Although multiple regions of the cerebral cortex have been implicated in swallowing, the functional contributions of each brain area remain unclear. The present study sought to clarify the roles of these cortical foci in swallowing by comparing brain activation associated with voluntary saliva swallowing and voluntary tongue elevation. Fourteen healthy right-handed subjects were examined with single-event–related functional magnetic resonance imaging (fMRI) while laryngeal movements associated with swallowing and tongue movement were simultaneously recorded. Both swallowing and tongue elevation activated 1) the left lateral pericentral and anterior parietal cortex, and 2) the anterior cingulate cortex (ACC) and adjacent supplementary motor area (SMA), suggesting that these brain regions mediate processes shared by swallowing and tongue movement. Tongue elevation activated a larger total volume of cortex than swallowing, with significantly greater activation within the ACC, SMA, right precentral and postcentral gyri, premotor cortex, right putamen, and thalamus. Although a contrast analysis failed to identify activation foci specific to swallowing, superimposed activation maps suggested that the most lateral extent of the left pericentral and anterior parietal cortex, rostral ACC, precuneus, and right parietal operculum/insula were preferentially activated by swallowing. This finding suggests that these brain areas may mediate processes specific to swallowing. Approximately 60% of the subjects showed a strong functional lateralization of the postcentral gyrus toward the left hemisphere for swallowing, whereas 40% showed a similar activation bias for the tongue elevation task. This finding supports the view that the oral sensorimotor cortices within the left and right hemispheres are functionally nonequivalent.
Recent neuroimaging studies have consistently shown that swallowing activates multiple regions of the human cerebral cortex (Hamdy et al. 1999a, b; Kern et al. 2001; Martin et al. 2001; Mosier et al. 1999a, 2001; Zald and Pardo 1999). The most prominent activation foci correspond to the lateral pericentral cortex, perisylvian cortex, anterior cingulate cortex (ACC), and right insula. These regions have been implicated in voluntary swallowing of a water bolus (Hamdy et al. 1999a, b; Martin et al. 1997a, 2001; Mosier et al. 1999a 2001), voluntary swallowing of saliva (Kern et al. 2001; Martin et al. 2001; Mosier et al. 1999a, 2001; Zald and Pardo 1999), and “autonomic” swallowing of saliva in naïve subjects (Martin et al. 2001). The latter finding suggests cortical mechanisms may play a role in regulating swallowing, even when swallowing is not a willed action.
Electrophysiologic studies in nonhuman primates and humans have implicated many of the same brain regions in swallowing (Martin and Sessle 1993). For example, swallowing can be evoked by intracortical microstimulation (ICMS) applied to 4 regions of the pericentral/perisylvian cortex in the awake primate: the face primary motor cortex (face-MI), face primary somatosensory cortex (face-SI), Brodmann's area (BA) 6 immediately lateral to face-MI, and the inner face of the frontal operculum (Martin et al. 1999). Furthermore, “swallow-related” neurons have been documented in the primate face-MI and lateral BA 6 (Martin et al. 1997b; Yao et al. 2002). Reversible cold block of the ICMS-defined “swallow cortex,” including BA 4 and 6, has been shown to alter swallowing electromyographic (EMG) activity in awake primates (Narita et al. 1999; Yamamura et al. 2002). In humans, cortical electrical stimulation (Penfield and Rasmussen 1950) and, more recently, transcranial magnetic stimulation (TMS) (Hamdy et al. 1996, 1997) studies have suggested that the muscles active in swallowing are represented within the lateral pericentral and premotor cortices.
Clinical reports of swallowing impairment after bilateral or unilateral hemispheric stroke also have implicated the cortex in swallowing regulation (Hamdy et al. 1997; Meadows 1973; Robbins and Levine 1988). However, the clinical literature is divided on the issue of an association between infarct location/size and patterns of swallowing deficits. Whereas some studies have implicated the lateral pericentral cortex, posterior inferior frontal gyrus, and anterior insula, and suggested patterns of swallowing abnormalities are dependent on stroke laterality (Daniels and Foundas 1997; Meadows 1973; Robbins and Levine 1988; Robbins et al. 1993; Tuch et al. 1941), others have reported only modest correlations between swallowing deficits and infarct characteristics (Alberts et al. 1992; Johnson et al. 1992; Smithard et al. 1997).
Taken together, these neuroimaging, electrophysiologic, and clinical findings suggest that a number of spatially discrete cortical and subcortical foci may be organized within a functional network for swallowing. Nevertheless, the specific contributions of each brain region in regulating swallowing remain unclear. Swallowing is a complex function involving sensory processing of ingested material, execution of oral, pharyngeal, and laryngeal movements, coordination with mastication and respiration, and, if performed volitionally, cognitive/attentional processing. Given these multiple subprocesses of swallowing, we propose that the discrete cortical and subcortical foci previously implicated in swallowing mediate functionally distinct components of the swallowing act. Some brain regions may process sensory attributes of the bolus, for example, whereas others mediate motor planning and/or execution.
