The Journal of Neurophysiology Vol. 78 No. 5 November 1997, pp. 2518-2530
Copyright ©1997 by the American Physiological Society

Quantal Organization of Reflex and Conditioned Eyelid Responses

José A. Domingo, Agnes Gruart, and José M. Delgado-García

Laboratorio de Neurociencia, Facultad de Biología, Universidad de Sevilla, 41012-Sevilla, Spain

	ABSTRACT

Abstract Introduction Methods Results Discussion References

Domingo, José A., Agnes Gruart, and José M. Delgado-García. Quantal organization of reflex and conditioned eyelid responses. J. Neurophysiol. 78: 2518-2530, 1997. Upper lid movements and the electromyographic activity of the orbicularis oculi muscle were recorded in behaving cats during spontaneous and experimentally evoked reflex blinks, and conditioned eyelid responses. Reflex blinks evoked by the presentation of air puffs, flashes, or tones consisted of a fast downward lid movement followed by late, small downward waves, recurring at 50-ms intervals. The latency, maximum amplitude, peak velocity, and number of late waves depended on the modality, intensity, and duration of the evoking stimulus. The power spectra of acceleration records indicated a dominant frequency of 20 Hz for air puff-evoked blinks. Flashes and tones usually evoked small and easily fatigable reflex responses of lower dominant frequencies (14-17 and 9-11 Hz, respectively). A basic 20-Hz oscillation was also noticed during lid fixation, and ramplike lid displacements evoked by optokinetic stimuli. Five classical conditioning paradigms were used to analyze the frequency-domain properties of conditioned eyelid responses. These learned lid movements differed in latency, maximum amplitude, and profile smoothness depending on the modality (air puff, tone), intensity (weak, strong), and presentation site (ipsi-, contralateral to the unconditioned stimulus) of the conditioned stimulus. It was found that the characteristic ramplike profile of a conditioned response was not smooth, but appeared to be formed by a succession of small waves at a dominant frequency of 20 Hz. The amplitude (and number) of the constituting waves depended on the characteristics of the conditioned stimulus and on the time interval until unconditioned stimulus presentation. Thus conditioned responses seemed to be formed from lid displacements of 2-6° in amplitude and 50 ms in duration, which increased in number throughout conditioning sessions, until a complete (i.e., lid closing) conditioned response was reached. It is suggested that a 20-Hz oscillator underlies the generation of reflex and conditioned eyelid responses. The oscillator is susceptible to being neurally modulated to modify the velocity of a given quantum of movement, and the total duration of the lid response. Learned eyelid movements are probably the result of a successively longer release of the oscillator as a function of the temporal-spatial needs of the motor response.

	INTRODUCTION

Abstract Introduction Methods Results Discussion References

The classical view of motor tremor as an exclusive, admonitory symptom of a pathological condition contrasts vividly with early descriptions of the effect of the electrical stimulation of cerebral motor cortex; namely, that "the frequency of muscular response to stimulation of the cortex does not vary with the rate of excitation, but maintains a nearly uniform rate of about 10 per second" (Horsley and Schäfer 1886). For the present view, tremor is considered to be not an unwanted byproduct of movement performance, or the exclusive result of the inertial and viscoelastic properties of moving body parts, but a necessary background for coordinated execution of movement (Elble 1996; Halliday and Redfearn 1956; Llinás 1991; Marshall and Walsh 1956).

The nictitating membrane/eyelid response has been used for many years as an experimental model for the study of motor learning (Gormezano et al. 1983; Thompson 1986; Woody 1986). A recent description (Gruart et al. 1995) of the kinetic properties of cat upper lid suggested the existence of a 25-Hz oscillator underlying reflex and conditioned responses (CRs). Human eyelid CRs present 25- to 30-Hz discontinuities in downward lid displacement (Marquis and Porter 1939), a fact also observed during conditioning of rabbit nictitating membrane (Welsh 1992) or upper lid (Gruart et al. 1997b) responses, and in the electromyographic (EMG) activity of the orbicularis oculi muscle during classically conditioning of eyelid blinks in rabbits (Berthier 1992), mostly when air puffs were used as conditioned (CS) and/or unconditioned (US) stimuli. A central neural support for those descriptions is the reported oscillatory behavior of pericruciate cortex neurons in the cat during the acquisition of conditioned blink movements (Aou et al. 1992) and of cerebellar interpositus neurons during reflex blinks also in cats (Gruart and Delgado-García 1994).

In spite of this convincing evidence, the possibility that eyelid movements might be organized in an oscillatory fashion has not been approached from an experimental point of view. For example, because attentive processes seem to be supported by a 40-Hz oscillator (Llinás and Ribary 1993), the lid motor system has to be fast enough to accomplish its motor tasks (emotional expression, corneal protection, etc.) without compromising vision. In this sense, blinks seem to accompany spontaneous eye saccades (Gordon 1951), particularly all saccade gaze shifts >15° (Evinger et al. 1994), and have been related to the resetting of visual information during alertness (Evinger et al. 1984; Gruart et al. 1995). On the other hand, the kinetics of this motor system imposes that blink amplitude can only be increased at the expense of a proportional increase in blink peak velocity (see Gruart et al. 1995 for references). This limitation suggests that a given angular trajectory of the lid can be accomplished by a single, fast downward lid displacement, as already reported for reflex blinks, or by a succession of downward steps, in a staircase manner, as recently described for cat eyelid CRs (Gruart et al. 1995).

The eyelid motor system is load free, has a negligible mass (i.e., less inertial damping), and is, apparently, free of proprioceptors (Porter et al. 1989). This latter point is very important because the system's oscillatory properties cannot be ascribed to local feedback circuits. Lid movement can be precisely recorded with the magnetic field search coil technique for a long time in chronic alert behaving animals, as well as the EMG activity of the orbicularis oculi muscle (Gruart and Delgado-García 1994; Gruart et al. 1995). In a previous paper we have described the kinetics of spontaneous, reflex and conditioned eyelid responses (Gruart et al. 1995). In the present paper we study their frequency-domain properties. It was found that both reflex and conditioned eyelid responses present movement discontinuities at a dominant frequency of 20 Hz. Moreover, CRs seemed to be generated from a basic 50-ms downward wave or sag. This quantum of movement was successively added in a staircaselike manner during conditioning sessions until reaching a more-or-less damped downward ramp movement of the lid, i.e., a complete CR. It was concluded that a 20-Hz oscillator underlies the generation of reflex blinks and allows the acquisition of entirely new motor lid responses in a quantal manner. Part of this work has been presented in abstract form (Domingo et al. 1996).

	METHODS

Abstract Introduction Methods Results Discussion References

Subjects

A total of 14 adult female cats, obtained from an authorized supplier (Iffa-Credo), were used as experimental subjects. Animals were prepared for the chronic bilateral recording of upper eyelid displacements and EMG activity of the orbicularis oculi muscle. Animals were handled in accordance with the guidelines of the European Union Council (86/609/EU) and following current Spanish legislation (BOE 67/8509-8512 1988) for the use of laboratory animals in chronic experiments.

Surgical procedures

Animals were anesthetized with pentobarbital sodium (Nembutal, 35 mg/kg ip) following a protective injection of 0.5 mg/kg of atropine sulphate to prevent unwanted vagal reflexes. A five-turn coil was implanted into the center of each upper eyelid at ~2 mm from the lid margin. Coils were made of Teflon-coated stainless steel wire, with an external diameter of 50 µm. Coils weighed a mean of 12.3 mg (i.e., 1.5% of the cat's upper lid weight) and did not produce any noticeable lid drooping or impairment of lid movements. Animals were also implanted with bipolar hook electrodes in both orbicularis oculi muscles. These electrodes were made of the same wire as the coils, and bared 1 mm at the tip. One of the wires was aimed at the zygomatic subdivision of the facial nerve, 1-2 mm posterior to the external canthus, and the other was implanted in the upper lid, 1 mm medial to the coil. A 1-mm-diam silver electrode was attached to the skull as a ground. Finally, animals were provided with a head-holding system for stability and coil reference purposes during the recording sessions. Three bolts were anchored to the skull in the vertical stereotaxic plane with the aid of six small screws fastened to the bone. The bolts were cemented to the skull with an acrylic resin. Eyelid coil, EMG, and ground wire terminals were soldered to a nine-pin socket also cemented to the holding system. Further details of this chronic preparation have been described elsewhere (Gruart et al. 1995).

Recording sessions

Recording sessions (a maximum of 18 per animal) were carried out for 3 h/day, beginning 2 wk after surgery. The animal was lightly wrapped with an elastic bandage and placed in a foam-coated Perspex box located on the recording table. The animal's head was immobilized by attaching the holding system to a bar affixed to the table. Data illustrated in Figs. 1-3 were obtained during the first two recording sessions.

