute threshold, and the body does not react. However, if the stimulus is strong enough to create an action potential in neurons away from the photoreceptor, the body will integrate the information and react appropriately. Visual information is processed in the occipital lobe of the CNS, specifically in the primary visual cortex. [3]

The absolute threshold for vision is the minimum amount of sensation needed to elicit a response from photoreceptors in the eye. This amount of sensation has a definable value and is often considered to be the amount of light present from someone holding up a single candle 30 miles away, if one's eyes were adjusted to the dark. [6]

Smell Edit

Smell allows the body to recognize chemical molecules in the air through inhalation. Olfactory organs located on either side of the nasal septum consist of olfactory epithelium and lamina propria. The olfactory epithelium, which contains olfactory receptor cells, covers the inferior surface of the cribiform plate, the superior portion of the perpendicular plate, the superior nasal concha. Only roughly two percent of airborne compounds inhaled are carried to olfactory organs as a small sample of the air being inhaled. Olfactory receptors extend past the epithelial surface providing a base for many cilia that lie in the surrounding mucus. Odorant-binding proteins interact with these cilia stimulating the receptors. Odorants are generally small organic molecules. Greater water and lipid solubility is related directly to stronger smelling odorants. Odorant binding to G protein coupled receptors activates adenylate cyclase, which converts ATP to camp. cAMP, in turn, promotes the opening of sodium channels resulting in a localized potential. [7]

The absolute threshold for smell is the minimum amount of sensation needed to elicit a response from receptors in the nose. This amount of sensation has a definable value and is often considered to be a single drop of perfume in a six-room house. This value will change depending on what substance is being smelled. [6]

Taste Edit

Taste records flavoring of food and other materials that pass across the tongue and through the mouth. Gustatory cells are located on the surface of the tongue and adjacent portions of the pharynx and larynx. Gustatory cells form on taste buds, specialized epithelial cells, and are generally turned over every ten days. From each cell, protrudes microvilli, sometimes called taste hairs, through also the taste pore and into the oral cavity. Dissolved chemicals interact with these receptor cells different tastes bind to specific receptors. Salt and sour receptors are chemically gated ion channels, which depolarize the cell. Sweet, bitter, and umami receptors are called gustducins, specialized G protein coupled receptors. Both divisions of receptor cells release neurotransmitters to afferent fibers causing action potential firing. [7]

The absolute threshold for taste is the minimum amount of sensation needed to elicit a response from receptors in the mouth. This amount of sensation has a definable value and is often considered to be a single drop of quinine sulfate in 250 gallons of water. [6]

Sound Edit

Changes in pressure caused by sound reaching the external ear resonate in the tympanic membrane, which articulates with the auditory ossicles, or the bones of the middle ear. These tiny bones multiply these pressure fluctuations as they pass the disturbance into the cochlea, a spiral-shaped bony structure within the inner ear. Hair cells in the cochlear duct, specifically the organ of Corti, are deflected as waves of fluid and membrane motion travel through the chambers of the cochlea. Bipolar sensory neurons located in the center of the cochlea monitor the information from these receptor cells and pass it on to the brainstem via the cochlear branch of cranial nerve VIII. Sound information is processed in the temporal lobe of the CNS, specifically in the primary auditory cortex. [7]

The absolute threshold for sound is the minimum amount of sensation needed to elicit a response from receptors in the ears. This amount of sensation has a definable value and is often considered to be a watch ticking in an otherwise soundless environment 20 feet away. [6]

Equilibrium Edit

Semi circular ducts, which are connected directly to the cochlea, can interpret and convey to the brain information about equilibrium by a similar method as the one used for hearing. Hair cells in these parts of the ear protrude kinocilia and stereocilia into a gelatinous material that lines the ducts of this canal. In parts of these semi circular canals, specifically the maculae, calcium carbonate crystals known as statoconia rest on the surface of this gelatinous material. When tilting the head or when the body undergoes linear acceleration, these crystals move disturbing the cilia of the hair cells and, consequently, affecting the release of neurotransmitter to be taken up by surrounding sensory nerves. In other areas of the semi circular canal, specifically the ampulla, a structure known as the cupula—analogous to the gelatinous material in the maculae—distorts hair cells in a similar fashion when the fluid medium that surrounds it causes the cupula itself to move. The ampulla communicates to the brain information about the head's horizontal rotation. Neurons of the adjacent vestibular ganglia monitor the hair cells in these ducts. These sensory fibers form the vestibular branch of the cranial nerve VIII. [7]

