Journal of Applied Physiology


The purpose was to compare patterns of brain activation during imagined handgrip exercise and identify cerebral cortical structures participating in “central” cardiovascular regulation. Subjects screened for hypnotizability, five with higher (HH) and four with lower hypnotizability (LH) scores, were tested under two conditions involving 3 min of 1) static handgrip exercise (HG) at 30% of maximal voluntary contraction (MVC) and 2) imagined HG (I-HG) at 30% MVC. Force (kg), forearm integrated electromyography, rating of perceived exertion, heart rate (HR), mean blood pressure (MBP), and differences in regional cerebral blood flow distributions were compared using an ANOVA. During HG, both groups showed similar increases in HR (+13 ± 5 beats/min) and MBP (+17 ± 3 mmHg) after 3 min. However, during I-HG, only the HH group showed increases in HR (+10 ± 2 beats/min; P < 0.05) and MBP (+12 ± 2 mmHg; P < 0.05). There were no significant increases or differences in force or integrated electromyographic activity between groups during I-HG. The rating of perceived exertion was significantly increased for the HH group during I-HG, but not for the LH group. In comparison of regional cerebral blood flow, the LH showed significantly lower activity in the anterior cingulate (−6 ± 2%) and insular cortexes (−9 ± 4%) during I-HG. These findings suggest that cardiovascular responses elicited during imagined exercise involve central activation of insular and anterior cingulate cortexes, independent of muscle afferent feedback; these structures appear to have key roles in the central modulation of cardiovascular responses.

  • human
  • imagery
  • single-photon-emission computed tomography
  • magnetic resonance imaging
  • autonomic nervous system

the idea of descending signals from higher brain centers has long been recognized within the overall scheme of cardiovascular regulation (16, 17,21). This concept has evolved into the widely accepted theory of “central command,” which is classically defined as a feed-forward mechanism involving a parallel activation of both motor and cardiovascular centers in the cerebral cortex during exercise (16, 18, 25). These descending signals are thought to converge with afferent signals arising from the working skeletal muscle at medullary centers of cardiovascular integration to yield an overall cardiovascular response in proportion to the intensity of physical activity or effort. Studies employing neuromuscular blockade during static exercise to weaken muscles and increase the level of central command needed to maintain force production report enhanced cardiovascular activation (3, 15, 23). These findings suggest that an individual's perceived exertion or “effort sense” during physical activity, independent of actual force production, largely dictates the magnitude of central command. A role for central command in cardiovascular regulation is evident. However, the specific region(s) of the higher brain responsible for these “effort-induced cardiovascular responses,” as opposed to those activated in response to task-specific “somatic motor programming,” have not been clearly identified.

In considering a region or regions of the brain that may be involved in the central regulation of autonomic function, Critchley et al. (9) compared brain regions activated by both exercise and a mental stressor designed to elicit cardiovascular responses. They reported common activation in the cerebellar vermis, brain stem, and anterior cingulate cortex. Other human studies specifically investigating cerebral cortical (and thalamic) structures involved in central command (as related to effort-induced cardiovascular responses) have reported activity in the sensorimotor cortex (30, 40, 44,45), cingulate cortex (45), insular cortex (19, 29, 44, 45), medial prefrontal cortex (19,40), as well as thalamic regions (19, 40, 45). These regions represent a combination of limbic, paralimbic, autonomic, and sensory regions, which together may serve as a central command network. However, a potential limitation of these central command studies is the difficulty in differentiating which cerebral cortical structures were actually responsible for the altered “cardiovascular responses” and which were activated (in parallel) in response to perceptual or cognitive processes involved in the task-specific somatic motor programming of the actual or attempted movement per se.

Toward isolating the cerebral cortical structures involved in the cardiovascular modulation by central command, an experiment was designed that permitted a contrast between patterns of brain activity during “imagined attempted movement,” both with and without cardiovascular activation. To accomplish this, brain activation was compared during actual and imagined handgrip exercise in groups of subjects screened for either high (HH) or low hypnotizability (LH). The goal of this investigation was to uncouple “central motor command” from “central cardiovascular command.” Based on previous findings (27, 45), an a priori assumption was made that only the subjects with higher hypnotizability would be able to experience an altered perception of effort and elicit cardiovascular responses during imagined handgrip. The hypothesis that the insular cortex and anterior cingulate cortex would be activated only when there were significant cardiovascular responses was tested. Regional cerebral blood flow (rCBF) distribution in several cerebral cortical regions was assessed during actual and imagined exercise by using single-photon-emission computed tomography (SPECT).