View larger version (20K): [in this window] [in a new window]   FIG. 1. Frequency-component analysis of experimentally evoked reflex blinks. A-C: upper eyelid displacement (EL), velocity (L) and acceleration (ËL), and electromyographic (EMG) activity of the orbicularis oculi muscle in response to air puff (A), flash (B), and tone (C) stimulations. Characteristics of applied stimuli are indicated at the top. Dashed lines in the bottom histograms represent the power spectra of the illustrated acceleration records, whereas continuous lines correspond to the mean power spectra averaged from 30 eyelid reflex responses to each stimulus. The 100% value for the illustrated power spectra corresponded to 5 × 107 (acceleration units, deg s2)2. Note the different Y-axis scale for A-C histograms. Calibrations in C are also for A and B.

Data illustrated in Figs. 4-9 were obtained from the classical conditioning of eyelid responses, using different conditioning paradigms. Animals were ascribed at random to one of five different conditioning paradigms. For one-half of the animals (n = 7), conditioning sessions were repeated for up to 12 days, being preceded by two habituation sessions and followed by two extinction sessions. The other animals were conditioned for 6 days, also preceded by two habituation sessions. After two extinction sessions these animals were conditioned again with a different paradigm for 6 days, not followed in this latter case by any extinction session (see below).

View larger version (21K): [in this window] [in a new window]   FIG. 4. Kinetic characteristics of conditioned eyelid responses (CRs) for 4 different conditioning paradigms. A-D: examples of eyelid position (EL), and acceleration (ËL), and of the electrical activity of the orbicularis oculi muscle (EMG) for CRs obtained (in 4 different animals, and during the 6th conditioning session) in response to the sole presentation of 4 different conditioning stimuli (CSs): A, tones (T); B, short (20 ms), strong (3 kg cm2) air puffs to the side ipsilateral to unconditioned stimulus (US) presentation (APi); C and D, short (20 ms), weak (0.8 kg cm2) air puffs presented to the side ipsilateral (C, api) or contralateral (D, apc) to the US. Histograms represent the mean power spectra for 10 CR records. The 100% value for the Y-axis of illustrated power spectra was 7 × 106 (deg s2)2. Calibrations in D are also for A-C.

Raw data (presented in Fig. 10) regarding lid movements following successful, de novo innervation of the orbicularis oculi muscle by a hypoglossal-facial anastomosis were collected from unpublished material of a recent work (Gruart et al. 1996). These recordings were made with similar eyelid coils and EMG electrodes to those reported for the present experiments.

View larger version (35K): [in this window] [in a new window]   FIG. 10. A comparison in the frequency domain of eyelid performance when innervated by the facial (control) or by the hypoglossal nerve (following a hypoglossal-facial anastomosis). A: from top to bottom are shown eyelid position (EL), and acceleration (ËL), and the electromyography of the orbicularis oculi muscle (EMG) during licking a few drops of milk. Records were taken 6 mo after an ipsilateral hypoglossal-facial anastomosis (see Gruart et al. 1996, for details). B: the continuous line indicates the mean power spectra of 30 acceleration records (1.024 s each) during licking. For comparison, the power spectra illustrated in Fig. 6B is also represented (- - -). The 100% value for the Y-axis was 2 × 106 (deg s2)2.

Special care was taken during recording sessions to avoid any discomfort to the experimental subjects. Indeed, the continuous monitoring of heart and respiratory rhythms with noninvasive procedures during the recording sessions yielded values not significantly different of those recorded with the cat resting in the arms of one of us.

Recording of eyelid movements and EMG activity of the orbicularis oculi muscle

Eyelid movements were recorded with the magnetic field search coil technique. As described elsewhere (Gruart and Delgado-García 1994; Gruart et al. 1995), eyelid coils were calibrated with the help of a transparent protractor placed sagittally to the head and with its center at the external canthus of the lids. For the sake of homogeneity, the gain of the recording system was set at 1 V = 10°. Lid opening ranged 38-42° for the 14 animals. The EMG activity of the orbicularis oculi muscle was recorded with differential amplifiers (AM 502, Tektronix) at a bandwidth of 10 Hz to 10 kHz.

Stimuli evoking reflex eyelid responses

Blinks were elicited with puffs of air directed to the cornea and tarsal skin, flashes of light, and tones. Air puffs were applied through the opening of a plastic pipette (3 mm diam) located 1 cm away from the cornea, at a lateral angle of 45° with respect to the central direction of gaze. Air puff duration ranged from 20 to 500 ms, and air pressure at the source was set at 0.5-3 kg cm². The precise onset of air puff stimuli to corneal and lid mechanoreceptors was determined with the help of a microphone located at the same distance as the animal's eye. The recorded signal was rectified, integrated, and fed into the computer as 1-V square pulses for latency measurements. Bright full-field flashes lasting for 1 ms were used as a blink-evoking visual stimulus. The xenon arc lamp was located 1 m in front of the animal. Tones of 600 or 6,000 Hz applied for 10-100 ms were used as acoustic stimuli. The loudspeaker was located 80 cm below the animal's head. Each stimulus modality was presented in blocks of 10-50 stimuli at variable intervals (5 ± 1 s). A minimum of 30 s was allowed between successive presentations of stimuli of different modalities.

Slow (ramp) eyelid movements were achieved by optokinetic stimulation in the vertical plane at constant velocity. The planetarium was located 25 cm above the animal's head and projected a random pattern of illuminated spots on a concave screen located 1 m in front of the animal's eyes. The planetarium was rotated in ramps of 0.1-10°/s in the upward or downward directions.

No experimental stimulus was used to evoke spontaneous eye blinks and "eyelid friendly displays" (Gruart et al. 1995). These eyelid motor responses were collected off-line for analysis from tape recordings with the help of a homemade window discriminator set to the detection of the initial fast increase in velocity of these eyelid responses.

Classical conditioning paradigms

The classical conditioning of eyelid movements was achieved by the use of one delayed (1) and four trace (2-5) conditioning paradigms. 1) Delayed tone-air puff (T-AP) paradigm: a 350-ms, 600-Hz, 90-dB tone was presented to the animal as CS. The tone was followed 250 ms from its onset by a 100-ms, 3-kg cm² air puff stimulation directed to the left cornea as US. The tone and air puff finished simultaneously. 2) Trace AP_i-AP paradigm: animals were presented with a short (20 ms), strong (3 kg cm²) air puff as CS, followed 250 ms later by the US described in 1. Both CS and US were applied to the same (left) cornea and eyelids. 3) Trace ap_i-AP paradigm: animals were presented with a short (20 ms), weak (0·8 kg cm²) air puff as CS, followed 250 ms later by the US described in 1. As for the AP_i-AP paradigm, both CS and US stimuli were presented to the left cornea. 4) Trace ap_c-AP paradigm: a short (20 ms), weak (0·8 kg cm²) air puff was presented to the contralateral (right) cornea as CS, followed 250 ms later by the US described in 1, presented to the left cornea. 5) Trace t-AP paradigm: in this case, animals were presented with a 20-ms, 600-Hz, 90-dB tone as CS, followed 250 ms later by the US indicated in 1. The use of multiple (n = 5) conditioning paradigms was intended to ascertain the different (if any) oscillatory properties of the eyelid motor system depending on the conditioning paradigm, and on the modality, intensity, and presentation side of stimuli used as a CS.

Each of the conditioning paradigms 1-4 (T-AP, AP_i-AP, ap_i-AP, ap_c-AP) was used to train three animals, and conditioning paradigm 5 (t-AP) a further two. An animal was considered to be conditioned when it was able to produce 95% of CRs per session to the CS-US paired presentation. This criterion was reached by the fourth or fifth conditioning session for T-AP and t-AP paradigms, and by the second to fourth session for AP_i-AP and ap_i-AP paradigms. Nevertheless, and as already reported (Gruart et al. 1995), animals trained with the ap_c-AP paradigm did not achieve >50% of conditioned responses along the conditioning paradigm. As indicated above, conditioning sessions were preceded by two habituation sessions and followed by two extinction sessions. Conditioning sessions were repeated up to 12 times in 7 of the animals and up to 6 in the other 7. The latter animals were reconditioned with an ap_i-AP paradigm at seven different CS-US interstimulus intervals (25, 50, 70, 150, 250, 500, and 750 ms) during six conditioning sessions.

In all cases, the conditioning session consisted of 12 blocks separated by a variable time interval (range 4-6 min). Each block consisted of 10 trials separated at random by intervals ranging from 20 to 40 s. In a trial within each block, the CS was presented alone, i.e., was not followed by the US. The complete conditioning session lasted for 2 h. Only the stimulus selected as CS was presented during habituation and extinction sessions, with the same number of blocks per session and trials per block, and with a similar random distribution of interblock and intertrial intervals.