In general, cellular response to stimuli is defined as a change in state or activity of a cell in terms of movement, secretion, enzyme production, or gene expression. [8] Receptors on cell surfaces are sensing components that monitor stimuli and respond to changes in the environment by relaying the signal to a control center for further processing and response. Stimuli are always converted into electrical signals via transduction. This electrical signal, or receptor potential, takes a specific pathway through the nervous system to initiate a systematic response. Each type of receptor is specialized to respond preferentially to only one kind of stimulus energy, called the adequate stimulus. Sensory receptors have a well-defined range of stimuli to which they respond, and each is tuned to the particular needs of the organism. Stimuli are relayed throughout the body by mechanotransduction or chemotransduction, depending on the nature of the stimulus. [3]

Mechanical Edit

In response to a mechanical stimulus, cellular sensors of force are proposed to be extracellular matrix molecules, cytoskeleton, transmembrane proteins, proteins at the membrane-phospholipid interface, elements of the nuclear matrix, chromatin, and the lipid bilayer. Response can be twofold: the extracellular matrix, for example, is a conductor of mechanical forces but its structure and composition is also influenced by the cellular responses to those same applied or endogenously generated forces. [9] Mechanosensitive ion channels are found in many cell types and it has been shown that the permeability of these channels to cations is affected by stretch receptors and mechanical stimuli. [10] This permeability of ion channels is the basis for the conversion of the mechanical stimulus into an electrical signal..

Chemical Edit

Chemical stimuli, such as odorants, are received by cellular receptors that are often coupled to ion channels responsible for chemotransduction. Such is the case in olfactory cells. [11] Depolarization in these cells result from opening of non-selective cation channels upon binding of the odorant to the specific receptor. G protein-coupled receptors in the plasma membrane of these cells can initiate second messenger pathways that cause cation channels to open.

In response to stimuli, the sensory receptor initiates sensory transduction by creating graded potentials or action potentials in the same cell or in an adjacent one. Sensitivity to stimuli is obtained by chemical amplification through second messenger pathways in which enzymatic cascades produce large numbers of intermediate products, increasing the effect of one receptor molecule. [3]

Nervous-system response Edit

Though receptors and stimuli are varied, most extrinsic stimuli first generate localized graded potentials in the neurons associated with the specific sensory organ or tissue. [7] In the nervous system, internal and external stimuli can elicit two different categories of responses: an excitatory response, normally in the form of an action potential, and an inhibitory response. [12] When a neuron is stimulated by an excitatory impulse, neuronal dendrites are bound by neurotransmitters which cause the cell to become permeable to a specific type of ion the type of neurotransmitter determines to which ion the neurotransmitter will become permeable. In excitatory postsynaptic potentials, an excitatory response is generated. This is caused by an excitatory neurotransmitter, normally glutamate binding to a neuron's dendrites, causing an influx of sodium ions through channels located near the binding site.

This change in membrane permeability in the dendrites is known as a local graded potential and causes the membrane voltage to change from a negative resting potential to a more positive voltage, a process known as depolarization. The opening of sodium channels allows nearby sodium channels to open, allowing the change in permeability to spread from the dendrites to the cell body. If a graded potential is strong enough, or if several graded potentials occur in a fast enough frequency, the depolarization is able to spread across the cell body to the axon hillock. From the axon hillock, an action potential can be generated and propagated down the neuron's axon, causing sodium ion channels in the axon to open as the impulse travels. Once the signal begins to travel down the axon, the membrane potential has already passed threshold, which means that it cannot be stopped. This phenomenon is known as an all-or-nothing response. Groups of sodium channels opened by the change in membrane potential strengthen the signal as it travels away from the axon hillock, allowing it to move the length of the axon. As the depolarization reaches the end of the axon, or the axon terminal, the end of the neuron becomes permeable to calcium ions, which enters the cell via calcium ion channels. Calcium causes the release of neurotransmitters stored in synaptic vesicles, which enter the synapse between two neurons known as the presynaptic and postsynaptic neurons if the signal from the presynaptic neuron is excitatory, it will cause the release of an excitatory neurotransmitter, causing a similar response in the postsynaptic neuron. [3] These neurons may communicate with thousands of other receptors and target cells through extensive, complex dendritic networks. Communication between receptors in this fashion enables discrimination and the more explicit interpretation of external stimuli. Effectively, these localized graded potentials trigger action potentials that communicate, in their frequency, along nerve axons eventually arriving in specific cortexes of the brain. In these also highly specialized parts of the brain, these signals are coordinated with others to possibly trigger a new response. [7]

