Elsevier

NeuroImage

Volume 49, Issue 2, 15 January 2010, Pages 1760-1768
NeuroImage

A meta-analysis of instructed fear studies: Implications for conscious appraisal of threat

https://doi.org/10.1016/j.neuroimage.2009.09.040Get rights and content

Abstract

In classical Pavlovian fear conditioning, a neutral stimulus (conditioned stimulus, CS) comes to be evaluated as threatening due to its association with an aversive stimulus (unconditioned stimulus, UCS), and elicits fear. In a subtype of fear conditioning paradigms, called instructed fear or anticipatory anxiety, subjects are made aware of the CS–UCS association prior to actually experiencing it. Initial fear elicitation during this type of conditioning results from the negative evaluation of the CS as a consequence of CS–UCS contingency awareness. Prior reports have suggested that this conscious appraisal process is mediated by a variety of brain regions, including rostral dorsomedial prefrontal/dorsal anterior cingulate cortex (dmPFC/dACC), lateral prefrontal cortex (lPFC), posterior cingulate, hippocampus/parahippocampus, and nucleus accumbens, but there is little overlap between results. We reasoned that a formal meta-analysis of existing instructed fear studies should help narrow down the search for conscious appraisal areas in fear conditioning to those consistently activated across studies. We found consistent activation in rostral dmPFC but not in the other candidate areas. These results allow for maintaining the theory that the rostral dmPFC is involved in conscious threat appraisal. We also report a meta-analysis of uninstructed (classical) fear conditioning studies in which we found notable activation in more posterior parts of the dmPFC/dACC that overlapped with some of the instructed fear activations. These data suggest that mid regions of the dmPFC/dACC are part of a “core” fear network that is activated irrespective of how fear was learnt.

Introduction

The appraisal of stimuli in terms of their emotional–motivational significance to the organism is thought to be causal in the generation of emotional responses (Roseman and Smith, 2001). Theorists have made a distinction between non-conscious, often quick and less elaborate, appraisal processes and conscious forms of appraisal which include propositional analysis (Leventhal and Scherer, 1987, Robinson, 1998). The working of non-conscious appraisals can be observed in those classical Pavlovian conditioning experiments in which the CS is presented below the perceptual threshold, yet evokes conditioned responses (CRs) (e.g., Critchley et al., 2002, Esteves et al., 1994, Morris et al., 1998, Ohman, 2005, Olsson and Phelps, 2004). As subjects cannot consciously predict the UCS from the CS in this type of paradigm, CRs must necessarily be a result of non-conscious processing. The opposite is true for instructed fear paradigms where subjects are told before the experiment that a given CS will be followed by a UCS. Because subjects have never experienced actual CS–UCS pairing, initial CRs must be caused by a conscious appraisal of the CS as threatening on the basis of the explicit knowledge about its relation to the UCS (i.e., CS–UCS contingency awareness). This conclusion is further supported by results from a recent instructed fear conditioning study in which subjects never actually received a UCS (preventing them from learning through experience) and produced skin conductance CRs to a CS when it was presented supra-threshold but, critically, not when it was presented sub-threshold and thus remained unperceived (Olsson and Phelps, 2004). Hence, contingency awareness is likely to be causal for CR generation in instructed fear.

Conscious appraisal of threat thus comprises explicit knowledge of the CS–UCS contingency as well as consequential cognitive elaborations about the CS and its implications. It may additionally include awareness of, and reflections about, the bodily and subjective responses induced by the CS (which have emotional stimulus quality in their own right and can thus also generate negative reactions). Various experimental approaches have been taken to dissociate the neural substrates of non-conscious and conscious appraisal in fear conditioning. Bechara et al., 1995, Clark and Squire, 1998 explicitly asked subjects about CS–UCS contingencies following uninstructed conditioning, showing that hippocampal patients cannot acquire contingency knowledge. Carter et al. (2006) tracked the emergence of contingency awareness online by having subjects rate shock expectancy while undergoing conditioning, finding activity correlated with explicit shock expectancy in the lPFC (bilateral middle frontal gyrus at MNI coordinates x, y, z =  36, 51, 30 and 36, 51, 36) and, just below the statistical threshold, in the parahippocampal gyrus. Kalisch et al. (2006) limited conscious processing by combining instructed fear conditioning with an attention- and working memory-demanding task, finding reduced conditioning-related rostral dmPFC/dACC activation (at − 8,38,28) in the high- compared to the low-load condition. Tabbert et al. (2006) compared an instructed to an uninstructed (classical) fear conditioning group, of which only the former developed contingency awareness. The authors found rostral dmPFC, temporal, and parietal activation in the aware group, which was, however, not significantly stronger than in the unaware group. Klucken et al. (2009a) compared subjects that accidentally developed contingency awareness during uninstructed conditioning to those that did not, finding stronger nucleus accumbens activation in the aware group. Finally, Knight et al. (2009) presented the same CS both sub- and supra-threshold during uninstructed conditioning. This manipulation was associated with explicit shock expectancy and hippocampal and parahippocampal as well as posterior cingulate activation in the supra-threshold trials. One lesion study (Funayama et al., 2001) showed that the left, but not right, amygdala is necessary to produce skin conductance CRs in instructed fear conditioning.