The present study was aimed at elucidating the specialized roles of the cortical and subcortical foci previously implicated in swallowing. Brain function for swallowing was inferred by comparing and contrasting brain activation associated with swallowing with that evoked by voluntary tongue elevation. Tongue elevation was selected as the comparison behavior because it represents one component of the integrated swallowing motor sequence that can be produced volitionally in isolation. In addition, previous brain imaging studies have reported that tongue movement activates the human pericentral cortex (Corfield et al. 1999; Wildgruber et al. 1996; Zald and Pardo 1999). Although one functional magnetic resonance imaging (fMRI) study compared tongue “rolling” with blocks of repetitive swallowing (Kern et al. 2001), there are no published reports of single-event–related fMRI studies comparing single trials of tongue movement with single trials of swallowing. However, the single-event–related paradigm is absolutely critical when comparing swallowing with other oral behaviors based on the well established fact that repetitive swallowing becomes progressively more difficult over time (i.e., deglutitive inhibition), related possibly to reduced saliva (Logemann 1998), central inhibitory mechanisms, and smooth muscle refractoriness (Bove et al. 1998; Tutuian et al. 2004). Because these inhibitory mechanisms certainly may influence the cortical processing of swallowing, the present study compared averaged single trials of tongue elevation with averaged single trials of swallowing, with successive swallowing trials separated by ≥2 min. Regions of brain activation common to both swallowing and tongue elevation tasks were interpreted as regions mediating processes shared by the 2 behaviors (e.g., motor planning and motor execution). Brain regions activated during swallowing but not tongue elevation were interpreted as playing functional roles more specific to swallowing regulation, that is, mediating processes not involved in tongue movement. Some of these data were previously reported briefly in abstract form (Martin et al. 2002).
Fourteen healthy volunteers, 12 female and 2 male (age, 28.0 ± 6.5 yr, mean ± SD) who were right-handed, as measured by the Edinburgh Handedness Inventory (Oldfield 1971), participated as subjects. Four subjects had previous fMRI experience. All subjects gave written informed consent before participating in this study. The study protocol was approved by the University of Western Ontario Review Board for Health Sciences research involving human subjects. The study adhered to MRI safety guidelines established for clinical scanners by the United States Food and Drug Administration.
Each subject participated in three 8-min functional imaging runs during a single experimental session. During each of the functional runs, 3 activation tasks were performed in semirandom order. During the first imaging run, the 3 activation tasks were voluntary swallowing of saliva, voluntary tongue elevation, and voluntary finger opposition. Although the main focus of the study was to compare and contrast swallowing and tongue elevation, the finger-opposition task was included as a means of determining whether areas of activation associated with swallowing and/or tongue elevation were specific to an oral sensorimotor task, or whether they were also activated in association with a limb motor task. In addition, we sought to obtain a “functional landmark,” corresponding to the hand/digit representation within the sensorimotor cortex, relative to which swallowing and tongue elevation activation maps could be examined (Colebatch et al. 1991; Grafton et al. 1991). The second imaging run consisted of voluntary saliva swallow, effortful saliva swallow, and finger-opposition tasks. During the third imaging run, the activation tasks were voluntary saliva swallow, voluntary apnea, and voluntary finger opposition. The present paper reports our findings on the comparison of voluntary swallowing, tongue elevation, and finger opposition performed during the first imaging run. Findings from the latter 2 imaging runs will be reported elsewhere. Therefore the present description of the methodology is limited to the first imaging run.
Each subject was trained on the activation tasks prior to the experiment. For the voluntary saliva swallow, the subject was instructed to swallow her/his accumulated saliva once without producing exaggerated oral movements. Cues for swallowing trials were separated by ≥2 other motor trials (i.e., ≥120 s) in an effort to increase the volume of saliva swallowed. The instruction for the tongue-elevation task was to raise the tongue body to the palate and maintain this elevated position for the duration of the visual cue (i.e., 2 s; see following text). For the finger opposition, the subject was trained to oppose the thumb and index finger of the right (dominant) hand at a rate of about 2 Hz for the duration of the visual cue. The 3 activation tasks were performed in response to visual cues that were back-projected onto a small mirror mounted above the subject's eyes. The visual cues for the swallowing, tongue-elevation, and finger-movement tasks were the words swallow, tongue, and finger, respectively. Between these task cues, the word rest was displayed. With an interstimulus interval of 40 s, each of the 3 activation tasks was performed 6 times during the 12-min imaging run.
IDENTIFICATION OF SWALLOWING AND TONGUE ELEVATION.
Single swallowing and tongue elevation trials were verified on the basis of their distinct profiles of laryngeal movement (Logemann et al. 1992; Martin et al. 2001). Laryngeal movements were recorded (PowerLab, v4.1.1, AD Instruments, Castle Hill, Australia) from the output signal of a pressure transducer driven from expanding magnetic resonance (MR)–compatible bellows (Siemens, Erlangen, Germany) positioned comfortably over the subject's thyroid cartilage. Laryngeal movements associated with the swallowing and tongue movement tasks could be distinguished on the basis of their distinct dynamics (see Fig. 1). As expected, the finger-apposition task was not accompanied by laryngeal movement.
All imaging experiments were performed on a Varian/Unity INOVA 4 Tesla (T) whole body imaging system (Varian, Palo Alto, CA) equipped with 40 mT/m Siemens Sonata actively shielded whole body gradients and amplifiers (Siemens). A whole head quadrature birdcage radio frequency (RF) coil was used to transmit and receive the MR signal (Barberi et al. 2000). The subject's head was immobilized with foam padding that was fit snugly between the head and a Plexiglas head cradle within the head coil.