If a signal from the presynaptic neuron is inhibitory, inhibitory neurotransmitters, normally GABA will be released into the synapse. [3] This neurotransmitter causes an inhibitory postsynaptic potential in the postsynaptic neuron. This response will cause the postsynaptic neuron to become permeable to chloride ions, making the membrane potential of the cell negative a negative membrane potential makes it more difficult for the cell to fire an action potential and prevents any signal from being passed on through the neuron. Depending on the type of stimulus, a neuron can be either excitatory or inhibitory. [13]

Muscular-system response Edit

Nerves in the peripheral nervous system spread out to various parts of the body, including muscle fibers. A muscle fiber and the motor neuron to which it is connected. [14] The spot at which the motor neuron attaches to the muscle fiber is known as the neuromuscular junction. When muscles receive information from internal or external stimuli, muscle fibers are stimulated by their respective motor neuron. Impulses are passed from the central nervous system down neurons until they reach the motor neuron, which releases the neurotransmitter acetylcholine (ACh) into the neuromuscular junction. ACh binds to nicotinic acetylcholine receptors on the surface of the muscle cell and opens ion channels, allowing sodium ions to flow into the cell and potassium ions to flow out this ion movement causes a depolarization, which allows for the release of calcium ions within the cell. Calcium ions bind to proteins within the muscle cell to allow for muscle contraction the ultimate consequence of a stimulus. [3]

Endocrine-system response Edit

Vasopressin Edit

The endocrine system is affected largely by many internal and external stimuli. One internal stimulus that causes hormone release is blood pressure. Hypotension, or low blood pressure, is a large driving force for the release of vasopressin, a hormone which causes the retention of water in the kidneys. This process also increases an individuals thirst. By fluid retention or by consuming fluids, if an individual's blood pressure returns to normal, vasopressin release slows and less fluid is retained by the kidneys. Hypovolemia, or low fluid levels in the body, can also act as a stimulus to cause this response. [15]

Epinephrine Edit

Epinephrine, also known as adrenaline, is also used commonly to respond to both internal and external changes. One common cause of the release of this hormone is the Fight-or-flight response. When the body encounters an external stimulus that is potentially dangerous, epinephrine is released from the adrenal glands. Epinephrine causes physiological changes in the body, such as constriction of blood vessels, dilation of pupils, increased heart and respiratory rate, and the metabolism of glucose. All of these responses to a single stimuli aid in protecting the individual, whether the decision is made to stay and fight, or run away and avoid danger. [16] [17]

Digestive-system response Edit

Cephalic phase Edit

The digestive system can respond to external stimuli, such as the sight or smell of food, and cause physiological changes before the food ever enters the body. This reflex is known as the cephalic phase of digestion. The sight and smell of food are strong enough stimuli to cause salivation, gastric and pancreatic enzyme secretion, and endocrine secretion in preparation for the incoming nutrients by starting the digestive process before food reaches the stomach, the body is able to more effectively and efficiently metabolize food into necessary nutrients. [18] Once food hits the mouth, taste and information from receptors in the mouth add to the digestive response. Chemoreceptors and mechanorceptors, activated by chewing and swallowing, further increase the enzyme release in the stomach and intestine. [19]

Enteric nervous system Edit

The digestive system is also able to respond to internal stimuli. The digestive tract, or enteric nervous system alone contains millions of neurons. These neurons act as sensory receptors that can detect changes, such as food entering the small intestine, in the digestive tract. Depending on what these sensory receptors detect, certain enzymes and digestive juices from the pancreas and liver can be secreted to aid in metabolism and breakdown of food. [3]

Clamping techniques Edit

Intracellular measurements of electrical potential across the membrane can be obtained by microelectrode recording. Patch clamp techniques allow for the manipulation of the intracellular or extracellular ionic or lipid concentration while still recording potential. In this way, the effect of various conditions on threshold and propagation can be assessed. [3]

Noninvasive neuronal scanning Edit

Positron emission tomography (PET) and magnetic resonance imaging (MRI) permit the noninvasive visualization of activated regions of the brain while the test subject is exposed to different stimuli. Activity is monitored in relation to blood flow to a particular region of the brain. [3]

Other methods Edit

Hindlimb withdrawal time is another method. Sorin Barac et al. in a recent paper published in the Journal of Reconstructive Microsurgery monitored the response of test rats to pain stimuli by inducing an acute, external heat stimulus and measuring hindlimb withdrawal times (HLWT). [20]