A total of 25 individuals initially volunteered to participate in this experiment, and each of these volunteers completed a battery of psychological questionnaires for screening purposes. Individuals scoring two or more standard deviations above the published norms for anxiety (39), depression, neuroticism, or psychoticism, as measured by these inventories (45), were not eligible for further participation in the study. All participants provided written, informed consent before their participation in this study, which was approved by the University of Texas Southwestern Medical Center Institutional Review Board and the Radiation Safety Committee.

Of the 25 individuals who completed the aforementioned screening, 18 volunteered to take part in further screening designed to measure hypnotizability. The Harvard Group Scale of Hypnotic Susceptibility (HGHS) (38) was administered to the subgroup of 18 subjects, and the mean hypnotizability score for this sample was 7.4 ± 3.1 units. Twelve of the eighteen subjects then participated in a hypnotic deepening session that was based on a modification of the induction procedures described in Form C of the Stanford Hypnotic Susceptibility Scale (43). The verbatim induction is available on request. Nine of the twelve possessed varying degrees of hypnotic responsiveness as measured by HGHS, and each of these individuals volunteered to complete the exercise portion of the study.

The nine subjects were then divided into two groups based on HGHS scores and responsiveness during the deepening procedures. The mean score for the five subjects in the HH group was 15.4 ± 1.1 units and was significantly greater than that for the four members in the LH group (7.5 ± 4.2 units). This classification of subjects into the HH and LH groups was performed before data collection and analysis, and the investigators responsible for evaluating and analyzing the cardiovascular and rCBF data were unaware of the subjects' group affiliation. Additionally, subjects were also unaware of their group assignment. The two groups, differentiated by their higher and lower hypnotizability, completed all protocols described in this study. The nine women (aged 23 ± 2.1 yr) undergoing the exercise protocols were all judged as being in good health. All were normotensive [resting blood pressure (BP) <140/90 mmHg], and none was taking any cardiovascular medications or reported any history of cardiovascular or neurological disorders. Poststudy medical examination of individual magnetic resonance (MR) scans showed no significant abnormalities.

Hypnotic procedures.

The induction procedure employed a standard script and, in each case, consisted of reminding the individual that it would be possible to maintain the required force for 3 min, provide verbal ratings of perceived exertion (RPE) periodically, and remain hypnotized throughout the session. Subjects were instructed to sit comfortably, breath normally, not hold their breath, and only use the hand and forearm muscles to maintain the required force. To help assess the depth of the hypnotic state, subjects were required to successfully respond to instructions resulting in hand levitation before engaging in the exercise tasks. After the imagined handgrip, a rating for the vividness of the imagery was obtained using a 10-point category scale described by Williamson et al. (45).

Exercise procedures.

Before actual testing, all subjects were well practiced squeezing the handgrip dynamometer for 3 min at 30% of their MVC with their dominant arm (eight subjects were right-handed). The dominant arm also served as the target arm for imagined handgrip. On a separate day, all subjects performed the actual handgrip exercise protocol in a control, nonhypnotic condition to serve as further familiarization with testing procedures. For each test, a venipuncture was made in the antecubital vein of the contralateral arm using a 21-gauge over-the-needle Teflon catheter ∼20 min before testing. The needle was removed, and the catheter was left in place and capped with an injectable site to facilitate the innocuous administration of the retained blood flow tracer.

Subjects were comfortably seated and then instrumented for data collection. A photoplethysmographic finger cuff (Finapres, Ohmeda 2300, Madison, WI) was placed on the middle finger on the contralateral (nondominant) hand and positioned at heart level for collection of BP and heart rate (HR) data. With the use of a handgrip dynamometer (Jamar, Asimow Engineering, Los Angeles, CA), each subject's MVC was recorded, and a 30% MVC value was calculated. Although the subjects had practiced squeezing on several occasions, they were provided verbal cueing as needed to help them maintain a constant force. Three recording electrodes were placed on each forearm to record muscle electromyographic (EMG) activity (Mespec 4001 EMG system, Kuopio, Finland). For integrated EMG (iEMG) comparisons between days, iEMG data during testing were calculated as a percentage of each subject's iEMG data obtained during their test day MVC for both test days. The iEMG signal averaged over the last 30 s of handgrip was used for analysis. This time frame coincided with assessments of cardiovascular responses and brain activity. Before actual and imagined exercise, subjects were read standardized instructions on the use of the 6–20 category scale developed by Borg (5) for RPE. These ratings were obtained every 30 s during testing.

rCBF assessment.