As a control procedure, four additional animals prepared for experiments not reported here were trained, before any other experimentation, to the unpaired presentation of the same CS and US described in paradigms 1-4. The session consisted of a total of 120 CS plus 120 US presented alone and distributed at random at a variable (30 ± 10 s) interval. Each animal was presented with one of the different CSs used in paradigms 1-4. The US was the same as described above. Up to three to five sessions were carried out per animal. No CRs were observed in those circumstances.

Data collection and analysis

The horizontal and vertical position of both upper eyelids, the unrectified EMG activity of both orbicularis oculi muscles, and 1-V rectangular pulses corresponding to blink-evoking stimuli, or to CSs and USs presented during conditioning sessions were stored digitally on an eight-channel videotape recording system. Data were transferred off-line, through an A/D converter (CED 1401-plus), to a computer for quantitative analysis. Data were collected at a sampling rate of 1,000 Hz, with an amplitude resolution of 12 bits.

Available computer programs (Spike2 and SIGAVG from CED, MATLAB, and Corel Draw) were modified, and new programs were developed by one of us to display single, overlapping, averaged, and raster representations of eyelid position, velocity, and acceleration, and of the EMG activity of the orbicularis oculi muscle. These programs also allowed the quantification, with the aid of cursors, of lid movement and EMG parameters, such as latency, amplitude, duration, and peak and mean values of recorded responses. Velocity and acceleration traces were computed digitally as the first and second derivative of lid position records, following low-pass filtering of the data (3 dB cutoff at 50 Hz and a zero gain at 100 Hz).

The power of the spectral density function (i.e., the power spectrum) of selected data were calculated using a fast Fourier transform to define the relative strength of the different frequencies present in evoked lid displacements and EMG records. Standard procedures appropriate for continuous processes were used (Bendat and Piersol 1986). According to Newton's second law, the force applied by the orbicularis oculi muscle to close the palpebral fissure will be directly reflected in the acceleration of the lids (Marshall and Walsh 1956). On the other hand, acceleration recordings enhance the higher frequency components of the movement (Halliday and Redfearn 1956; Wessberg and Vallbo 1995). For these two reasons, power spectra were calculated from the acceleration recordings. Acceleration recordings were divided in 1.024-s segments, starting 100 ms in advance to the presentation of either the blink-evoking stimulus or the CS. Segments containing CRs were selected exclusively from those obtained during the sole presentation of the five different CSs. This design allowed the complete evoked (reflex or conditioned) lid response to be contained in the segment, with a spectral resolution of 0.97 Hz. A similar procedure was followed to determine the power spectra of EMG records illustrated in Fig. 6C. For ramp lid displacements during optokinetic stimulus, only segments free of compensatory fast phases were selected. Unless otherwise stated, illustrated power spectra correspond to the mean value of power spectra computed from 10 different acceleration (or EMG) segments. For comparative purposes, the 100% value of power spectra (Y-axis) in acceleration (deg s²)² or voltage (V)² units is indicated in each figure legend.

View larger version (32K): [in this window] [in a new window]   FIG. 6. Frequency- and time-domain analyses of eyelid conditioned responses (CRs). A: from top to bottom are illustrated the displacement (EL), velocity (L), and acceleration (ËL) of eyelid during a CR, and the electromyographic activity of the orbicularis oculi muscle (EMG). Illustrated records were obtained after the sole CS presentation (a short, weak ipsilateral air puff) during the 9th conditioning session, in an api-AP trace conditioning paradigm. B: mean power spectra from 10 acceleration profiles recorded during the 9th conditioning session for the same animal and paradigm illustrated in A. The 100% value in the Y-axis was 2 × 106 (deg s2)2. C: mean power spectra obtained from the 10 EMG records was acceleration profiles analyzed in B. The 100% value for the power spectra was 1·2 × 103 (mV)2. D: coherence spectra between acceleration (data illustrated in B) and EMG (data in C) records, normalized to 0 to 1 values. E: autocorrelation function of the acceleration record in A. Note peaks repeated at 50-ms intervals. FFT, fast Fourier transform.

Autocorrelation of acceleration recordings as illustrated in Fig. 6E, and cross-spectral values between acceleration and EMG recordings as illustrated in Fig. 6D, were calculated according to available statistical tools (Bendat and Piersol 1986). The coherence spectrum was normalized to a 0 to 1 scale. Statistical analyses were carried out using the SPSS/PC + package, for a statistical significance level of P = 0.05. Mean values are followed when necessary by their standard deviation (SD). Statistical differences of mean values were determined with the help of the Student's t-test or with the analysis of variance (ANOVA), according to the number of categories. Peaks of power spectra were tested with the ²-distributed test for spectral density functions.

	RESULTS

Abstract Introduction Methods Results Discussion References

The objectives of the present experiments were to analyze the frequency-domain and time-domain properties of spontaneous, reflex, and conditioned eyelid responses. Because the kinetics of cat blinks have been described recently (Gruart et al. 1995), they will be referred to only when necessary. Because this paper is concerned mostly with the study of the oscillations present in eyelid learned movements, no systematic presentation of the acquisition process or of the mechanisms of association will be made (for a review see Gormezano et al. 1983). The oscillatory properties of spontaneous and reflexively evoked eye blinks will be presented first, followed by a description of periodicities observed in eyelid profiles for different conditioning paradigms and during the acquisition of the CR. A comparison with lid rhythmicity following a successful hypoglossal-facial anastomosis will also be made.

Power spectra of reflexively evoked blinks

As illustrated in Fig. 1, a reflex blink consisted of a fast downward lid displacement followed by a longer-lasting, wavy upward phase. The initial fast downward movement and the later smaller sags were the result of the phasic contraction of the orbicularis oculi muscle. Air puff-induced blinks showed a significantly (P < 0.01) shorter (11.6 ± 0.8 ms, mean ± SD, n = 50, for 100 ms, 3 kg cm² air puffs) latency than those evoked by flashes of light (48.2 ± 5.9 ms, n = 50), or tones (51.3 ± 7.4 ms, n = 50). According to the position, velocity, and acceleration profiles illustrated in Fig. 1, and for the range of stimulus intensities used here, air puff-induced blinks exhibited a larger peak lid displacement and velocity than flash- and tone-evoked blinks.

The successive downward waves following the initial down phase of air puff-evoked blinks outlasted the duration of the stimulus and, when measured by hand from peaks in velocity profiles, showed a duration of 40-60 ms. The power spectra of the acceleration profile illustrated in Fig. 1A (dashed line) presented a significant (P < 0.01) peak at 20 Hz. The mean power spectra of 30 air puff-evoked blinks also presented a significant (P < 0.01) peak at 17-20 Hz, over a broadband of frequencies between 10 and 30 Hz. The mean power spectra of flash-evoked blinks showed a dominant peak (P < 0.01) at a lower (14-16 Hz) frequency. Blinks evoked reflexively by tone presentations seemed to oscillate at a dominant frequency (P < 0.05) of 9-11 Hz. It should be noted that owing to the difference in the number of successive waves following each type of reflexively evoked blink and to the noticeable difference in their acceleration profiles, the dominant frequency in the mean power spectra for air puff-induced blink was one order of magnitude larger than that for flash-evoked blinks, and three orders of magnitude larger than that corresponding to blinks evoked by tone presentations. Thus the latency, profile, kinematics, and frequency content of reflex blinks seemed to depend on the sensory channel used to evoke them.

Dependence of air puff-evoked blinks on stimulus parameters

As already described (Gruart et al. 1995; Manning and Evinger 1986) and further illustrated in Fig. 2, air puff-evoked blinks in the cat were highly dependent on the intensity and duration of the applied stimulus. A complete set of different air pressures (0.5, 1, 2, and 3 kg cm²) and durations (10, 20, 50, 100, 200, and 500 ms) was presented at random, for 10 times at least, to 6 of the animals. The increase in pressure of the applied stimulus decreased the latency of the blink from 15.3 ± 0.3 ms for 0.5-kg cm² air puffs to 11.6 ± 0.8 ms for 3-kg cm² air puffs. Maximum amplitude of the first downward component of lid reflex response increased by three to four times from the minimum (0.5 kg cm²) to the maximum (3 kg cm²) air pressures applied in these experiments. Peak velocity also increased three to five times for the same set of air pressures. As illustrated in Fig. 2B, for a fixed air pressure (3 kg cm²), the increase in stimulus duration did not modify the kinetics of the first down phase of the blink, only lengthening its duration.