Recent results suggest that transcranial alternating current stimulation (tACS) can noninvasively alter brain activity [1–4], but the physiological mechanisms behind these exciting findings remain poorly understood. Traditionally, tACS is thought to produce oscillating electric fields within the brain that hyperpolarize and depolarize neurons, so that they fire synchronously with the stimulation. Small-animal experiments demonstrate that the fields generated by applying current to the bare skull can entrain neurons [1, 4, 5], consistent with the idea that intracranial electric fields have a direct effect on brain activity. In humans, however, the tACS electrodes are placed on the participant’s intact scalp, not within the skull. Since the skin is highly conductive, but the skull beneath is not, much of that current is shunted away from the brain and stimulates neurons in the skin instead [6]. Rhythmic activation of somatosensory fibers could thus indirectly entrain central neurons by providing them with temporally structured sensory input. Since shunting also weakens electric fields in the brain, this indirect mechanism has been frequently proposed to be the dominant mode of action in humans [5, 7–10]. If this were true, it would have dramatic implications for how tACS is used and studied: brain areas would need to be targeted on the basis of somatosensory connectivity, rather than physical location, and brain regions that received little or no somatosensory input would be unreachable.

These competing hypotheses can be distinguished through the use of topical anesthesia. Pretreatment of the skin under and around the tACS electrodes with topical anesthetic blocks cutaneous afferents [11] and prevents them from generating somatosensory percepts [12]. If tACS acts indirectly via somatosensory inputs, topical anesthesia should reduce or abolish its effects by blocking transmission from the periphery. Conversely, if the electric fields directly affect neurons, applying topical anesthesia should produce little or no changes in the effects of tACS, as the electric fields produced within the brain remain the same. Previous attempts to test the indirect hypothesis have used proxy measurements for neural activity, with mixed results: topical anesthesia appears to prevent tACS from affecting nociception [10] and tremor [5], but effects on motor-evoked potentials [13] and language processing [14, 15] persist when somatosensory inputs are blocked. Interpreting these results is challenging, because the neural mechanisms behind these readouts are not well understood and each may involve multiple brain regions, only some of which may have been affected by the tACS used in each study.

An unambiguous test of the role of somatosensory input is to directly measure neural entrainment during tACS, with and without topical anesthetic. Here, we perform that decisive experiment in nonhuman primates, a highly realistic model for human neurostimulation. Using single-unit recordings of neurons in the hippocampus and visual cortex, we find that blocking somatosensory input has little effect on neural entrainment by tACS. Instead, our data support claims of a direct effect on neurons in the stimulated regions.

Materials and Methods


Eight healthy volunteers (3 females, 7 right-handed mean age, 30 ± 4 years age range, 22–32 years) took part in the study after providing written informed consent. They all had musical experience, either in performance (3 participants with 15–25 years of practice) or as amateur listeners or dancers (5 participants). They had no history of hearing, neurological, or psychiatric disorder, and were not taking any drug at the time of the experiment. The study was approved by the local ethics committee.

Auditory stimulation.

Each auditory stimulus lasted 33 s. The stimulus consisted of a 333.3 Hz pure tone in which a 2.4 Hz auditory beat was introduced by modulating the amplitude of the tone with a 2.4 Hz periodicity (i.e., 144 beats/min), using an asymmetrical Hanning envelope (22 ms rise time and 394 ms fall time, amplitude modulation between 0 and 1). A 2.4 Hz periodicity was chosen because (1) pilot participants were comfortable in imagining binary (1.2 Hz) and ternary (0.8 Hz) rhythms using this 2.4 Hz tempo and (2) these tempi lie in the ecological range of tempo perception and production (Drake and Botte, 1993). The sound was then amplitude modulated using an 11 Hz sinusoidal function oscillating between 0.3 and 1. Because the 2.4 Hz frequency was not an integer ratio of the 11 Hz frequency, the convolution of the two frequencies generated subtle irregularities in terms of amplitude and occurrence of the beats, thus resulting in a pseudo-periodic beat structure (Fig. 1A). Importantly, the frequency content of the sound envelope obtained by convoluting the two different amplitude modulation frequencies (2.4 and 11 Hz) contained a peak at the frequency of the beat (2.4 Hz) but did not contain any sideband frequencies corresponding to the frequencies of the binary or ternary meters (i.e., 1.2 and 0.8 Hz, respectively) (Fig. 1C). The subtle irregularities of the beat were perceived by all subjects and were purposely created to avoid induction of an involuntary binary subjective meter in the control condition (Bolton, 1894 Vos, 1973). Furthermore, the pseudo-periodicity of the beat structure, resulting from these irregularities, was closer to the more ecological situation where beat perception refers to the perception of periodicity from a non-strictly periodic framework (Large, 2008).

