INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The detailed characterization of tactile afferent fibers in several species, in particular, in the cat, the macaque monkey, and in human subjects, has permitted classification of this broad group of sensory nerve fibers into discrete functional subsets that are similar but not identical across species. The subdivision into a number of discrete functional classes has been crucial for analyzing the central processing of tactile information, in particular, the extent to which information from the different peripheral classes is processed in discrete channels through the sensory pathway and within the cortical primary and secondary somatosensory processing areas (SI and SII), and second, the extent to which the quantifiable features of sensory coding in the peripheral fibers are retained or modified at synaptic junctions in the central pathways.

In the case of tactile fibers that arise from the distal glabrous skin of the limbs in the cat, it is well known that there are three major functional classes, which correlative morpho-functional studies have linked to three distinct histological receptor types. One class consists of slowly adapting (SA) tactile afferents that appear to be associated with Merkel cell receptor endings in the epidermal pegs (Ferrington and Rowe 1980a; Iggo and Ogawa 1977; Jänig 1971; Jänig et al. 1968), whereas the other fibers are sensitive only to the dynamic components of tactile stimuli and can be divided into two distinct classes according to their sensitivity and responsiveness to cutaneous vibration. One class, most sensitive to vibration at 20-50 Hz, appears to be associated with intradermal, encapsulated receptors known as the Krause corpuscles (Iggo and Ogawa 1977; Jänig 1971), while fibers in the other class are exquisitely sensitive to cutaneous vibration at 200-400 Hz and are presumably associated with Pacinian corpuscles (Hunt 1961; Hunt and McIntyre 1960; Jänig et al. 1968; Lynn 1969; Sato 1961).

In the macaque and other old-world monkeys, a similar subset of three major classes has been identified (Johnson 1974; Lindblom 1965; Lindblom and Lund 1966; Talbot et al. 1968), although with minor differences in the classification terminology. These include a single SA class, again thought to be associated with Merkel receptors, and two purely dynamically sensitive classes, one, a rapidly adapting (RA) or quickly adapting (QA) class of fibers whose functional properties appear very similar to those of feline RA fibers, although the associated receptors in the macaque glabrous skin appear to be Meissner's corpuscles, rather than Krause corpuscles, which are found only in nonprimates. The third major class is again a very distinctive Pacinian corpuscle (PC)-related class whose functional properties appear identical to the feline class.

For human subjects, there appear to be some differences as the broad class of SA tactile fibers appear to form two classes. The first of these, designated the SA type I (SAI) fibers, appears to correspond with the single broad SA class in both the cat and macaque monkey and is probably associated with Merkel receptor complexes in the human glabrous skin. The second class, the SA type II (SAII) fibers, appears to be associated with Ruffini receptor endings in the human glabrous skin, in particular, in association with the skin around nail beds and skin creases near metacarpophalangeal joints (Johansson and Vallbo 1979; Knibestöl and Vallbo 1970). Although this SAII class has previously been identified in the cat, they appear in that species to be confined to the hairy regions of skin where they are also associated with Ruffini endings (Chambers et al. 1972; Gynther et al. 1992).

In the present study, with the use of quantitative stimulus-response analyses, we have classified the tactile afferent fibers that supply the glabrous skin of the hand in the marmoset monkey. This characterization of peripheral tactile mechanisms was undertaken as the marmoset is increasingly used as a primate model for the analysis of somatic sensory and other perceptual mechanisms (e.g., Brysch et al. 1990; Dick et al. 1991; Garraghty et al. 1990; Kaske et al. 1991; Krubitzer and Kaas 1992; Wilson et al. 1999; Zhang et al. 1996, 2001a,b). The present study was reported in abstract form (Coleman et al. 1996).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were performed on adult marmosets (Callithrix jacchus) of either sex, between 2 and 5 yr old, and with an average weight of around 300 g. Each was the subject of electrophysiological study at thalamocortical levels (Zhang et al. 1996, 2001a,b) prior to the peripheral recording analysis. All experiments conformed with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. In three animals, anesthesia was induced using a combination of halothane by inhalation and an intramuscular injection of ketamine (25 mg kg¹) and xylazine (2 mg kg¹) and then maintained by means of a continuous infusion of ketamine (~20 mg · kg¹ · h¹) and either diazepam (1 mg · kg¹ · h¹) or xylazine (1 mg · kg¹ · h¹) diluted in 0.18% sodium chloride and 4% dextrose solution. In 10 animals, anesthesia was induced with a combination of halothane and an intramuscular injection of alfaxalone and alfadolone acetate mixture (Saffan, 18 mg kg¹) and was followed with periodic intravenous injections of chloralose or pentobarbitone sodium (Nembutal) to maintain surgical anesthesia. Anesthesia was maintained for the duration of the experiments, and the animals were not allowed to recover consciousness between experiments carried out at the thalamocortical level (Zhang et al. 2001a,b) and those described in the present paper. As no systematic differences in peripheral tactile fiber properties were found for animals with different anesthetic regimes, data from the different animals were treated together. Atropine sulfate (0.05 mg kg¹) was administered subcutaneously following anesthesia to suppress respiratory secretions. At the end of the experiment, an overdose of sodium pentobarbitone was administered.