Injection of the blood flow tracer occurred during the last minute of the 3-min handgrip bouts. To determine the rCBF distributions during each testing condition, 20 mCi of freshly reconstituted99mTc ethylcysteinatedimer (Neurolite, DuPont Pharmaceuticals, Billerica, MA) were injected intravenously. This retained brain blood flow tracer is a photon emitter with a physical half-life of 6 h. Increases in rCBF to a particular region of the brain subsequently lead to an increase in the amount of radioactivity recorded from that region (33). Given the normal variation in brain activity, only significant changes in rCBF of at least ≥5% are deemed of clinical relevance (14). A technician administered the blood flow tracer and flushed the catheter with normal saline. The subjects continued their activity for an additional 1 min to facilitate appropriate distribution of the tracer. Because their eyes were closed during exercise, subjects were unaware of the exact time of injection and reported no noticeable side effects.

All subjects were taken to the SPECT camera room 20 min after exercise, and scanning was completed within 50 min of injection for all but one subject. In one case, a subject had to be rescanned later the same day, which increased total time from injection to completion to 110 min. Subjects were placed in the scanner, cameras were optimally positioned by a trained technician, and the exact coordinates were recorded for subsequent scans. Whole brain scans were obtained using a three-headed, fast-rotating scanner (Picker 3000, Cleveland, OH). Data were uniformity and fan-beam corrected. This process was repeated for both test protocols on separate days. These specific brain scanning procedures have been previously reported in detail (45).

Image processing and statistical analysis.

Each individual's brain images were aligned in three dimensions by computer using an automated volume coregistration algorithm widely used for positron emission tomography-positron emission tomography coregistration (47). Once the SPECT scans for a given subject were coregistered, normalization of total radioactive count variability was obtained by rescaling each volume so that total counts were equal for all volumes. After SPECT-SPECT coregistration for each individual, SPECT-MR coregistration was obtained by using an interactive coregistration algorithm (24) implemented on the workstation, after the SPECT voxel size was made to match the MR voxel size. Absolute and percent count differences for each pixel were obtained between “actual” and “imagined” scans. These differences were then displayed for a selected slice within the volume, as a color overlay superimposed on the MR scan.

Specific brain regions and structures were located by using the coregistered MR as an anatomical reference. By using the computer, regions of interest were drawn around these areas as seen on the MR slice. This procedure was repeated on contiguous transaxial slices until the entire area had been assessed. The number of 1.5-mm slices assessed varied by specific region and subject but ranged from 10 (thalamus) to 20 (insular cortices) slices. The total number of radioactive SPECT counts within each region was then compared between conditions for each subject as absolute counts and as a percent change from the actual handgrip condition. The areas and structures analyzed included dominant and nondominant hand and forearm sensorimotor regions, anterior cingulate cortex, thalamus (divided bilaterally), right and left inferior insular cortices (divided into equal halves to define superior and inferior regions), cerebellar vermis, and a white matter region encompassing the anterior corpus callosum. The SPECT data were corrected for any changes in rCBF to white matter regions between conditions. During the processing and rCBF data assessment, data were coded such that the researchers performing these analyses were blinded with regard to the subject identity, group affiliation, and order of experimental conditions.

A Friedman analysis of variance was used to compare differences in dependent variables from rest to 3 min (data for RPE were compared at 30 s and 3 min) for subjects with high and low hypnotizability between conditions. If significance was detected, a Wilcoxon post hoc analysis was performed to determine specific differences for pairwise comparisons. Changes in brain activation and between actual and imagined handgrip were compared between groups using a Mann-WhitneyU-test for ranked data with a Bonferroni correction for multiple comparisons. The α-level was set at P < 0.05 for all analyses. Data are presented as means ± SD.



During the debriefing session after the imagined exercise, subject's were asked to rate the vividness of the imagined exercise on a 10-point category scale ranging from 1 (not vivid at all) to 10 (very vivid), and this scale has been described in an earlier report by Williamson et al. (44) dealing with exercise and hypnosis. The mean rating for the HH group was 6.2 ± 1.5 compared with a rating of 4.8 ± 1.9 for the LH group, and this difference yielded an effect size (d) = 0.87. Hence, the perceived vividness of the imagined exercise was significantly greater for the HH group.