The auditory stimuli were generated using the PsychToolbox extensions (Brainard, 1997) running under Matlab 6.5 (The MathWork), and presented binaurally through earphones at a comfortable hearing level (BeyerDynamic DT 990 PRO).

Meter mental imagery and control conditions.

Participants were asked to perform three different tasks: a control task, a binary meter imagery task, and a ternary meter imagery task, in separate conditions (Fig. 1). Each condition consisted of 10 trials during which the 33 s auditory stimulus was presented after a 3 s foreperiod. Stimulus presentation was self-paced. During the first condition, participants performed the control task. They were asked to detect a very short (4 ms) sound interruption that was inserted at a random position in two additional trials interspersed within the block. This control task required a sustained level of attention as the stimulus had a complex structure. The two trials containing a short interruption were excluded from further analyses. During the second condition, participants performed the binary meter imagery task. They were asked to imagine a binary metric structure onto the perceived beat (f/2 = 1.2 Hz). During the third condition, they performed the ternary meter imagery task, by imagining a ternary metric structure onto the beat (f/3 = 0.8 Hz). Before the binary and ternary meter conditions, to ensure that participants understood the task, they were asked to perform overt movements (e.g., hand tapping, aloud counting) paced to the imposed metric structure, first with the help of the experimenter, and then alone. Subjective evaluation by the experimenter of the synchrony of those movements with the meter indicated that all participants performed the task without difficulty. The participants were then asked to begin their meter imagery as soon as they heard the first auditory beat of the stimulus, and to maintain this imagery as consistently as possible throughout the entire trial. During debriefing, participants reported that they had performed the mental imagery task without difficulty, although it did require a relatively high level of attention.

EEG recording.

Subjects were comfortably seated in a chair with their head resting on a support. They were instructed to relax, avoid any head or body movement during the recordings, and keep their eyes fixated on a point displayed on a computer screen in front of them. The experimenter remained in the recording room to monitor compliance to these instructions. The EEG was recorded using 64 Ag-AgCl electrodes placed on the scalp according to the International 10/10 system (Waveguard64 cap, Cephalon A/S). Vertical and horizontal eye movements were monitored using four additional electrodes placed on the outer canthus of each eye and in the inferior and superior areas of the right orbit. Electrode impedances were kept at <10 kΩ. The signals were amplified, low-pass filtered at 500 Hz, digitized using a sampling rate of 1000 Hz, and referenced to an average reference (64 channel high-speed amplifier, Advanced Neuro Technology).

EEG analysis.

Continuous EEG recordings were filtered using a 0.1 Hz high-pass Butterworth zero-phase filter to remove very slow drifts in the recorded signals. EEG epochs lasting 32 s were then obtained by segmenting the recordings from +1 to +33 s relative to the onset of the auditory stimulus at the beginning of each trial, thus yielding 10 epochs for each subject and condition. The EEG recorded during the first second of each epoch was removed (1) to discard the transient auditory evoked potential related to the onset of the stimulus (Saupe et al., 2009), (2) because previous studies have shown that steady-state EPs require several cycles of stimulation to be steadily entrained (Regan, 1989), and (3) because previous studies have shown that several repetitions of the beat are required to elicit a steady perception of beat and meter (Repp, 2005). These EEG processing steps were performed using Analyzer 1.05 (Brain Products).

The following EEG processing steps were performed using Letswave (Mouraux and Iannetti, 2008), Matlab (The MathWorks), and EEGLAB (

Artifacts produced by eye blinks or eye movements were removed using a validated method based on an independent component analysis (Jung et al., 2000), using the runica algorithm (Bell and Sejnowski, 1995 Makeig et al., 1996), as implemented in EEGLAB. For each subject and condition, EEG epochs were averaged across trials to reduce the contribution of activities non-phase locked to the stimulation train. The time-domain-averaging procedure was used to enhance the signal-to-noise ratio by attenuating the contribution of activities that were not strictly phase locked across trials (i.e., activities that were not phase locked to the sound stimulus). The obtained average waveforms were then transformed in the frequency domain using a discrete Fourier transform (Frigo and Johnson, 1998), yielding a frequency spectrum of signal amplitude (μV) ranging from 0 to 500 Hz with a frequency resolution of 0.031 Hz (Bach and Meigen, 1999).