Imaging planes for the functional scans were prescribed with the aid of a high-resolution [256 × 256, 28-cm field of view (FOV)] sagittal anatomic image with gray/white matter contrast (i.e., T1-weighted) acquired using a longitudinal magnetization–prepared fast low-angle shot (FLASH) imaging sequence [inversion time (TI) = 750 ms, echo time (TE) = 3.5 ms, repetition time (TR) = 12 ms, flip angle = 11°, 5 mm slice thickness]. From this scout image, 15 contiguous axial slices were prescribed with a slice thickness of 5 mm oriented in a plane approximately parallel to the anterior commissure (AC)–posterior commissure (PC) line and extending from the superior extent of the paracentral lobule to about 10 mm below the AC–PC plane. During each functional task described above, blood oxygenation level–dependent (BOLD) images (i.e., T2*-weighted) were acquired continuously using an interleaved, centrically ordered 4-segment, echoplanar imaging (EPI) sequence (64 × 64 matrix size, TR = 500 ms, TE = 10 ms, flip angle = 30°, 19.2 cm FOV, volume collection time = 2 s). Each image was corrected for physiologic fluctuations using a navigator echo that was collected at the beginning of every image segment (Hu and Kim 1994). At the end of the experimental session, anatomic reference images were acquired along the same orientation as the functional images using an inversion-prepared 3-dimensional (3D) turbo-FLASH sequence (256 × 256 × 128 matrix size, 1.25 mm reconstructed slice thickness, TI = 600 ms, TR = 10 ms, TE = 5.5 ms, 19.2 cm FOV).
fMRI data analysis
Previous studies have shown that swallow-related movements of the tongue and mandible occurring outside the imaging FOV can disturb the magnetic field in nearby imaging slices and thus contribute to false-positives in the BOLD signal (Birn et al. 1999). These swallow-related movements produce both magnitude and phase changes in the complex-valued MRI signal, with the phase change corresponding in time to swallow-related laryngeal movements (Martin et al. 2001). In contrast, hemodynamic events within the microvasculature (i.e., any vasculature smaller than intracortical veins) are expected to produce only magnitude changes in the MR signal (Menon 2002). Based on these considerations, we applied a motion-suppression algorithm developed by Menon (2002) in which the fraction of the BOLD signal arising from motion is estimated and removed by measuring its influence on the phase angle of the complex-valued fMRI time series. The algorithm determines a maximum likelihood estimator based on a linear least-squares fit of the MRI signal phase to the BOLD signal magnitude in each voxel. Baseline drift in the MR time course of each voxel was then removed by applying a band-pass filter. To verify the effects of the motion-suppression algorithm on the activation maps, the analyses were repeated for 3 subjects without applying the motion-suppression procedure. One of the 3 subjects had previous fMRI experience and had been shown to have minimal head movement based on previous analyses of 3D motion correction. The other 2 subjects had no fMRI experience and thus we anticipated that they would demonstrate relatively greater task-related motion.
Subsequent image analyses were performed using BrainVoyager v4.9 (Brain Innovation, Maastricht, The Netherlands; Goebel et al. 1996). Volume–time courses were generated by coregistering the 2-dimensional (2D) functional slices with the 3D anatomic images. Anatomic images were aligned with the AC–PC plane and transformed to standard stereotaxic space (Talairach and Tournoux 1988). Small head movements were removed by applying a 3D motion correction sinc interpolation to the volume–time courses. A Levenberg–Marquardt nonlinear least-squares method fit 6 parameters (i.e., 3 translation, 3 rotation) of each image volume to a reference volume (Press et al. 1992). The data were then spatially convolved with a Gaussian filter (full width at half-maximum = 4 mm) to facilitate intersubject analyses.
Both group and single-subject data analyses were performed. Brain activation associated with swallowing, tongue elevation, and/or finger opposition was identified using multiple regression analyses. Predictors were generated by convolving a square-wave function representing the time course of the experimental tasks with a γ function (δ = 1.25; τ = 2.5) representing the hemodynamic impulse response (Boynton et al. 1996; Cohen 1997). The experimental time course was defined from the task triggers or, alternatively, the laryngeal movement signal, in 3 separate analyses as follows. In the first analysis, a value of 1 in the square-wave function was assigned to the single-imaging volumes during which the triggers for the visual task cues were delivered, 0 being assigned to all other imaging volumes. Because the time courses of the swallowing and tongue elevation tasks differed slightly in terms of response latency and/or duration for some subjects on certain trials (see results), the regression analyses were repeated in 2 ways. To address task latency, a second analysis was performed in which the reference function was assigned a value of “1” during the single-imaging volume that corresponded to a swallow-related or tongue-related laryngeal movement peak for each trial and subject. Finally, to examine the effect of task duration, a third analysis was conducted in which the reference time course was assigned a value of “1” during the consecutive-imaging volumes beginning when the task trigger was generated and ending when the laryngeal signal reached its peak for a given task trial. With the imaging volume duration set at 2 s, 2% of the swallows and 14% of the tongue elevation trials for the entire group were assigned a duration of 2 imaging volumes (i.e., 4 s) in this last analysis.
The first phase of the statistical analysis used a random-effects model and tested, on a voxel-by-voxel basis, for significant activation associated with the swallowing, tongue-elevation, and finger-apposition tasks for the group of 14 subjects. A P < 0.0001 (uncorrected for the total number of voxels tested) was considered statistically significant for each task. Activation maps for each task also were generated for individual subjects (P < 0.05, Bonferroni corrected for the total number of voxels tested).
Second, to determine regions of activation common to both the swallowing and tongue elevation tasks, we then tested for the significant conjunction of activations associated with swallowing and tongue elevation across subjects (random effects, P < 0.0001; Price and Friston 1997) and within individual subjects (P < 0.0001). A similar analysis was performed to examine activations common to the swallowing and finger-opposition tasks.
