Journal Logo

Topical review

Imaging cognitive modulation of pain processing

Petrovic, Predrag; Ingvar, Martin

Author Information
doi: 10.1016/S0304-3959(01)00467-5
  • Free

1. Introduction

The intensity and unpleasantness of a painful experience is often described as correlating well with the degree of noxious stimulation. However, the perception of pain is not a linear phenomenon, reflecting the signal from the peripheral neuron. Rather, the noxious input may be modulated at every level of the neural axis. One of the most potent sources of modulation is the brain—although these mechanisms have only sparsely been studied. The supraspinal modulatory influences involve both lower order automatic response schemata and higher order dynamic cognitive mechanisms. This organizational pattern has developed as an evolutionary driven adaptation, in which both fast hardwired responses and slower dynamic responses increased the chance for survival. In line with this hypothesis, it has been suggested that the brain is initially processing noxious input in the brainstem supporting the demand for a fast response (Petrovic et al., 2000a; Price, 2000). Apart from autonomic changes and a wide range of defense reactions, the brainstem may induce powerful analgesia in direct response to noxious stimuli (Fanselow, 1994). At a higher level, cognitive processes may dramatically modulate the perception of pain (Melzack and Casey, 1968; Weisenberg et al., 1996). Recently, functional imaging tools, such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), have described some of the possible underlying mechanisms that are involved in cognitive modulation of pain perception.

Several functional imaging studies have indicated that pain processing may be modulated by cognitive mechanisms (Bantick et al., 2001; Longe et al., 2001; Petrovic et al., 2001a,b; Rainville et al., 1997, 1999; Willoch et al., 2000). Rainville and colleagues used hypnotic suggestion to modulate the perception of unpleasantness during noxious stimulation. When the subjects were suggested to perceive the noxious stimulation as highly unpleasant there was a concomitant increase in the activity in the anterior cingulate cortex (ACC) significantly more than when the subjects were suggested to perceive the same stimulation as less unpleasant (Rainville et al., 1997). However, the activity in the somatosensory areas was unaltered. Since lesion studies and animal studies have indicated that the ACC is involved in processing pain unpleasantness (Vogt et al., 1993) this finding indicates that cognitive mechanisms may specifically modulate sub-systems of the pain network. We used a different but classical approach in order to show that pain networks may be modulated by cognitive demands (Petrovic et al., 2000b). Most people have probably experienced that pain perception can decrease and even disappear when actively engaging the mind in a distracting task. We tested this mechanism by involving the subjects in a highly attention demanding task (computerized perceptual maze test) during noxious stimulation. We were able to show that when subjects solved the maze task and we induced a painful stimulation they perceived less pain as compared with when there was no competition for attentional space. At a neural level, activity was significantly attenuated in somatosensory regions and the PAG in this condition. Recently, it has been shown that cognitive distraction also may attenuate the pain-evoked activity in the ACC, the insula and the thalamus (Bantick et al., 2001; Longe et al., 2001). One intriguing study has demonstrated the opposite cognitive modulation in which the pain network was activated without any noxious stimulation being given (Willoch et al., 2000). In this study, painful perception was induced in patients with phantom-limb pain using hypnotic suggestion that the missing limb was in a painful position.

All the regions discussed above are involved in pain processing, and modulation in their activity coincides with the changes in a pain perception. Apart from attention dependent changes in the network processing the perception, another distinct set of structures may act as sources for modulating the activity in these sites (Coull, 1998; Frith and Dolan, 1997). In line with this, several of the studies of cognitive interaction with pain (Bantick et al., 2001; Faymonville et al., 2000; Petrovic et al., 2001a,b; Rainville et al., 1999; Willoch et al., 2000) also indicate cortical structures, which may act as sources of the activity modulation. In this review, we focus on the lateral orbitofrontal cortex and the rostral ACC, which are activated specifically during cognitive modulation of pain processing.