This rather divergent pattern of areas associated with conscious threat appraisal led us to wonder if any of these areas are consistently activated by instructed fear conditioning across studies. As subjects are aware of the CS–UCS contingency throughout an instructed fear experiment, and thus likely to reflect upon the threatening situation, we reasoned that those areas most consistently activated by instructed fear are the strongest candidates for mediating conscious appraisal of threat. This does not imply that non-conscious processing does not contribute to the fear reaction in instructed fear paradigms, in particular when these involve CS–UCS pairings and thus also permit direct learning from experience. Likewise, the mere consistent activation of an area X in instructed fear studies does not prove the area subserves conscious appraisal (as it may also, for example, subserve non-conscious processing). However, it is reasonable to argue that areas that are not consistently active across instructed fear studies are unlikely candidates for this function.

The most appropriate method for determining such overlap is formal quantitative meta-analysis since, unlike the traditional method of providing scatter plots of activation maxima, formal meta-analysis generates quantitative scores of activation consistency that can be tested for significance using established statistical methods (Wager et al., 2007). Further, random between-study variance can be taken into account (Wager et al., 2007).

One prior meta-analysis had analyzed a mixed sample of instructed and uninstructed conditioning studies (Etkin and Wager, 2007). In addition to reporting a meta-analysis of instructed fear, we therefore also calculated a meta-analysis of “pure” uninstructed fear and attempted to determine regional overlap between both. Our rationale was that areas that are activated by both instructed and uninstructed fear conditioning, irrespective of the contribution of conscious appraisal processes, can be regarded as belonging to a “core” fear network.

Section snippets

Terminology and study selection

In a typical uninstructed fear conditioning experiment, a subject learns over repeated trials that a CS (often a visual or auditory stimulus) is often or always followed by a UCS such as electric shock, heat pain, painful pressure, aversive sounds or pictures, unpleasant odor, or loss of money. Human fear conditioning paradigms usually involve discrimination between one or several such CSs (termed CS+) and one or several CS−'s which is/are never paired with the UCS, thereby controlling for

Instructed fear conditioning

At an FDR threshold of q < 0.01, 15 clusters showed consistent activation across instructed fear studies (Table 3 and Fig. 1). The most consistently activated areas, with an activation density score of Pˆv>0.5, were a large cluster in the bilateral mid dmPFC/dACC which extended into the presupplemental motor area (preSMA); a cluster in the bilateral anterior insulae, with an extension into the right putamen; and a cluster in the bilateral caudate–putamen which extended into the anterior thalamus

Discussion

The main result of this study is that instructed fear consistently activates a rostral part of the dorsomedial prefrontal cortex and that activation in the other conscious threat appraisal candidate areas is considerably less consistent. Consistent activation in this paradigm as such does not prove an involvement in conscious appraisal, since many cognitive processes other than conscious appraisal can be assumed to operate during instructed fear conditioning. In particular, during most

Acknowledgments

We thank T. Wager for meta-analysis code, F. Eippert, U. Herwig, J. Jensen, J. Nitschke, I. Sarinopoulos, D. Schiller, and R. Stark for providing additional activation data from contrasts not included in their original articles, and T. Egner for critical suggestions. This work was funded by the Deutsche Forschungsgemeinschaft (DFG Emmy Noether grant KA1623/3-1 to R.K.). We also thank C. Büchel for initial financial support.

References (91)

  • GenoveseC.R. et al.

    Thresholding of statistical maps in functional neuroimaging using the false discovery rate

    Neuroimage

    (2002)
  • HerwigU. et al.

    Expecting unpleasant stimuli—an fMRI study

    Psychiatry Res.

    (2007)
  • JensenJ. et al.

    Direct activation of the ventral striatum in anticipation of aversive stimuli

    Neuron

    (2003)
  • KalischR.

    The functional neuroanatomy of reappraisal: time matters

    Neurosci. Biobehav. Rev.

    (2009)
  • KalischR. et al.

    Levels of appraisal: a medial prefrontal role in high-level appraisal of emotional material

    Neuroimage

    (2006)
  • KluckenT. et al.