Perceived exertion.

The RPE for actual and imagined handgrip for both groups are presented in Fig. 1. During actual exercise, there were no significant group differences for RPE. The HH group had a change from 9 ± 2 units (30-s value) to 15 ± 4 units (3-min value), and the LH group increased from 8 ± 2 to 13 ± 2 units for the same time frame. During the imagined handgrip, the increase in RPE for the HH group was significant. The HH group increased from 9 ± 2 to 12 ± 3 units, whereas values for LH were from 7 ± 1 to 8 ± 1 units at 30 s and 3 min, respectively. Thus the HH group had higher RPE during imagined exercise.

Fig. 1.

Cardiovascular and perceived exertion responses during actual and imagined handgrip exercise. Data (means ± SD) are presented for heart rate, mean blood pressure, and ratings of perceived exertion at rest and for 30-s averages across the 3 min of actual and imagined handgrip. * Significance between groups, P< 0.05.

Force and EMG.

Overall, there were no differences between groups for force production or iEMG responses. Force produced by the HH group during the MVC on the day of actual handgrip was 32 ± 3 kg and was not different on the day of imagined handgrip (33 ± 3 kg). During actual handgrip at 30% of MVC, force was maintained at 11 ± 2 kg. The iEMG activity for the dominant arm was 31 ± 5% of the daily MVC, whereas the iEMG for the contralateral arm was 3 ± 1%. During imagined handgrip, with subjects simply holding the dynamometer, there was no measurable force production; the iEMG for this arm was 3 ± 1% of the daily MVC compared with 2 ± 1% in the contralateral arm. Thus the only significant change in iEMG from resting levels for the HH group was noted for the dominant arm during actual handgrip.

For the LH group, maximal force on the day of actual handgrip was 31 ± 4 kg and was 30 ± 3 kg on the day of imagined handgrip. During handgrip at 30% of MVC, force was 10 ± 2 kg. The iEMG during actual handgrip in the dominant arm was 34 ± 7 and 2 ± 1%, respectively, in the contralateral arm. No measurable force was detected during imagined handgrip, and iEMG activity in the target arm was 4 ± 2% and was 2 ± 1% in the contralateral arm. Thus the only significant change in iEMG from resting levels for LH was noted for the dominant arm during actual handgrip.

HR and BP.

The cardiovascular responses to actual and imagined handgrip for both groups are presented in Fig. 1. For actual handgrip, the mean HR at rest for the HH group was 74 ± 7 beats/min compared with 71 ± 7 beats/min for the LH group. After 3 min of exercise, HR was at 88 ± 10 and 84 ± 12 beats/min for HH and LH groups, respectively. The HH group had increases in HR from resting values of 75 ± 9 beats/min to 85 ± 10 beats/min (P < 0.05) after 3 min of imagined handgrip. The resting HR for the LH was 73 ± 10 beats/min at rest and 74 ± 12 beats/min after 3 min of imagined handgrip.

Mean BP (MBP) for actual handgrip was similar between groups at rest (HH = 99 ± 6 mmHg vs. LH = 98 ± 6 mmHg) and at 3 min of actual handgrip (HH = 118 ± 9 vs. LH = 114 ± 4 mmHg). The HH group increased from 98 ± 6 mmHg at rest to 110 ± 6 mmHg (P < 0.05) after 3 min of imagined handgrip. The MBP for the LH group was 97 ± 4 mmHg at rest and 100 ± 5 mmHg after 3 min of imagined handgrip.

Brain activation.

Changes in rCBF distribution are presented in Table1 for the two groups as radioactive counts (for the identical regions of interest) and as the percent change in brain activation from actual exercise. When the activation patterns of specific brain regions are compared between actual and imagined exercise, there were no group differences for dominant or contralateral hand and arm sensorimotor regions, thalamic regions, or cerebellar vermis. There were significant group differences for the anterior cingulate and insular cortices (P < 0.004). The LH group, which had no significant increase in cardiovascular responses or RPE responses, showed a significant reduction for rCBF distribution (e.g., deactivation) in the anterior cingulate (−6.3 ± 2%), right insular cortex (−9.4 ± 4%), and left insular cortex (−8.2 ± 5%) as shown in Fig.2.

