Elsevier

NeuroImage

Volume 22, Issue 1, May 2004, Pages 447-455
NeuroImage

The neural correlates of placebo effects: a disruption account

https://doi.org/10.1016/j.neuroimage.2004.01.037Get rights and content

Abstract

The neurocognitive pathways by which placebo effects operate are poorly understood. Positron emission tomography (PET) imaging was used to assess the brain response of patients with chronic abdominal pain (irritable bowel syndrome; IBS) to induced intestinal discomfort both before and after a 3-week placebo regimen. A daily symptom diary was used to measure symptom improvement. Increases in right ventrolateral prefrontal cortex (RVLPFC) activity from pre- to post-placebo predicted self-reported symptom improvement, and this relationship was mediated by changes in dorsal anterior cingulate (dACC), typically associated with pain unpleasantness. These results are consistent with disruption theory [Lieberman, M.D., 2003. Reflective and reflexive judgment processes: a social cognitive neuroscience approach. In: Forgas, J.P., Williams, K.R., von Hippel, W. (Eds.), Social Judgments: Explicit and Implicit Processes. Cambridge Univ. Press, New York, pp. 44–67], which proposes that activation of prefrontal regions associated with thinking about negative affect can diminish dACC and amygdala reactivity to negative affect stimuli. This is the first study to identify a neural pathway from a region of the brain associated with placebos and affective thought to a region closely linked to the placebo-related outcome of diminished pain unpleasantness.

Introduction

Placebo effects and the power of belief over physical outcomes in the body have a history as long as medicine itself. Only recently, however, have the neural bases of these psychophysiological phenomena begun to be examined. Current research has shown that placebos can produce changes in brain activity similar to the pharmacological agents they are replacing. For instance, placebos administered in the treatment of depression have been shown to alter brain activity in much the same way as fluoxetine, an effective antidepressant medication (Mayberg et al., 2002). Such findings suggest that the final result of placebo administration may mimic the effects of active agents; however, the mediating neurocognitive process by which ingesting a placebo and placebo-related beliefs lead to these final neural results has not yet been examined. In other words, it is unknown what placebo-specific neural activity sets in motion the subsequent symptom-specific changes in brain activity. In a positron emission tomography (PET) study of placebo effects on visceral pain, we investigated a possible mediating mechanism derived from disruption theory Eisenberger et al., 2003, Lieberman, 2003, Lieberman et al., 2003, which suggests that thinking about affective processes has the unintended effect of diminishing the reactivity of brain regions involved in the automatic generation of negative affect.

Neuroimaging research on placebo effects has focused primarily, though not exclusively, on the neural outcome of placebo effects—identifying changes in the brain regions that most directly relate to subjective changes in symptoms, and comparing these changes with those stemming from active pharmacological agents de la Fuente-Fernandez et al., 2001, Leuchter et al., 2002, Mayberg et al., 2002. This research has been critical in demonstrating that top-down, belief-related placebo effects modulate the activity of brain regions ordinarily affected by other treatments that presumably operate through bottom-up mechanisms that are not belief-related. Such work speaks to the longstanding concern regarding whether placebo effects and other expectancy effects are real, experimental artifacts, or self-presentational effects (Hrobjartsson and Gotzsche, 2001). It is not clear from previous neuroimaging studies, however, whether the neural regions that have been identified are the neural correlates of the placebo effect or the neural correlates of the placebo effect's downstream consequences on symptom outcomes. Though a drug and a placebo may both affect brain region X, the drug may do so directly, whereas placebo effects are typically mediated by placebo-induced thoughts.1

In previous studies, right prefrontal cortex, anterior cingulate and striatal dopamine have each been implicated in placebo effects, but it is not clear that these are all plausible neurocognitive candidates for transforming placebo-induced thoughts into altered experience. Petrovic et al. (2002) recently compared placebo-induced analgesia to opioid-induced analgesia and found that whereas both placebos and induced opioids led to similar changes in the rostral anterior cingulate and brainstem, only the placebo led to increased activity in right prefrontal cortex (x = 30, y = 30, z = −6). Though Petrovic et al. suggested this region may be involved in cognitive regulation of pain, a neurocognitive model of this process has yet to emerge. We begin with a brief review of the neuroanatomy of pain and then describe how disruption theory (Eisenberger et al., 2003; Lieberman, 2003), a model of the cognitive regulation of affect, may be applied in the context of pain and placebo effects.

