- Email: [email protected]
The anticipation of pain modulates spatial attention: evidence for pain-specificity in high-pain catastrophizers
Pain 111 (2004) 392–399 www.elsevier.com/locate/pain
The anticipation of pain modulates spatial attention: evidence for pain-specificity in high-pain catastrophizers Stefaan Van Dammea,*, Geert Crombeza, Christopher Ecclestonb a
Department of Experimental, Clinical and Health Psychology, Ghent University, Henri Dunantlaan 2, 9000 Ghent, Belgium b University of Bath, Bath BA2 7AY, UK Received 4 May 2004; received in revised form 7 July 2004; accepted 19 July 2004
Abstract Recent studies have suggested that the anticipation of pain may modulate spatial attention. However, it is possible that this modulation reflects a general effect of anticipating somatosensory stimulation, without being pain-specific. In the present study, we therefore compared the effect of the anticipation of somatosensory stimulation on spatial attention between two groups, using conditioned signals in a spatial cueing paradigm. In the pain group, signals predicted painful electrocutaneous stimulation, whereas in the control group, signals predicted non-painful vibrotactile stimulation. Tests between both groups showed that attentional engagement was equally facilitated by the anticipation of somatosensory stimulation in both groups. Interestingly, disengagement of attention was more retarded by the anticipation of pain than by the anticipation of non-painful vibrotactile stimulation in participants high in catastrophic thinking about pain. Theoretical and clinical implications of these findings are discussed. q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. Keywords: Spatial attention; Experimental pain; Anticipation; Catastrophizing
1. Introduction The attentional demand of pain has been documented by an impressive number of studies in clinical and nonclinical populations (Eccleston and Crombez, 1999; Pincus and Morley, 2001). Furthermore, it has been demonstrated that attention to pain is intensified by the threat value of pain, and particularly by catastrophic thinking about pain (Crombez et al., 1998a; Van Damme et al., 2004a). Recently, research interest has also been moving to the attentional effects of pain anticipation. The anticipation of pain may have an important protective function, allowing the avoidance of bodily harm by the initiation of adaptive behaviour. In a number of studies, behavioural evidence was found for the attentional demand of pain anticipation, particularly in high-pain catastrophizers (Spence et al., * Corresponding author. Tel.: C32-9-2649-105; fax: C32-9-2649-149. E-mail address: [email protected] (S. Van Damme).
2002; Van Damme et al., 2002b). Furthermore, neurophysiological evidence has indicated that the anticipation of pain may trigger cortical systems closely involved in the experience of pain itself (Ploghaus et al., 1999; Porro et al., 2002), and several studies have begun now to map the cortical representation of pain anticipation (Jensen et al., 2003; Ploghaus et al., 2003; Porro et al., 2003). One particular way in which pain can be anticipated is the allocation of attention to the expected location of threat. Indirect evidence for this view was provided by Honore´ et al. (1995), who found that sensitivity to experimental pain stimuli was extended when participants directed their eyes to the stimulated location. However, no studies have investigated whether pain anticipation modulates spatial attention. In order to examine this, we adapted a spatial cueing paradigm (Posner et al., 1987; Van Damme et al., 2004c). In this paradigm, participants detect visual targets, preceded by ipsilateral (valid) and contralateral (invalid) cues. Valid cues typically lead to response time benefits (due to attentional engagement at the validly cued
0304-3959/$20.00 q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2004.07.022
S. Van Damme et al. / Pain 111 (2004) 392–399
location), whereas invalid cues lead to response time costs (due to disengagement from the invalidly cued location). In order to investigate the effect of pain anticipation on spatial attention, we used conditioned signals of impending pain as spatial cues. In a recent study with this paradigm, we demonstrated that signals of impending pain enhanced spatial cueing (Van Damme et al., 2004c). However, this finding can be alternatively explained, possibly reflecting a pain-unspecific modulation of spatial attention by the anticipation of somatosensory stimulation in general (Spence et al., 2001). In the present study, we therefore compared the effects of somatosensory anticipation on spatial attention between two groups. In the pain group, we used signals of impending pain as spatial cues. In the control group, we used signals predicting non-painful vibrotactile stimulation as spatial cues. Three hypotheses were tested. First, we expected that attentional engagement would be more facilitated in the pain group compared to the control group. Second, we expected that disengagement would be more retarded in the pain group than in the control group. Third, we expected that the hypothesized effects would be more pronounced in participants high in catastrophic thinking about pain.
2. Methods 2.1. Participants Fifty-four undergraduate psychology students (8 males and 46 females; mean ageZ18.22 years, SDZ0.74; range 17–21 years) from Ghent University participated in order to fulfil course requirements. All participants gave informed consent and were free to terminate the experiment at any time. Participants were randomly assigned to the pain group (nZ28; 4 men and 24 women; ageZ18.36 years) and to the control group (nZ26; 4 men and 22 women; ageZ18.08 years). Each person had normal or corrected-to-normal eyesight. Experimental duration was approximately 30 min. Two participants from the pain group were excluded from the analyses (see later). 2.2. Exogenous cueing task The exogenous cueing task was programmed and presented by the INQUISIT Millisecond software package (Inquisit 1.32, 2001) on a S710 Compaq Deskpro computer with a 72 Hz, 17-inch colour monitor. INQUISIT measures response times with millisecond accuracy (De Clercq et al., 2003). Target stimuli consisted of black squares (1.1!1.1 cm width), presented on a white background. Two coloured frames (green and pink; 4.8 cm high!6.5 cm wide) served as exogenous cues. Each trial began with a fixation cross in the middle of the screen (duration of
393
Fig. 1. Schematic illustration of validly and invalidly cued trials. On each trial, a fixation cross was presented for 1000 ms. This was immediately followed by the presentation of a cue (200 ms duration). Immediately after cue offset, the target was introduced at an SOA of 200 ms. Targets were presented at the same location of cues (validly cued trials), or at the opposite location of cues (invalidly cued trials).