View this table:
Table 1.

Number of radioactive counts for actual and imagined exercise for specific regions of interest

Fig. 2.

Differences in brain activation for actual and imagined handgrip for groups with higher (A) and lower hypnotizability (B). Coregistered single-photon-emission computed tomography and magnetic resonance imaging data representing group averages, for a single transaxial slice showing relevant anatomy, from groups with higher and lower hypnotizability. Top and bottom correspond to anterior and posterior orientation, respectively. Changes in regional cerebral blood flow distribution were mapped on the magnetic resonance image by using an arbitrary color scale with a positive range from 5 to 25% (from green through yellow to red) and negative range from −5 to −25% (from purple through dark blue to light blue). The white lines denote the specific regions of interest assessed (in this brain slice) and encompass the insular cortices, thalamic regions (bilaterally), and anterior cingulate cortex. A: single-photon-emission computed tomography data for the higher hypnotizable group shows the similarities between actual and imaged handgrip (no significant changes). B: image represents the significant decreases in activation for both anterior cingulate and insular regions for the lower hypnotizable group, who did not elicit cardiovascular responses during imagined handgrip.


The goal of this investigation was to uncouple the cerebral cortical regions involved in task-specific central motor command (somatic motor programming) from those involved in central cardiovascular command (effort-induced cardiovascular activation). These findings suggest that the anterior cingulate and insular cortex are involved with the effort-induced cardiovascular modulation, whereas the sensorimotor and thalamic regions and the cerebellar vermis are involved in motor programming associated with the specific motor task. The hypothesis that the anterior cingulate cortex and insular cortex would have higher rCBF only if there were significant cardiovascular responses during handgrip was supported. In other words, rCBF distribution was lower in those subjects not eliciting cardiovascular responses during the imagined exercise. The data are consistent with the concept that the level of central command is related more to an individual's sense of effort than to force production, as increases in RPE, HR, and BP during imagined handgrip were only noted in the group with higher hypnotizability.

Although the potential effects of hypnosis on brain activation pose interesting questions, hypnosis was only employed as a tool to facilitate the isolation of cerebral cortical structures involved in cardiovascular regulation. The use of hypnosis to modify perception of effort during actual and imagined exercise and concomitant cardiorespiratory responses is well established (1, 2, 10, 20,26-28, 40, 45). The present study was designed in an attempt to characterize responsiveness to imagined exercise in individuals differing in hypnotizability. The HH and LH groups did not differ in their perception of effort during actual exercise in the hypnotic condition. This finding is also consistent with the results of earlier research demonstrating that hypnosis per se does not influence the perception of effort during exercise (26, 27, 45). The a priori assumption that perceptual differences would occur during and after the hypnotic suggestion of imagined exercise for HH and LH subjects was confirmed. The HH group rated the perception of effort as being significantly higher during imagined exercise compared with the LH group, and, furthermore, the HH group rated the imaginary exercise as being significantly more vivid than did the LH group. The differences in patterns of brain activation and cardiovascular responses between conditions of actual and imagined handgrip, in subjects differing in hypnotizability, is an important finding of this study.

Cardiovascular responses.

Central command is generally thought to have a greater effect on HR and cardiac output during static exercise (23, 46) than on BP, with the latter being attributed to a pressor reflex response generated from the contracting muscle (for reviews, see Refs. 18,25). In the present study, subjects with higher hypnotizabiltiy had significant increases in HR during imagined handgrip, and these increases were of similar magnitude to those seen during actual handgrip (Fig. 1). Decety et al. (12) have reported increases in HR during mental simulation of motor actions that were proportional to the amount of simulated exercise. Furthermore, in a study involving the comparison of dynamic arm exercise with imagined exercise, it was reported by Wang and Morgan (42) that increases in ventilatory minute volume were comparable in the actual and imagery conditions. However, it should be noted that these investigators did not employ hypnosis with the imagery condition. In the present study, BP changes during imagined handgrip for the HH group were lower than those seen during actual handgrip and may have been related more to an increase in cardiac output via increased HR. Thornton et al. (40) reported increases in ventilation (30% of actual exercise) and HR (12% of actual exercise) during imagined cycling exercise. Additionally, Decety et al. (12) noted that HR (and ventilation) increased immediately after the “mental exercise” was initiated; this rapid-onset response for HR and BP to imagined handgrip was evident for the HH group (Fig. 1).