Third, to determine activations specific to either swallowing or tongue elevation, but not common to both, we tested for the significant contrast between the swallow and tongue conditions for the group (random effects, P < 0.0001) and for single subjects (P < 0.0001). In addition to this statistically based method of determining activations specific to swallowing or tongue movement, differences between conditions were represented graphically as follows. Regions of activation associated with the swallowing, tongue-elevation, and finger-opposition tasks, as well as regions determined from the conjunction analyses to be significantly activated by swallowing and tongue elevation, or swallowing and finger opposition, were also superimposed as color-coded regions of activation on selected anatomic slices and on the inflated brain representation of a single subject.
A region-of-interest (ROI) analysis was also performed in an attempt to 1) explore the extent to which the group activation maps for swallowing and tongue movement accurately reflected activation patterns within individual subjects, and 2) further examine activation differences as a function of task and cerebral hemisphere. Two types of ROI were defined. First, the spatial extents (i.e., total volumes) of 5 prominent activations obtained from the group map for swallowing were defined as “functional swallow ROIs”: a postcentral ROI, precentral ROI, anterior cingulate BA 32 ROI, anterior cingulate BA 24 ROI, and precuneus ROI. Second, the Brodmann areas and gyri corresponding to these functional ROIs were defined as “anatomic swallow ROIs.” These included BA 4, 6, 3, 24, 32, 40, 43, 7, and the precentral, postcentral, cingulate, and supramarginal gyri. In addition, for each functional or anatomic ROI corresponding to the precentral and postcentral gyri, a second ROI was defined with respect to the identical stereotaxic location in the opposite hemisphere. This analysis, aimed at comparing activation in the 2 hemispheres, was not applied to regions of activation along the midline (e.g., ACC) because of the relative difficulty in assigning a midline voxel to one hemisphere.
For each ROI so defined, the number of significantly activated voxels and the percentage MR signal change were calculated for each subject and task. The number of activated voxels in a ROI was determined using the multiple regression analyses described above at P < 0.05 for an individual subject activation map, with the Bonferroni correction applied to account for the total number of voxels tested within a ROI. Percentage signal change was calculated over a period of 7 volumes immediately after each task cue, relative to a baseline period of 3 volumes (i.e., 6 s) immediately before each task cue. Both the number of activated voxels and the percentage MR signal change were submitted to pairwise t-test to determine differences between swallowing and tongue-elevation tasks, and between right and left hemispheres. The significance level was set familywise at 0.05, with ROI defined as the unit of analysis, yielding a Bonferroni-corrected P < 0.0125 (i.e., 4 t-test per family: swallow vs. tongue task left hemisphere, swallow vs. tongue task right hemisphere, left vs. right hemisphere swallow task, and left vs. right hemisphere tongue task).
The number of activated voxels in left–right hemisphere ROI pairs were used in calculating laterality indices (LI) for the anatomically defined precentral gyrus and postcentral gyrus as follows (1)
Effects of motion-suppression algorithm
Comparison of the fMRI swallowing data analyzed with and without the motion correction algorithm indicated the following. First, for voxels that were significant in the vast majority of single subjects and in the group maps, there was a close correspondence between the motion-corrected and -uncorrected maps, suggesting that these significant regions reflected the BOLD effect rather than motion (see Fig. 2b). Within these regions, which included the pericentral cortex and ACC (see following text), correspondence between the motion-corrected and -uncorrected maps did vary across subjects, suggesting intersubject variation in swallow-coupled motion. Second, for significant voxels that were less consistent across subjects and not identified as significant in the group map, such as those located along the edges of the brain, along the ventricles, and in the lower slices, there was far less correspondence between the motion-corrected and -uncorrected data (see Fig. 2a). Given previous findings that these areas are highly susceptible to motion artifacts (Birn et al. 1999), the results of our motion-correction analysis suggest that these latter MR signal changes were associated with motion and not brain activation. Thus the motion-suppression algorithm appeared to selectively reduce significant voxels in brain regions previously documented to be most affected by motion.
Task response latency
Across subjects, the mean latency from the computer-generated trigger for the visual cue to the peak amplitude of the swallow-related signal recorded from the laryngeal movement transducer was 1.46 ± 0.31 s (mean ± SD, n = 13). The mean latency for the tongue elevation task was 1.94 ± 0.72 s (n = 13). As these data indicate, the subjects responded immediately to the visual cues in both task conditions, with the tongue task showing a somewhat longer response latency.
Activation associated with swallowing
The voluntary saliva swallow activated a number of brain regions across subjects (see Fig. 3a and Table 1). The total activated brain volume in the group map was 2,635 mm3 (P < 0.0001). The largest groupwise activation was located at the junction of the lateral postcentral gyrus, parietal operculum, and supramarginal gyrus (i.e., BA 43, 40) in the left hemisphere only, suggesting a strong functional lateralization for swallowing. A second prominent area of activation corresponded to the ACC, including BA 32 and 24, as well as a small area adjacent to the SMA. Activation was identified within the precuneus and cuneus, as well as the right frontoparietal operculum and insula. There was modest activation of the right precentral gyrus (i.e., BA 4, z = 54), as well as a smaller region of activation of the left lateral pericentral cortex (i.e., BA 4, 3; z = 25). Other small areas of activation corresponded to the middle frontal gyrus (i.e., BA 46, 9) and superior temporal gyrus bilaterally (i.e., BA 22).
In addition to these regions of activation, small areas of significant MR signal decrease were also found. These corresponded to the posterior cingulate gyrus, left middle temporal gyrus, and the depth of the central sulcus, near the superior surface of the brain (i.e., z = +61). These areas showing MR signal decrease represented only 4% of the total number of significant voxels.