Surgical procedures

A tracheostomy was performed routinely, and animals were allowed to breathe spontaneously except for two that were ventilated artificially with a gas mixture of 75-80% N₂O and 20-25% O₂. End-tidal CO₂ was monitored and maintained at 3.75 ± 0.25%. The femoral artery and vein were cannulated. Blood pressure and heart rate were monitored via a cannula in the femoral artery, while a cannula inserted into the femoral vein was used to administer anesthetic agents and fluid supplementation during the experiment.

Recording and stimulating procedures

An incision was made on the volar surface of the forearm from the axillary region to the elbow and the median or ulnar nerve exposed for recording. A pool filled with warm paraffin was made using the skin flaps from the incision and the exposed nerve kept submerged in the pool. The nerve was transected and its proximal end placed on an insulated platform to isolate it electrically from the underlying tissue. Small strands of nerve were dissected from the trunk of the transected nerve and placed over a silver hook electrode. The response of each nerve strand to stimulation of the glabrous skin of the hand was tested, and the nerve strand further subdivided until unequivocal single fiber recording was achieved for individual tactile fibers with receptive fields on the glabrous skin of the hand. Recorded signals were fed to a preamplifier and filter, thence to an audio amplifier and speaker, and to an oscilloscope display. A magnetic tape recorder was used to record responses for analysis after the experiment. The signals were also passed to a differential amplitude discriminator from which constant output pulses could be relayed to a counter unit and a computer for further analysis.

Mechanical stimulation of tactile afferent fibers

When single tactile afferent fibers were isolated from the median or ulnar nerve strands their cutaneous receptive fields were mapped using von Frey hairs of 0.5 g weight. Precise and reproducible mechanical stimuli, derived from a servo-controlled mechanical stimulator, were delivered to the most sensitive part of the fiber's receptive field using a circular probe (usually a 1- or 2-mm tip diameter), as described in previous studies from this laboratory (e.g., Ferrington and Rowe 1980a,b; Turman et al. 1992; Zhang et al. 1996). The probe of the mechanical stimulator was attached to a shaft mounted on a moving coil vibrator. The shaft passed through a cylindrical linear voltage displacement transducer whose output varied as a function of shaft displacement and provided the feedback signal to the stimulator control unit. Stimuli were repeated at a rate of no more than 1 per 8 s to allow time for skin recovery.

All fibers were tested initially for their responsiveness to steady indentation of the skin lasting 1.5 s (amplitude 1 mm) and were classified into two broad groups: slowly adapting fibers that responded with sustained discharge throughout this indentation and purely dynamically sensitive fibers that responded only to the onset and offset of the stimulus.

Slowly adapting fibers were characterized quantitatively by constructing stimulus-response relations (see Fig. 1B) based on responses recorded to a range of indenting skin stimuli; whereas the purely dynamically sensitive fibers were characterized with the use of controlled dynamic forms of tactile stimuli, in particular, sinusoidal vibration usually delivered to the skin in trains lasting 1 s and superimposed on a background 1.5-s-long step indentation of 400 µm. The use of sinusoidal vibration enabled comparisons to be made with the functional capacities of dynamically sensitive tactile afferents analyzed in the cat, the macaque, and human subjects, in particular, in measures of vibrotactile thresholds (both absolute, and the tuning or 1:1 threshold, which was based on the minimum vibration amplitude needed to elicit a regular impulse discharge on successive cycles of the vibration train), and the frequency bandwidth over which vibrotactile responsiveness operated. In addition, the dynamically sensitive fibers were characterized quantitatively for the precision of impulse patterning in their responses to the vibration stimuli (see RESULTS), again permitting comparison with the dynamically sensitive fibers identified in other species.

View larger version (35K): [in this window] [in a new window]   Fig. 1. Response traces and stimulus-response relations for slowly adapting (SA) tactile afferent fibers supplying the glabrous skin of the marmoset hand. A: step indentations of 1.5 s duration (represented by the bottom trace) were applied to the receptive field focus on the palm (see figurine in B inset) at a series of amplitudes ranging from 0 to 1,500 µm for 1 SA afferent fiber. B: stimulus-response relations for 8 SA tactile fibers in which the mean response level (imp/s ± SD) for each fiber is plotted (ordinate) as a function of the steady step indentation amplitude (abscissa). In all cases, an increase in step amplitude resulted in a graded increase in response, although the response level of some fibers reached a plateau at indentation amplitudes of >800-1,000 µm. All 8 fibers had receptive fields confined to small areas on the glabrous skin of the fingers or palm. The relation plotted as the open diamond symbol in B was obtained for the SA fiber in A.