View larger version (28K): [in this window] [in a new window]   FIG. 2. Frequency-component analysis of reflex blinks as a function of the duration and pressure of air puffs applied to the ipsilateral eye. A and B: eyelid position (EL) and acceleration (ËL) during reflex blinks in response to air puffs of increasing pressure (A) or duration (B). For homogeneity of the results, the stimulus was always triggered from the same eyelid position. Histogram in A shows the mean power spectra of 10 acceleration profiles recorded during air puff presentations for 20 ms and with 0.5 (1), 1 (2), 2 (3), and 3 (4) kg cm2 of air pressure. Histogram in B shows the mean power spectra of 10 acceleration profiles recorded during 3 kg cm2 air puffs of 10 (1), 20 (2), 50 (3), 100 (4), 150 (5), 200 (6), and 500 (7) ms. The 100% value for the Y-axis of illustrated power spectra was 1·5 × 107 (deg s2)2. Amplitude calibration for EL and ËL records in A is also for B.

Interestingly enough, the rise time (i.e., the time expended by the lid, during the 1st down phase, from the initial 10% to the final 90% of its total displacement) of the evoked-blinks was not significantly modified by the different air pressures and durations, and remained around a mean value of 17.7 ± 1.3 ms. This finding indicated that 1) the increased amplitude of the first down phase of eyelid blinks in response to stronger air puffs was almost exclusively supported by a corresponding increase in lid peak velocity, with a nonsignificant change in the rise time of the reflex response; and 2) further increases in blink duration in response to longer puffs of air had to be accomplished by the generation of new, subsequent downward waves (Fig. 2B).

Power spectra of acceleration profiles corresponding to reflex blinks evoked by puffs of air of increasing pressure also increased in their relative strength (Fig. 2A). In the same way, the increase in the duration of the reflex-evoking air puff also increased the height of the corresponding dominant frequency (20 Hz) power spectra until an optimal duration for the stimulus of 100-150 ms (Fig. 2B).

Power spectra of spontaneous eyelid responses

In the control conditions of the recording room, softly illuminated and with a 60-dB white noise, the animals produced spontaneous blinks at a frequency of 0.2-0.5 times min¹. At a slightly higher frequency (1-2 min¹) they also made a sort of peering lid movement, previously described as a "friendly eyelid display" (see Gruart et al. 1995 for references). These two types of spontaneous eyelid response (Fig. 3) were not the exclusive result of the motor activity of orbicularis oculi muscle fibers as checked in the EMG records (not illustrated). Furthermore, these spontaneous eyelid responses were rather variable in maximum amplitude, duration of the down phase, and profile. Consequently, their corresponding mean spectra displayed several peaks. Spontaneous blinks showed three peaks at 2-6, 12-13, and 24-26 Hz, whereas friendly eyelid displays appeared as composed of a broad frequency band of 10-30 Hz.

View larger version (19K): [in this window] [in a new window]   FIG. 3. Frequency-component analysis of spontaneous eyelid displacements. A and B: eyelid position (EL), velocity (L), and acceleration (ËL) records corresponding to a spontaneous blink (A), and to a friendly eyelid display (B). Histograms at the bottom illustrate the mean power spectra, for each behavior, averaged from >10 acceleration records. The 100% value for the Y-axis of the illustrated power spectra was 2 × 106(deg s2)2. Calibrations of the 2 sets of records are indicated in B.

Power spectra of slow lid movements during ramp optokinetic stimulation

As usual for our recording system, the animal's head was set 21° nose down with respect to the horizontal stereotaxic plane. In this position, and with the animal looking straight ahead, the lower margin of the upper lid was located 5-10° over the primary position of the eye by the activity of the levator palpebrae muscle, with almost no electrical activity in the orbicularis oculi muscle (Evinger et al. 1984; Fuchs et al. 1992; Gruart et al. 1995). Power spectra of 55 segments taken from this lid position in the absence of any noticeable eyelid blink or saccade demonstrated the presence of a significant (P < 0.01) peak at 19-20 Hz. It is important to indicate that this peak was almost four orders of magnitude smaller than that already described as present during air puff-evoked blinks.

The orbicularis oculi muscle was also inactive during slow, ramp lid displacements following upward and downward optokinetic stimuli (see also Fuchs et al. 1992; Gruart et al. 1995). As a consequence, the downward movement of the lid during optokinetic stimulation was completely passive, whereas upward lid displacement was produced by the active, tonic contraction of levator palpebrae muscle fibers. Power spectra of 40 segments taken from upward lid ramp displacements presented a dominant frequency of 18-20 Hz (P < 0.01), whereas a similar sample of downward lid movements showed a significant (P < 0.05) peak in the power spectra in the range 20-22 Hz (not illustrated). Nevertheless, the power spectra of the 20-Hz frequency observed during slow, ramp lid displacements were approximately two orders of magnitude smaller than those reported for air puff-evoked blinks. The small-amplitude periodic discontinuities present during straight ahead lid fixation and during slow, ramp eyelid displacement suggest that a 20-Hz oscillation is also present in the overall activity of levator palpebrae motoneurons. Alternatively, a rather small number of orbicularis oculi motor units (unnoticeable for our EMG recording electrodes) remained active during those lid responses.

To determine the resonant frequency of eyelids, the amplitude of lid displacement during 1-s trains of electrical stimulation of the zygomatic subdivision of the facial nerve (50-µs, square, cathodic pulses of <0.5 mA) at frequencies ranging from 1 to 200 Hz was checked in two of the animals. Power spectral density analysis indicated the presence of a significant (P < 0.001) peak at 26 Hz. By comparison, the power spectra appeared as almost completely flat for frequencies of stimulation >50, and <10 Hz (not illustrated). Accordingly, the cat upper lid seems to have a resonant frequency slightly above those reported here for reflex and conditioned eye blinks. The fusion frequency for the muscle was achieved at 60-70 pulses s¹.

Kinetic characteristics of eyelid conditioned responses

Representative profiles of eyelid CRs obtained during the sixth conditioning session for T-AP, AP_i-AP, ap_i-AP, and ap_c-AP paradigms are illustrated in Fig. 4, A-D. As already described (Gruart et al. 1995), latency, maximum amplitude, and profile of these CRs were very different, depending on the modality, intensity, and presentation side of the sensory cue used as CS (see Rescorla 1988). Thus CRs evoked by tones when used as CS presented smaller amplitudes and a smoother, ramplike profile than CRs evoked by ipsilateral short air puffs used as CS. The reason was that, in the latter case, the evoked CR presented a larger amplitude and a staircaselike appearance. It should be noted that the US and the CS-US intervals were the same for the four conditioning paradigms. Furthermore, the maximum amplitude of the four sorts of CR was reached at the time of US presentation.

The power spectra of acceleration profiles corresponding to the four types of CRs showed significant (P < 0.01) peaks of 20 Hz. The largest amplitude of these CRs was reached after the presentation of short (20 ms), strong (3 kg cm²) air puffs as CS (Fig. 4B). Also, these CRs displayed the highest peak in power spectra as compared with the other three types of CR, which was ~50 times larger than that measured when tones (T) were used as CS. The height of the 20-Hz peak decreased successively for short (20 ms), weak (0.8 kg cm²) air puffs ipsilateral to the US (ap_i, Fig. 4C); short (20 ms), weak (0.8 kg cm²) air puffs contralateral to the US (ap_c, Fig. 4D); and 350-ms, 600-Hz, 90-dB tones (T, Fig. 4A), used as CSs.

To rule out the possibility that the noticeable differences in CR profiles, when using tones or short air puffs as CSs, were related to the different conditioning paradigms (i.e., "delayed" for tones and "trace" for short air puffs), an additional conditioning paradigm was designed. This new tone, trace paradigm (t-AP) consisted of a 20-ms, 600-Hz, 90-dB tone as CS, followed 250 ms later by the same US stimulus as for the other paradigms, that is, a long (100 ms), strong (3 kg cm²) air puff. Acquired CRs and their profiles were similar to those obtained during delayed T-AP as described above. Power spectra of 10 acceleration segments selected from the sixth conditioning session of this t-AP trace paradigm also presented a significant peak at 20 Hz, of a height similar to those observed for T-AP, delayed paradigms (not illustrated).

Figure 5 illustrates the variability in the frequency component of the CR of different animals for T-AP, AP_i-AP, ap_i-AP, and ap_c-AP conditioning paradigms. Although the power spectra demonstrated a significant (P < 0.05 at least) peak at 20 Hz for all the animals (n = 3 for each paradigm), a considerable variability in height was observed for the different paradigms. Also, broad frequency bands were observed for some of the animals, a fact that could be ascribed to the paucity of the experimental data (n  10) that had to be collected from the same experimental session, and in response to the sole CS presentation.