Within the obtained frequency spectra, signal amplitude may be expected to correspond to the sum of (1) EEG activity induced by the auditory beat and/or the meter imagery task (i.e., beat- and meter-related steady-state EPs) and (2) unrelated residual background noise due, for example, to spontaneous EEG activity, muscle activity, or eye movements. Therefore, to obtain valid estimates of the beat- and meter-related steady-state EPs, the contribution of this noise was removed by subtracting, at each bin of the frequency spectra, the average amplitude measured at neighboring frequency bins (two frequency bins ranging from −0.15 to −0.09 Hz and from +0.09 to +0.15 Hz relative to each frequency bin). The validity of this subtraction procedure relies on the assumption that, in the absence of a steady-state EP, the signal amplitude at a given frequency bin should be similar to the signal amplitude of the mean of the surrounding frequency bins.

Finally, the magnitude of beat- and meter-related steady-state EPs was estimated by averaging the signal amplitude measured at the three frequency bins centered on the target frequency of each steady-state EP (i.e., 2.4 Hz: bins ranging from 2.356 to 2.418 Hz 1.2 Hz: bins ranging from 1.178 to 1.240 Hz 0.8 Hz: bins ranging from 0.775 to 0.837 Hz 1.6 Hz: bins ranging from 1.581 to 1.643 Hz), thereby accounting for a possible spectral leakage due to the fact that the discrete Fourier transform did not estimate signal amplitude at the exact frequency of each steady-state EP.

Statistical analyses.

For each participant, condition, and target frequency, the magnitude of steady-state EPs was averaged across all scalp electrodes, thus excluding any electrode selection bias (see Figs. 2, 3). This approach was used because there was no a priori assumption on the scalp topography of the beat- and meter-induced responses. Group-level results were expressed using the median and interquartile range (see Fig. 4). To examine whether the beat and meter induced a significant steady-state response, one-sample t tests were used to determine whether the noise-subtracted amplitudes measured at the target frequencies were significantly different from zero. Indeed, in the absence of a steady-state response, the average of the subtracted signal amplitude may be expected to tend toward zero.

To compare the beat- and meter-induced steady-state responses across experimental conditions, for each target frequency a one-way repeated-measures ANOVA was used to compare the noise-subtracted amplitudes obtained in the control, binary meter, and ternary meter conditions. Degrees of freedom were corrected using the Greenhouse–Geisser correction for violations of sphericity. Size effects were expressed using the partial η 2 . When significant, post hoc pairwise comparisons were performed using paired-sampled t tests. The significance level was set at p < 0.05.

Transient auditory event-related potentials.

To examine whether the beat and meter elicited transient auditory event-related potentials that could be identifiable in the time domain, average waveforms were computed after bandpass filtering (0.3 Hz to 30 Hz) and epoch segmentation from +1 s to +33 s relative to the onset of the sound stimulus (see Fig. 5).

Materials and methods

Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Biological sample (Macaca fuscata)Macaca fuscataNational Bio Resource Project of the MEXT, Japan
Software, algorithmMATLABMathWorksRRID:SCR_001622


Two adult macaque monkeys (Macaca fuscata monkey H, female, 6.6 kg, 8 years old monkey P, female, 6.8 kg, 7 years old) were used for the experiments. All procedures for animal care and experimentation were approved by the University of Tsukuba Animal Experiment Committee (permission number 14–137), and were carried out in accordance with the guidelines described in Guide for the Care and Use of Laboratory Animals published by the Institute for Laboratory Animal Research.

Behavioral tasks

Behavioral tasks were controlled using MATLAB (Mathworks, MA) with Psychtoolbox, a freely available toolbox for MATLAB. The monkeys sat in a primate chair facing a computer monitor in a sound-attenuated and electrically shielded room. Eye movements were monitored using an infrared eye-tracking system (EYE-TRAC 6 Applied Science Laboratories, MA) with sampling at 240 Hz. A liquid reward was delivered through a spout that was positioned in front of the monkey's mouth. Licking of the monkeys was monitored during the recording of 68 of the 99 recorded dopamine neurons. To monitor licking, a strain gauge was attached to the spout and measured the strain of the spout caused by licking.