Evaluation of phase locking in the responses of dynamically sensitive fibers to vibration

The capacity of the marmoset dynamically sensitive tactile fibers to respond to vibrotactile stimuli with precise temporal patterning was evaluated by constructing peristimulus time histograms (PSTHs) and cycle histograms. In the PSTHs the time of occurrence of each impulse was plotted in relation to a fixed starting point, and responses were accumulated usually for five successive stimuli. Cycle histograms (CHs) were used to derive quantitative measures of the phase locking of impulse activity to the applied vibration stimulus. The CHs use a pulse associated with the onset of each vibration cycle as a stimulus marker and show the time of occurrence of impulses occurring during each cycle. In all cases, CHs were generated from the accumulated response of a fiber to five successive trains of vibration that each lasted 1 s. Responses that are tightly synchronized to a single portion or phase of the vibration cycle (or tightly phase locked) will appear in a CH as a narrow peak. When impulses occur independent of the phase of the applied vibration waveform, the histogram will appear flat.

The resultant (R) derived from the CH distributions has been used as a measure of phase locking in the responses of tactile neurons to vibrotactile stimuli (for example, Vickery et al. 1992) and of auditory neurons to tonal stimuli (for example, Bledsoe et al. 1982; Lavine 1971). Values for the resultant range from 0 to 1. A uniform, or random distribution, in which there is no preferred response phase will have a resultant approaching zero, while a distribution with perfect phase locking, in which all points or impulse occurrences are aligned to a particular phase, will have a resultant approaching the theoretical maximum value of 1. For a sample size (number of impulses accumulated in the cycle histogram) of 100 or more, a value of R < 0.17 would indicate that the response was not phase locked, at the 95% confidence level, whereas values of R > 0.3 indicate very significant phase locking (P < 0.0001 for n = 100) (Durand and Greenwood 1958; and Table B32 in Zar 1984).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Forty-seven fibers with tactile receptive fields on the glabrous skin of the marmoset hand were examined to classify them according to response characteristics and to establish their relations to classes identified in other mammalian species. In response to static indentation of the skin, 26 fibers (55%) responded in a sustained manner and were classified as SA fibers. The remaining 21 fibers (45%) were purely dynamically sensitive, responding only to the ON and OFF components of the step indentation. The dynamically sensitive fibers could be further subdivided according to their sensitivity and responsiveness to sinusoidal vibratory stimuli and formed two classes that appeared to correspond to the RA and PC classes described for other species. The numbers and proportions of the three tactile fiber classes innervating the glabrous regions of the marmoset hand are shown in Table 1.

Although other fibers were isolated in the course of this study, with receptive fields on the hairy skin of the forearm or in deep structures such as the forearm muscles, these were not studied further or included in the above fiber count. Muscle spindle units could be differentiated from slowly adapting cutaneous fibers on several criteria: 1) they could rarely be activated by von Frey hairs of <1.0 g wt; 2) their spontaneous or resting discharge was higher than that of cutaneous units, often in excess of 15 imp/s; 3) step indentations of the skin of <800 µm produced little or no change in the fiber's impulse rate; and 4) the effect of forearm probing or manipulation was consistent with the response arising from muscle stretch.

SA primary afferent fibers

The twenty-six SA fibers that responded to steady skin indentation with sustained discharge had small receptive fields (typically 2-3 mm diam) with well-defined boundaries when mapped with von Frey hairs. They had little or no spontaneous activity (range 0-6 imp/s) and responded to steady indentation with an irregular pattern of discharge (Fig. 1A). No systematic attempt was made to classify SA fibers into the SAI and SAII subgroups identified in the human hand (Chambers et al. 1972; Johansson and Vallbo 1979; Knibestöl 1975); however, the clearly defined boundaries of the receptive fields of all 26 SA fibers and their irregular discharge pattern in response to steady skin indentation is consistent with the behavior of SA type I fibers found in other species (see DISCUSSION).

Responses of SA fibers to steady indentation

Individual SA fibers responded with an initial high-frequency burst of impulses at the onset of a step indentation, followed by a sustained discharge at a lower rate throughout the static component of the stimulus. The responsiveness of individual SA fibers to static indentation was examined by recording responses to a 1- or 1.5-s step indentation and constructing stimulus-response relations that plotted the average response rate (imp/s ± SD) as a function of indentation amplitude. In all cases, the average response rate was obtained over the first 1 s of a step indentation, for five repetitions of the stimulus. As the amplitude of the applied step was increased, SA fibers responded with a graded increase in impulse rate (Fig. 1A). The stimulus-response relations obtained for eight SA fibers (Fig. 1B) illustrate this tendency of the SA fiber impulse rate to increase as a function of indentation amplitude. For three or four of the fibers, a response plateau was reached, where increases in amplitude above 800-1,000 µm produced little change in impulse rate. We did not undertake an analysis of temporal patterning within SA fiber responses to static skin indentation because any such patterning is unlikely to be of importance for coding the intensity of the static stimulus.

Dynamically sensitive primary afferent fibers

Twenty-one tactile fibers that displayed a pure dynamic sensitivity in responding only to the onset and offset of a steady indentation were isolated in the course of this study. Their responses to trains of sinusoidal cutaneous vibration permitted a more comprehensive examination of their response characteristics, and division of this broad group into two distinct functional classes. The first of these, comprising 15 fibers, responded best to low-frequency vibration, ranging from ~10 to 50 Hz, whereas the second class (6 of 21 fibers) responded preferentially to high-frequency vibration in the range from 50 to 700 Hz, with maximal sensitivity around 200-300 Hz. These two classes in the marmoset displayed the general characteristics of the RA and PC fiber classes described previously in other mammalian species (see DISCUSSION).

