INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Despite extensive research efforts, the role of the somatosensory cortex (SI and SII) in human pain processing remains largely unclear and controversial. Historically, clinical observations of focal brain lesions and electrical stimulation of the cortex suggested that cortical involvement in pain perception was minimal or absent and indicated, instead, a predominant role for thalamic and subcortical contributions to the experience of pain (Head and Holmes 1911; Penfield and Boldrey 1937). More recently, however, neurophysiological studies have documented that SI and SII receive noxious and innocuous cutaneous input from somatosensory thalamus (Friedman and Murray 1986; Gingold et al. 1991; Kenshalo et al. 1980; Rausell and Jones 1991; Shi and Apkarian 1995) and contain neurons that code spatial, temporal, and intensive aspects of innocuous and noxious somatosensory stimuli (Chudler et al. 1990; Dong et al. 1994; Kenshalo and Isenesee 1983; Kenshalo et al. 1988, 2000). These experimental data provide a possible neural substrate for the processing of cutaneous information within SI and SII; however, the scarcity of nociceptive neurons within these areas has led to questions concerning their functional significance in pain perception.

Adding to this controversy have been the inconsistent results observed in different human brain imaging experiments. Although activation of SII by noxious stimuli has been consistently reported in both positron-emission tomography (PET) and functional magnetic resonance imaging (fMRI) studies (e.g., Peyron et al. 2000), involvement of SI in the processing of nociceptive information has been less obvious. Imaging studies over the past decade have confirmed, at most, that SI activation is a variable finding when human subjects are presented painful stimuli (Bushnell et al. 1999).

Even if activation of somatosensory cortices is observed during the presentation of painful stimuli, the functional significance of this activation for the perception of pain remains in question. Although some studies have demonstrated that cognitive manipulation of pain perception is paralleled by changes in the activation of SI (Carrier et al. 1998; Hofbauer et al. 1998), other reports suggest that these pain-related changes in SI activation may also have an important function in modulating tactile activity rather than specifically encoding perceptual features of pain perception (Apkarian et al. 1992, 1994; Backonja et al. 1991; Tommerdahl et al. 1996; see also Stohler et al. 2001 for the influence of muscle pain on touch perception). Considering the substantial overlap in activity evoked by noxious and innocuous stimuli within both SI and SII (Coghill et al. 1994; Gelnar et al. 1999), these regions may, indeed, be involved in both perception and modulation of both painful and nonpainful somatosensory sensations. The present study uses fMRI to address the hypothesized role of somatosensory cortices in pain perception by comparing the time course of cortical activity evoked by innocuous brush and noxious heat stimuli with the temporal features of touch perception and pain perception that are associated with those respective stimuli. Implicit within this hypothesis is the assumption that the distinctive, time-dependent changes in the conscious perception of heat pain (and innocuous touch) must be mirrored by similar changes in the dynamic response of neuronal activity observed within those cerebral structures that participate in this process. While negative findings might be difficult to interpret (and could simply reflect a lack of sensitivity in the recording technique, fMRI), a positive temporal relationship between cerebral activity and conscious perception would lend strong support to the hypothesized role of the somatosensory cortices in the appreciation of pain sensation. A subset of these data has been previously reported in abstract form (Chen et al. 1999).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Six normal volunteers (4 males, 2 females, ages 23-47) participated in the imaging study; however, data from two subjects were excluded due to the presence of motion artifacts. Eight normal volunteers (6 males, 2 females, ages 21-52, including 2 from the imaging study) participated in psychophysical experiments that incorporated stimuli identical to those presented in the scanning sessions. All subjects were instructed in the basic design of the experiment and were fully aware of the duration and intensity of pain. The study was approved by the Montreal Neurological Institute and Hospital (MNI) Research Ethics Committee, and written informed consent was obtained from each subject prior to each study session.