Activation associated with tongue elevation
The tongue-elevation task activated a greater total brain volume than the swallowing task [i.e., tongue-task–activated brain volume: 11,143 mm3 (P < 0.0001); see Fig. 3b and Table 2]. The largest activation corresponded to the lateral precentral and postcentral gyrus (i.e., BA 4/6/3/2), with the activation volume in the left hemisphere being greater than twice the volume in the right hemisphere. Furthermore, the centroid of the precentral gyrus activation in the right hemisphere was about 15 mm superior to that in the left hemisphere (i.e., Talairach coordinates, right vs. left: x = 47, y = −6, z = 37 vs. −46, −7, 22). There was a large activation focus within the ACC, including BA 24 and 32, and the SMA. The right putamen was also identified as a large activation focus, as well as the thalamus bilaterally. Tongue elevation also activated the precuneus and supramarginal gyrus (i.e., BA 40). Smaller activations corresponded to the pars opercularis and inner face of the frontal operculum, middle frontal gyrus (i.e., BA 9/10), the depth of the precentral sulcus, the superior frontal gyrus, and the right superior temporal gyrus (STG).
Activation associated with finger–thumb opposition
The finger–thumb opposition task activated the SMA and ACC, the putamen particularly in the left hemisphere, and the left supramarginal and left precentral and postcentral gyrus. Smaller activations were found within the precuneus and cuneus, insula bilaterally, and right STG (see Fig. 3c and Table 3).
Comparison of swallowing, tongue elevation, and finger opposition
ACTIVATION COMMON TO SWALLOWING AND TONGUE ELEVATION.
The group conjunction analysis identified 2 large regions of activation common to swallowing and tongue elevation (see Fig. 4 and Table 4). These corresponded to 1) the left lateral postcentral and precentral gyri, parietal operculum, and supramarginal gyrus (i.e., BA 43, 40, 3, 1, 2, 4), and 2) the ACC and immediately adjacent SMA. Substantially smaller activations common to swallowing and tongue elevation were identified in the right precuneus and cuneus.
Results of the single-subject conjunction analyses are presented in Table 5. In addition to the areas of activation common to swallowing and tongue elevation identified in the group analysis, at least half of the subjects displayed activation within the superior and middle temporal gyri, superior and middle frontal gyri, and the lingual gyrus.
ACTIVATION COMMON TO SWALLOWING AND FINGER OPPOSITION.
Areas of activation common to the swallowing and finger opposition tasks were limited to the SMA and a very small area of the caudal ACC (i.e., BA 24) (see Table 4). The small activation focus within the ACC was limited to the caudal extent of the ACC activation focus common to swallowing and tongue elevation. Thus the region of ACC activated by both swallowing and tongue movement was not activated in association with the finger task.
ACTIVATION COMMON TO TONGUE ELEVATION AND FINGER OPPOSITION.
Regions of activation common to the tongue-elevation and finger-opposition tasks included the SMA and ACC (principally BA 24), as well as a very small activation within the right STG (see Table 4). Again, the ACC and SMA activation foci were caudal to those regions activated by both the swallowing and tongue-movement tasks.
Contrast of swallowing and tongue elevation
The contrast analysis identified a number of large regions where activation associated with tongue elevation was greater than that associated with swallowing (see Fig. 5). The most prominent of these areas corresponded to the SMA and ACC (i.e., BA 24). Thus whereas the conjunction analysis identified regions of ACC and adjacent SMA activated in common by both swallowing and tongue movement, the contrast analysis indicated that other regions of the SMA and ACC were preferentially activated by the tongue task. Other prominent regions where tongue-movement–related activation exceeded swallow-related activation included the precentral gyrus bilaterally, right postcentral gyrus, premotor cortex, right putamen, and thalamus. However, the contrast analysis failed to identify any regions where activation for swallowing was significantly greater than that for tongue elevation. Because the contrast analysis was performed using a random-effects model, yielding a conservative statistical test, the data were reanalyzed for a significant contrast using a fixed-effects model (P < 0.05, Bonferroni correction applied to account for the total number of voxels analyzed). Two very small areas were identified in which activation for swallowing was greater than that for tongue elevation. These corresponded to the precuneus and cuneus (see Fig. 5). Reanalysis of the data using the reference function constructed relative to the laryngeal movement signal produced almost identical results, with the addition of a very small (i.e., 4 mm3) region of activation within the SFG and adjacent ACC (i.e., BA 9, 10, 32).
In distinction to the results of the contrast analysis, the superimposed maps showing activation associated with the swallowing, tongue-elevation, and finger-opposition tasks indicated that the following areas were activated only in association with swallowing: 1) a region of the left lateral precentral/pericentral cortex immediately lateral to the region activated by both swallowing and tongue movement, 2) a region corresponding to the junction of the left postcentral and supramarginal gyri, 3) a rostral region of the ACC, 4) the precuneus and cuneus, 5) the middle frontal gyrus, and 6) the right parietal operculum/insula (see Figs. 5 and 6). Similar results were obtained when the regression analysis was repeated with respect to the laryngeal movement time series.
Results of the single-subject contrast analyses are presented in Table 6. There was substantial variation across subjects in the number of areas for which activation associated with swallowing exceeded that associated with tongue movement. Four of the 14 subjects showed several areas of greater activation for swallowing, whereas other subjects showed a single area. In agreement with the group contrast analysis, the cuneus and superior frontal gyri showed greater activation for swallowing than for tongue movement in over half the subjects. Activation of the middle and inferior frontal gyri also was greater for swallowing than for tongue elevation in several subjects.