RA afferent fibers: stimulus-response characteristics

All 15 RA afferent fibers, isolated from either the median or ulnar nerve, had receptive fields on the glabrous skin of the hand that were no more than a few millimeters in diameter when mapped with 0.5 g von Frey hairs. These RA fibers displayed vibrotactile responsiveness of the form shown in Fig. 2, where the impulse traces obtained in response to 30 Hz (A) and 50 Hz (B) vibration reveal a high sensitivity, with absolute thresholds below 5 µm at each frequency. Furthermore, this RA fiber attained its 1:1 response threshold by 15 µm at 30 Hz and by 30 µm at 50 Hz and displayed a precisely patterned discharge in response to each of these frequencies.

View larger version (30K): [in this window] [in a new window]   Fig. 2. Responses of a representative rapidly adapting (RA) fiber to different amplitudes of cutaneous vibration. Vibration frequency in A was 30 Hz and in B, 50 Hz, and was delivered with a 5-mm-diam probe within the RA fiber's receptive field on the palm, proximal to the thumb (indicated in the figurine). In each case the vibration train lasted 1 s and started 300 ms after the onset of a 1.5-s step indentation as represented by the stimulus waveforms beneath A and B. The amplitude of vibration is indicated to the left of each trace.

Stimulus-response relations were constructed in Fig. 3A by plotting the mean response (imp/s) as a function of vibration amplitude at a series of frequencies ranging from 10 to 200 Hz. The fiber responded to stimulation at amplitudes as low as 5 µm for all frequencies below 200 Hz. At any given frequency, the response level increased as a function of vibration amplitude, reaching a plateau when the discharge rate of the fiber attained the 1:1 pattern (1 impulse per cycle) reflecting the applied vibration frequency. This 1:1 response plateau could be retained over a broad range of amplitudes as indicated in particular by the graphs at 50, 80, and 100 Hz. If the vibration amplitude was further increased, it was sometimes possible to drive the fiber's response above the 1:1 level, in particular, at low vibration frequencies (30 Hz; e.g., Fig. 3A).

View larger version (15K): [in this window] [in a new window]   Fig. 3. Stimulus-response relations and tuning curves for representative RA fibers. A: stimulus-response relations have been plotted at a range of vibration frequencies (10-300 Hz) for an RA fiber that was most sensitive to low-frequency vibration (30-50 Hz). At each frequency the points represent the mean of 5-10 successive responses (imp/s) to 1-s trains of vibration at the indicated amplitude (abscissa). B: tuning curves showing the 1:1 response threshold and bandwidths of vibration sensitivity for 5 RA fibers supplying the glabrous skin of the marmoset hand (fiber in A plotted as open diamonds). They were constructed from the stimulus-response data by plotting the lowest vibration amplitude at which 1:1 following (1 impulse per vibration cycle) was obtained for each frequency. Arrows on the right-hand side of 3 of the curves indicate that the 1:1 thresholds exceeded the maximum vibration amplitudes available for testing at these higher frequencies. RA fibers typically were most sensitive to vibration frequencies around 10-30 Hz.

The lower limit of the 1:1 plateau, corresponding to the lowest vibration amplitude that produced a 1:1 following in the afferent fiber response over the 1-s sample period, is called the tuning point or 1:1 threshold (Talbot et al. 1968). Tuning curves constructed by plotting amplitude values for the 1:1 thresholds against vibration frequency for five RA afferent fibers are shown in Fig. 3B. While the 1:1 response thresholds varied from fiber to fiber, the marmoset RA fibers tested showed greatest sensitivity to vibration frequencies between 10 and 50 Hz. At frequencies below 10 Hz or above 50 Hz, the 1:1 threshold amplitude increased markedly. The most sensitive RA fibers could be activated in a 1:1 response pattern by vibration amplitudes as low as 10 µm at their best frequencies and displayed a level and range of sensitivity consistent with RA fibers studied in other species (Ferrington and Rowe 1980a; Ferrington et al. 1984; Iggo and Ogawa 1977; Johnson 1974; Talbot et al. 1968). They also resembled RA fibers in the macaque (Talbot et al. 1968) in displaying an initial entrainment to just the first few cycles of vibration at the lower amplitudes (e.g., at 5-10 µm at 30 Hz, and 10-20 µm at 50 Hz in Fig. 2) before becoming entrained over the whole 1-s segment of the vibration train at higher amplitudes.