Stimuli

THERMAL. Thermal stimulation consisted of noxious (45-46°C) and neutral (35-36°C) stimuli applied to the inner left calf via contact thermodes (9-cm² aluminum blocks connected to recirculating water baths under thermostatic control). These temperatures were chosen prior to imaging experiments during a preliminary session in which subjects were acclimated to the thermal stimuli and trained to rate the perceived pain intensity using a five-point verbal scale ("0" = no sensation of pain; "5" = the most intense pain sensation that the subject would tolerate). For each subject, the temperature of the noxious heat stimulus was determined as that which produced a moderate but tolerable level of pain (a rating of 4 out of 5 on the pain intensity scale). The temperature of the neutral stimulus was chosen as that which produced a sensation of innocuous warmth. During imaging sessions, the thermal stimuli (noxious heat and neutral warmth) were applied in an alternating cyclic fashion (described below) within the same scanning run. Each of the two thermodes was maintained at a constant preset temperature (by means of the circulating water baths) and applied to the skin during stimulation periods and withdrawn during the interstimulus interval. The presentation of the neutral warm stimuli served as a control for tactile and cognitive aspects of the phasic stimulation paradigm.

MECHANICAL. Mechanical stimulation was presented to the same site used for heat stimulation and consisted of light manual brushing at 2 Hz, using a 2-cm wide soft artist's paint brush moving back and forth in a proximal-distal orientation, over a 10-cm region of the skin. The brush stimuli were also presented during the preliminary session during which subjects practiced rating the intensity of stimulation using a similar five-point scale, where 0 represented no sensation and 5 represented very intense, but nonpainful sensation. During imaging sessions, the brush stimuli were presented in separate scanning runs, without thermal stimulation; periods of stimulation and rest (interstimulus intervals) were identical to those used in the thermal stimulation experiments.

Experimental protocol

Each subject participated in three imaging sessions conducted on different days. Before being placed in the scanner, subjects were instructed to attend to the stimuli, to keep their eyes closed, and to refrain as much as possible from moving throughout the imaging session. After being placed in a comfortable position, the head was immobilized with padded ear-muffs, a foam headrest, and a plastic bar across the bridge of the nose; an additional bite bar was used in two subjects. Each imaging session consisted of five to eight functional scanning runs and one high-resolution anatomical scan. During the scanning, thermal and tactile stimuli were applied to the left calf on separate runs. Imaging sessions always started with tactile runs followed by thermal runs to avoid the possible effect of sensitization induced by the noxious stimuli. Thermal runs consisted of 10 cycles of rest, painful heat, rest, and neutral heat stimulation, with each condition lasting 3 complete full-brain scans, ~10 s (Fig. 1A). Tactile runs contained 20 cycles of brushing and rest, each with the same duration as that used during the thermal runs (Fig. 1B). Auditory signals, generated by the scanner at the beginning of each full-brain scan sequence, were used to cue the manual presentation of both thermal and brush stimuli, thus synchronizing stimuli with data acquisition.

View larger version (10K): [in this window] [in a new window]   Fig. 1. Stimulation paradigm. Each study session consisted of 5-8 functional scanning runs. Thermal runs (A) consisted of 10 cycles of rest, painful heat, rest, and neutral heat stimulation delivered manually via a 9-cm2 contact thermode. Tactile runs (B) contained 20 cycles of brushing and rest. Each stimulus condition was presented for ~10 s during which 3 brain scans were acquired.

To assess the consistency of stimulation and to control for possible stimulus sensitization and/or habituation, subjects were instructed, following each run, to rate the intensity of the stimuli as perceived at the beginning and at the end of the run, in the manner described previously. Subjects were also asked to rate any discomfort arising from sources other than the stimulus. All ratings were given nonverbally, using the fingers of one hand, to minimize head movement.

Separate psychophysical experiments, following the same stimulus paradigms used in fMRI, were also performed by two of the subjects who participated in the imaging sessions and by six other healthy volunteers of similar age and gender. During these psychophysical experiments, subjects gave continuous ratings throughout the stimulus period using a mechanical visual analogue scale (VAS), sampled at 10 Hz, which allowed instantaneous changes in the ratings according to the changing perception of the stimulation.

Data acquisition

Imaging was performed in the McConnell Brain Imaging Center at the Montreal Neurological Institute using a 1.5 Tesla Siemens Vision scanner with a standard head coil. BOLD fMRI images were obtained using a T2*-weighted gradient echo (GE) echo planar imaging (EPI) sequence (TR = 3.36 s, TE = 51 ms, flip angle = 90°, FOV = 300 mm, matrix = 128 × 128). Images were taken in 120 whole-brain volumes (or "frames") per run (3.36 s/frame, ~7 min/run) with 10-13 contiguous axial slices of 7-mm thickness parallel to the AC-PC line (in-plane resolution 2.3 × 2.3 mm), covering the brain from the vertex to the base of the thalamus. High-resolution T1-weighted anatomical scans (3-dimensional gradient recalled echo, TR = 22 ms, TE = 10 ms, flip angle = 30°, 1-mm isotropic sampling) were acquired for all scanning sessions.