TASK EFFECTS ON PERCENTAGE MR SIGNAL CHANGE.
The ROI analyses revealed 2 brain regions for which the tongue task evoked a significantly greater average percentage MR signal change than the swallow task. These correspond to 1) the functionally defined postcentral ROI within the right hemisphere (t = 3.01, df = 13, P = 0.0038; see Fig. 7), and 2) the functionally defined ACC area 24 ROI (t = 3.01, df = 13, P = 0.008). These findings are consistent with the results of the contrast analysis reported above.
HEMISPHERE EFFECTS ON PERCENTAGE MR SIGNAL CHANGE.
Comparison of left and right hemisphere percentage signal change showed a significant left > right hemisphere difference for the swallowing task for the functionally defined postcentral gyrus (t = 2.65, df = 13, P = 0.01; see Fig. 7), and a left > right difference for BA 3 for the tongue task (t = 3.01, df = 13, P = 0.009; see Fig. 8).
HEMISPHERE EFFECTS ON NUMBER OF ACTIVATED VOXELS: LATERALITY INDICES.
Postcentral gyrus. Laterality indices (LI) showed that the swallow-related activation of the postcentral gyrus was strongly lateralized to the left hemisphere in 8 of the 14 subjects who had significant activation in this region (i.e., LI > +0.5; see Fig. 9a). This ROI-based finding is consistent with the group activation map for swallowing, indicating strong lateralization of the postcentral gyrus to the left. However, the LI also indicated that swallow-related activation of the postcentral gyrus was strongly lateralized to the right hemisphere for 2 subjects, and not lateralized for 4 subjects. These findings suggest that, whereas the processing of swallowing within the postcentral gyrus typically is lateralized to the left hemisphere, there is substantial intersubject variation in swallow-related laterality.
For the tongue-elevation task, activation of the postcentral gyrus was strongly lateralized to the left hemisphere for 6 of 14 subjects, and strongly lateralized to the right for 2 subjects. Six subjects showed no clear asymmetry. Thus although the tongue-elevation task was not a laterally directed movement, over half the subjects showed a strong functional lateralization of the postcentral gyrus, typically toward the left hemisphere.
Precentral gyrus. The LI analysis indicated that activation of the precentral gyrus in association with swallowing was strongly asymmetric toward the left hemisphere in 5 subjects, strongly lateralized to the right hemisphere in 4 subjects, and bilateral in 5 of 14 subjects. For the tongue-elevation task, 12 of the 14 subjects showed no clear asymmetry, whereas 2 showed a strong asymmetry toward the right hemisphere (see Fig. 9b).
Overall, there appeared to be greater lateralization of the postcentral gyrus than the precentral gyrus for both the swallowing and tongue-elevation tasks. In addition, lateralization of the postcentral and precentral gyri was greater for the swallowing task than for the tongue-elevation task.
Correlation analyses revealed a moderate negative relationship between lateralization of activation for the swallowing and tongue tasks for the postcentral gyrus (r = −0.55), but no relationship for activation of the precentral gyrus (r = −0.07). Thus for the postcentral gyrus only, there was a tendency for lateralization for swallowing and tongue elevation to be toward opposite hemispheres. For both the swallowing and tongue tasks, functional lateralization of the precentral and postcentral gyri were moderately positively correlated (i.e., swallow task: r = +0.49; tongue task: r = +0.64). Thus although there was a tendency for the precentral and postcentral gyri to be lateralized to the same side for a given task, there was not a 1:1 correspondence. Some subjects showed a dissociation of hemispheric lateralization of the precentral and postcentral gyri.
Activation common to swallowing and tongue movement
The aim of the present study was to compare the cortical representations of swallowing and tongue movement as a means of inferring brain function for swallowing. We found that a number of cortical regions that were activated by swallowing also were activated by voluntary tongue elevation, the most consistent being the lateral pericentral cortex, frontoparietal operculum, and ACC. Based on their common activation for swallowing and tongue movement, we propose that these cortical areas mediate processes shared by the 2 oral sensorimotor tasks; they do not appear to be sites of “swallow-specific” processing.
Although future studies are needed to elucidate the specific functional contributions of these cortical regions in swallowing and tongue movement, the present results provide insights into the functional neuroanatomy of swallowing, particularly when interpreted within the context of previous animal and clinical studies. For example, single-neuron recording studies have shown that neurons in the ICMS-defined primate tongue MI fire in relation to both voluntary tongue protrusion and swallowing, with some neurons firing in advance of tongue electromyographic (EMG) activity (Martin et al. 1997b; Murray et al. 1992). Furthermore, swallowing can be evoked by ICMS applied to 4 pericentral regions including the lateral aspects of tongue-MI and face-SI (Martin et al. 1999). Together with the present finding of activation common to swallowing and tongue movement, this evidence suggests that the lateral pericentral cortex mediates the execution of tongue movements produced within a variety of behavioral contexts. Alternatively, pericentral activation during both tasks may reflect mechanical sensory stimulation of the oral cavity by the moving bolus during the swallow and by the moving tongue contacting other oral tissues during both tasks. Oral sensory processing would be expected to activate both the postcentral and precentral gyri based on primate studies showing that neurons in both areas have orofacial mechanoreceptive fields (Lin et al. 1994; Martin et al. 1997b; Murray et al. 1992). Cortical sensory processing in swallowing is thought to modulate the intensity of muscle contractions, superimposing this regulation on the sequential control of the brain stem swallowing network (Miller 1999). Future brain-mapping studies using high temporal resolution techniques are required to determine whether the pericentral activation is related to motor execution or reafference.