1000 ms). The cue was presented 9.28 from a fixation cross for a duration of 200 ms. Targets followed cues at a Stimulus Onset Asynchrony (SOA: time between cue onset and target onset) of 200 ms. Cue location correctly predicted target location on two-thirds of the test trials (validly cued trials). On the remaining test trials, cue location incorrectly predicted target location (invalidly cued trials). These trial types are illustrated in Fig. 1. Participants were instructed to respond to the target as quickly as possible without sacrificing accuracy, by pressing the corresponding key on a standard AZERTY computer keyboard (left index finger at ‘q’ key for left targets; right index finger at ‘5’ key for right targets). A trial ended when a participant responded or 2000 ms had elapsed. In order to control for anticipatory responses (responses to cues instead of targets), a number of catch trials were presented, in which the cue was not followed by a target. Furthermore, in order to ensure that participants maintained gaze at the middle of the screen, a number of digit trials were presented. On these trials, the fixation cross was followed only by a randomly selected digit between 1 and 9 for a duration of 50 ms. Participants were instructed to report the digit aloud. If participants were not able to report the digits (correctly), this indicated that they did not maintain gaze at the fixation cross. 2.3. Experimental manipulation Cues were differentially conditioned using two colours. In half of the presentations, the conditioned cue (CSC) was 300 ms after offset followed by a somatosensory stimulus
394
S. Van Damme et al. / Pain 111 (2004) 392–399
(UCS), whereas the other cue (CSK) was never followed by the UCS. The colours of the CSC and CSK were counterbalanced across participants. The CSC and CSK were presented equally often, in a fixed random order with a maximum of three consecutive presentations of the same cue. In the pain group, the UCS was an aversive electrocutaneous stimulus, delivered by an AC stimulator with an internal frequency of 50 Hz. Electrocutaneous stimuli were delivered at the external side of the left forearm by two lubricated Fukuda standard Ag/AgCl electrodes (1 cm diameter). Intensity of the electrocutaneous stimulus was 0.63 mA, with an instantaneous rise and fall time, and a duration of 300 ms. Using the sensory pain words of the Dutch McGill Pain Questionnaire (Vanderiet et al., 1987), it was found that the characteristics of this stimulus were best described as pricking, boring, flickering, electric, and cutting. Previous studies have further demonstrated that an electrocutaneous stimulus of this intensity is painful but easy to tolerate, and that the attentional demand of this stimulus is strongly related to the specific threat value of pain (Crombez et al., 1998a). In the control group, the UCS was a non-painful vibrotactile stimulus, delivered by a vibration element. This element consisted of a Nokia 3210 vibramotor, enveloped by a plastic cylinder (1.3 cm in diameter and 3.0 cm long), which was attached to the left forearm with a velcro. Vibrotactile stimuli had an instantaneous rise and fall time, and a duration of 300 ms. In a previous study (Van Damme et al., 2004b), it was found that the vibrotactile stimulus is perceived as equally intense as, but significantly less threatening than the electrocutaneous stimulus used in the pain group. 2.4. Procedure Participants were tested individually in a sound-attenuated room designed for psychophysiological experiments. Preparation phase. In the pain group, participants were informed that an electrocutaneous stimulus would be used during the experiment. They were told that this stimulus ‘stimulates the pain fibres’ and that ‘most people find this kind of stimulation unpleasant’. Next, participants gave their informed consent. Then, the electrodes were attached to the left forearm. The skin at the electrode sites was first abraded with a peeling cream (Nihon Kohden) in order to reduce skin resistance. To familiarise the participants with the electrocutaneous stimuli, they were given a series of stimuli with increasing intensity (0.032, 0.16, 0.32, and 0.63 mA). The intensity of the final stimulus (0.63 mA) was used during the experiment with all participants. In the control group, participants were informed that a vibrotactile stimulus would be used, and that this stimulus ‘stimulates the touch fibres and that most people do not find this kind of stimulation unpleasant’. Next they were familiarised with the vibrotactile stimulus. Participants were seated in front of a computer, 60 cm from the screen, to perform