Data analysis

STATISTICAL ACTIVATION MAP. Functional data were motion corrected and low-pass filtered with a 6-mm FWHM Gaussian kernel in order to increase the signal-to-noise ratio. Activation maps, comparing painful heat to neutral heat conditions and tactile to rest conditions, were generated using fMRISTAT-MULTISTAT software developed at the MNI. This analysis yields t-statistics based on a linear model using random field theory, correlated errors, and Bonferroni correction; data are also corrected for temporal correlation, artifactual drift, and random effects. We have recently described the procedures in detail (Worsley et al. 2002; technical support available at http://www.math.mcgill.ca/keith/fmristat).

CONSTRUCTION OF REGIONS OF INTEREST (ROIs) AND EXTRACTION OF TIME-COURSE INFORMATION. For each subject, the central sulcus and Sylvian fissure were identified relative to other easily recognizable cortical landmarks, i.e., the superior frontal sulcus, the precentral sulcus, and the ascending marginal branch of the cingulate sulcus (Kido et al. 1980; Sobel et al. 1993). Then, directed searches were performed on identified somatosensory regions reliably activated by the brush and noxious heat stimuli. For each subject, the session showing the strongest activation by these two stimulus modalities was chosen for further comparison of their temporal components. ROIs were thus defined for the two stimulus modalities in each subject as the cluster of significant voxels surrounding the highest peak of stimulus-related activation within the SI and SII search areas. In general, the inclusion criterion for the ROI was conservatively defined as the global threshold t-value, as calculated for each individual subject (i.e., the minimum t-value, approximately 4.7, associated with a significance of P = 0.05 for a global search of the entire brain volume, defined as 1,000 cm³). In 3 of the 16 cases (4 subjects, 2 stimulus modalities, contralateral SI and SII), in which stimulus-related activation did not meet the strict global search threshold, ROIs were determined based on the appropriate threshold for a directed search (i.e., a minimum t-value, approximately 1.7, associated with a significance of P = 0.05 for a directed search of the target region). Subsequently, the original raw data (i.e., unanalyzed) from functional runs of like modality were averaged for individual subjects. The time course of activation was then extracted from these averaged functional runs by using the corresponding ROI as a selection mask. These data were further averaged, within the mean experimental run, to examine the time course of activation per stimulus cycle.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Psychophysical results

FUNCTIONAL SCANNING ESTIMATES. During the fMRI sessions, all subjects rated noxious heat stimuli as painful (mean rating 4.1 ± 0.1 out of 5.0 on the pain scale) and innocuous tactile stimuli (light mechanical brushing) as moderately intense but nonpainful (mean rating 2.0 ± 0.5 out of 5 on the nonpain scale). In addition, no significant differences were found between the ratings of pain experienced at the beginning and at the end of scanning runs (P = 0.12, paired t-test).

CONTINUOUS RATINGS. Figure 2 illustrates perceptual ratings evoked by noxious heat and innocuous brush stimuli obtained in a separate psychophysical experiment following the same stimulus paradigm as that used for fMR scanning. During the noxious heat condition, the temporal profile describing the perception of pain intensity (Fig. 2A, black line) was characterized by a delayed peak response, in which the perceived pain intensity gradually increased over time, exceeding the duration of the stimulation period, and finally reaching its maximum approximately 12 s after the onset of the stimulus. By contrast, during the brush condition, the subjects' perception of innocuous mechanical stimulation quickly reached its maximum response and remained at this level throughout the stimulation period (Fig. 2B, black line). Statistical comparison of these continuous ratings confirmed that perceived intensity of the heat pain stimulus is significantly delayed compared with that recorded for the brush stimulus (paired t-test, 2-tailed comparison of peak responses relative to stimulus onset: heat pain intensity = 11.86 s, brush intensity = 6.45 s, t = 6.87, P < 0.001).

View larger version (22K): [in this window] [in a new window]   Fig. 2. Continuous visual analogue scale (VAS) intensity ratings. Mean (±SE) intensity ratings recorded from 8 normal subjects during presentation of noxious heat stimuli (A: 45-46°C) and innocuous brushing stimuli (B: 2-cm soft artist's paint brush) to the lower leg. Arrows and the gray lines illustrate onset and duration of stimulation, respectively.