View larger version (20K): [in this window] [in a new window]   FIG. 5. Power spectra of the acceleration records for conditioned responses (CRs) obtained in 4 different conditioning paradigms. A-D: histograms representing the mean power spectra of 10 CRs obtained during the sole presentation of the corresponding conditioning stimulus (CS) during the 5th conditioning session. Each record is for a different animal. The 4 different CS were A, 350 ms, 600 Hz, 90 dB tones; B, short (20 ms), strong (3 kg cm2) air puffs ipsilateral to unconditioned stimulus (US) presentation; and C and D, short (20 ms), weak (0.8 kg cm2) air puffs ipsilateral (C) or contralateral (D) to US presentation. As explained in METHODS, T-AP conditioning illustrated in A corresponded to a delayed paradigm, whereas APi-AP (B), api-AP (C), and apc-AP (D) conditionings corresponded to trace paradigms. The 100% value for the Y-axis of illustrated power spectra was 7 × 106 (deg s2)2.

Frequency- and time-domain analyses of conditioned eyelid and orbicularis oculi EMG responses

Cross-correlation studies between the unrectified EMG activity of the orbicularis oculi muscle and acceleration profiles corresponding to reflex and CRs evoked by air puffs showed that the electrical activity of muscle fibers preceded the negative acceleration by 5.5 ms (99% confidence interval, 4.9-6.1 ms). No difference in latency was observed between reflex and CRs (not illustrated). These values are in agreement with a previous report of EMG activity of the orbicularis oculi preceding movement initiation by 4 ms (Gruart et al. 1995).

As illustrated in Fig. 6, the power spectra of selected segments of EMG activity of the orbicularis oculi muscle during the performance of CRs evoked by short (20 ms), weak (0.8 kg cm²) air puffs (ap_i) used as CS also showed a dominant peak at 20 Hz, accompanied by other peaks at lower (14 Hz) and higher frequencies. The fact that the EMG signal was not filtered as eyelid position profiles left them with high-frequency components not present in the eyelid acceleration signal. In any case, the coherence between acceleration and EMG power spectra from CRs evoked during T-AP, AP_i-AP, and ap_i-AP conditioning paradigms (n = 1 animal each) was significant (99% confidence limit) for frequencies >10 Hz.

The rhythmicity of acceleration profiles (i.e., their variability from eyelid sag to sag) corresponding to different types of CR was also checked. As illustrated in Fig. 6E, the autocorrelation function of acceleration records showed repeated peaks at 50-ms intervals. Nevertheless, the autocorrelation time of the signal seemed to be rather short, a fact that could be ascribed to the shortness of the evoked response, and/or to the presence of a random variation in the duration of successive eyelid downward sags or waves (however, see Fig. 9, B and C).

View larger version (23K): [in this window] [in a new window]   FIG. 9. Variability in frequency and peak velocity of the successive downward steps composing a conditioned response (CR). A: 2 eyelid CRs (EL) and their corresponding velocity profiles (L) obtained during an api-AP trace conditioning paradigm. These CRs were evoked with the sole presentation of the CS consisting of a short (20 ms), weak (0.8 kg cm2) air puff to the eye ipsilateral to the US presentation. Note variability in peak velocity and frequency during the CR. B: evolution of sag frequency (in Hz) during the CR as a function of time from CS presentation. Frequency was computed as the reciprocal of the time interval (in s) between successive velocity peaks. C: evolution of peak velocity of the successive eyelid downward sags as a function of time from CS presentation. The equation corresponding to the exponential line best fitting the data are shown. For both B and C, illustrated data were obtained from 10 different CRs.

Evolution of eyelid conditioned-response profiles during successive conditioning sessions

It was noticed through conditioning sessions that CRs did not occur on an all-or-none fashion, but that they followed a slow building up that allowed their successive increase in amplitude and time. When short air puffs (AP_i, ap_i, and ap_c) or tones (T, t) were used as CSs, the CR appeared as a small downward lid movement that (usually) radiated from its onset, close to the beginning of the CS, toward the US, where it reached its maximum amplitude. As illustrated in Fig. 7B, for nine conditioning sessions with an ap_i-AP paradigm, CRs seemed to increase by the addition of waves, or quantum of lid displacement, which added to the downward movement produced by the preceding one. In this way, the profile of CRs showed evident discontinuities, producing a staircase profile. The increase in the number of waves present in CRs during successive conditioning sessions produced a corresponding increase in the height of the dominant 20-Hz component (see right set of records in Fig. 7B). Power spectra of habituation and extinction sessions also illustrate the transformation undergone by acceleration records as a consequence of the reduction in the number of components (i.e., quanta or waves) of habituated or extinguished CRs (Fig. 7, A and C).

View larger version (21K): [in this window] [in a new window]   FIG. 7. Raster display of session evolution during the unpaired or paired presentation of 2 successive air puffs as conditioned (CS) and unconditioned (US) stimuli. A: 2 eyelid responses recorded during 2 sessions of habituation to a stimulus consisting of a 20-ms, 0.8-kg cm2 air puff. Trial number of each record (T) is indicated. Records at the right illustrate mean power spectra of 10 acceleration profiles obtained during each habituation session. B: sample of eyelid CRs evoked by the sole CS presentation during 9 conditioning sessions. The CS consisted of the above-mentioned short, weak air puff followed 250 ms later by a long (100 ms), strong (3 kg cm2) air puff as a US. As for A, the trial number of each eyelid position record is indicated (left set of records) as well as the mean power spectra of 10 acceleration records from each conditioning session (right set of records). C: at the left are illustrated 2 CRs during 2 extinction sessions. The mean power spectra of 10 acceleration records during those sessions are illustrated at the right. The 100% power spectra value for the histograms illustrated in the figure was 3·5 × 106 (deg s2)2. Calibration for eyelid position traces is presented in C.

The intrinsic neural properties that characterized the formation of CRs are further analyzed in Fig. 8. It was observed that when CRs were grouped according to the number of quanta or waves present in the CR profile (Fig. 8A), the power spectra for frequencies 20 Hz increased in height (and in significance with respect to surrounding 20 frequencies) with the number of waves. It should be noted that this precise elaboration of the CR is entirely from a central origin, because no stimulus was present during the CS-US interval during trace conditioning paradigms (Figs. 6A, 7B, and 8A).

View larger version (25K): [in this window] [in a new window]   FIG. 8. Frequency characteristics and kinetics of eyelid conditioned responses (CRs) as a function of CS-US repetition and interval. A: at the top is shown a sample of four eyelid CRs and their corresponding acceleration profiles obtained, from top to bottom, during successive conditioning sessions. Records correspond to the sole presentation of a CS consisting of a short (20 ms), weak (0·8 kg cm2) air puff. Note the successive increase in the number of downward steps or sags with training. The bottom histogram illustrates mean power spectra of 4 sets of 10 acceleration records for CRs showing 1-4 downward sags (numbered 1-4 in the histogram). B: at the top are shown 4 CRs of the same total amplitude obtained during api-AP trace conditioning with different CS-US interstimulus intervals. Illustrated CRs correspond to the sole presentation of the CS (a short, weak air puff). The corresponding acceleration records are arranged from the shortest (top) to the longest (bottom) CS-US interval. Note the decrease in peak amplitude of acceleration profiles and the increase in the number of acceleration waves crossing the middle line. The histogram at the bottom illustrates the mean power spectra of >10 acceleration records computed from CRs evoked with the following CS-US intervals: 1, 750 ms; 2, 500 ms; 3, 250 ms; 4, 150 ms; and 5, 70 ms. The 100% value for the Y-axis was 6 × 106 (deg s2)2. Calibrations in B are also for A.

As explained in METHODS, seven animals were reconditioned again with the same trace paradigm ap_i-AP but at different CS-US time intervals (25, 50, 70, 150, 250, 500, and 750 ms). For the shorter (25, 50 ms) intervals, it was not possible to obtain any conditioning of the two animals. For the other intervals (>50 ms), the conditioning criterion was reached at the fourth through sixth conditioning session. In this situation, CRs showed similar amplitudes, also reaching their maximum displacement values at the time of US presentation, but their durations were obviously different, in function of the selected CS-US interval. Thus, and as illustrated in Fig. 8B, it was observed that the duration of waves composing each CR was 50 ms, independently of the selected interval, i.e., the duration of the waves was not related to the duration of the interval. Apparently, for very short intervals (25 and 50 ms) there was not enough time to generate even a single wave (or quantum) to start the elaboration of a proper CR, and, as a consequence, no CR was observed for these two trace ap_i-AP paradigms (not illustrated). It was also found that the "optimum" definition of the 20-Hz component of the power spectra was achieved for 150 and 250 ms CS-US interstimulus intervals; that is, those values demonstrated by others as the most favorable for the classical conditioning of the nictitating membrane/eyelid response (Gormezano et al. 1983; Rescorla 1988).

Internal structure of a conditioned response

The internal structure of well-defined CRs, selected mostly from those performed after the criterion was reached, is illustrated in Fig. 9. The succession of small (2-6° in amplitude) waves that composed a CR showed no significant tendencies with respect to their duration or that of the duration reciprocal, the instantaneous frequency (Fig. 9B).