the exogenous cueing task. All instructions were presented on the computer screen. Practice phase. The experiment began with a practice phase without differential conditioning, and which consisted of 15 trials: eight validly cued trials, four invalidly cued trials, two catch trials, and one digit trial. Participants were made aware of the fact that no UCS would be presented during this phase. Baseline phase. A baseline phase was included in which 90 trials of the exogenous cueing task were presented without differential conditioning of the cues: 48 validly cued trials, 24 invalidly cued trials, 12 catch trials, and 6 digit trials. Participants were informed that no UCS would be presented. Acquisition phase. Participants were informed that one of the cues would be followed sometimes by an UCS, and that no UCS would follow the other cue. The acquisition phase consisted of 180 trials of the exogenous cueing task: 96 validly cued trials, 48 invalidly cued trials, 24 catch trials, and 12 digit trials. Half of the CSC presentations was followed by an UCS. In order to facilitate the differential conditioning, the acquisition phase started with two buffer trials in which the CSC was followed by the UCS. Manipulation check. After the acquisition phase, participants rated the extent to which they expected that an UCS would follow the CSC and the CSK and how fearful they were during the presentation of the CSC and the CSK, on 11-point numerical rating scales (anchored 0Znot at all and 10Zvery strongly). The painfulness of the UCS was assessed using a similar rating scale. Furthermore, participants rated the (un)pleasantness of the UCS on 11-point numerical rating scale (anchored K5Zvery unpleasant; C 5Zvery pleasant). Finally, participants completed the Dutch version of the Pain Catastrophizing Scale (PCS; Crombez et al., 1998a; Sullivan et al., 1995). This is a 13item scale that measures the level of catastrophic thinking about pain in both non-clinical and clinical populations. Participants were asked to reflect on past painful experiences and to indicate the degree to which they experienced each of the 13 thoughts or feelings during pain (e.g. ‘I become afraid that the pain may get worse’) on a five-point scale from 0 (not at all) to 4 (all the time). The Dutch version of the PCS has been shown to be valid and reliable (Van Damme et al., 2002a). In this study, we used the median of the PCS (medianZ18) to differentiate between high-pain catastrophizers (nZ25, MZ23.84, SDZ5.27, range 19–38) and low-pain catastrophizers (nZ27, MZ 12.56, SDZ4.26, range 4–18). Available norms show that the distribution between high- and low-pain catastrophizers in our sample is representative for the population of undergraduate students (Van Damme et al., 2000). 2.5. Statistical analyses A 2 (pain catastrophizing: high, low)!2 (group: pain, control)!2 (cue validity: valid, invalid)!2 (signal: CSC,
S. Van Damme et al. / Pain 111 (2004) 392–399
395
Table 1 Means and standard deviations of self-reports as a function of pain catastrophizing and experiment group Pain group
UCS aversiveness UCS painfulness CSC expectancy CSK expectancy CSC fear CSK fear PCS
Control group
Total group, NZ26
High PCS, NZ13
Low PCS, NZ13
Total group, NZ26
High PCS, NZ12
Low PCS, NZ14
K1.77 (1.61) 2.65 (2.31) 7.15 (2.60) 1.34 (2.00) 4.54 (3.31) 1.12 (1.97) 17.85 (7.05)
K2.08 (1.75) 3.00 (2.58) 7.92 (2.33) 0.92 (1.44) 5.23 (3.65) 0.46 (1.13) 23.00 (5.52)
K1.46 (1.45) 2.31 (2.06) 6.38 (2.72) 1.77 (2.42) 3.85 (2.91) 1.77 (2.42) 12.69 (3.92)
0.00 (1.79) 0.31 (1.12) 7.46 (1.50) 1.38 (2.26) 2.12 (2.45) 0.42 (1.06) 18.11 (7.87)
K0.25 (2.34) 0.67 (1.61) 7.08 (1.51) 2.25 (2.77) 3.00 (2.83) 0.58 (1.44) 24.75 (5.05)
0.21 (1.19) 0.00 (0.00) 7.79 (1.48) 0.64 (1.45) 1.36 (1.86) 0.29 (0.61) 12.43 (4.70)
CSK) ANOVA with repeated measures was performed upon the mean response times in the acquisition phase.1 The trials in which the CSC was followed by an UCS were not analysed. In these trials, response times could be affected by both the CSC and the UCS, because there was a temporal overlap between the presentation of the UCS and the response to the target. As we were interested in the pure effect of the CSC, we omitted these trials from the analyses. Because we were interested in comparing attentional processes between the pain group and the control group, and between high- versus low-pain catastrophizers, group and pain catastrophizing were included as between-subject factors. Throughout the paper, Greenhouse–Geisser corrections (with corrected degrees of freedom) were presented when the sphericity assumption was violated (Mauchly’s Test of Sphericity; P!0.05). As an estimate of effect size, percentage of variance (PV) was reported for the hypothesized effects. Following Cohen’s (1988) guidelines, effect sizes of 0.01, 0.10, and 0.25 were used as thresholds to define small, medium and large effects, respectively.
3. Results 3.1. Self-report data All self-report data are shown in Table 1. The experimental manipulation was successful. First, participants in the pain group rated the UCS as more aversive and painful than participants in the control group (aversiveness: F(1,51)Z14.07, P!0.001; painfulness: F(1,51)Z21.63, P!0.001). Second, in both the pain group and the control group, differential conditioning effects emerged as a result of the acquisition phase. Participants expected the UCS 1
For reasons of clarity, baseline RTs were not included in this ANOVA. As there were no UCS presentations in this phase, inclusion of these RTs analysis would result in very complex and useless interactions. A separate analysis of the baseline phase showed a significant main effect of cue validity, indicating that responses were faster to validly cued targets compared to invalidly cued targets. No other interpretable effects were found.