Somatosensory activation related to noxious heat and innocuous brush stimuli

Single-session analyses revealed significant pain- and brush-related activation within both SI and SII in all four subjects (Tables 1 and 2). Three of the four subjects demonstrated pain-related activation within SI contralateral to the stimulated body site, while one subject (subject 4) showed bilateral pain-related activation of SI (Fig. 3A). Gentle mechanical brushing produced significant activity within contralateral SI in all subjects (Fig. 3B, Table 1B); however, this innocuous tactile stimulation showed no indication of activating ipsilateral SI in these subjects. All stimulus-related activation sites observed within SI were limited to a 4.2-cm³ region approximating the area associated with cutaneous input from the leg.

View larger version (60K): [in this window] [in a new window]   Fig. 3. Pain- and brush-related SI activation. Activation sites within contralateral SI cortex were observed in all subjects in response to noxious heat (A) and innocuous brush stimulation (B) applied to the left leg. One subject (subject 4) showed bilateral activation during noxious heat stimulation. All coordinates are expressed in millimeters according to the Talairach atlas (Collins et al. 1994).

Within SII, pain-related activation (Fig. 4A) was restricted to the contralateral side in two subjects (subjects 1 and 3) and was bilateral in two subjects (subjects 2 and 4). Brush-related activation within SII (Fig. 4B) was observed bilaterally in all subjects.

View larger version (58K): [in this window] [in a new window]   Fig. 4. Pain- and brush-related SII activation. Activation sites within contralateral SII cortex were observed in all subjects in response to noxious heat (A) and innocuous brush stimulation (B) applied to the left leg. Two subjects (subjects 2 and 4) showed bilateral activation during noxious heat stimulation. All 4 subjects demonstrated bilateral activation of SII during brush stimulation. Coordinates are expressed in millimeters according to the Talairach atlas (Collins et al. 1994).

Examination of pain- and brush-related activations within contralateral somatosensory cortices showed some degree of variability across the four subjects. Within SI, pain- and brush-related activation sites were in close spatial proximity for two of the four subjects (Table 1 and Fig. 3); however, the remaining two subjects (subjects 3 and 4) demonstrated heat-related activation 8-14 mm displaced from that associated with their respective brush-related activation sites. Although the mean coordinates for SI sites of activation indicate a tendency for heat-pain activity to be displaced medial and inferior to the corresponding sites activated during brush stimulation, these differences are not consistent across all subjects. Within SII, pain- and brush-related activation sites were also in close spatial proximity for two of the four subjects (Table 2 and Fig. 4); however, the remaining two subjects (subjects 1 and 2), who showed up to 14-mm differences for sites activated by the two stimulus modalities, were those who had demonstrated the least spatial variability in SI activation. In general, activation sites in SII for pain and brush stimulation showed no consistent differences in spatial localization across all subjects.

Temporal analysis of SI and SII responses

Examination of the time course of somatosensory activation associated with the noxious heat and innocuous brush stimuli revealed a consistent and distinctive temporal signature for each of these stimulus modalities. The temporal profiles extracted from the fMRI data (averaged over one experimental session for each subject) are illustrated in Fig. 5. For activity evoked by noxious heat stimulation, all subjects demonstrated a maximum response approximately 15 s following stimulus onset, as recorded in SI and in SII (Fig. 5, A and B, top graphs). In contrast, brush-evoked activity for each subject approached its maximum response within 5-8 s following stimulus onset, and generally remained at this plateau for approximately 10 s before returning to baseline levels (Fig. 5, A and B, middle graphs).

View larger version (39K): [in this window] [in a new window]   Fig. 5. Temporal profiles of somatosensory activation per stimulus cycle. Each subject's mean time course of activation for noxious heat (top graphs) and innocuous brush (middle graphs) was extracted from contralateral SI and SII regions of interest centered on significant stimulus-evoked activity as described in Tables 1 and 2. Data are normalized, expressed as % change in signal intensity (±SE) relative to the prestimulation baseline, and averaged across all experimental runs in which the relevant stimulus was presented. Time is expressed relative to stimulus onset (arrow); data recorded at each full-brain scan (3.4-s intervals) are temporally registered at the midframe latency (scan onset + 1.7 s). The bottom graphs illustrate the mean time course of SI and SII stimulus-related activity, averaged across the 4 subjects.