We designed a Pavlovian procedure with three different conditions (Figure 1A–C) and applied it to the two monkeys. In each condition, trials started with the presentation of a central fixation point (0.8°×0.8°). The monkey was required to fixate this point. After a 400 ms fixation period, a bar stimulus (2.9°×9.4°) with red and green areas was presented as a CS. The size of the green area indicated the amount (i.e., the value) of a liquid reward that the monkey would receive. The larger the green area became, the larger the associated reward value. In the first condition (value-increase condition, Figure 1A), the green area was minimal at the beginning and gradually increased (3.8°/s), that is, the reward value gradually increased (0.082 ml/s) (Video 1). The gradual increase randomly stopped within 2450 ms after the onset of the bar stimulus (uniform distribution from 0 to 2450 ms) so that the monkey was unable to precisely predict the reward value until the green area stopped increasing. In the second condition (value-decrease condition, Figure 1B), the green area was maximal at the beginning and gradually decreased (3.8°/s), that is, the reward value gradually decreased (0.082 ml/s) (Video 2). The gradual decrease randomly stopped within 2450 ms after the bar onset (uniform distribution from 0 to 2450 ms). In the third condition (value-fixed condition, Figure 1C), the size of the green area did not change and was instead fixed at the minimum, half, or maximum that predicted a 0.1, 0.2, or 0.3 ml reward, respectively (these reward amounts were the same as those indicated by the green area in the value-increase and value-decrease conditions). In addition to these CSs, a CS constituting of vertical red and green bars was included, and this was randomly followed by a 0.1-0.3 ml reward (i.e., the reward value was uncertain). These four CSs were presented with an equal probability (25% each). In each condition, the total time during which the CS was presented was fixed at 2850 ms, and the monkey was required to maintain the central fixation during this period. Immediately after the offset of the bar stimulus, the reward indicated by the green area was delivered simultaneously with a tone (1000 Hz). The delay between the CS onset and the reward delivery did not change across trials. Each condition consisted of a block of 50 trials, and we collected data by repeating the three conditions once or more for each neuron (Figure 1D). The order of the three conditions, (1) the value-fixed, (2) value-increase, and (3) value-decrease conditions, was fixed across recording sessions. The total amount of reward was the same (10 ml) among blocks. Trials were aborted immediately if the monkey (1) did not start the central fixation within 4000 ms after the onset of the fixation point or (2) broke the central fixation during the initial 400 ms fixation period or the 2850 ms CS period (i.e., a continuous 3250 ms fixation was required). These error trials were signaled by a beep tone (100 Hz) and excluded from analyses.

We also used a choice task (Figure 1E). Each trial started with the presentation of a central fixation point (0.8°×0.8°), and the monkey was required to fixate this point. After a 400 ms fixation period, two bar stimuli with red and green areas (2.9°×9.4°) were presented on the right and left sides of the fixation point (eccentricity: 8.8°). The monkey was required to choose one of the bar stimuli by making a saccade immediately after the presentation of the bar stimuli. These bar stimuli were identical to those used in the Pavlovian procedure except that the green area did not increase or decrease over time. The size of the green area indicated the value of the reward that the monkey would obtain by choosing that bar stimulus and was randomly assigned to each bar stimulus (uniform distribution from 0.1 to 0.3 ml). Immediately after the monkey chose a bar stimulus, the other bar stimulus disappeared. Then, the reward indicated by the green area of the chosen bar stimulus was delivered simultaneously with a tone (1000 Hz). Trials were aborted immediately if the monkey (1) did not start the central fixation within 4000 ms after the onset of the fixation point or (2) broke the central fixation during the 400 ms fixation period. These error trials were signaled by a beep tone (100 Hz) and excluded from analyses.

Single-unit recording

A plastic head holder and recording chamber were fixed to the skull under general anesthesia and sterile surgical conditions. The recording chamber was placed over the midline of the frontoparietal lobes, and aimed at the SNc and VTA in both hemispheres. The head holder and recording chamber were embedded in dental acrylic that covered the top of the skull and were firmly anchored to the skull by plastic screws.

Single-unit recordings were performed using tungsten electrodes with impedances of approximately 3.0 MΩ (Frederick Haer, ME) that were introduced into the brain through a stainless steel guide tube using an oil-driven micromanipulator (MO-97-S Narishige, Tokyo, Japan). Recording sites were determined using a grid system, which allowed recordings at every 1 mm between penetrations. For a finer mapping of neurons, we also used a complementary grid that allowed electrode penetrations between the holes of the original grid.