The experience of pain can be divided into sensory and affective aspects, corresponding to the sensory intensity and the unpleasantness of pain, respectively Melzack and Casey, 1968, Price, 2000. In recent years, neuroimaging work has suggested that these two aspects of pain are neurally dissociable as well. Whereas the sensory aspects of pain, including its intensity and location, are represented primarily in somatosensory cortex and the insula (Peyron et al., 1999; cf. Coghill et al., 1999), pain unpleasantness is represented in the dorsal anterior cingulate (dACC; Peyron et al., 2000, Rainville et al., 1997, Tolle et al., 1999). Additionally, patients who have had their anterior cingulate surgically removed report that they are still able to feel the intensity of pain, but are no longer bothered by it (Foltz and Lowell, 1962). In contrast, a patient who had his somatosensory cortex removed could still report pain distress despite difficulties in reporting on sensory aspects of the pain (Ploner et al., 1999).

We take the established relationship between dACC's response to painful stimulation and experienced pain unpleasantness as a starting point for our investigation. We assume that changes in the dACC represent one possible end result of a placebo's neurocognitive effects in the domain of pain analgesia.2 Using dACC changes as a neural anchor or endpoint indicative of changes in the unpleasant experience of pain, functional connectivity analyses Lieberman et al., in press, Poldrack et al., 2001 can be used to determine which other brain regions may be modulating the reactivity of dACC and the subjective reports of pain. Functional connectivity analyses are entirely correlational, and thus cannot determine causal links directly. Such links can be inferred with greater confidence to the extent that a strong theoretical rationale and preestablished neural pathways suggest a route by which the effects may be achieved. Disruption theory (Lieberman, 2003) posits that right ventrolateral prefrontal cortex (RVLPFC), the same region identified by Petrovic et al. (2002) in the cognitive regulation of pain, is well-suited to transform placebo-induced beliefs into at least some of its known subjective and neural outcomes.

Disruption theory (Lieberman, 2003) proposes that the reflective conscious processes that are engaged in response to automatic negative affective processes tend to inhibit or ‘disrupt’ the very same negative affective processes by virtue of a hardwired feedback mechanism. Reflective, nonautomatic, conscious processes, marked by intentionality, effort, awareness, and propositional structure, are typically used sparingly because they are fragile and limited, despite their flexibility (Wegner and Bargh, 1998). For example, keeping one phone number in mind is relatively easy, yet keeping two in mind simultaneously is nearly impossible. Thus, people tend to be ‘cognitive misers’ (Fiske and Taylor, 1991), using as little of this limited resource as possible so that it will be available when it is really needed. If the cognitive miser view is correct, then it would be advantageous to have an alarm system that signals when reflective conscious processes are really needed. Previous theories LeDoux, 1996, Schwarz and Bless, 1991 suggest that negative affect serves precisely this function. Disruption theory hypothesizes that once negative affect has sounded the alarm and set reflective processing in motion to respond to the current situation, it is adaptive to have the alarm turned off or at least muted a bit. If the negative affect alarm continues to sound at the same ‘volume’, it is likely that some reflective processing resources will be consumed by attending to the alarm, rather than operating on the environment to respond to whatever caused the negative affect. Consider the extreme case of panic when the sense of anxiety and alarm is so overwhelming that the individual is essentially paralyzed, unable to respond to real threats in the environment effectively.

Disruption theory suggests that primates evolved a hardwired feedback mechanism such that the activation of brain regions associated with reflective processing responses to negative affect will disrupt, at least to some extent, the activity of brain regions that generate the negative affect alarm. Such a disruption mechanism would keep the distracting effect of an automatic affect alarm to a minimum once the alarm has been initially registered, promoting a more efficient and focused response to the event(s) activating the alarm.

Disruption theory posits RVLPFC as a focal point because it is involved both in reflective conscious responses to negative affect and the disruption of activity in the neural sources of negative affect in the amygdala and anterior cingulate. RVLPFC has been implicated in numerous studies of inhibition generally Aron et al., 2003, Garavan et al., 1999, Iversen and Mishkin, 1970, Konishi et al., 1999, and overcoming negative affect more specifically (Eisenberger et al., 2003; Hariri et al., 2000, Monchi et al., 2001, Small et al., 2001).