2. The lateral orbitofrontal region

The lateral orbitofrontal cortex has shown relatively increased activity in several studies involving modulation of pain. This effect was noted when pain perception was attenuated (Bantick et al., 2001; Petrovic et al., 2001a,b), augmented (Willoch et al., 2000) and changed in both directions (Rainville et al., 1999; Fig. 1B). Apart from the data showing that the lateral orbitofrontal cortex increases in activity during modulation of pain processing by cognitive means, this hypothesis dwells on knowledge from other fields in cognitive neuroscience. There are several lines of evidence that implicate the lateral orbitofrontal cortex in modulating distant neural activity in emotional contexts. First, the orbitofrontal cortex is involved in response suppression when a value based stimulus–response association has to be suppressed (Elliott et al., 2000). This mechanism may be dependent on a representation of the magnitude of punishment in the lateral orbitofrontal cortex (O'Doherty et al., 2001). The second line of evidence is based on the involvement of the lateral orbitofrontal cortex in depression. Although both the amygdala and the lateral orbitofrontal cortex show increased activity during major depression, only the amygdala activity correlates with the severeness of the disease (Drevets, 2000). In contrast, the lateral orbitofrontal cortex activity is inversely related to several indices of depression. Drevets and colleagues have therefore suggested that the lateral orbitofrontal cortex is involved in suppressing a network (possibly including the amygdala), which has a pathologically increased activity. Thus, the present hypothesis states that the lateral orbitofrontal cortex is involved in modulating value based processes in emotional networks or response systems. Moreover, electrical stimulation of the lateral orbitofrontal cortex in both the rat and primate supports its involvement in analgesia (Oleson et al., 1980; Zhang et al., 1997). We interpret the relative increase of activity in the lateral orbitofrontal cortex during pain as representing a source of cognitive modulation of emotional components that are produced by or interact with pain processing.

Fig. 1:
Studies showing the voxel with the most expressed relative increase of activity in the rostral ACC (A) and the lateral orbitofrontal cortex (B) during pain modulation (gray dots). In study 1 (Rainville et al., 1999) the plot represents increased rCBF activity during modulated pain perception vs. unchanged pain perception (all under hypnosis). In study 2 (Petrovic et al., 2000b) the plot represents placebo analgesia (i.e. placebo analgesia vs. untreated pain). In study 3 (Petrovic et al. 2001a) the plot represents relatively increased activity when pain was altered due to a distracting cognitive task. In this task an interaction analysis was employed since the (pain-) distracting Maze task itself induces decreased activity in the orbitofrontal regions. Similarly, in study 4 (Bantick et al., 2001) the plot indicates the activity, which relatively increases during an attention demanding (pain-) distracting tasks. An approximate division of the ACC based on several meta-studies (Bush et al., 2000; Hsieh 1995; Willoch 2001) show that the ACC may be divided into a region which is often activated by pain per se (marked region in the caudal ACC), an adjacent area which is involved in general attention (marked region in the middle ACC) and an area which is involved in emotional tasks and in stimuli induced pain modulation (marked region in the rostral ACC). The contrasts involving cognitive pain modulation activated the emotional subdivision of the ACC in many studies (Bantick et al., 2001; Petrovic et al., 2001a (sub-significant effect with a Z-value=2.8); Petrovic et al., 2001b; Rainville et al., 1999) (gray dots). When the effects of pain (crossed dots) from the same studies are plotted on the image they predominately activate the caudal ACC. In study 4 (Bantick et al., 2001) the gray dot represents the ACC region, which showed the ACC area that was relatively more activated during pain with a low distraction (as compared with pain during high distraction). The effect in (Willoch et al., 2000) is not plotted since it was not possible to test the relative increase during pain modulation. The result of (Faymonville et al., 2000) is not plotted since it did not show any relative increase in the orbitofrontal cortex during pain modulation, and the interaction in the rostral ACC was somewhat uncertain to interpret since it appeared in the corpus callosum below the ACC. The Talairach space is represented in neuroradiological convention (i.e. left is right on image in part B).