Effort, force, EMG.

Whereas both the HH and LH subjects were asked to imagine exercise, only the former group reported an increase in perceived exertion. This increase in effort sense was independent of any increase in force or iEMG activity. During the imagined exercise, with the subjects simply holding the dynamometer as it rested on their thigh, the iEMG activity was not different from that observed for the contralateral limb (3 ± 1% of maximal iEMG). If imagined iEMG is calculated as a percentage of the iEMG activity occurring during actual handgrip (∼8%), this level of activity is consistent with the level of EMG activity (∼6% of EMG for actual movement) for mental imagery of foot movement (4). The magnitude of iEMG change was similar for both the HH and LH groups, but only the subjects with higher hypnotizability elicited cardiovascular responses. Thus it would appear that changes in cardiovascular responses achieved during imagined handgrip were independent of significant afferent input from the target limb. This finding is in agreement with observations that respiratory responses to imagined exercise can occur without movement feedback or changes in CO2 (40).

Brain activation.

There is evidence to suggest that imagined and executed actions share, to some extent, the same patterns of activation for central structures, with variations resulting from differences in the complexity and type of motor sequences employed (11). In the present study, a protocol involving 3 min of static handgrip exercise at 30% MVC was selected because of its relative simplicity with regard to motor programming and its ability to yield cardiovascular responses. Findings of similar rCBF distributions for hand and arm sensorimotor regions between actual and imagined handgrip (Table 1) parallel findings of primary sensorimotor activation in subjects attempting handgrip after regional anesthesia (30) and in subjects of Thornton et al. (40), who reported sensorimotor activation to imagined cycling. Roth et al. (35) reported significant activation of the primary motor cortex during a mentally executed finger-to-thumb opposition task. Whereas it is probable that the sensorimotor cortex is involved in the somatic motor programming of both actual and imagined handgrip and may be involved in central command, its activation does not appear requisite for eliciting cardiovascular responses. In a previous investigation, subjects performing constant-load cycling while under hypnosis were asked to imagine that they were riding “uphill” (45). There were increases in their perceived exertion, HR, and BP but no further activation of the leg sensorimotor regions. In agreement with this finding, Nowak et al. (30) have previously concluded that it is unlikely that activation of sensorimotor cortex represents a central command influence on the cardiovascular system.

Similar rCBF distributions for the thalamus were found between actual and imagined effort for both HH and LH groups. The thalamus represents a complex region of nuclei involved in sensory integration. The thalamus is activated during handgrip exercise (19, 44) and may be further involved in cardiovascular regulation (19,32). It has also been implicated as part of a neural network involved in motor imagery (11). Current findings seem to support this role, as thalamic rCBF distributions were similar for both actual and imagined handgrip for both groups. Not unlike the sensorimotor cortex, the thalamus (or discrete nuclei within) may also be involved in the somatic motor programming or sensory integration of both actual and imagined handgrip and may be involved in central command. However, the similarity in thalamic rCBF distributions between conditions, coupled with the differences in cardiovascular response between conditions, do not support a critical role for the thalamus in eliciting cardiovascular responses.

Toward defining the cerebral cortical structures related to cardiovascular activation, or more specifically the absence of cardiovascular responses in the LH group, the LH group showed a significantly lower rCBF distribution in the anterior cingulate and insular cortices during imagined handgrip. Both of these regions have been implicated in the central modulation of autonomic function (6, 9, 19, 31, 36, 37, 41). With regard to a specific role of the anterior cingulate cortex as related to central command, it is largely involved in the discrimination of peripheral somatosensory input. Thus it could serve to interpret an individual's sensation of effort. This region, defined as the anterior cingulate cortex by both Critchley et al. (9) and the present study, is included within the region termed the medial prefrontal cortex by King et al. (19). They reported that activation of this medial prefrontal region during handgrip exercise was associated with cardiovascular activation. Reviews by Cechetto and Saper (6) and Verberne and Owens (41) have defined a role for the medial prefrontal cortex in cardiovascular regulation.

Deactivation of the anterior cingulate cortex has been reported in studies using hypnotic analgesia to reduce pain (22, 34). This finding parallels that of anterior cingulate deactivation in response to hypnotically induced decreases in perceived effort during constant-load exercise (45). It has been suggested that the anterior cingulate cortex could be involved in the immediate behavioral and autonomic responses associated with central command (7) and with changes in painful or unpleasant stimuli (34). Based on present and prior findings, we would add changes in effort sense to these stimuli as well. Furthermore, it has been suggested that these autonomic responses are modulated by an interactive neural network, including the anterior cingulate and insular cortices (34). This same basic premise has been forwarded for the medial prefrontal cortex (which encompasses the anterior cingulate cortex) and insular cortex (6, 41).