    Neural, electrodermal and behavioral response patterns in contingency aware and unaware subjects during a picture–picture conditioning paradigm

    Neurosci.

    (2009)
  • KnightD.C. et al.

    Neural substrates of explicit and implicit fear memory

    Neuroimage

    (2009)
  • KoberH. et al.

    Functional grouping and cortical–subcortical interactions in emotion: a meta-analysis of neuroimaging studies

    Neuroimage

    (2008)
  • LaBarK.S. et al.

    Human amygdala activation during conditioned fear acquisition and extinction: a mixed-trial fMRI study

    Neuron

    (1998)
  • MenonM. et al.

    Temporal difference modeling of the blood-oxygen level dependent response during aversive conditioning in humans: effects of dopaminergic modulation

    Biol. Psychiatry

    (2007)
  • MiladM.R. et al.

    A role for the human dorsal anterior cingulate cortex in fear expression

    Biol. Psychiatry

    (2007)
  • NagaiY. et al.

    Activity in ventromedial prefrontal cortex covaries with sympathetic skin conductance level: a physiological account of a “default mode” of brain function

    Neuroimage

    (2004)
  • NitschkeJ.B. et al.

    Functional neuroanatomy of aversion and its anticipation

    Neuroimage

    (2006)
  • O'DohertyJ.P. et al.

    Neural responses during anticipation of a primary taste reward

    Neuron

    (2002)
  • OhmanA.

    The role of the amygdala in human fear: automatic detection of threat

    Psychoneuroendocrinol

    (2005)
  • PattersonJ.C. et al.

    Task-independent functional brain activity correlation with skin conductance changes: an fMRI study

    Neuroimage

    (2002)
  • PhanK.L. et al.

    Neural correlates of individual ratings of emotional salience: a trial-related fMRI study

    Neuroimage

    (2004)
  • PhelpsE.A. et al.

    Extinction learning in humans: role of the amygdala and vmPFC

    Neuron

    (2004)
  • SchaeferA. et al.

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

    Neuroimage

    (2003)
  • SeminowiczD.A. et al.

    A re-examination of pain–cognition interactions: implications for neuroimaging

    Pain

    (2007)
  • StarkR. et al.

    Influence of the stress hormone cortisol on fear conditioning in humans: evidence for sex differences in the response of the prefrontal cortex

    Neuroimage

    (2006)
  • TabbertK. et al.

    Hemodynamic responses of the amygdala, the orbitofrontal cortex and the visual cortex during a fear conditioning paradigm

    Int. J. Psychophysiol.

    (2005)
  • TabbertK. et al.

    Dissociation of neural responses and skin conductance reactions during fear conditioning with and without awareness of stimulus contingencies

    Neuroimage

    (2006)
  • TaylorS.F. et al.

    Subjective rating of emotionally salient stimuli modulates neural activity

    Neuroimage

    (2003)
  • VeitR. et al.

    Brain circuits involved in emotional learning in antisocial behavior and social phobia in humans

    Neurosci. Lett.

    (2002)
  • VuilleumierP. et al.

    Neural response to emotional faces with and without awareness: event-related fMRI in a parietal patient with visual extinction and spatial neglect

    Neuropsychologia

    (2002)
  • BecharaA. et al.

    Double dissociation of conditioning and declarative knowledge relative to the amygdala and hippocampus in humans

    Science

    (1995)
  • BirbaumerN. et al.

    Deficient fear conditioning in psychopathy: a functional magnetic resonance imaging study

    Arch. Gen. Psychiatry

    (2005)
  • BlackwoodN.J. et al.

    Persecutory delusions and the determination of self-relevance: an fMRI investigation

    Psychol. Med.

    (2004)
  • BuchelC. et al.

    Amygdala–hippocampal involvement in human aversive trace conditioning revealed through event-related functional magnetic resonance imaging

    J. Neurosci.

    (1999)
  • CarterR.M. et al.

    Working memory and fear conditioning

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

    (2003)
  • ClarkR.E. et al.

    Classical conditioning and brain systems: the role of awareness

    Science

    (1998)
  • CritchleyH. et al.

    Explicit and implicit neural mechanisms for processing of social information from facial expressions: a functional magnetic resonance imaging study

    Hum. Brain Mapp.

    (2000)
  • CritchleyH.D. et al.

    Human cingulate cortex and autonomic control: converging neuroimaging and clinical evidence

    Brain

    (2003)
  • CritchleyH.D. et al.

    Neural systems supporting interoceptive awareness

    Nat. Neurosci.

    (2004)
  • Cited by (316)

    View all citing articles on Scopus
    View full text