3. The Anterior cingulate cortex

Functional imaging studies have implicated the rostral ACC both in analgesia induced by several electro-stimulation methods (see Willoch, 2001), but also in analgesia induced by cognitive manipulations (Bantick et al., 2001; Faymonville et al., 2000; Petrovic et al., 2001a,b; Rainville et al., 1999; Fig. 1A). One of the most intriguing analgesia methods described is placebo analgesia. Only the belief that a noxious stimulus will hurt less may lower pain ratings dramatically. It is a challenge to understand why placebo analgesia may be induced but also which mechanisms are involved. Therefore, it was a major breakthrough when Levine and colleagues were able to show that placebo analgesia could be abolished by the opioid receptor antagonist naloxone (Levine et al., 1978; Price, 1999). This observation implicated the involvement of the endogenous opioid system in placebo analgesia. In line with this, we were able to show that both a short acting opioid, i.e. Remifentanil, and placebo analgesia activated the rostral ACC (Petrovic et al., 2001a). The rostral ACC seems to have one of the highest cortical concentrations of opioid receptors in the brain (Willoch, 2001; Willoch et al., 1999), suggesting that similar opioid dependent systems may have been induced in placebo analgesia as when opioids were given. Opioid analgesia is dependent on a lower opioid system that resides in the brainstem and involves the periaqueductal gray (PAG), the parabrachial nucleus and the ventromedial medulla (Fields and Basbaum, 1999). The lower opioid system is under control by the opioid receptor rich areas in the ACC and the amygdala through direct or indirect fibers stretching to the brainstem (Fanselow, 1994; Vogt et al., 1993; Fig. 2). We propose that cognitively induced opioid dependent analgesia is likely to use this cortical control mechanism on lower hierarchy opioid systems. In support of this hypothesis, we observed a covariation between the rostral ACC and the brainstem both during opioid analgesia and placebo analgesia, but not during the pain only condition (Petrovic et al., 2001a).

Fig. 2:
The brainstem opioid network includes the periaqueductal gray (PAG), the parabrachial nucleus and the ventromedial medulla (Fields and Basbaum, 1999). This network may be controlled by the amygdala during stress related analgesia (Fanselow, 1994), and by the rostral ACC during opioid analgesia (Vogt et al., 1993). The cortical control may be especially important when higher order cognitive tasks interact with the endogenous opioid system such as in placebo analgesia (Petrovic et al., 2001a).

The specific role of the rostral ACC in pain modulation is compatible with the general function of this region. Previous meta-analyses of functional imaging studies have divided the functions of the ACC into three distinct sub-regions. One of these has shown that pain as such preferentially activates a caudal part of the ACC, while pain modulation also involves a rostral region of the ACC (Willoch, 2001). Another subdivision has shown that strictly attentional tasks activate an adjacent anterior part of the ACC as compared with pain per se (Hsieh, 1995). The ACC has also been divided between this strictly attentional sub-region and a more rostral part that is activated by emotional tasks (Bush et al., 2000). Thus, the ACC may be divided into a caudal region, showing increased activity during pain per se, an adjacent part preferentially involved in general attention and a rostral region involved in emotional processing. When the activities induced by pain modulation are plotted on a map of the ACC, it is apparent that these increases reside in the emotional sub-region (Fig. 1A) although the effect of pain is expressed in the caudal ACC in the same studies. An interesting functional division has been noted in that the counting Stroop task has been shown to increase activity in the attentional sub-region of the ACC while the emotional Stroop task that require similar attentional effort but in the emotional domain activate the emotional sub-region. This suggests that the emotional sub-region also is involved in attentional processes, but under emotional influences.