significantly more after the presentation of the CSC than after the presentation of the CSK (pain group: t(25)Z9.94, P!0.001; control group: t(25)Z11.13, P!0.001). Furthermore, participants reported more fear during the presentation of the CSC than during the presentation of the CSK (pain group: t(25)Z5.29, P!0.001; t(25)Z4.20, P! 0.001). The pain group had no stronger expectations than the control group that the CSC would be followed by the UCS (F!1). However, the pain group reported significantly more fear than the control group during the presentation of the CSC, F(1,51)Z8.98, P!0.01. Finally, it should be noted that there were no significant differences in PCS scores between the pain group and the control group, neither for the total groups, nor for high- and low-pain catastrophizers separately (all F!1) (see also Table 1). 3.2. Response time data After visual inspection, RTs smaller than 150 ms and larger than 750 ms were considered as outliers and omitted (1.80%). Trials with errors were also excluded from the RT analyses (2.00%). Two participants were excluded from the analyses. The first frequently responded to catch trials (66.67%), had a large number of outliers (41.67%), and made a large number of errors (16.67%). The second frequently responded to catch trials (45.83%). A 2 (pain catastrophizing: high, low)!2 (group: pain, control)!2 (cue validity: valid, invalid)!2 (signal: CSC, CSK) ANOVA with repeated measures was performed. Mean response times are shown in Table 2. We were interested in the cue validity!signal interaction, indicating effects of differential conditioning upon attentional engagement and disengagement. More particularly, we wanted to investigate the effect of the UCS type (pain versus control) upon this interaction. Surprisingly, the three-way interaction of group!cue validity!signal failed to reach significance, F(1,48)Z 0.27, MSEZ133.92, n.s. However, the four-way interaction effect of pain catastrophizing!group!cue validity!signal, proved to be significant, F(1,48)Z5.57, MSEZ133.92, P!0.05, indicating that the effect of group is dependent upon the level of pain catastrophizing. In order to disentangle this interaction, we used a three-step approach. First, we performed a 2 (group: pain, control)!2 (cue validity: valid,
396
S. Van Damme et al. / Pain 111 (2004) 392–399
Table 2 Mean RTs (and standard errors) as a function of cue validity, signal, catastrophic thinking, and experiment group Total group
VC IC CSC CSK VC CSC VC CSK IC CSC IC CSK
Pain group
Control group
High PCS
Low PCS
High PCS
Low PCS
High PCS
Low PCS
319 (8) 355 (8) 334 (7) 339 (8) 313 (8) 324 (8) 356 (7) 355 (9)
327 (7) 367 (7) 344 (7) 350 (8) 321 (8) 333 (8) 367 (7) 367 (8)
324 (11) 359 (10) 341 (10) 342 (11) 318 (11) 330 (11) 364 (10) 354 (12)
326 (11) 374 (10) 347 (10) 353 (11) 321 (11) 330 (11) 373 (10) 376 (12)
313 (11) 351 (11) 328 (10) 337 (12) 308 (11) 319 (11) 347 (10) 355 (12)
329 (10) 360 (10) 341 (9) 347 (11) 321 (11) 336 (10) 361 (9) 358 (11)
VCZvalidly cued; ICZinvalidly cued.
invalid)!2 (signal: CSC, CSK) ANOVA separately in high- and low-pain catastrophizers. Second, we calculated the effect of pain signals and vibrotactile signals on attentional engagement and disengagement. The engagement component relates to the validly cued trials. We tested whether the presentation of the CSC as a valid cue facilitated target detection compared to the presentation of the CSK. The disengagement components relates to the invalidly cued trials. We tested whether the presentation of the CSC as an invalid cue impaired target detection compared to the presentation of the CSK. Third, we compared attentional engagement and disengagement between the pain group and the control group. High-pain catastrophizers. All effects are illustrated in Table 2. There was a significant main effect of cue validity, F(1,23)Z158.04, MSEZ211.72, P!0.001. Responses were significantly faster to validly cued targets compared to invalidly cued targets. Also the main effect of signal was significant, F(1,23)Z5.14, MSEZ120.64, P!0.05. Responses were significantly faster to targets preceded by the CSC compared to targets preceded by the CSK. There were two significant two-way interactions (group!signal, F(1,23)Z4.30, MSEZ120.64, P!0.05; cue validity! signal, F(1,23)Z7.16, MSEZ132.98, P!0.05), but these were subsumed under the hypothesized three-way interaction of group!cue validity!signal, F(1,23)Z4.03, MSEZ132.98, PZ0.057. This interaction indicates a group-dependent modulation of spatial attention by the anticipation of somatosensory stimulation. More particularly, we expected that the effects of somatosensory signals on attentional engagement and disengagement would be more pronounced in the pain group than in the control group. In order to test this, we calculated the effects of signals of somatosensory stimulation on engagement (RT to validly cued CSK targets minus RT to validly cued CSC targets) and disengagement (RT to invalidly cued CSC targets minus RT to invalidly cued CSK targets), and compared these effects between the pain and control group, using t-tests. The effects are illustrated in Fig. 2. Attentional engagement was facilitated both by signals of impending pain (MZ11.23, SDZ19.37; t(12)Z2.09, PZ0.05) and by signals of impending vibrotactile stimulation (MZ11.08,
SDZ11.58; t(11)Z3.32, P!0.01). However, engagement was equally facilitated by somatosensory signals in the pain group and in the control group, t(23)!1 (PVZ0.00). Attentional disengagement was significantly retarded by signals of impending pain (MZ10.38, SDZ14.79; t(12)Z 2.53, P!0.05), but not by signals of impending vibrotactile stimulation (MZK8.00, SDZ16.68; t(11)Z1.66, PZ 0.125). Disengagement was significantly more retarded by somatosensory signals in the pain group than in the control group, t(23)Z2.92, P!0.01. This effect had a large effect size (PVZ0.27). Low-pain catastrophizers. All effects are illustrated in Table 2. There was a significant main effect of cue validity,
Fig. 2. Indices of attentional engagement (ENG) and disengagement (DISENG) as a function of experiment group and level of catastrophic thinking about pain. The engagement index was calculated by subtracting the mean RT to validly cued CSC targets from the mean RT to validly cued CSK targets. Positive difference scores indicate stronger attentional engagement to the CSC compared to the CSK. The disengagement index was calculated by subtracting the mean RT to invalidly cued CSKtargets from the mean RT to invalidly cued CSC targets. Positive difference scores indicate slower disengagement of attention from the CSC compared to the CSK.