The mean velocity of CRs depends on the CS-US interval, and also on the stimulus sensory modality, intensity, and presentation side (Fig. 4). Although the mean velocity of CR responses for ap_i-AP paradigms was 50°/s, for 250 CS-US intervals (see Fig. 6A), the peak velocity of individual waves within a given CR was variable, reaching up to 300°/s during ap_i-AP paradigms at 250 CS-US intervals (Figs. 6A and 9C), and up to 500°/s during AP_i-AP paradigms at the CS-US interval checked in the present experiments (250 ms), and during ap_i-AP paradigms at short (70, 100 ms) CS-US intervals (Fig. 8B). Within a given CR, the successive waves seemed to decrease progressively in peak velocity. Data quantified from four different CRs obtained from ap_i-AP paradigms showed an exponential decay in the peak velocity of the successive waves or sags that formed the CR profile (Fig. 9C).

Lid kinetic and frequency-domain properties following a hypoglossal-facial anastomosis

To further prove that lid oscillatory responses reported here are the result of motor commands arriving to the orbicularis oculi muscle, unpublished records corresponding to a recent work from our group (Gruart et al. 1996) were reanalyzed. Data shown in Fig. 10 illustrate lid position and acceleration profiles, as well as the EMG activity of the orbicularis oculi, while the animal licks a few drops of milk delivered from a nearby pipette. Records were taken 6 mo after left hypoglossal-facial anastomosis. The fast Fourier transform analysis of 1.024-s acceleration records showed a significant (P < 0.01) peak at 5 Hz, similar to previous descriptions for rat licks (7-8 Hz) (Welsh et al. 1995), and quite different from the frequency components present in lid acceleration profiles during reflex and conditioned eyelid responses (see Fig. 10B).

	DISCUSSION

Abstract Introduction Methods Results Discussion References

Discontinuous versus continuous nature of eyelid movements

Frequency-domain data collected in the present experiments, together with a recent study of eyelid kinetics (Gruart et al. 1995), allow some insights into how reflex and conditioned motor responses are generated. With the animal looking straight ahead, the lid presented a principal frequency oscillation of 20 Hz almost unnoticeable in eyelid position records. This 20-Hz dominant frequency became more evident during upward and downward eyelid displacements following ramp, constant optokinetic stimulus. Lid optokinetic responses presented very low (a few deg s¹) mean velocities. In contrast, air puff-evoked blinks presented a peak velocity of up to 2,000°/s during the first down phase of the blink (see also Gruart et al. 1995) and its power spectra presented a well-defined principal frequency component of 20 Hz. These results indicate that the lid performs these two reflex responses with different kinetic requirements within the same frequency-domain framework. The initial large movement required for the first down phase of a reflex blink is accomplished within the same time window (50 ms) as the tiny oscillations observed during optokinetic eyelid ramp movements. On the other hand, a complete blink reflex response to a puff of air presented, after the initial, fast downward displacement, successive small downward waves or sags, resulting from the active and precise contraction of (a few) orbicularis oculi motor units at the same period (50 ms), and as a function of the duration of the stimulus. Thus the basic framework for lid movements consists of the selective interplay of larger or shorter lid displacements as a function of the number of motor unit simultaneously active within a time window of 50 ms, and ofthe repetitive activation of a few motor units at 20 cycles/s.Accordingly, data presented here suggest a possible quantal nature of reflex and conditioned eyelid responses. In fact, a recent report on the firing activities of identified facial motoneurons indicates that a single motor unit is able to evoke a downward lid movement of 2° (Trigo et al. 1997), that is, an amplitude similar to those reported here for minimum reflex and conditioned eyelid responses (see Figs. 2A and 8A).

Previous studies in other motor systems support the above contentions. By definition, a ramp (i.e., as during reaching) continuous movement should present, from its beginning to its end, a sole velocity peak and only one zero crossing of the acceleration record (Brooks 1974). In practice, slow elbow, wrist, and finger extension, or flexion, movements appear clearly as discontinuous; i.e., they are composed of successive, more-or-less fused, twitches that generate repeated peaks in the velocity profile interrupted by periods without movement (Brooks 1974; Wessberg and Vallbo 1995). The discontinuity of movement execution is made more evident during learning of a new motor task, or during ramplike displacements conditioned to the presentation of sensory cues for its initiation and conclusion. Even apparently continuous movements, such as fast alternations of body appendages (hand, arm), present some discontinuities in their acceleration profiles, and, in whatever case, cannot be repeated above a determined maximum frequency characteristic for each motor system (Llinás 1991). There appears to be a phase relationship between movement initiation and physiological tremor, as it has been shown that finger movements are initiated at points of maximum angular velocity during the oscillation, that is, at the moment of greatest kinetic energy (Goodman and Kelso 1983).

Quantal mechanisms underlying eyelid responses

As reported for different species of mammals (Evinger et al. 1984, 1991; Gruart et al. 1995; Guitton et al. 1991; Manning and Evinger 1986) the down phase of a reflex blink is linearly related to its peak velocity but shows no relation with the duration of the movement. Because the rise time of the down phase remains constant, changes in blink amplitude seem to be the exclusive result of proportional changes in blink velocity. In other words, the initial phase of air puff-evoked blinks can be modified in amplitude, depending on stimulus strength, without any modification of its total duration that remains around 50 ms. Double pulse (10 ms of pulse duration) air puff stimulation in alert cats showed a facilitatory response at an interval of 50 ms (Gruart et al. 1995).

Motor strategies aimed at the generation of a CR also have to take advantage of the same neural motor design. A consolidated CR could be considered the optimal solution of a temporal-spatial problem. The lids have to be close together at the moment of US presentation. According to the present results, CRs are elaborated in the same step-wise manner as the late components of reflexively evoked blinks; that is, successive waves of 2-5° in amplitude and 50 ms in duration are generated increasingly throughout conditioning sessions until the appropriate motor response is reached. Here again, the number of waves and their amplitude seem to be a function of the stimulus modality, intensity, and presentation side of the CS, and of the CS-US interval (Gormezano et al. 1983; Gruart et al. 1995; Rescorla 1988). The effect on CR profiles of different US has not been considered here, but has also been reported from a kinetic point of view (see Gormezano et al. 1983; Rescorla 1988 for references).

The fact that CRs depend on both kinetics and frequency-domain properties could be hypothesized as that they are dependent on the peculiar organization of the pathways involved. When short air puffs are used as CS, the stimulus probably arrives, by intermediation of second-order trigeminal neurons, at large, phasic motoneurons, while when tones were presented as CS, the stimulus arrives, by a polysynaptic pathway, at small, tonic motoneurons. The presence of both pools of motoneurons within the orbicularis oculi subdivision (i.e., the dorsolateral complex of the facial nucleus) has been proved on both morphological and electrophysiological grounds (Fanardjian and Manvelyan 1987; Shaw and Baker 1985), and is also reflected in the functional properties and distribution of motor units within the orbicularis oculi muscle (Gordon 1951), with the more phasic motor units placed in the tarsal lid and the tonic ones occupying a more peripheral location. The more direct access to phasic units from trigeminal inputs could be explained by their more direct involvement in reflex blinks, while tones are behaviorally more prone to evoking emotional and socially relevant eyelid responses, which require a more precise performance. This hypothesis also explains the fact that tone-evoked CRs present both a shorter maximum amplitude as compared (for the same conditioning session) with air puff-evoked CRs, and a less-defined 20-Hz component, as tonic motor units fire at frequencies close to or even above the fusion frequency of the orbicularis muscle (Gordon 1951 and the present results). Tone-evoked CRs also presented a longer latency because of the polysynaptic pathway to facial motoneurons. As a result, tone-evoked CRs (both during delayed and trace paradigms) showed a smoother profile than air puff-evoked ones (see also Gruart et al. 1995). It can be proposed, then, that the latency, kinematics and frequency-domain characteristics of a learned movement depend on the sensory channel used to generate it.

According to the present results, lid movements (mostly those evoked by trigeminal inputs) are the consequence of a 20-Hz oscillator in which both the amplitude of a given wave or sag and the repetitive sequence of 50-ms waves can be precisely controlled by the CNS. For reflex blinks, the amplitude of the initial down phase can be modified increasing its peak velocity, but further increases in the duration of the response are achieved by adding more 50-ms quanta of downward waves or sags. For CRs, the motor strategy appeared to be different, probably because, by definition, CSs have to be unable to produce an initial strong eyelid response. Moreover, for trace conditioning, the CR has to be elaborated in the absence of any sensory input; that is, with the exclusive participation of central neural processes. In this situation, the quantal nature of movement acquisition became more evident (see Figs. 7 and 8), as well-defined CRs were achieved during successive CS-US trials, by the adding of an increasing number of small, downward lid movements until the desired motor response was achieved; that is, to have the lid closed at the moment of US presentation. It could be proposed that neural mechanisms are organized in such a manner as to be able to regulate both the duration and the intensity of the neural volley. In this way, both amplitude and duration of the lid displacement can be precisely controlled during the learning process. An alternative suggestion is that, in general, neuronal circuits have to be tuned to the mechanic properties of the systems they must move. In this sense, the dominant frequency of a given circuit should increase as body size decreases, or the metabolic rate increases. Indeed, available data show that the maximum spectral power of reflex blinks occurred at 7.8 Hz in humans (C. Evinger, personal communication), 10-15 Hz in rabbits (Berthier 1992; Gruart et al. 1997b; Welsh 1992), 20 Hz in cats (present results), and 29.29 Hz in guinea pigs (C. Evinger, personal communication).