The sensorimotor activation focus for swallowing and tongue movement was distinct from the region activated by finger tapping. This is consistent with previous evidence that the orofacial representation within the sensorimotor cortex is lateral to that of the hand (Lin et al. 1994; Mosier et al. 1999a; Murray et al. 1992; Penfield and Rasmussen 1950).
The frontoparietal operculum also has been implicated in orofacial function. The frontal operculum is thought to control the temporal organization of oral movements, for example, controlling the timing of swallowing within the larger sequence of masticatory or other oral movements (Miller 1999; Miller and Bowman 1977). The common activation of this region for swallowing and tongue movement in the present study suggests that this region may mediate similar processing during even fairly simple voluntary oral movements such as tongue elevation. The somatosensory association areas (BA 40, 43) have been implicated in the processing of sensory stimulation applied to the human face (Hodge et al. 1998) as well as to taste sensation in monkeys and humans (Burton et al. 1971; Cerf et al. 1998; Faurion et al. 1998). The present finding of common activation of this region during both swallowing and tongue movement suggests that the activation reflects mechanical sensory processing, rather than gustatory processing, given that the tongue-elevation task did not involve the manipulation or swallowing of a bolus.
In the present study, both swallowing and tongue elevation activated the cingulate motor area (CMA) of the ACC and immediately adjacent SMA. Both the SMA and ACC are thought to play important roles in movement planning and execution (Deiber et al. 1996; Devinsky et al. 1995; Dum and Strick 1991, 2002; Richter et al. 1997), with the ACC potentially coding mode of movement initiation, movement type, and rate (Deiber et al. 1999). The ACC has also been implicated in response selection (Milham et al. 2003; Paus 2001) and attention allocation (Luks et al. 2002). Indeed, it appears that the SMA and CMA may be activated even in the absence of a motor response (Downar et al. 2000). Because our swallowing and tongue movement tasks involved similar attentional and response-selection components, as well as oral movement planning and execution, it remains unclear whether the role of the CMA and SMA in the swallowing task is to mediate movement planning and/or execution or, alternatively, response selection and attentional processes. Additional swallowing studies, incorporating a “go, no-go” paradigm, are under way in our laboratory in attempts to clarify this issue.
The region of the ACC and SMA activated by finger–thumb opposition task was caudal to that activated by tongue movement and swallowing. This finding is consistent with previous reports that the SMA and ACC are somatotopically organized, with the face and speech processes represented rostral to the upper limb and manual tasks (Fried et al. 1991; Paus et al. 1993; Salmelin and Sams 2002).
Swallowing and tongue elevation both activated a region of the precuneus that was not activated by the finger-tapping task. The precuneus has been implicated in sequence processing (Catalan et al. 1998; Jenkins et al. 1994), multimodal integration of sensory inputs (Grafton et al. 1992; Stephan et al. 1995), and visuomotor integration. The role of the precuneus in oral motor processing remains unclear. Future studies in which elements of the swallowing task are varied, such as task cueing and feedback regarding response accuracy, may be important in elucidating the function of the precuneus in oral behavior.
Activation specific to swallowing or tongue elevation
The tongue task activated a substantially greater cortical and subcortical brain volume than the swallowing task. Consistent with this, the contrast analysis revealed a number of brain areas for which tongue-related activation was greater than swallow-related activation. One possibility is that these differences in activation resulted from the somewhat different time courses of the swallowing and tongue-movement tasks. However, our finding of greater activation for tongue movement than swallowing, regardless of whether the regression analyses were performed with respect to the visual triggers, the swallow and tongue-task latencies, or the swallow and tongue-task durations, argues against this interpretation. Another interpretation that we cannot completely rule out is that the differential overall activation volumes associated with the swallowing and tongue elevation are related to greater subject motion during the swallowing trials, compared with the tongue-elevation trials, possibly producing fewer significant voxels for swallowing activation, particularly after the application of the phase-magnitude motion-suppression algorithm. Greater total activation during the tongue task may have occurred because it involved greater motor effort than the swallow, although EMG studies are needed to test this hypothesis. Finally, given that the central control of swallowing is mediated by a brain stem integrative network (Jean 1990; Miller 1999), the differential cortical activation for swallowing and tongue movement may reflect the fact that much of the processing for swallowing is mediated by brain stem mechanisms, whereas regulation of voluntary tongue movements relies more heavily on cortical/subcortical networks.
In contrast, the present findings regarding brain regions selectively activated by swallowing were equivocal. Although our group contrast analyses failed to identify areas activated more strongly by swallowing than by tongue elevation, consistent with Kern et al. (2001), our superimposed maps identified some brain areas activated by swallowing only. These corresponded to the left lateral pericentral cortex immediately lateral to the area activated by tongue movement, left postcentral gyrus/supramarginal gyrus, ACC, precuneus/cuneus, and the right insula/operculum. This finding on the pericentral cortex is generally consistent with a positron emission tomography (PET) study, which showed that the precentral gyrus representation of swallowing was 1 cm lateral to the tongue motor representation (Zald and Pardo 1999). Furthermore, ICMS findings in primates have localized the “cortical swallowing” and “cortical masticatory” areas to the lateral aspects of face-MI and -SI, as well as the area immediately lateral and/or anterior to face-MI (Martin et al. 1998, 1999). Together these findings suggest that the lateral pericentral and adjacent opercular cortex may mediate processes specific to swallowing.