Several neuro-psychological theories of attention, including those proposed by Mesulam and Posner, have pointed to the importance of the ACC in attentional processes (see Coull, 1998). It is suggested that the ACC may be involved in orchestrating cognitive resources and adding a motivational value for different processes. Shallice and colleagues have proposed that ACC is involved in supervisory attentional control, which may include both suppression and potentiation of distant neural activity when several processes are competing for cognitive resources (Stuss et al., 1995). Likewise, the known modulatory mechanisms of opioid receptors that reside in the ACC and the analgesic effect due to stimulation of this region (Hardy, 1985) support an active role for the ACC during pain modulation. In line with these suggestions, we propose that the rostral ACC is executing attentional control during pain modulation. In fact, the presented data are highly supportive for an active role of the ACC during attentional tasks in general.

4. Decreases of limbic activity during pain—indications of a cognitive coping mechanism?

If a behavioral response to avoid an aversive context is not possible, cognitive strategies to deal with the situation are often used (Folkman and Lazarus, 1988). It is well established that coping may change not just the perception of pain but also autonomic responses during noxious stimulation (Thompson, 1981; Weisenberg et al., 1996). Most imaging studies have to date been conducted in the context of unavoidable pain, i.e. the subject knows about the upcoming stimulus and simply has to endure it. Some recent studies have addressed this issue by manipulating the context in which the noxious stimulus was administered. These studies have indicated that one possible cognitive mechanism, which is used to cope with such situations, involves suppression of activity in limbic and paralimbic structures (Hsieh et al., 1999; Petrovic et al., 2001b; Simpson et al., 2001).

From a processing perspective, a coping strategy has the greatest potential if initiated before the actual stimulus (Tulving 1985). Therefore, it is especially interesting to study the anticipation of a painful event. In one of these studies, it was shown that after subjects had gone through several practice sessions in which a noxious stimulus was used, the activity in the caudal ACC and subgenual ACC/medial orbitofrontal cortex was down-regulated during the anticipation phase (Hsieh et al., 1999). In contrast, in another group of subjects that had not been acquainted with the noxious stimulus before the imaging session, activity was up-regulated in similar regions in the caudal ACC and subgenual ACC/medial orbitofrontal cortex, as well as in the PAG, in the same situation (Hsieh et al., 1999). Thus, while the subjects that knew what to expect suppressed the activity, the subjects that did not know what to expect increased the activity in regions involved in pain processing. It has recently been shown that this down-regulation in the subgenual ACC/medial orbitofrontal cortex (and the hypothalamus) during the anticipation of pain was inversely correlated with the degree of anxiety ratings (Simpson et al., 2001). Thus, the subjects that rated the least anxiety also had the most expressed decrease of activity in these regions. Both papers suggest that efficient cognitive coping involves a suppression of activity preferentially in limbic and paralimbic structures.

Indices of different strategies to adapt to pain have also been shown during noxious stimulation. The amygdala may be seen as a nexus orchestrating behavioral, autonomic and arousal responses during fear (Davis, 1997) but is also involved directly in pain processing (Fanselow, 1994). This is apparently at variance with the results from several studies that show a decreased activity in the amygdala during pain (e.g. Petrovic et al., 1999). We have shown that when subjects expect a longer noxious stimulus, the activity in the amygdala decreases as compared with their expectation of a shorter stimulus (Petrovic et al., 2001b). This finding was observed during the same time of the noxious stimuli, and can thus only be explained by different expectations in the two contexts. One interpretation of the findings is that a suppression of the amygdala output also decreases the subjective distress during the stimulation itself.

Thus, there is data indicating that human subjects may apply different cognitive adaptations to aversive situations, including pain, which involve suppression of activity in limbic and paralimbic structures. However, these studies are still few and several critical experiments remain to be performed.

5. Conclusions

The second generation of functional imaging studies has started to reveal the mechanisms that involve cognitive modulation of pain processing. A wide knowledge in this field may introduce new powerful analgesic therapies in the future, give us a better theoretical knowledge of the cognitive mechanisms important in pain perception, and also shed new light upon general theories of cognition and attention. One of the most interesting questions to be resolved in the near future is how different regions that are involved in the modulation of pain, e.g. the rostral ACC and the lateral orbitofrontal cortex, interact with each other to modulate the perception of pain.