Origin of the 20-Hz oscillation underlying reflex and conditioned eyelid responses

Llinás (1991) has provided elegant evidence for a role of the inferior olive in the coherent organization of movement on the basis of the 10-Hz, intrinsic oscillatory properties of inferior olive neurons. In his view, the climbing fiber system operating in cerebellar circuits will act as a timing device able to synchronize the appropriate onset of muscle activity during movement. Nevertheless, different motor systems seem to have different oscillation frequencies, a fact that is related not only to their resonant properties (i.e., resonant frequency is inversely proportional to the square root of the mass), but also to their function. For example, 6-7 licks/s have been described in rat licking (Welsh et al. 1995), 8- to 10-Hz cycles are present in voluntary finger movements in humans (Wessberg and Vallbo 1995), synchronous 30-Hz EMG activity in cat neck muscles have been recorded during paw licking (Loeb et al. 1987), and the eye seems to oscillate at 40-70 Hz (see Marshall and Walsh 1956). Such a broadband of oscillatory frequencies suggests different sources governing action tremor, but at the same time reinforces a central neural control of it. The fact that lid principal oscillation frequency was dramatically modified after a hypoglossal-facial anastomosis (Gruart et al. 1996 and the present experiments) further reinforces the suggestion that there are different oscillatory systems controlling motor responses in accordance with their functional needs.

Gordon (1951) reported that orbicularis oculi motor units in humans can be divided in two groups. One group consists of phasic motor units, preferentially located under the tarsal skin, and able to fire in short bursts of 1-3 spikes at a peak frequency of up to 180 spikes/s. The other group, located in more peripheral orbital areas, is composed of tonic fibers capable of a sustained firing of >50 spikes/s. A third, intermediate group has also been described (Gordon 1951; Van der Werf et al. 1996). Phasic units seem more involved in reflex, fast lid responses, whereas tonic units are active during lid movements related to emotional and attentive states. The uneven distribution of phasic and tonic motor units (McLoon and Wirtschafter 1991) could explain why EMG records from the orbicularis oculi muscle do not exactly match lid acceleration profiles and suggest that frequency-domain analysis should be better made from the total output of the muscle, that is from actual lid displacement. On the basis of both morphological (Porter et al. 1989) and physiological (Gruart et al. 1996) experiments, it has been proposed that orbicularis oculi muscle lacks the proprioceptors, so the rhythmic activity present in its motor units has to be entirely of a neural central origin.

As shown with the present data and confirmed by the above reports, the eyelid motor system seems to work at frequencies higher than that reported for the characteristic 10-Hz activity of the inferior olive/climbing fiber system (Llinás 1991; Welsh et al. 1995). Nevertheless, oscillatory responses at frequencies matching those reported here have been recorded in other CNS structures. For example, the so-called intralaminar thalamic nuclei "fixation" neurons fire at a sustained rate of 20 spikes/s during attentive states in monkeys (Schlag and Schlag-Rey 1986). Synchronous 25- to 35-Hz oscillations have also been recorded in the sensorimotor cortex and related areas of rats (Nicolelis et al. 1995), cats (Bouyer et al. 1987), and monkeys (Murthy and Fetz 1992; Sanes and Donoghue 1993). There are also early reports of a 40-cycle/s electroencephalographic activity in the amygdala-prepyriform area related to arousal states in cats (Pagano and Gault 1964). Blinks are related to the generation of spontaneous eye saccades in humans (Evinger et al. 1994; Gordon 1951) and could accompany the resetting of visual information during the awake state (Evinger et al. 1984; Gruart et al. 1995). In this line of thought, some reflex and voluntary lid movements in humans could be synchronized with the 40-Hz coherent magnetic activity recorded in humans during awareness (Llinás and Ribary 1993).

This cortical and basal nuclei information seems to reach facial motoneurons by polysynaptic (motor cortex, superior colliculus, amygdala complex) and monosynaptic (amygdala) pathways (Basso et al. 1996; Fanardjian and Manvelyan 1987). Recently, it has been reported that amygdala stimulation enhances reflex eyelid responses in the rat through a short-latency mechanism (Canli and Brown 1996). Thus lid responses, because of their narrow relationship with attentive (mostly of visual origin) processes and with the precise expression of emotional states, apparently need to be controlled by a fast (20-Hz) oscillatory neural motor system.

The putative neural site where learning of new eyelid motor responses takes place has been a subject of controversy in recent years (see Aou et al. 1992; Bloedel 1992; Thompson 1986). The oscillatory properties of the eyelid motor system seems to preclude the inferior olive/climbing fiber system as a neural carrier of such high-frequency messages. However, cat interpositus neuron firing follows in a time-locked manner the recurring waves composing reflex and conditioned blinks (Gruart and Delgado-García 1994; Gruart et al. 1997a). It could be suggested that these fast (20 Hz) efferent-copy commands arrive at cerebellar structures via mossy fiber afferents. On the other hand, it has been reported that pericruciate cortex neurons fire at frequencies high enough (25-30 Hz) and latencies short enough (Aou et al. 1992) to be involved in the generation and/or modulation of eyelid learned responses, and that amygdala neurons have direct (i.e., monosynaptic) access to facial motoneurons (Canli and Brown 1996; Fanardjian and Manvelyan 1987). As a consequence, cortical and amygdalar structures should be considered for their putative involvement in learned eyelid responses in future experiments.

	ACKNOWLEDGEMENTS

  We thank R. Churchill for help in the editing of the manuscript.

  This work was supported by grants from the Spanish Comisión Interministerial de Ciencia y Tecnologia (SAF 96-0160) and Dirección General de Investigación Científica y Técnica (PB93-1175) and Junta de Andalucía (PAI-3045).

	FOOTNOTES

  Address for reprint requests: J. M. Delgado-García, Laboratorio de Neurociencia, Facultad de Biología, Avda. Reina Mercedes 6, 41012 Sevilla, Spain.

  Received 3 March 1997; accepted in final form 14 July 1997.