The insular cortex has long been recognized for its involvement in autonomic modulation and ties with the limbic system (6, 36, 37,41). Regarding its role in central command, insular activation in accordance with exercise intensity (44) and changes in effort sense (29, 45) have been previously reported. Consistent with current findings during imagined exercise for the LH group, insular activation does not usually occur during hand movements that do not result in cardiovascular activation (8, 13).


The cerebral cortical regions identified in this study may not be inclusive of all brain regions involved in central command. Additionally, the spatial limitations of the SPECT technique (∼8–10 mm) do not allow us to assess, with confidence, smaller regions that may also play an important role in cardiovascular regulation, such as specific nuclei of the thalamus, medulla, or other subcortical structures. Thus it is possible that autonomic signals from these subcortical regions may activate higher brain structures. Although the hypnotic induction procedure was identical under both experimental conditions, subjects were told in advance of their testing order. A nonhypnotic familiarization trial was always conducted first, and the actual exercise and imagined exercise trials were rotated (i.e., actual-imagined or imagined-actual) and randomly assigned. Participants were informed before a given trial that they would be asked to actually exercise or imagine exercising. It was necessary to take this approach because participants would always know what the second trial would involve once the first trial had been completed. It is possible that individuals may have had an anticipatory response as a consequence of this prior knowledge, but this approach was deemed necessary to facilitate consistency in perceptual-cognitive responses between trials. Additionally, there is not a precise measure of the depth of the hypnotic state during actual testing. However, the objective measure of hand levitation, the subjective measure of imagery vividness and altered time perception after hypnosis, unsolicited comments made by subjects during the posthypnotic debriefing, and the fact that subjects classified as HH and LH actually differed in physiological responses and perception of effort during and after imagined exercise serve to argue in favor of a true hypnotic effect.


It appears that central command during exercise does indeed result in a parallel activation of cardiovascular and motor centers in the higher brain, as originally defined. However, the present study suggests that a network of structures exist that is involved in the centrally induced cardiovascular activation and does not require a “parallel motor activation” to exert its influence. In other words, activation of sensorimotor, thalamic regions, and the cerebellar vermis during motor activity appears to be more related to the execution of the movement itself, whereas activation of the anterior cingulate (or medial prefrontal cortex) and insular cortices is more closely related to the effort-induced cardiovascular response. An individual's sense of effort as interpreted by the anterior cingulate cortex could potentially serve as a feedback signal, which is routed through the insular cortex to effect appropriate autonomic modulation. Such a system could explain the close coupling between physical effort (or its perception) and cardiovascular responses (25). Furthermore, this network could serve to modulate autonomic function under a variety of circumstances. These findings suggest that cardiovascular regulation during imagined exercise involves “central” activation of the anterior cingulate cortex and insular cortex, independent of muscle activity and muscle afferent feedback; these structures appear to have significant roles in the central modulation of cardiovascular responses commonly termed central command.


We thank the subjects for cooperation and also acknowledge the expert technical assistance of Michael Viguet, as well as the cooperation of Zale Lipshy University Hospital, Dallas, TX.


  • This study was supported by American Heart Association Grant-in-Aid 0050726Y from the Texas Affiliate (to J. W. Williamson), National Heart, Lung, and Blood Institute Grant R01 HL-59145–03 (to J. W. Williamson), and the Harry S. Moss Heart Center at the University of Texas Southwestern Medical Center. Support was also provided by the University of Wisconsin Sea Grant Institute (W. P. Morgan) under grants from the National Sea Grant Program, National Oceanic and Atmospheric Administration, US Department of Commerce, and from State of Wisconsin Federal Grant NA86RG0047, project R/NI-29-PD (to W. P. Morgan).

  • Address for reprint requests and other correspondence: J. W. Williamson, UT Southwestern Allied Health Sciences School, Department of Physical Therapy, 5323 Harry Hines Blvd., Dallas, TX 75390–8876 (E-mail: jon.williamson{at}

  • 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.

  • 10.1152/japplphysiol.00939.2001


View Abstract