A number of studies have also implicated VLPFC in negative affective processes Eugene et al., 2003, Gottfried et al., 2002, Markowitsch et al., 2003; however, disruption theory suggests that this region supports the labeling, interpretation, and thought about negative affect rather than the generation or experience of negative affect itself. Five studies have shown greater VLPFC activation when individuals were required to make explicit judgments of affect present in pictures compared to when the affective valence of the pictures was incidental to the judgment dimension Cunningham et al., 2003, Gorno-Tempini et al., 2001, Gur et al., 2002, Nakamura et al., 1999, Narumoto et al., 2000. One of these studies found VLPFC to be more active for negative judgments than positive judgments (Cunningham et al., 2003). A sixth study found VLPFC activity when individuals were required to make propositional appraisals of affective stimuli (Schaefer et al., 2003). There are other areas that have also been implicated in reflective processing of affect including dorsomedial PFC and ACC Crosson et al., 1999, Teasdale et al., 1999; however, these regions have not been implicated in most studies of inhibition. It is the joint involvement in inhibition and reflective emotional processing that is central to disruption theory.

RVLPFC also has the neuroanatomical connections necessary to inhibit negative affective processes. RVLPFC comprises Brodmann area (BA) 47, as well as BA44 and BA45 in the human, and is considered homologous with primate area 12 in lateral orbitofrontal cortex (LOFC; Ongur and Price, 2000). LOFC has dense projections to the amygdala and dACC Carmichael and Price, 1995, Cavada et al., 2000, Vogt and Pandya, 1987. Most specific to a disruption account of placebo analgesia, electrical stimulation of LOFC in rats diminishes pain behavior in response to painful stimulation (Zhang et al., 1997) and has been implicated in some aspect of pain processing in nearly a dozen imaging studies (see Petrovic et al., 2000, Table 2).

In the context of placebo effects on pain, we propose that the thoughts, beliefs, and expectations associated with taking a placebo, such as thoughts about the expected reduction in pain unpleasantness, lead to increased activation in RVLPFC. This activation, according to disruption theory, should to some degree inhibit the activations in dACC associated with pain unpleasantness, which should bring about symptom improvements and diminished experience of pain. In other words, increased activity in RVLPFC should be related to self-reported symptom improvement, but this relation should be mediated by changes in dACC activity, which is more proximally related to the experience of pain.

Patients with irritable bowel syndrome (IBS) were scanned using PET both before and following 3 weeks of taking a placebo (or active drug) for their gastrointestinal symptoms. IBS is a common syndrome characterized by chronic, recurrent abdominal pain and discomfort (associated with altered bowel habits). It is distinguished from inflammatory bowel disease by the lack of any detectable organic manifestation in the gut (Drossman et al., 2000). This has led researchers to posit that IBS symptoms may be a consequence of a hyperactive neural response to normal visceral activity (Mayer et al., 2001). One benefit of recruiting IBS patients for a study examining placebo effects is that they respond to placebos at a high rate (Spiller, 1999) relative to the range of placebo response rates found in other domains and clinical populations Beecher, 1955, Brown et al., 1992, Kissel and Barrucand, 1964.

On scan day 1 (‘pre-placebo’), patients were scanned both at rest and during controlled rectal stimulation, known to produce physical discomfort similar to IBS symptoms (Mertz et al., 1995). Patients were then given pharmacologically inactive pills to take daily for the next 3 weeks, and were told that the pill might diminish their IBS symptoms. Patients kept a symptom diary starting a week before scan day 1 and continuing throughout the 3 weeks of placebo administration. On scan day 2 (‘post-placebo’), after having taken the placebo for 3 weeks, patients were again scanned multiple times at rest and during rectal stimulation.

We hypothesized that the dACC response to rectal stimulation would be diminished post-placebo relative to pre-placebo, but only to the extent that a placebo response occurred, operationalized as reported symptom improvements. Habituation to the experimental paradigm may account for general drops in dACC activity across the time span; however, this should not relate to the changes in self-reported symptoms. We also hypothesized that the RVLPFC response would be greater post-placebo relative to pre-placebo as a function of symptom improvement. Finally, we predicted that the relation between RVLPFC and symptom improvement would be mediated by changes in dACC, which would suggest that RVLPFC has its impact on symptoms by diminishing the reactivity of the dACC, which in turn is more directly related to symptom improvements.

Section snippets

Participants

All participants met Rome I criteria for IBS (Thompson et al., 1989) and were clinically and endoscopically without inflammatory or other structural intestinal disease. Potential participants were also excluded if they suffered any major psychiatric disorder, including clinical anxiety or depression, or scored above 63 on any two of the nine subscales or global symptom indices on the SCL-90R (Derogatis and Lazarus, 1994). Patients discontinued peripherally acting treatments for IBS at least 7

Question 1: Was there a placebo response?