Temporal differences between the pain- and brush-related responses are clearly illustrated in the bottom graphs of Fig. 5. Statistical comparison, across all subjects, of the peak responses evoked by these two stimulus modalities in SI and in SII confirmed a significant temporal delay in the response to noxious heat compared with that evoked by the innocuous brush stimuli (linear model, t₍₅₄₎ > 5.0, Ps < 0.001).

A smaller initial peak of activity observed during noxious heat stimulation is evident in the mean response recorded in SII approximately 5 s after application of the stimulus (Fig. 5B, bottom graph). This initial response to painful heat, which approximates temporally the onset of activity following application of the innocuous brush stimulus, is a variable feature in the SI response pattern (Fig. 5A, top graph), but is consistently observed in all subjects in SII (Fig. 5B, top graph). Statistical analysis of activity levels observed during the three components of this SII double response (initial and final peaks, separated by the relative decline in activity) reveals a highly significant main effect (ANOVA, 4 subjects, 3 conditions; F = 52.2, P < 0.001) and significant differences between each of the three components (peak 1 vs. interpeak response: F = 5.4, P = 0.002; interpeak response vs. peak 2: F = 12.2, P < 0.001; peak 1 vs. peak 2: F = 34.1, P < 0.001).

Perceptual versus fMRI response

The sensory experiences evoked by these two stimulus modalities (noxious heat and innocuous brush) differed not only in perceived intensity, but also in the temporal domain, as described above in the psychophysical results. Figure 6 illustrates, for the two stimulus modalities, the mean time course of perceptual estimates recorded during the continuous-rating experiment and the mean temporal profile of fMRI responses recorded during the comparable stimulation cycles. As described above, the maximum responses to noxious heat are significantly delayed relative to the maximum responses associated with innocuous brush stimulation, for both the perceptions evoked by these stimuli and the fMRI responses recorded in the somatosensory cortices during the presentation of each stimulus modality. Likewise, the 6 to 9-s difference between the perceptual maxima associated with noxious heat and innocuous brush (Fig. 6A) closely approximates the differences in the maximum fMRI responses to noxious heat pain and innocuous brush observed in SI (Fig. 6B) and SII (Fig. 6C), respectively. In addition, for each stimulus modality, the time lag between perceptual maxima and the associated fMRI responses is consistent with the accepted hemodynamic delay between neuronal activity and the related changes in deoxyhemoglobin detected by fMRI.

View larger version (23K): [in this window] [in a new window]   Fig. 6. Temporal comparison of SI activity and perceived intensity evoked by experimental stimuli. The functional magnetic resonance imaging (fMRI)-defined SI and SII activations evoked by noxious heat and innocuous brush (B and C) occur ~3 to 5 s following the corresponding peaks of the perceptual responses (A). This difference corresponds to the hemodynamic delay between neuronal activation and the de-oxyhaemoglobin-dependent fMRI response and is thus consistent with a role for this activation in cortical processes underlying the perception of these stimuli. fMRI activity is expressed as mean % change in signal intensity (described in Fig. 5)].

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study assessed the time course of stimulus-related activity (BOLD-fMRI) observed in human somatosensory cortices in order to examine the possible relevance of this activation to the human perception of pain. We hypothesized that, if stimulus-evoked activation of SI and/or SII contributes to the contemporaneous perception of pain intensity, then the time course of this activation should encode, or at least differentiate, the distinct temporal features that characterize the ongoing perceptions evoked by innocuous and noxious stimulation.

Other fMRI pain studies have examined temporal aspects of pain-related cerebral activation, but results have been less than definitive. Porro et al. (1998) demonstrated activation in a number of cortical areas (including SI) that was correlated with the perception of pain intensity measured over a period of 10-15 min following subcutaneous injection of ascorbic acid. However, the control group received only 20 s of innocuous cutaneous touch stimulation, thus limiting any direct comparison between the two stimulus modalities. In addition, individual subjects received only one stimulus modality, permitting only between-group comparisons (with a limited temporal resolution; two slices/21 s). More recently, Apkarian and colleagues (Apkarian et al. 1999) assessed cortical activity in the middle third of the brain during vibratory and heat pain stimulation and found that activation in posterior areas of parietal cortex were better related to temporal properties of pain perception compared with that observed in more anterior areas. In their study, however, both vibration and pain tasks required the subject to move his hand onto the stimulating surface, thus confounding motor and sensory components in each task. In addition, response variability in the thermal pain task necessitated averaging activation over all ROIs identified in the middle third of the brain, prohibiting any site-specific temporal analyses or direct comparisons of vibratory and heat-pain responses in SI cortex.