S. Van Damme et al. / Pain 111 (2004) 392–399
F(1,25)Z186.47, MSEZ231.45, P!0.001. Responses were significantly faster to validly cued targets compared to invalidly cued targets. The significant group!cue validity interaction shows that this effect was stronger in the pain group compared with the control group. Also the main effect of signal was significant, F(1,25)Z4.65, MSEZ 216.61, P!0.05. Responses were significantly faster to targets preceded by the CSC compared to targets preceded by the CSK. There was a significant cue validity!signal interaction effect, F(1,25)Z6.81, MSEZ134.78, P!0.05. Decomposition of this effect shows that signals of impending somatosensory stimulation affect attentional engagement but not disengagement. Responses to validly cued targets were significantly faster in trials with the CSC as a cue compared to trials with the CSK as a cue, t(26)Z3.27, P!0.01, reflecting facilitated engagement. Responses to invalidly cued targets were not slower in trials with the CSC as a cue compared to trials with the CSK as a cue, t(26)!1, indicating no retarded disengagement. The threeway interaction of group!cue validity!signal failed to reach significance, F(1,25)Z1.74, MSEZ134.78, n.s., indicating that the effects of signals of somatosensory stimulation on spatial cueing were not group-dependent. In order to test this, we compared the effects of signals of somatosensory stimulation on attentional engagement and disengagement between the pain and control group, using a similar procedure as in high-pain catastrophizers. The effects are illustrated in Fig. 2. Attentional engagement was significantly facilitated by signals of impending vibrotactile stimulation (MZ14.43, SDZ17.78; t(13)Z 3.04, PZ0.01), but not by signals of impending pain (MZ9.46, SDZ20.94; t(12)Z1.63, PZ0.129). However, the difference between the pain group and the control group was not significant, t(25)!1 (PVZ0.02). Attentional disengagement was not retarded by signals of impending pain (MZK3.69, SDZ20.41; t(12)!1), and by signals of impending vibrotactile stimulation (MZ3.14, SDZ15.69; t(13)!1). Furthermore, there was no difference between the pain group and the control group, t(25)!1 (PVZ0.04).
4. Discussion The main objective of this study was to investigate whether pain anticipation modulates spatial attention, and to examine the pain-specificity of this modulation. The results partially supported the hypotheses, and can be readily summarised: spatial attention was more strongly modulated by the anticipation of pain than by the anticipation of vibrotactile stimulation. However, this was only the case for participants high in catastrophic thinking about pain. First, there were no differences in attentional engagement when comparing signals predicting pain with signals predicting vibrotactile stimulation. Second, disengagement was more retarded by signals predicting pain than by signals predicting vibrotactile stimulation for high-pain
397
catastrophizers. This effect was not found for low-pain catastrophizers. We will now discuss the findings more in detail. The experimental manipulation proved to be successful. In line with Van Damme et al. (2004b), somatosensory stimulation was rated as more aversive and painful in the pain group than in the control group. Furthermore, selfreported expectation and fear indicated conditioning effects in both groups. There was no difference in the expectation of somatosensory stimulation after the conditioned signal between the pain group and the control group. However, in the pain group, more fear was reported during the conditioned signal than in the control group. Finally, reaction time data showed an overall faster detection of targets preceded by valid cues compared to targets preceded by invalid cues. This strong cue validity effect is in line with classic spatial cueing studies (Posner et al., 1987). Our results replicate and extend the findings of previous studies. First, in line with Van Damme et al. (2004c), we found that the anticipation of pain facilitated attentional engagement. However, this effect was not pain-specific, because engagement was equally facilitated by signals of impending pain as by signals predicting vibrotactile stimulation. This finding is in line with a recent crossmodal cueing study by Van Damme et al. (2004b), who found that the detection of pain stimuli and vibrotactile stimuli was equally facilitated by a cue correctly predicting the sensory modality of the stimuli. No differences were found between high- and low-pain catastrophizers. Second, we found that the anticipation of pain retarded disengagement of attention in high-pain catastrophizers. This effect proved to be painspecific, because disengagement was significantly more retarded by signals of impending pain than by signals predicting vibrotactile stimulation. This finding is in line with the crossmodal cueing study of Van Damme et al. (2004b), in which the detection of auditory stimuli was more retarded by cues predicting pain than by cues predicting vibrotactile stimulation. Our findings have a number of theoretical implications. Facilitated engagement by the anticipation of pain is not a pain-specific process. This indicates that there is no threatrelated bias in the initial processing of signals predicting somatosensory stimulation. This suggests that the amount of attention initially allocated to signals is probably determined by their salience and not by their threat value. In contrast to this, retarded disengagement by the anticipation of pain is a pain-specific process. This indicates that the amount of attention held on signals is determined by their threat value. Once a signal is perceived as threatening, attention is kept on the threat location, in order to allow the initiation of protective behaviour and escape strategies. When the signal is tagged as non-threatening, further processing of the signal is inhibited, and attention is directed to other relevant locations. Furthermore, this would explain the differential effects of high- and low-pain catastrophizers. High-pain catastrophizers probably perceive the signal of
398
S. Van Damme et al. / Pain 111 (2004) 392–399
impending pain as highly threatening. As pain responses of high catastrophizers are activated very swiftly (Sullivan et al., 2001), these participants will worry more about, and be more fearful of, anticipated pain compared with low-pain catastrophizers (Aldrich et al., 2000). Consequently, they will process the signal predicting pain thoroughly, and hold attention to its location. In contrast, low-pain catastrophizers, for whom the anticipated pain is not threatening, will inhibit further processing of the signal, and direct attention to other relevant locations. Our findings may have some clinical relevance. Most patients suffering from chronic pain will perceive their pain and signals of impending pain as highly threatening. Consequently, the initial interruption by and processing of such pain signals should be considered a normal process. However, based on our findings, it could be expected that these patients will have pronounced difficulties disengaging attention from the anticipated pain location and directing it to other environmental stimuli. Such fixation of attention by the anticipation of pain might become dysfunctional, resulting in hypervigilance, fear, worry, and avoidance behaviour (Aldrich et al., 2000; Crombez et al., 1998b; Peters et al., 2002). However, our ideas about the relation between these concepts and difficulty disengaging are speculative, and require further investigation with patient populations. This study has a number of limitations, which require further consideration. First, this study was conducted with pain-free volunteers, using experimental pain stimuli. Therefore, one must be cautious in generalizing the results to chronic pain patients. Our findings need further confirmation, both in non-clinical and clinical populations. Second, the majority of the participants in our study were female, limiting the generalizability of the results. Third, although we were able to demonstrate that the pattern of results for high-pain catastrophizers is robust (Van Damme et al., 2002b, 2004a), our conclusions were drawn from relatively small sample sizes. Therefore, it is possible that the statistical power in our study was low, resulting in the detection of large effects but leaving undetected differences with small effect size. In future research, work with a priori selected high- and low-pain catastrophizers could be considered, thereby requiring smaller sample sizes. In sum, evidence was found that spatial attention is modulated by the anticipation of pain. Furthermore, only the modulation of attentional disengagement and not engagement proved to be pain-specific. Finally, painspecificity was only found in individuals who perceived pain as highly threatening. One challenge for future research might be the combination of behavioural and neurophysiological measures, to foster our understanding of the differential representation of attentional engagement and disengagement in the brain (Legrain et al., 2002; Lorenz et al., 2003).