Phase locking and impulse patterning in vibrotactile responses of RA fibers

Although the impulse patterning in response to cutaneous vibration is apparent in impulse traces recorded in response to low vibration frequencies (Fig. 2), this becomes less clear as the frequency increases and permits only a qualitative impression of the patterning. To evaluate more precisely the capacity of the RA fibers for responding in a phase-locked manner to the vibrotactile stimuli, PSTHs and CHs were constructed from responses over a range of frequencies. The PSTHs in Fig. 4 were constructed from six consecutive responses of a representative RA fiber to vibrotactile stimulation at 10, 50, 100, and 200 Hz, all at the fixed amplitude of 30 µm. Those on the left were constructed over a 2.5-s segment commencing 0.5 s before the onset of a 1.5-s step on which the 1-s vibrotactile stimulus was superposed. These histograms show the overall profile of the RA fiber response, at the ON and OFF phase of the step and a well-maintained response throughout the 1-s train of vibration. The top PSTH reveals the phase-locked pattern of response at 10 Hz, but at the higher frequencies of 50, 100, and 200 Hz the phase locking becomes apparent only in the right hand PSTHs with the expanded time scales that display response details over the first 10 cycles of vibration. In each case the response peaks are separated at intervals corresponding to the cycle period of the vibration; that is, 100 ms at 10 Hz and 20, 10, and 5 ms, respectively, at the three higher frequencies.

View larger version (25K): [in this window] [in a new window]   Fig. 4. Peristimulus time histograms (PSTHs) showing the profile and pattern of RA fiber responses to vibration. The paired PSTHs on the left and right sides were constructed from 6 successive responses of an RA fiber to vibration at 10, 50, 100, and 200 Hz at the fixed amplitude of 20 µm. Those on the left show the distribution of activity over a 2-s segment starting 0.5 s prior to the onset of a 1.5-s step indentation on which the 1-s train of vibration was superimposed (stimulus waveform shown below). These histograms show the response at the ON and OFF phases of the step and the profile of maintained response across the 1 s of vibration at 10, 50, and 100 Hz, but some decline after ~0.5 s for the 200-Hz vibration. Clear phase locking of the response is apparent in the left hand PSTHs only at 10 Hz, but the PSTHs on the right, with the expanded time scales showing the response to the 1st 10 cycles of vibration, reveal that the tight phase locking is present at all 4 vibration frequencies. The height of each column represents the number of impulse counts accumulated in each address, which was 20, 20, 10, and 10 ms, respectively, for the 4 left hand histograms and 10, 2, 1, and 0.5 ms, respectively, for the right hand histograms.

Quantified estimates of phase locking in this RA fiber's responses were derived from the cycle histograms constructed in Fig. 5 in which the analysis time on each abscissa corresponds with the vibration cycle period. The tight grouping of impulse occurrences in the CHs constructed from responses at 10, 30, 50, and 100 Hz is reflected in the high values of the resultant (R  0.95). However, at the highest frequencies the R value declines to 0.88 at 200 Hz, and to 0.28 at 300 Hz, where the double peak in the distribution indicates that there are two disparate phases of the vibratory waveform on which impulses are discharged.

View larger version (17K): [in this window] [in a new window]   Fig. 5. Cycle histograms (CHs) showing the phase locking of RA responses to vibration. The CHs show the distribution of impulse occurrences within the vibratory cycle period for 6 vibration frequencies from 10 to 300 Hz. The analysis period in each case corresponds to the vibration cycle period, and the accumulated counts (ordinate) show the preferred grouping of impulses at each frequency. The quantitative measure of phase locking, the resultant, R, is given for each CH distribution.

Functional capacities of PC afferent fibers supplying the marmoset hand

The afferent fibers whose vibrotactile sensitivity implied an association with Pacinian corpuscle receptors had larger receptive fields on the glabrous skin than SA or RA fibers, when mapped with von Frey hairs, and could also be activated from more remote locations on the limb by manual tapping. Quantification of their vibrotactile sensitivity and responsiveness was undertaken by means of controlled stimulation at the most sensitive location within their receptive fields and stimulus-response relations constructed that enabled absolute and 1:1 thresholds to be derived. The representative relations in Fig. 6 reveal the extremely low absolute thresholds (<0.5 µm) for the marmoset PC fibers at the best vibrotactile frequencies of 200 and 300 Hz, and the very low thresholds (<2-3 µm) over a broad range of frequencies from 100 to 600 Hz. At the low end of the vibrotactile range, absolute thresholds rose steeply, with values of ~4 µm at 80 Hz, ~10 µm at 50 Hz, >20 µm at 30 Hz, and >40 µm at 10 Hz (Fig. 6). Once the absolute threshold was exceeded, the stimulus-response relations for the PC fibers rose abruptly over a broad range of vibration frequencies (e.g., 80-300 Hz in Fig. 6) to attain a 1:1 plateau level of response corresponding to the discharge of one impulse on each cycle of the vibration. This behavior conferred on the PC fibers a very narrow dynamic range, defined as the range of vibration amplitudes over which the fiber displayed a graded responsiveness. In Fig. 6 this dynamic range was no more than ~5 µm at each frequency in the range 80-300 Hz, whereas the stimulus-response plateau extended over most of the amplitude range (50 µm) plotted in Fig. 6 for these frequencies. At 400 Hz the 1:1 response plateau was achieved by <10 µm but was not attained at 500 and 600 Hz, although at the latter frequency the fiber became locked to a 1:2 pattern of response generating a plateau at 300 imp/s. At the low frequencies (e.g., 50 and 80 Hz) the fiber displayed a second plateau corresponding to a 2:1 response level.