The present fMRI study utilized a within-subject design to directly compare activation observed during the presentation of noxious heat and innocuous brush stimuli of the same duration. Our results demonstrate that, although SI and SII are each activated by both innocuous brush and noxious heat stimulation, the time course of these activations is characterized by separate temporal signatures consistent with the time-dependent changes in perceived intensity reported by the subjects for these two modes of stimulation. The following discussion describes results of the current study in light of previous investigations and the implications of these results to understanding the role of somatosensory cortices in the perception of pain.

Spatial aspects of somatosensory cortex activation

In the present study, significant pain- and brush-related activation loci were observed in somatosensory areas located within the postcentral gyrus (SI) and along the upper bank of the Sylvian fissure (SII, parietal operculum). Within SI, stimulus-evoked activation was predominantly contralateral for both pain and brush stimuli, in agreement with the majority of previous studies using PET or fMRI imaging techniques (for review, see Bushnell et al. 1999). Although all SI activation sites fell within a 4.2-cm³ region of the postcentral gyrus (spatial coordinates normalized across all subjects), the relative positions of pain- and brush-related loci for individual subjects were variable, ranging from partial overlap to approximately 14 mm of separation between sites activated by the two stimulus modalities. Although the mean coordinates for SI sites of activation indicate a tendency for heat-pain activity to be displaced medial and inferior to the corresponding sites activated during brush stimulation, these differences were not consistent, nor significant, across all subjects.

Previous studies have not yielded a consistent view of the possible relationship between sites of SI activation evoked by noxious and innocuous stimuli. For example, Coghill et al. (1994) observed a substantial overlap between the activation sites associated with noxious heat and innocuous vibrotactile stimulation; however, the relatively poor spatial resolution of the PET technique may not resolve activation loci that differ by less than the spatial smoothing required for averaging across subject groups. Using magnetoencephalography (MEG) in human subjects, Ploner et al. (2000) described a single dipole source of laser pain-related activation in area 1 compared with two sources (in areas 3b and 1) evoked by innocuous electrical stimulation of a peripheral nerve. This approach, however, was restricted to an analysis of A fiber-mediated responses and could not assess the localization of the more slowly conducting C-fiber responses likely to contribute to the longer duration of pain perception described in the present study. A somewhat conflicting result was described by Tommerdahl et al., using intrinsic optical signal imaging of SI cortex; this study, in anesthetized monkey, demonstrated a distinct response to noxious heat within area 3a, rather than area 1, and showed a separate region of vibrotactile-evoked activation within areas 3b and 1 (Tommerdahl et al. 1996).

Compared with the predominantly contralateral activation that we observed in SI, stimulus-related activity detected in SII showed a stronger tendency to be bilateral, with all subjects exhibiting bilateral activation by innocuous brush stimuli and half of the subjects showing bilateral activation by the noxious heat stimuli (a nonsignificant trend toward bilateral activation by noxious heat was observed in one additional subject: subject 3). As seen in SI, the relative positions of pain- and brush-related loci for individual subjects did not differ significantly, but were variable, with a similar range of overlap or disparity. Interestingly, the two subjects who demonstrated the most disparity between innocuous- and noxious-related sites of activation within SII were those who showed the least spatial variability in SI activation evoked by these stimuli. Previous studies have likewise found little difference in the spatial extent of SII activation evoked by innocuous tactile and noxious thermal stimuli (Coghill et al. 1994; Ploner et al. 2000); however, the complex folding of the Sylvian fissure and the limited spatial resolution of MEG (and PET) may not allow for fine determination of spatial differences in activation sites. An answer to this issue awaits high-resolution fMRI studies of a larger number of subjects.