Acknowledgements This study was supported by a research grant (G.0107.00) to Geert Crombez by the Fund for Scientific Research Flanders. The authors would like to thank Charlotte Nollet for her help with the data collection, and Ernst Koster for his always inspiring comments.
References Aldrich S, Eccleston C, Crombez G. Worrying about chronic pain: vigilance to threat and misdirected problem solving. Behav Res Ther 2000;38:457–70. Cohen J. Statistical power analysis for the behavioural sciences. San Diego, CA: McGraw-Hill; 1988. Crombez G, Eccleston C, Baeyens F, Eelen P. When somatic information threatens, catastrophic thinking about pain enhances attentional interference. Pain 1998a;75:187–98. Crombez G, Vervaet L, Lysens R, Baeyens F, Eelen P. Avoidance and confrontation of painful back straining movements in chronic back pain patients. Behav Modif 1998b;22:62–77. De Clercq A, Crombez G, Roeyers H, Buysse A. A simple and sensitive method to measure timing accuracy. Behav Res Methods Instrum Comput 2003;35:109–15. Eccleston C, Crombez G. Pain demands attention: a cognitive-affective model on the interruptive function of pain. Psychol Bull 1999;125: 356–66. Honore´ J, He´non H, Naveteur J. Influence of eye orientation on pain as a function of anxiety. Pain 1995;63:213–8. Jensen J, McIntosh AR, Crawley AP, Mikulis DJ, Remington G, Kapur S. Direct activation of the ventral striatum in anticipation of aversive stimuli. Neuron 2003;40:1251–7. Legrain V, Gue´rit J, Bruyer R, Plaghki L. Attentional modulation of the nociceptive processing into the human brain: selective spatial attention, probability of stimulus occurrence, and target detection effects on laser evoked potentials. Pain 2002;99:21–39. Lorenz J, Minoshima S, Casey KL. Keeping pain out of mind: the role of the dorsolateral prefrontal cortex in pain modulation. Brain 2003;126: 1079–91. Peters ML, Vlaeyen JWS, Kunnen AMW. Is pain-related fear a predictor of somatosensory hypervigilance in chronic low back pain patients? Behav Res Ther 2002;40:85–103. Pincus T, Morley S. Cognitive processing bias in chronic pain: a review and integration. Psychol Bull 2001;127:599–617. Ploghaus A, Tracey I, Gati JS, Clare S, Menon RS, Mathews PM, Rawlins JNP. Dissociating pain from its anticipation in the human brain. Science 1999;284:1979–81. Ploghaus A, Becerra L, Borras C, Borsook D. Neural circuitry underlying pain modulation: expectation, hypnosis, placebo. Trends Cogn Sci 2003;7:197–200. Porro CA, Baraldi P, Pagnoni G, Serafini M, Facchin P, Maieron M, Nichelli P. Does anticipation of pain affect cortical nociceptive systems? J Neurosci 2002;22:3206–14. Porro CA, Cettolo V, Francescato MP, Baraldi P. Functional activity mapping of the mesial hemispheric wall during anticipation of pain. NeuroImage 2003;19:1738–47. Posner MI, Inhoff A, Friedrich FJ, Cohen A. Isolating attentional systems: a cognitive-anatomical analysis. Psychobiology 1987;15:107–21. Spence C, Nicholls MER, Driver J. The costs of expecting events in the wrong sensory modality. Percept Psychophys 2001;63:330–6. Spence C, Bentley DE, Phillips N, McGlone FP, Jones AKP. Selective attention to pain: a psychophysical investigation. Exp Brain Res 2002; 145:395–402.
S. Van Damme et al. / Pain 111 (2004) 392–399 Sullivan MJL, Bishop SR, Pivik J. The Pain Catastrophizing Scale: development and validation. Psychol Assess 1995;7:524–32. Sullivan MJL, Thorn B, Haythornthwaite JA, Keefe F, Martin M, Bradley LA, Lefebvre JC. Theoretical perspectives on the relation between catastrophizing and pain. Clin J Pain 2001;17: 52–64. Van Damme S, Crombez G, Vlaeyen JWS, Goubert L, Van den Broeck A, Van Houdenhove B. De Pain Catastrophizing Scale: psychometrische karakteristieken en normering [The Pain Catastrophizing Scale: psychometric characteristics and norms]. Gedragstherapie 2000;33: 209–22. Van Damme S, Crombez G, Bijttebier P, Goubert L, Van Houdenhove B. A confirmatory factor analysis of the Pain Catastrophizing Scale: invariant factor structure across clinical and non-clinical populations. Pain 2002a;96:319–24.