The present finding that the right insula/operculum was activated by swallowing is consistent with previous brain-imaging and clinical studies (Daniels and Foundas 1997; Hamdy et al. 1999a; Martin et al. 2001). However, the precise location of insular activation remains unclear. Previous findings suggested that swallowing activates the anterior insula and tongue movement activates the posterior insula (Martin et al. 2001; Zald and Pardo 1999), whereas our present findings localize the swallow-related activation to the right posterior insula (i.e., y = −6). Further brain-imaging studies in which the insula is examined at high spatial resolution are needed to clarify the functional organization of the insula for oropharyngeal function.
Given the inconsistency of the evidence on brain activation specific to swallowing, it would be premature to rule out the possibility that certain brain areas are activated selectively by swallowing. Additional studies focusing on the “swallowing-specific” brain regions identified by our superimposed maps in which larger numbers of subjects and swallowing/oral movement trials are used should help to resolve this issue.
Hemispheric lateralization for swallowing and tongue movement
A novel finding of the present study is that the representations of both swallowing and tongue movement within the left and right hemispheres are highly distinct. With respect to swallowing, activation of the lateral pericentral/opercular/anterior parietal cortex is strongly lateralized to the left hemisphere in the majority of individuals. This was evident from our group activation map for swallowing as well as our ROI analyses. That said, the ROI analyses also showed that, whereas left-hemisphere lateralization of the postcentral gyrus is the typical pattern for swallowing, there is substantial intersubject variation, with a minority showing strong right hemisphere lateralization. Although previous reports also have suggested that the sensorimotor swallowing representation is strongly asymmetric within individual subjects (Hamdy et al. 1999a, b; Mosier et al. 1999a), most previous studies have failed to find a groupwise lateralization of the sensorimotor cortex for swallowing, some subjects showing relatively greater activation in the left hemisphere whereas others exhibit right-hemisphere asymmetry (Hamdy et al. 1999a, b; Martin et al. 2001). However, in support of our finding of lateralization to the left hemipshere, a recent magnetoencephalography study by Dziewas et al. (2003) showed that activation of the lateral primary sensorimotor cortex was strongly lateralized to the left for execution of voluntary water swallowing, and less lateralized for reflexive swallowing. Similarly, Mosier et al. (1999b) reported a left-hemisphere dominance of sensorimotor cortex activity for saliva swallowing in 5/8 subjects. Regarding our tongue movement task, although the tongue elevation was produced about the midline, activation of the postcentral gyrus was strongly lateralized (typically to the left) in approximately half of the subjects. Whereas some previous brain-imaging studies of tongue movement have reported bilateral activation of the precentral gyrus (Wildgruber et al. 1996) and sensorimotor cortex (Corfield et al. 1999), Shinagawa et al. (2003) found that subjects who reported a strong chewing-side preference had greater tongue-movement–related activity in the hemisphere contralateral to the preferred chewing side.
Other lines of evidence also suggest the possibility that the left and right postcentral and anterior parietal cortices are functionally nonequivalent for oral sensorimotor function. Pardo et al. (1997) reported that sensory stimulation of the left side of the tongue produced both contralateral and ipsilateral activation in the somatosensoty cortex, whereas stimulation of the right side produced only a contralateral response. They attributed these “inherent asymmetries of (lingual) somatosensory processing” to the specialization of the human tongue for spoken language. Patients with stroke involving the anterior parietal lobe of the dominant hemisphere have been reported to exhibit apraxia of speech, further implicating the left postcentral/parietal cortex in oral movement regulation (Square et al. 1997). Specialization of the left parietal lobe for the preparation and regulation of limb movements (Astafiev et al. 2003), and the formation of tactile representations of shape (Hadjikhani and Roland 1998) also have been reported. This is consistent with evidence that damage to left inferior parietal lobule causes errors in planning bilateral voluntary movements (Freund 2001).
Our study suggests that swallowing and tongue movement recruit overlapping but distinct interhemispheric networks of cortical and subcortical foci. Approximately 60% of our right-handed subjects showed strong functional lateralization of the postcentral gyrus toward the left hemisphere for swallowing, whereas 40% showed a similar activation bias toward the left postcentral gyrus for symmetric tongue movement. This finding suggests the possibility that the orofacial sensorimotor cortices within the left and right hemispheres are functionally nonequivalent. The left postcentral gyrus may be specialized for processing oral sensory input based on its phylogenetic specialization for oral language. This left hemisphere specialization would be expected to be influenced by a number of ontogenetic, environmental, and constitutional factors. This may explain why right-handed individuals display different degrees of hemispheric lateralization for nonspeech oral sensorimotor processing.
Unlike limb and hand movements, oral movements are produced by muscles organized in pairs about the midline. Little is known about the cortical control of this class of movements. Oral motor behavior may serve as a model through which we can better understand the neural control of axial movements. Such studies may be particularly important to the neural plasticity literature where the vast majority of motor studies of healthy controls or patients have involved limb movements. Application of fMRI motion-suppression methods, such as the phase-magnitude algorithm applied in the present study, increase the feasibility of such studies.
This research was supported by Ontario Ministry Health Career Scientist Award to R. E. Martin, Natural Sciences and Engineering Research Council Grant to R. E. Martin, Canada Research Chair Support to R. S. Menon, and Canadian Institutes of Health Research Maintenance Grant to R. S. Menon.
The authors acknowledge the valuable contributions of Dr. Christopher Thomas in computer programming and data analysis.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 by the American Physiological Society