View larger version (18K): [in this window] [in a new window]   Fig. 6. Stimulus-response curves for a representative Pacinian corpuscle (PC) fiber across a range of vibrotactile frequencies. At each of the 10 frequencies (10-600 Hz), the points represent the mean (imp/s) for 5 or 10 responses to 1-s trains of vibration at the indicated amplitude (abscissa). The receptive field for this PC fiber covered most of digit 2 sinusoidal vibration. Stimuli were applied using a 2-mm probe tip.

Temporal patterning in the vibrotactile responses of marmoset PC fibers

The characteristic plateau attained in the stimulus-response relations of PC fibers at a response level matching the frequency of the applied vibration (Fig. 6) meant that the impulse patterns, as in PC fibers of other species, could potentially reflect the periodicity of the vibration and therefore provide a pattern code for the frequency parameter of the vibrotactile event. This was confirmed qualitatively by examining impulse trace records for the PC fibers on a time scale that revealed the regularity of the impulse pattern for a broad range of vibration frequencies (Fig. 7A). At each frequency from 50 to 400 Hz in Fig. 7A, the PC fiber responded with a 1:1 pattern in which the impulses were tightly phase locked to the vibration allowing the inter-impulse intervals to closely approximate the cycle period of the vibration. The only minor exception to this metronome-like reflection of the vibration periodicity in the impulse pattern occurred on the first two cycles of the vibration (in particular, at 50 and 100 Hz), where the onset spike was initiated at a different phase of the first vibration cycle (reflecting presumably the abrupt onset of this 1st vibration cycle from the null position). At the highest vibration frequency in Fig. 7, 600 Hz, the PC fiber response remained tightly phase locked, but, at the fixed 20-µm amplitude used here (and at all 6 frequencies illustrated), the response attained only a 1:2 pattern (that is, 1 impulse on every 2nd cycle of the vibration), which is the same temporal pattern displayed in the 1:1 response at the lower frequency of 300 Hz. However, the potential ambiguity of signaling in these two impulse records may be resolved by the population behavior of the PC afferent fibers.

View larger version (32K): [in this window] [in a new window]   Fig. 7. Temporal patterning in PC fiber responses to a range of vibration frequencies (50-600 Hz). A: spike traces show the PC fiber responses to each vibration frequency at an amplitude that produced a 1:1 response level (except at 600 Hz, where the maximum response level obtained in the fiber was 1:2). Responses are shown for the 1st 100 ms of a 1 s duration vibration stimulus, superimposed on a 1.5-s step indentation. B: cycle histograms show the phase locking of responses for a typical PC fiber at the 6 frequencies (50-600 Hz; amplitude fixed at 20 µm). The CHs were accumulated from the 1st 500 ms of 10 repetitions of the stimulus at each frequency. The receptive field location on the index finger is indicated in the figurine.

Quantification of the tightness of phaselocking in the PC fiber responses to vibration was derived from cycle histogram distributions of the type shown in Fig. 7B, which show that impulse occurrences were locked to a narrow segment (<5-10%) of the cycle period at all frequencies from 100 to 600 Hz. Values for the Resultant, R, were in excess of 0.99 at each of these frequencies and >0.95 at 50 Hz, reflecting near-perfect phase locking.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Classification of tactile sensory fibers supplying the marmoset hand

The classification of tactile sensory fibers innervating the glabrous skin of the marmoset hand was undertaken with the use of mechanical stimuli, in particular, static indentation and sinusoidal vibration, that were employed in earlier studies of peripheral tactile mechanisms in the cat (e.g., Ferrington and Rowe 1980a; Ferrington et al. 1984; Iggo and Ogawa 1977; Jänig 1971; Jänig et al. 1968; Lynn 1969), the macaque monkey (Johnson 1974; Talbot et al. 1968) and human subjects (Johansson and Vallbo 1979; Knibestöl and Vallbo 1970). These stimuli provided the most effective way of achieving comparability of classification for marmoset tactile fibers with classification schemes in other species. Furthermore, the different functional classes identified in other species with these stimuli appear to have the additional utility, or "validation," of an actual, or implied morphological correlate in the form of a distinctive peripheral receptor structure.

Tactile fibers in the present study were not characterized in terms of conduction velocity to avoid possible damage to the skin of the hand associated with the insertion of stimulation electrodes. Furthermore, to help maintain temperature control for the peripheral nerve, we did not expose additional segments of the nerve for electrical stimulation. However, from the concatenation of their functional properties, it appears that the 47 low-threshold mechanosensitive fibers studied belong to the group II category of somatic afferent nerves whose conduction velocities should be in the range ~30-70 m s¹. The fibers fell into three discrete functional classes, one SA class, and two purely dynamically sensitive classes that could be subdivided according to their vibrotactile sensitivity. As the two dynamically sensitive classes resembled those identified in the cat, macaque and human distal glabrous skin they were designated the RA and PC-related classes. The classification therefore appears to conform to those established for tactile fibers supplying the glabrous skin of the footpads in the cat (Ferrington and Rowe 1980a; Ferrington et al. 1984; Iggo and Ogawa 1977; Jänig 1971; Jänig et al. 1968; Rowe 1982) and the raccoon (Pubols and Pubols 1973; Pubols et al. 1971), and the hand in the macaque (Johnson 1974; Talbot et al. 1968), where, in each case there are three major classes, an SA class, an RA (or QA, for quickly adapting), and a PC class.