399
Van Damme S, Crombez G, Eccleston C. Retarded disengagement from pain cues: the effects of pain catastrophizing and pain expectancy. Pain 2002b;100:111–8. Van Damme S, Crombez G, Eccleston C. Disengagement from pain: the role of catastrophic thinking about pain. Pain 2004a;107:70–6. Van Damme S, Crombez G, Eccleston C, Goubert L. Impaired disengagement from threatening cues of impending pain in a crossmodal cueing paradigm. Eur J Pain 2004b;8:227–36. Van Damme S, Lorenz J, Eccleston C, Koster EHW, De Clercq A, Crombez G. Fear-conditioned cues of impending pain facilitate attentional engagement in an emotional modification of the exogenous cueing paradigm. Neurophysiol Clin 2004c;34:33–9. Vanderiet K, Adriaensen H, Carton H, Vertommen H. The McGill Pain Questionnaire constructed for the Dutch language (MPQ-DV): preliminary data concerning reliability and validity. Pain 1987;30:395–408.
The anticipation of pain modulates spatial attention: evidence for pain-specificity in high-pain catastrophizers Stefaan Van Dammea,*, Geert Crombeza, Christopher Ecclestonb a
Department of Experimental, Clinical and Health Psychology, Ghent University, Henri Dunantlaan 2, 9000 Ghent, Belgium b University of Bath, Bath BA2 7AY, UK Received 4 May 2004; received in revised form 7 July 2004; accepted 19 July 2004
Abstract Recent studies have suggested that the anticipation of pain may modulate spatial attention. However, it is possible that this modulation reflects a general effect of anticipating somatosensory stimulation, without being pain-specific. In the present study, we therefore compared the effect of the anticipation of somatosensory stimulation on spatial attention between two groups, using conditioned signals in a spatial cueing paradigm. In the pain group, signals predicted painful electrocutaneous stimulation, whereas in the control group, signals predicted non-painful vibrotactile stimulation. Tests between both groups showed that attentional engagement was equally facilitated by the anticipation of somatosensory stimulation in both groups. Interestingly, disengagement of attention was more retarded by the anticipation of pain than by the anticipation of non-painful vibrotactile stimulation in participants high in catastrophic thinking about pain. Theoretical and clinical implications of these findings are discussed. q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. Keywords: Spatial attention; Experimental pain; Anticipation; Catastrophizing
1. Introduction The attentional demand of pain has been documented by an impressive number of studies in clinical and nonclinical populations (Eccleston and Crombez, 1999; Pincus and Morley, 2001). Furthermore, it has been demonstrated that attention to pain is intensified by the threat value of pain, and particularly by catastrophic thinking about pain (Crombez et al., 1998a; Van Damme et al., 2004a). Recently, research interest has also been moving to the attentional effects of pain anticipation. The anticipation of pain may have an important protective function, allowing the avoidance of bodily harm by the initiation of adaptive behaviour. In a number of studies, behavioural evidence was found for the attentional demand of pain anticipation, particularly in high-pain catastrophizers (Spence et al., * Corresponding author. Tel.: C32-9-2649-105; fax: C32-9-2649-149. E-mail address: [email protected] (S. Van Damme).
2002; Van Damme et al., 2002b). Furthermore, neurophysiological evidence has indicated that the anticipation of pain may trigger cortical systems closely involved in the experience of pain itself (Ploghaus et al., 1999; Porro et al., 2002), and several studies have begun now to map the cortical representation of pain anticipation (Jensen et al., 2003; Ploghaus et al., 2003; Porro et al., 2003). One particular way in which pain can be anticipated is the allocation of attention to the expected location of threat. Indirect evidence for this view was provided by Honore´ et al. (1995), who found that sensitivity to experimental pain stimuli was extended when participants directed their eyes to the stimulated location. However, no studies have investigated whether pain anticipation modulates spatial attention. In order to examine this, we adapted a spatial cueing paradigm (Posner et al., 1987; Van Damme et al., 2004c). In this paradigm, participants detect visual targets, preceded by ipsilateral (valid) and contralateral (invalid) cues. Valid cues typically lead to response time benefits (due to attentional engagement at the validly cued
0304-3959/$20.00 q 2004 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2004.07.022
S. Van Damme et al. / Pain 111 (2004) 392–399
location), whereas invalid cues lead to response time costs (due to disengagement from the invalidly cued location). In order to investigate the effect of pain anticipation on spatial attention, we used conditioned signals of impending pain as spatial cues. In a recent study with this paradigm, we demonstrated that signals of impending pain enhanced spatial cueing (Van Damme et al., 2004c). However, this finding can be alternatively explained, possibly reflecting a pain-unspecific modulation of spatial attention by the anticipation of somatosensory stimulation in general (Spence et al., 2001). In the present study, we therefore compared the effects of somatosensory anticipation on spatial attention between two groups. In the pain group, we used signals of impending pain as spatial cues. In the control group, we used signals predicting non-painful vibrotactile stimulation as spatial cues. Three hypotheses were tested. First, we expected that attentional engagement would be more facilitated in the pain group compared to the control group. Second, we expected that disengagement would be more retarded in the pain group than in the control group. Third, we expected that the hypothesized effects would be more pronounced in participants high in catastrophic thinking about pain.