Time course of somatosensory activations

The most prominent finding revealed by the present fMRI time-course analysis was the extended period of activation associated with noxious heat stimuli, compared with that of innocuous brush stimuli (Fig. 5). These distinct temporal patterns of somatosensory cortex activity are consistent with the temporal patterns of perceived intensity observed in the psychophysical experiment, which showed that the perception of pain intensity increases gradually over time, reaching its maximum following stimulus offset, whereas the perception of touch (brushing) reaches its maximum soon after the onset of stimulation and terminates immediately following the offset of stimulation (Fig. 2). These data suggest that somatosensory cortex activation observed during noxious heat stimulation is indeed related to the processing of painful heat, and not simply to tactile input occasioned by contact of the thermode with the skin. Such observations are similar to results shown in primate studies using intrinsic optical imaging (Tommerdahl et al. 1996), where nociceptive neurons in area 3a of SI cortex were found to exhibit slow temporal summation and poststimulus response persistence after repeated cutaneous heat stimulation, paralleling the perceptual consequences of such stimulation in humans (Price et al. 1977).

Another temporal characteristic of somatosensory cortex activity evoked by noxious heat stimulation was the biphasic nature of the fMRI response. Although not robust within SI, this double-peak response was substantial and significant in SII across all subjects. This biphasic response is consistent with findings reported in a recent fMRI study using a similar type of thermal stimulus. Becerra et al. (1999) observed both early and late responses associated with noxious heat stimulation (46°C) during the initial two presentations of a repeated-stimulation paradigm; however, only the late response was seen (in several regions of the cortex, including SI and SII) during subsequent stimulation periods. Since only a single early peak had been observed during a neutral heat (40°C) condition, the authors concluded that the late portion of the double response was uniquely associated with the sensation of pain (Becerra et al. 1999).

In the present study, the early peak of activity observed in the pain condition approximates that observed with innocuous brushing and is thus consistent with a response to mechanical consequences of applying the thermode. In addition, the absence of a perceptual distinction in the psychophysical results, between "first pain" and "second pain," would argue against involvement of a nociceptive A component in the early portion of the byphasic cortical response. However, the 9-s stimulation paradigm used in the present study was not optimized for separating perceptual components of first- and second-pain (see Price et al. 1977), and possible contributions of thermal information to the early fMRI response observed in somatosensory areas during noxious heat stimulation cannot be ruled out. Although the 3-s resolution of the present fMRI paradigm limits identification of specific afferent contributions to cortical activation, the first peak in the SII response may well be related to myelinated input (conveying either touch or A-dependent heat information), while the second peak may reflect a contribution by the slower C-fiber afferents, which are responsible for temporal summation of noxious information observed in second-order nociceptive neurons (Campbell and Lamotte 1983; Lewis and Pochin 1937).

Hemodynamic response versus perception

A comparison of perceptual and fMRI hemodynamic responses revealed a close temporal relationship between pain perception and blood flow. This finding suggests that the activation observed in somatosensory cortices during noxious stimulation is related to the nociceptive component of the stimuli, providing evidence for the involvement of SI (and SII) in pain, as well as in innocuous tactile processing. Our results also show that the fMRI-defined activations, evoked by the innocuous tactile and noxious heat stimulation, followed the peak perceptual responses by approximately 3-5 s, a time lapse consistent with the hemodynamic delay that follows neuronal activation in the cortex (Bandettini et al. 1995).

Functional significance of somatosensory cortices in pain perception

The possible roles of SI and SII in pain perception have been addressed by studies in a number of different disciplines. As discussed above, neurophysiological studies in animals indicate that nociceptive pathways lead to the somatosensory cortices and that lesions to some of these cortical areas may alter the animal's ability to use noxious stimuli in discriminative tasks (see especially Kenshalo and Willis 1991). Although much has been learned from the rigorous experimental approach that can be used in animal studies, the relationship between these identified nociceptive responses and the human perception of pain remains indirect.

Early reports of human brain lesions indicated little or no involvement of the cortex in pain perception; however, more recent studies, incorporating detailed psychophysical assessments, have suggested a role for SI and SII in certain aspects of pain perception. Greenspan and Winfield (1991) described a patient with deficits in pain and tactile perception associated with a tumor located near the most posterior portion of the insula and parietal operculum. Following surgical removal of the tumor, the patient's perceptual capacities returned to normal, suggesting to the authors that SII and the posterior insular region are essential for the normal pain and tactile perception (Greenspan and Winfield, 1991). In a more recent report, Ploner et al. (1999) described an unusual example of pain deficits in a patient with an ischemic lesion of the right postcentral gyrus and parietal operculum, comprising the hand area of SI and SII. Noxious laser stimuli, presented to the patient's involved left hand, evoked a "clearly unpleasant" feeling that the patient wanted to avoid, but he could not localize the stimulus to the hand, and denied any qualitative characteristics like warm, hot, cold, touch, or even "pain-like." The authors argue that these results indicate an essential role of SI and SII for the sensory-discriminative aspects of pain perception, and suggest that pain affect does not require the integrity of SI and SII (Ploner et al. 1999). While these studies lend support to the notion of a critical role for the somatosensory cortices in pain perception, their implications are limited by problems inherent in human lesion data, i.e., difficulty in determining the precise delineation of the cortical lesions, possible damage to fibers of passage altering the function of distant regions, and the problematic situation of predicting normal cortical function based on abnormal and possibly compensatory behavior observed consequent to cortical damage.