	REFERENCES

Abstract Introduction Methods Results Discussion References

• AOU, S., WOODY, C. D., BIRT, D. Changes in the activity of units of the cat motor cortex with rapid conditioning and extinction of a compound eye blink movement. J. Neurosci. 12: 549-559, 1992.[Abstract]
• BASSO, M. A., POWERS, A. S., EVINGER, C. An explanation for reflex blink hyperexcitability in Parkison's disease. I. Superior colliculus. J. Neurosci. 16: 7308-7317, 1996.[Abstract/Free Full Text]
• BENDAT, J. S., PIERSOL, A. G. In: Random Data., . New York: Wiley, 1986
• BERTHIER, N. E. Muscle activity during unconditioned and conditioned eye blinks in the rabbit. Behav. Brain Res. 48: 21-28, 1992.[Medline]
• BLOEDEL, J. R. Functional heterogeneity with structural homogeneity: how doas the cerebellum operate? Behav. Brain Sci. 15: 666-678, 1992.
• BOUYER, J. J., MONTARON, F., WAHNEE, J. M., ALBERT, M. P., ROUGEUL, A. Anatomical localization of cortical beta rhythms in cat. Neuroscience 22: 863-869, 1987.[Medline]
• BROOKS, V. B. Some examples of programmed limb movements. Brain Res. 71: 299-308, 1974.[Medline]
• CANLI, T., BROWN, T. H. Amygdala stimulation enhances the rat eyeblink reflex through a short-latency mechanism. Behav. Neurosci. 110: 51-59, 1996.[Medline]
• DOMINGO, J. A., GRUART, A., DELGADO-GARCÍA, J. M. A 20-Hz oscillator underlies the acquisition of conditioned eyelid responses. Soc. Neurosci. Abstr. 22: 1875, 1996.
• ELBLE, R. J. Central mechanisms of tremor. J. Clin. Neurophysiol. 13: 133-144, 1996.[Medline]
• EVINGER, C., MANNING, K. A., PELLEGRINI, J. J., BASSO, M. A., POWERS, A. S., SIBONY, P. A. Not looking while leaping: the linkage of blinking and saccadic gaze shifts. Exp. Brain Res. 100: 337-344, 1994.[Medline]
• EVINGER, C., MANNING, K. A., SIBONY, P. A. Eyelid movements. Mechanisms and normal data. Invest. Ophthalmol. Vis. Sci. 32: 387-400, 1991.[Abstract/Free Full Text]
• EVINGER, C., SHAW, M. D., PECK, C. K., MANNING, K. A., BAKER, R. Blinking and associated eye movements in humans, guinea pigs, and rabbits. J. Neurophysiol. 52: 323-339, 1984.[Abstract/Free Full Text]
• FANARDJIAN, V. V., MANVELYAN, L. R. Mechanisms regulating the activity of facial nucleus motoneurons. III. Synaptic influences from the cerebral cortex and subcortical structures. Neuroscience 20: 835-843, 1987.[Medline]
• FUCHS, A. F., BECKER, W., LING, L., LANGER, T. P., KANEKO, C.R.S. Discharge patterns of levator palpebrae superioris motoneurons during vertical lid and eye movements in the monkey. J. Neurophysiol. 68: 233-243, 1992.[Abstract/Free Full Text]
• GOODMAN, D., KELSO, J.A.S. Exploring the functional significance of physiological tremor: a biospectroscopic approach. Exp. Brain Res. 49: 419-431, 1983.[Medline]
• GORDON, G. Observations upon the movements of the eyelids. Br. J. Ophthalmol. 35: 339-351, 1951.
• GORMEZANO, I., KEHOE, E. J., MARSHALL, B. S. Twenty years of classical conditioning research with the rabbit. Prog. Psychobiol. Physiol. Psychol. 10: 197-275, 1983.
• GRUART, A., BLÁZQUEZ, P., DELGADO-GARCÍA, J. M. Kinematics of spontaneous, reflex, and conditioned eyelid movements in the alert cat. J. Neurophysiol. 74: 226-248, 1995.[Abstract/Free Full Text]
• GRUART, A., DELGADO-GARCÍA, J. M. Discharge of identified deep cerebellar nuclei neurons related to eye blinks in the alert cat. Neuroscience 61: 665-681, 1994.[Medline]
• GRUART, A., GUNKEL, A., NEISS, W. F., ANGELOV, D. N., STENNERT, E., DELGADO-GARCÍA, J. M. Changes in eye blink responses following hypoglossal-facial anastomosis in the cat: evidence of adult mammal motoneuron unadaptability to new motor tasks. Neuroscience 73: 233-247, 1996.[Medline]
• GRUART, A., PASTOR, A. M., ARMENGOL, J. A., DELGADO-GARCÍA, J. M. Involvement of cerebellar cortex and nuclei in the genesis and control of unconditioned and conditioned eyelid motor responses. In: The Cerebellum: From Structure to Control, edited by C. I. de Zeeuw, P. Strata, and J. Voogd . Amsterdam: Elsevier, 1997a, p. 511-528
• GRUART, A., SCHREURS, B. G., DOMINGUEZ DEL TORO, E., ALKON, D. L., AND DELGADO-GARCÍA, J. M. Kinematic, time-domain and frequency-domain analyses of unconditioned and conditioned eyelid responses in the alert behaving rabbit. Soc. Neurosci. Meeting, New Orleans, 1997b.
• GUITTON, D., SIMARD, R., CODÈRE, F. Upper eyelid movements measured with a search coil during blinks and vertical saccades. Invest. Ophthalmol. Vis. Sci. 32: 3298-3305, 1991.[Abstract/Free Full Text]
• HALLIDAY, A. M., REDFEARN, J.W.T. An analysis of the frequencies of finger tremor in healthy subjects. J. Physiol. (Lond.) 134: 600-611, 1956.
• HORSLEY, V., SCHÄFER, E. A. Experiments on the character of the muscular contractions which are evoked by excitation of the various parts of the motor tract. J. Physiol. (Lond.) 7: 96-110, 1886.
• LLINÁS, R., RIBARY, U. Coherent 40-Hz oscillation characterizes dream state in humans. Proc. Natl. Acad. Sci. USA 90: 2078-2081, 1993.[Abstract/Free Full Text]
• LLINÁS, R. R. The noncontinuous nature of movement execution. In: Motor Control: Concepts and Issues, edited by D. R. Humphrey, and H.-J. Freund . New York: Wiley, 1991, p. 223-242
• LOEB, G. E., YEE, W. J., PRATT, C. A., CHANAUD, C. M., RICHMOND, F.J.R. Cross-correlation of EMG reveals widespread synchronization of motor units during some slow movements in intact cats. J. Neurosci. Methods 21: 239-249, 1987.[Medline]
• MANNING, K. A., EVINGER, C. Different forms of blinks and their two-stage control. Exp. Brain Res. 64: 579-588, 1986.[Medline]
• MARQUIS, D. G., PORTER, J. M. Differential characteristics of conditioned eyelid responses established by reflex and voluntary reinforcement. J. Exp. Psychol. 24: 347-365, 1939.
• MARSHALL, J., WALSH, E. G. Physiological tremor. J. Neurol. Neurosurg. Psychiatry 19: 260-267, 1956.
• MCLOON, L. K., WIRTSCHAFTER, J. D. Regional differences in the orbicularis oculi muscle: conservation between the species. J. Neurol. Sci. 104: 197-202, 1991.[Medline]
• MURTHY, V. N., FETZ, E. E. Coherent 25- to 35-Hz oscillations in the sensorimotor cortex of awake behaving monkeys. Proc. Natl. Acad. Sci. USA 89: 5670-5674, 1992.[Abstract/Free Full Text]
• NICOLELIS, M.A.L., BACCALA, L. A., LIN, R.C.S., CHAPIN, J. K. Sensorimotor encoding by synchronous neural ensemble activity at multiple levels of the somatosensory system. Science 268: 1353-1358, 1995.[Abstract/Free Full Text]
• PAGANO, R. R., GAULT, F. P. Amygdala activity: a central measure of arousal. Electroencephalogr. Clin. Neurophysiol. 17: 255-260, 1964.[Medline]
• PORTER, J. D., BURNS, L. A., MAY, P. J. Morphological substrate for eyelid movements: innervation and structure of primate levator palpebrae superioris and orbicularis oculi muscles. J. Comp. Neurol. 287: 64-81, 1989.[Medline]
• RESCORLA, R. A. Behavioral studies of pavlovian conditioning. Annu. Rev. Neurosci. 11: 329-352, 1988.[Medline]
• SANES, J. N., DONOGHUE, J. P. Oscillations in local field potentials of the primate motor cortex during voluntary movement. Proc. Natl. Acad. Sci. USA 90: 4470-4474, 1993.[Abstract/Free Full Text]
• SCHLAG, J., SCHLAG-REY, M. Role of the central thalamus in gaze control. In: Progress in Brain Research. The Oculomotor and Skeletalmotor Systems: Differences and Similarities, edited by H.-J. Freund, U. Büttner, B. Cohen, and J. Noth . Amsterdam: Elsevier, 1986, vol. 64, p. 191-201
• SHAW, M. D., BAKER, R. Morphology of motoneurons in a mixed motor pool of the cat facial nucleus that innervate orbicularis oculi and quadratus labii superioris, stained intracellularly with horseradish peroxidase. Neuroscience 14: 627-643, 1985.[Medline]
• THOMPSON, R. F. The neurobiology of learning and memory. Science 233: 941-947, 1986.[Abstract/Free Full Text]
• TRIGO, J. A., GRUART, A., AND DELGADO-GARCÍA, J. M. Discharge properties of abducens, ccessory abducens and orbiculari oculi motoneurons during unconditioned and conditioned eye blinks in the alert cat. Soc. Neurosci. Meeting, New Orleans, 1997.
• VAN DER WERF, F., ARAMIDEH, M., ONGERBOER DE VISSER, B. W., OTTO, A. J., BALJET, B., SPEELMAN, J. D. A fluorescent tracing study of the orbicularis oculi muscle in the monkey. Vision Res. 36: S194, 1996.
• WELSH, J. P. Changes in the motor pattern of learned and unlearned responses following cerebellar lesions: a kinematic analysis of the nictitating membrane reflex. Neuroscience 47: 1-19, 1992.[Medline]
• WELSH, J. P., LANG, E. J., SUGIHARA, I., LLINÁS, R. Dynamic organization of motor control within the olivocerebellar system. Nature 374: 453-457, 1995.[Medline]
• WESSBERG, J., VALLBO, Å. B. Coding of pulsatile motor output by human musce afferents during slow finger movements. J. Physiol. (Lond.) 485: 271-282, 1995.[Abstract/Free Full Text]
• WOODY, C. D. Understanding the cellular basis of memory and learning. Annu. Rev. Psychol. 37: 433-493, 1986.[Medline]

0022-3077/97 $5.00 Copyright ©1997 The American Physiological Society