In contrast to the above classification scheme for glabrous skin tactile afferents in the marmoset, macaque, and cat, that for the human being, the baboon, and the rat appears to comprise four distinct classes as the SA fibers have been subdivided into SAI and SAII classes (Dykes and Terzis 1979; Knibestöl and Vallbo 1970; Leem et al. 1993), a subdivision that is well established for the hairy skin in several species including the cat where the SAI fibers are known to be associated with Merkel receptor complexes beneath the touch domes (Iggo and Muir 1969) and the SAII fibers with Ruffini end organs (Chambers et al. 1972). A dichotomy in functional properties of SAI and SAII fibers in the hairy skin (Chambers et al. 1972) is consistent with the morphological dichotomy in their respective Merkel and Ruffini receptor endings. In the human glabrous skin, where this functional distinction between SAI and SAII fibers has also been reported, the SAII class is concentrated in its innervation to regions around the nail beds and skin creases (Johansson and Vallbo 1979; Knibestöl 1973; Knibestöl and Vallbo 1970). However, the evidence for an association of the SAII fibers in human glabrous skin with Ruffini end organs is based only on light-microscopy studies (Miller et al. 1958).

In contrast to this SA dichotomy in the human and baboon hand, there is no evidence for an SAII class in the glabrous skin of the macaque hand or the cat foot pads where SA fibers form a functional continuum whose properties conform to the SAI class. Furthermore, in the cat footpads there is no evidence for Ruffini endings, and the SA fibers have been shown by Jänig's correlative morpho-functional studies (Jänig 1971) to be associated with Merkel receptor complexes in the epidermal pegs. The behavior of the SA fibers supplying the marmoset hand was also consistent with SAI characteristics (and therefore a probable association with Merkel receptors), as they displayed an absence or low levels of spontaneous activity (Fig. 1), irregular interspike intervals in response to skin deformation (Fig. 1), and small, punctate receptive fields (Chambers et al. 1972; Iggo and Muir 1969; Vickery et al. 1992).

As our observations favor a single SA class for the marmoset glabrous skin, it appears that the tactile innervation of the hand for this species conforms to the triad of sensory fiber classes present in both the cat and macaque monkey rather than the four-class grouping found in the baboon and human hand. In this respect it is perhaps surprising that there should be three classes with very similar functional properties across representatives of nonprimate placental mammals (the cat and the raccoon), new-world primates (the marmoset), and old-world primates (the macaque monkey), when other representatives of old-world primates (the human being and baboon) and the rat have four recognized classes.

Proportions of different classes represented among marmoset tactile afferent fibers

Approximately one-half (55%) of the sample of tactile afferent fibers supplying the marmoset hand displayed slowly adapting properties (Table 2). This is in the range of the proportions reported for the human hand (75% in Knibestöl and Vallbo 1970; 44% in Johansson and Vallbo 1979) and was a little higher than those reported for both the macaque (40% in Talbot et al. 1968; 36% in Darian-Smith and Kenins 1980) and the cat (33% in Iggo and Ogawa 1977).

However, differences in sample sizes and search stimuli used may contribute in part to these differences, in addition to any genuine species differences in innervation densities. Another factor that may have a bearing on the proportions found for different species could be the skin region sampled. For example, the majority of tactile fibers in our sample (30 of 47) had receptive fields on the palmar surface, where the proportion of SA fibers in the overall sample of the three classes was higher (~60%) than was the case on the fingers (~40%). This regional difference is consistent with the finding for the human hand of a higher proportion of SA afferents in the sample from the palmar surface compared with that from the finger tip (Johansson and Vallbo 1979).

Among purely dynamically sensitive fibers there has been a consistent preponderance of RA (or QA or FAI) fibers over PC (or FAII) fibers in all studies (Table 2), although as might be expected, the ratio is somewhat variable. The present study in the marmoset shares a similar ratio of 2.5-3 to 1 (RA to PC) to that found in human subjects (Johansson and Vallbo 1979), the baboon (Dykes and Terzis 1979), and the cat (Iggo and Ogawa 1977), whereas in the macaque and the rat it was closer to 5 to 1 (Leem et al. 1993; Talbot et al. 1968).