To check for significant symptom improvement, mean symptom levels from pre-treatment (week 1) and post-treatment (week 4) were computed from daily diary ratings of frequency, severity, and duration of symptoms. Symptom improvement was seen in 71% of the participants, with an average improvement of 8.5%, t(12) = 2.21, P < 0.05.4 This change is in

Discussion

This is the first neuroimaging study to investigate the functional contribution of RVLPFC in the placebo response. Petrovic et al. (2002) had previously observed that RVLPFC was activated during placebo analgesia but not during opioid analgesia. In the current study, we observed a strong positive correlation between pre- to post-placebo increases in RVLPFC and subjective reports of symptom improvement. Mediational analyses indicated that this relationship was mediated by dACC such that

Acknowledgements

We thank Leah Fitzgerald, N.P, Ph.D., and Jean Stains, R.N., for invaluable assistance in executing the study; and Molly Crockett, Naomi Eisenberger, Sarah Master, Jennifer Pfeifer, Ajay Satpute, and Robert Spunt for comments on an earlier draft. This work was supported in parts by grants from National Institute of Diabetes and Digestive and Kidney Diseases (DK48351 and DK64539) to E.A.M., from the National Institute of Mental Health to B.N. (R01NR07768), M.D.L (R21MH66709-01), and J.M.J.

References (71)

  • M. Ploner et al.

    Pain affect without pain sensation in a patient with a postcentral lesion

    Pain

    (1999)
  • A. Schaefer et al.

    Neural correlates of disruption account of placebos 30 “hot” and “cold” emotional processing: a multilevel approach to the functional anatomy of emotion

    NeuroImage

    (2003)
  • L. Vase et al.

    The contributions of suggestion, desire, and expectation to placebo effects in irritable bowel syndrome patients. An empirical investigation

    Pain

    (2003)
  • A.D. Wagner et al.

    Recovering meaning: left prefrontal cortex guides controlled semantic retrieval

    Neuron

    (2001)
  • Y. Zhang et al.

    Inhibitory effects of electrically evoked activation of ventrolateral orbital cortex on the tail-flick reflex are mediated by periaqueductal gray in rats

    Pain

    (1997)
  • M. Amanzio et al.

    Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioning-activated specific subsystems

    J. Neurosci.

    (1999)
  • A.R. Aron et al.

    Stop-signal inhibition disrupted by damage to right inferior frontal gyrus in humans

    Nat. Neurosci.

    (2003)
  • R.M. Baron et al.

    The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations

    J. Pers. Soc. Psychol.

    (1986)
  • H.K. Beecher

    The powerful placebo

    J. Am. Med. Assoc.

    (1955)
  • F. Benedetti et al.

    Conscious expectation and unconscious conditioning in analgesic, motor, and hormonal placebo/nocebo responses

    J. Neurosci.

    (2003)
  • S.T. Carmichael et al.

    Limbic connections of the orbital and medial prefrontal cortex in macaque monkeys

    J. Comp. Neurol.

    (1995)
  • C. Cavada et al.

    The anatomical connections of the Macaque monkey orbitofrontal cortex. A review

    Cereb. Cortex

    (2000)
  • R.C. Coghill et al.

    Pain intensity processing within the human brain: a bilateral, distributed mechanism

    J. Neurophysiol.

    (1999)
  • J. Cohen et al.

    Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences

    (1983)
  • B. Crosson et al.

    Left-hemisphere processing of emotional connotation during word generation

    NeuroReport

    (1999)
  • W.A. Cunningham et al.

    Neural components of social evaluation

    J. Pers. Soc. Psychol.

    (2003)
  • R. de la Fuente-Fernandez et al.

    Expectation and dopamine release: mechanism of the placebo effect in Parkinson's disease

    Science

    (2001)
  • L.R. Derogatis et al.

    SCL-90-R, Brief Symptom Inventory, and matching clinical rating scales

  • D.A. Drossman et al.

    ROME: II. The Functional Gastrointestinal Disorders. Diagnosis, Pathophysiology and Treatment: A Multinational Consensus Consensus

    (2000)
  • N.I. Eisenberg et al.

    Does rejection hurt? An fMRI study of social exclusion

    Science

    (2003)
  • Eisenberg, N.I., & Lieberman, M.D. (in press). Why it hurts to be left out: the neurocognitive overlap between physical...
  • S.T. Fiske et al.

    Social Cognition

    (1991)
  • E.L. Foltz et al.

    Pain “relief” by frontal cingulomotomy

    J. Neurophysiol.

    (1962)
  • S.D. Forman et al.

    Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of cluster-size threshold

    Magn. Reson. Med.

    (1995)
  • H. Garavan et al.

    Right hemispheric dominance of inhibitory control: an event-related functional MRI study

    Proc. Natl. Acad. Sci. U. S. A.

    (1999)
  • Cited by (0)

    View full text