Electrophysiological signals were amplified, band-pass filtered (200 Hz to 3 kHz RZ5D, Tucker-Davis Technologies, FL), and stored in a computer at the sampling rate of 24.4 kHz. Single-unit potentials were isolated using a window discrimination software (OpenEx, Tucker-Davis Technologies, FL).

Localization and identification of dopamine neurons

We recorded single-unit activity from putative dopamine neurons in the SNc and VTA. To localize the recording regions, the monkeys underwent an MRI scan to determine the positions of the SNc and VTA (Figure 2—figure supplement 1). Putative dopamine neurons were identified based on their well-established electrophysiological signatures: a low background firing rate at around 5 Hz, a broad spike potential in clear contrast to neighboring neurons with a high background firing rate in the substantia nigra pars reticulata (Figure 2—figure supplement 2), and a phasic excitation in response to free reward.

Data analysis

For null hypothesis testing, 95% confidence intervals (p<0.05) were used to define statistical significance in all analyses.

To examine whether the monkeys accurately predicted the reward values according to the size of the green area (Figure 1F), the choice rate of the right bar stimulus was fitted by the following logistic function:

where P indicates the choice rate of the right bar stimulus, Vright and Vleft indicate the reward values obtained by choosing the right and left bar stimuli, respectively, and β0 and β1 indicate the coefficients determined by logistic regression.

To calculate spike density functions (SDFs), each spike was replaced by a Gaussian curve (σ = 15 ms).

To calculate SDFs aligned by the CS onset in the value-increase and value-decrease conditions (Figure 2A,B,E,F and Figure 5A), spikes that occurred before the reward value stopped increasing or decreasing were used.

To calculate averaged SDFs across neurons (Figure 2E,F, Figure 3C, and Figure 5A,B), the baseline firing rate of each neuron was measured using a window from 500 to 0 ms before the fixation point onset and was subtracted from the original firing rate of each neuron.

To quantify tonic activity changes in the value-increase and value-decrease conditions (Figure 2C,D), the slope of the regression line between firing rate and time was calculated for each neuron. First, a window from 650 to 2450 ms after the CS onset was divided into 200 ms bins, and the firing rate in each bin was measured. Then, the regression line between the firing rate in each bin and the time at the center of each bin was calculated. The calculation window from 650 to 2450 ms after the CS onset was determined to exclude the effect of phasic responses evoked immediately after the CS onset.

To examine phasic activity modulations evoked by the reward value in the value-fixed condition, the regression coefficient between firing rate and reward value was calculated for each neuron. The firing rate was measured using a window from 100 to 400 ms after the CS onset. The calculation window was determined such that the window includes the major part of the neural modulation in the averaged activity.

To calculate SDFs aligned by the onset of each saccade (Figure 2—figure supplement 7B,F), the onset was determined as the time when angular eye velocity exceeded 40°/s.

To examine phasic activity modulation evoked when the reward value stopped increasing or decreasing in the value-increase and value-decrease conditions (Figure 3B,D), the regression coefficient between firing rate and reward value was calculated for each neuron. First, all trials were divided into three groups based on the reward value (large 0.23-0.3 ml, medium 0.16-0.23 ml, small 0.1-0.16 ml), and the firing rate was measured using a window from 150 to 500 ms after the stop onset for each trial group. Then, the regression coefficient between the firing rate and the mean reward value in each trial group was calculated. The calculation window was determined such that the window includes the major part of the neural modulation in the averaged activity.

To statistically test the difference in firing rate (original firing rate – baseline firing rate) between burst and non-burst spike firing (left in Figure 5A), a bootstrap procedure was applied. We first divided the calculation time window (650-2450 ms after the CS onset), which we used to detect the significantly positive regression slope of the 19 dopamine neurons (horizontal gray bar in the left column of Figure 2E), into initial (650-1250 ms), middle (1250-1850 ms), and late (1850-2450 ms) periods. The 19 dopamine neurons were randomly resampled with replacements to form a new bootstrap dataset that had the same number of neurons as the original dataset. Using the new dataset, we compared the firing rate (original firing rate – baseline firing rate) between burst and non-burst spike firing for the initial, middle, and late periods. This random resampling and comparison process was repeated 1000 times. If the firing rate was larger for burst spike firing than for non-burst spike firing or vice versa in more than 975 repetitions, the difference in firing rate was regarded as significant (p<0.05 bootstrap test with 1000 repetitions).

Watch the video: Πώς να αποφύγετε τον αυχενικό πόνο; (December 2021).