Central representation of peripheral tactile afferent classes

The existence of discrete tactile neuron classes appears to be maintained to a large extent within the central tactile pathways as many studies, in particular, in the cat and macaque, have reported that three major classes of tactile neuron related to the distal glabrous skin are present at the level of dorsal column nuclei (Bystrzycka et al. 1977; Douglas et al. 1978; Dykes et al. 1982), the thalamic ventroposterolateral (VPL) nucleus (Dykes et al. 1981; Ghosh et al. 1992, 1994; Herron and Dykes 1986; Warren et al. 1986), and at the level of the primary (SI) and secondary (SII) areas of somatosensory cortex (Bennett et al. 1980; Burton and Sinclair 1990, 1991; Ferrington and Rowe 1980b; Mountcastle et al. 1969; Sinclair and Burton 1993). While some evidence for modality mixing has been reported (e.g., Ghosh et al. 1992; Waldron et al. 1989), this is hardly surprising on anatomical grounds as individual axons at different levels ramify over considerable distances within the relay nuclei (Fyffe et al. 1985; Rainey and Jones 1983). However, the preponderant compartmentalization of tactile modalities at central levels may be achieved in part by functional mechanisms, such as afferent inhibition, which act to limit the effectiveness of subsidiary or nondominant sources of input to central neurons (Bystrzycka et al. 1977; Carmody and Rowe 1974; Ferrington et al. 1987; Mountcastle and Powell 1959).

Functional properties of marmoset tactile afferent fibers

SA AFFERENT FIBERS. Individual SA fibers supplying the marmoset hand displayed considerable variation in sensitivity, as measured by their absolute thresholds to step indentations of the skin, and in the magnitude of their responses, in impulses/second (Fig. 1B). However, this variability has also been observed for SA fibers in other species (e.g., Ferrington and Rowe 1980a; Iggo and Ogawa 1977; Knibestöl 1975). No attempt was made to establish an exact mathematical characterization for the SA fibers' stimulus-response relations (Fig. 1B), for example, whether they were best fitted by a linear or power function relation, as the validity of such attempts has been questioned (Kruger and Kenton 1973). Nevertheless, the graded form of the relations and the consistency in response level at fixed indentation amplitudes (reflected in the low SD values in Fig. 1B) indicate that the individual SA fibers can signal discriminative information about the magnitude of static skin displacement in the marmoset hand. However, the central interpretation of information about this parameter of tactile contact may depend on the total level of impulse traffic in the population of responding fibers, as appears to be the case for intensity coding in other sensory continua (e.g., Johnson 1974) and for the coding of static tactile indentation in other species (e.g., Ferrington and Rowe 1980a; Mountcastle et al. 1966).

DYNAMICALLY SENSITIVE TACTILE AFFERENT FIBERS. The division of marmoset dynamically sensitive tactile afferent fibers into the PC- and RA-related classes reflects, as in other species, the differential dynamic sensitivity of the two classes of associated receptors (the Pacinian corpuscles and the presumed intradermal receptors, respectively), in particular, in response to vibrotactile stimuli that reveal that marmoset RA fibers are most sensitive to low vibration frequencies, 50 Hz (Figs. 2 and 3). Furthermore, they show marked fiber-to-fiber variability in vibration sensitivity as do the RA classes in the macaque monkey and the cat. For example, the 1:1 threshold variations in Fig. 3B for marmoset RA fibers are consistent with those of cat RA fibers (Ferrington and Rowe 1980a; Ferrington et al. 1984), and of macaque RA fibers that show 1:1 or tuning thresholds that vary by a factor of ~100 (from ~7 to ~700 µm) over the best vibration frequency range, 10-40 Hz (Fig. 22 in Talbot et al. 1968). Tuning curves obtained for the marmoset PC fibers were also consistent with those obtained in cats (Ferrington and Rowe 1980a; Ferrington et al. 1984) and macaques (Talbot et al. 1968). However, quantitative comparison of vibrotactile sensitivity and bandwidths of marmoset RA and PC fibers with the human RA and PC fibers is not possible, as investigations of their responses to vibration have generally been qualitative rather than quantitative (e.g., see Johansson and Vallbo 1979; Knibestöl and Vallbo 1970).

Impulse patterning and temporal coding in marmoset RA and PC fibers

The broad 1:1 response plateau in the vibration-induced responses of marmoset RA and PC fibers, together with the precise phaselocking of their responses (Figs. 4, 5, and 7) ensures a metronome-like temporal precision in the impulse train (Figs. 2 and 7) that accurately reflects the periodic nature of the vibrotactile stimulus. As this temporal patterning appears to be the crucial requirement for signaling vibratory frequency information (Ferrington and Rowe 1980b; La Motte and Mountcastle 1975; Mountcastle et al. 1969, 1990; Talbot et al. 1968), it appears that the marmoset RA and PC fibers are as well equipped to signal this information reliably to central neurons as their counterparts in the macaque monkey (Talbot et al. 1968) and the cat (Ferrington and Rowe 1980a; Ferrington et al. 1984).

In summary, it appears that for the hand of the marmoset monkey, the tasks of tactile perception and prehension are based on a triad of tactile sensory fiber classes, comprising an SA, RA, and PC class, with similar functional capacities to those present in the human being, macaque, and cat. However, in contrast to the human hand there is no clear evidence for a subdivision of the SA fibers into type I and type II classes.