The writing of this article was supported by MFR (8246), Stockholm County Council and Hedlunds Stiftelse. The authors are very thankful to all collaborators in the primary studies underlying this review.


Bantick SJ, Wise RG, Ploghaus A, Clare S, Smith SM, Tracey I. Imaging how attention modulates pain in humans using FMRI. Brain 2001. in press.
Bush G, Luu P, Posner MI. Cognitive and emotional influences in anterior cingulate cortex. Trends Cogn Sci. 2000;4:215-222.
Coull JT. Neural correlates of attention and arousal: insights from electrophysiology, functional neuroimaging and psychopharmacology [Review]. Prog Neurobiol. 1998;55:343-361.
Davis M. Neurobiology of fear responses: the role of the amygdala [Review]. J Neuropsychiatry Clin Neurosci. 1997;9:382-402.
Drevets WC. Neuroimaging studies of mood disorders. Biol Psychiatry. 2000;48:813-829.
Elliott R, Dolan RJ, Frith CD. Dissociable functions in the medial and lateral orbitofrontal cortex: evidence from human neuroimaging studies [Review]. Cereb Cortex. 2000;10:308-317.
Fanselow MS. Neural organization of the defensive behavior system responsible for fear. Psychon Bull Rev. 1994;1:429-438.
Faymonville ME, Laureys S, Degueldre C, DelFiore G, Luxen A, Franck G, Lamy M, Maquet P. Neural mechanisms of antinociceptive effects of hypnosis. Anesthesiology. 2000;92:1257-1267.
Fields H, Basbaum A., 1999. Central nervous system mechanisms of pain modulation. In: Wall P, Melzack R, editors., Textbook of pain. Churchill Livingstone, Edinburgh, pp. 309-329.
Folkman S, Lazarus RS. The relationship between coping and emotion: implications for theory and research [Review]. Soc Sci Med. 1988;26:309-317.
Frith C, Dolan RJ. Brain mechanisms associated with top–down processes in perception [Review]. Philos Trans R Soc Lond B Biol Sci. 1997;352:1221-1230.
Hardy SGP. Analgesia elicited by prefrontal stimulation. Brain Res. 1985;339:281-284.
Hsieh J-C. Central processing of pain; functional brain imaging studies with PET, vol. ISBN: 91-628-1722-1, Thesis, Karolinska Institute, Stockholm, 1995.
Hsieh JC, Stone-Elander S, Ingvar M. Anticipatory coping of pain expressed in the human anterior cingulate cortex: a positron emission tomography study. Neurosci Lett. 1999;262:61-64.
Levine JD, Gordon NC, Fields HL. The mechanism of placebo analgesia. Lancet. 1978;2:654-657.
Longe SE, Wise R, Bantick S, Lloyd D, Johansen-Berg H, McGlone F, Tracey I. Counter-stimulatory effects on pain perception and processing are significantly altered by attention: an fMRI study. NeuroReport. 2001;12:2021-2025.
Melzack R, Casey KL., 1968. Sensory, motivational and central control determinants of pain. In: Kenshalo DR, editor., The skin senses. Thomas, Springfield, IL, pp. 423-439.
O'Doherty J, Kringelbach ML, Rolls ET, Hornak J, Andrews C. Abstract reward and punishment representations in the human orbitofrontal cortex [see comments]. Nat Neurosci. 2001;4:95-102.
Oleson TD, Kirkpatrick DB, Goodman SJ. Elevation of pain threshold to tooth shock by brain stimulation in primates. Brain Res. 1980;194:79-95.
Petrovic P, Ingvar M, Stone-Elander S, Petersson KM, Hansson P. A PET activation study of dynamic mechanical allodynia in patients with mononeuropathy. Pain. 1999;83:459-470.