Over the past decade, brain imaging studies in healthy normal subjects have added additional perspective to our understanding of the possible role of somatosensory cortices in pain perception. Studies using PET or fMRI techniques have generally shown activation, in response to painful stimuli, within a number of cortical areas, including the somatosensory cortices; however, inconsistent findings, especially in regards to pain-related activation of SI, have left the topic open to ongoing debate. The varying results among brain imaging studies may actually reflect the specialized role of SI in processing certain sensory aspects of noxious stimulation, i.e., differences among experimental paradigms in the quality, intensity, location, spatial extent, and timing of noxious stimuli may impact substantially on the "pain-related" activation observed in SI. Evidence supporting a critical role of stimulus intensity in evoking SI and SII responses was recently demonstrated by Timmermann et al. (2001). Using magnetoencephalography in healthy subjects, they found that SI activity closely matched subjects' ratings of pain intensity over a range of noxious stimuli; SII showed stronger pain-evoked responses, but only at the higher stimulus intensities. The authors suggested that many pain imaging paradigms would be more likely to detect the larger differential of activity demonstrated by SII, even though SI responses were more consistent with a sensory-discriminative assessment of noxious input. While this hypothesis may explain some discrepancies in the literature, it is not altogether consistent with those studies demonstrating SII activation with innocuous or mildly noxious stimuli (e.g., Coghill et al. 1994; Jones et al. 1991) nor with those demonstrating pain-intensity encoding within SII (see especially Coghill et al. 1999).

Varying cognitive states (resulting from specific prescan instructions) may also differentially modulate SI activity, giving further clues to the functional importance of this structure to the perception of pain. Considerable evidence indicates that SI responsiveness to noxious stimuli is highly susceptible to cognitive manipulations such as attention and hypnotic induction. Directing attention toward or away from a painful heat stimulus not only modifies the subjective intensity of pain, but also modulates activity within SI (Carrier et al. 1998). Likewise, hypnotic suggestions directed specifically toward changing the perception of pain intensity result in a preferential modulation of pain-related activity in SI (Hofbauer et al. 1998, 2001), whereas a selective change in pain unpleasantness by hypnosis has no effect on pain-related activity in SI (Rainville et al. 1997). Thus the influence of attentional and cognitive factors on both the sensation of pain and on SI activation provides further support for the role of this structure in the perception of pain.

Recent evidence suggests that nociceptive input to SI may also serve to modulate tactile perception by means of inhibition, as originally proposed in the "touch gate" theory (Apkarian et al. 1992, 1994). Results from intrinsic optical imaging studies in primates demonstrate that noxious heat stimulation evokes an increase in the intrinsic signal in area 3a, as well as a decrease in activation of areas 3b and 1 associated with low-threshold mechanical stimulation (Tommerdahl et al. 1996). Such findings suggest that SI activation, observed during noxious stimulation, may be related to a modulation of touch sensation that is proportional to the perception of pain; however, a potential role for SI nociceptive activity in the modulation of innocuous cutaneous sensations does not rule out or preclude a role for this structure in sensory aspects of pain perception.

In conclusion, different lines of evidence have pointed toward a role for the somatosensory cortices in pain perception, including a growing number of studies that have utilized functional imaging. However, inconsistent results among some of these studies had led to controversy and suggestions that activation observed during noxious stimulation might only reflect spatial cues or attentional factors related to innocuous aspects of stimulation. Our data now show, in normal intact human subjects, that activation in both SI and SII encodes a temporal signature specific to the perceptual characteristics of both noxious thermal and innocuous tactile stimulation, thus indicating a role for these cortical areas in processing sensory information inherent to the perception of pain intensity. Future studies will continue to investigate whether the processing of this "pain-related" information by the somatosensory cortices is sufficient or even necessary for the overall, multidimensional phenomenon of pain perception.