Petrovic P, Petersson KM, Hansson P, Ingvar M. Habituation to pain – rCBF and autonomic effects of stimulus duration during cold pressor test. Neuroimage. 2000;11:S717.
Petrovic P, Petersson KM, Ghatan PH, Stone-Elander S, Ingvar M. Pain-related cerebral activation is altered by a distracting cognitive task. Pain. 2000;85:19-30.
Petrovic P, Kalso E, Petersson KM, Ingvar M. Shared processing in the rostral ACC during opioid and placebo treatment (Abstract). Neuroscience. 2001;120:10.
Petrovic P, Carlsson K, Peterson KM, Hansson P, Ingvar M. Context dependent amygdala deactivation during pain. Neuroimage. 2001;13:S457.
Price D., 1999. Placebo analgesia. In: Price DD, editor., Psychological mechanisms of pain and analgesia. IASP Press, Seattle, WA, pp. 155-181.
Price DD. Psychological and neural mechanisms of the affective dimension of pain [Review]. Science. 2000;288:1769-1772.
Rainville P, Duncan GH, Price DD, Carrier B, Bushnell MC. Pain affect encoded in human anterior cingulate but not somatosensory cortex. Science. 1997;277:968-971.
Rainville P, Hofbauer RK, Paus T, Duncan GH, Bushnell MC, Price DD. Cerebral mechanisms of hypnotic induction and suggestion. J Cogn Neurosci. 1999;11:110-125.
Simpson JR Jr, Drevets WC, Snyder AZ, Gusnard DA, Raichle ME. Emotion-induced changes in human medial prefrontal cortex: II. During anticipatory anxiety. Proc Natl Acad Sci USA. 2001;98:688-693.
Stuss DT, Shallice T, Alexander MP, Picton TW. A multidisciplinary approach to anterior attentional functions [Review]. Ann N Y Acad Sci. 1995;769:191-211.
Thompson SC. Will it hurt less if I can control it? A complex answer to a simple question. Psychol Bull. 1981;90:89-101.
Tulving E. Memory and Consciousness. Can. Psychol. 1985;26:1-12.
Vogt BA, Sikes RW, Vogt LJ., 1993. Anterior cingulate cortex and the medial pain system. In: Vogt BA, Gabriel M, editors., Neurobiology of cingulate cortex and limbic thalamus: a comprehensive handbook. Birkhäuser, Boston, MA, pp. 313-344.
Weisenberg M, Schwarzwald J, Tepper I. The influence of warning signal timing and cognitive preparation on the aversiveness of cold-pressor pain. Pain. 1996;64:379-385.
Willoch F. PET studies on pain and analgesia: brain activity changes and opioidergic mechanisms. Oslo: Department of Pharmacology, University of Oslo; 2001. p. 624.
Willoch F, Tolle TR, Wester HJ, Munz F, Petzold A, Schwaiger M, Conrad B, Bartenstein P. Central pain after pontine infarction is associated with changes in opioid receptor binding: a PET study with 11C-diprenorphine. AJNR Am J Neuroradiol. 1999;20:686-690.
Willoch F, Rosen G, Tolle TR, Oye I, Wester HJ, Berner N, Schwaiger M, Bartenstein P. Phantom limb pain in the human brain: unraveling neural circuitries of phantom limb sensations using positron emission tomography. Ann Neurol. 2000;48:842-849.
Zhang Y-Q, Tang J-S, Yuan B, Jia H. Inhibitory effects of electrically evoked activation of ventrolateral orbital cortex on the tail-flick reflex are mediated by periaqueductal gray in rats. Pain. 1997;72:127-135.

Anterior cingulate cortex; Attention; Orbitofrontal cortex; Positron emission tomography; Regional cerebral blood flow; Pain

© 2002 Lippincott Williams & Wilkins, Inc.