2. Methods 2.1. Participants Fifty-four undergraduate psychology students (8 males and 46 females; mean ageZ18.22 years, SDZ0.74; range 17–21 years) from Ghent University participated in order to fulfil course requirements. All participants gave informed consent and were free to terminate the experiment at any time. Participants were randomly assigned to the pain group (nZ28; 4 men and 24 women; ageZ18.36 years) and to the control group (nZ26; 4 men and 22 women; ageZ18.08 years). Each person had normal or corrected-to-normal eyesight. Experimental duration was approximately 30 min. Two participants from the pain group were excluded from the analyses (see later). 2.2. Exogenous cueing task The exogenous cueing task was programmed and presented by the INQUISIT Millisecond software package (Inquisit 1.32, 2001) on a S710 Compaq Deskpro computer with a 72 Hz, 17-inch colour monitor. INQUISIT measures response times with millisecond accuracy (De Clercq et al., 2003). Target stimuli consisted of black squares (1.1!1.1 cm width), presented on a white background. Two coloured frames (green and pink; 4.8 cm high!6.5 cm wide) served as exogenous cues. Each trial began with a fixation cross in the middle of the screen (duration of
393
Fig. 1. Schematic illustration of validly and invalidly cued trials. On each trial, a fixation cross was presented for 1000 ms. This was immediately followed by the presentation of a cue (200 ms duration). Immediately after cue offset, the target was introduced at an SOA of 200 ms. Targets were presented at the same location of cues (validly cued trials), or at the opposite location of cues (invalidly cued trials).
1000 ms). The cue was presented 9.28 from a fixation cross for a duration of 200 ms. Targets followed cues at a Stimulus Onset Asynchrony (SOA: time between cue onset and target onset) of 200 ms. Cue location correctly predicted target location on two-thirds of the test trials (validly cued trials). On the remaining test trials, cue location incorrectly predicted target location (invalidly cued trials). These trial types are illustrated in Fig. 1. Participants were instructed to respond to the target as quickly as possible without sacrificing accuracy, by pressing the corresponding key on a standard AZERTY computer keyboard (left index finger at ‘q’ key for left targets; right index finger at ‘5’ key for right targets). A trial ended when a participant responded or 2000 ms had elapsed. In order to control for anticipatory responses (responses to cues instead of targets), a number of catch trials were presented, in which the cue was not followed by a target. Furthermore, in order to ensure that participants maintained gaze at the middle of the screen, a number of digit trials were presented. On these trials, the fixation cross was followed only by a randomly selected digit between 1 and 9 for a duration of 50 ms. Participants were instructed to report the digit aloud. If participants were not able to report the digits (correctly), this indicated that they did not maintain gaze at the fixation cross. 2.3. Experimental manipulation Cues were differentially conditioned using two colours. In half of the presentations, the conditioned cue (CSC) was 300 ms after offset followed by a somatosensory stimulus
394
S. Van Damme et al. / Pain 111 (2004) 392–399
(UCS), whereas the other cue (CSK) was never followed by the UCS. The colours of the CSC and CSK were counterbalanced across participants. The CSC and CSK were presented equally often, in a fixed random order with a maximum of three consecutive presentations of the same cue. In the pain group, the UCS was an aversive electrocutaneous stimulus, delivered by an AC stimulator with an internal frequency of 50 Hz. Electrocutaneous stimuli were delivered at the external side of the left forearm by two lubricated Fukuda standard Ag/AgCl electrodes (1 cm diameter). Intensity of the electrocutaneous stimulus was 0.63 mA, with an instantaneous rise and fall time, and a duration of 300 ms. Using the sensory pain words of the Dutch McGill Pain Questionnaire (Vanderiet et al., 1987), it was found that the characteristics of this stimulus were best described as pricking, boring, flickering, electric, and cutting. Previous studies have further demonstrated that an electrocutaneous stimulus of this intensity is painful but easy to tolerate, and that the attentional demand of this stimulus is strongly related to the specific threat value of pain (Crombez et al., 1998a). In the control group, the UCS was a non-painful vibrotactile stimulus, delivered by a vibration element. This element consisted of a Nokia 3210 vibramotor, enveloped by a plastic cylinder (1.3 cm in diameter and 3.0 cm long), which was attached to the left forearm with a velcro. Vibrotactile stimuli had an instantaneous rise and fall time, and a duration of 300 ms. In a previous study (Van Damme et al., 2004b), it was found that the vibrotactile stimulus is perceived as equally intense as, but significantly less threatening than the electrocutaneous stimulus used in the pain group. 2.4. Procedure Participants were tested individually in a sound-attenuated room designed for psychophysiological experiments. Preparation phase. In the pain group, participants were informed that an electrocutaneous stimulus would be used during the experiment. They were told that this stimulus ‘stimulates the pain fibres’ and that ‘most people find this kind of stimulation unpleasant’. Next, participants gave their informed consent. Then, the electrodes were attached to the left forearm. The skin at the electrode sites was first abraded with a peeling cream (Nihon Kohden) in order to reduce skin resistance. To familiarise the participants with the electrocutaneous stimuli, they were given a series of stimuli with increasing intensity (0.032, 0.16, 0.32, and 0.63 mA). The intensity of the final stimulus (0.63 mA) was used during the experiment with all participants. In the control group, participants were informed that a vibrotactile stimulus would be used, and that this stimulus ‘stimulates the touch fibres and that most people do not find this kind of stimulation unpleasant’. Next they were familiarised with the vibrotactile stimulus. Participants were seated in front of a computer, 60 cm from the screen, to perform
the exogenous cueing task. All instructions were presented on the computer screen. Practice phase. The experiment began with a practice phase without differential conditioning, and which consisted of 15 trials: eight validly cued trials, four invalidly cued trials, two catch trials, and one digit trial. Participants were made aware of the fact that no UCS would be presented during this phase. Baseline phase. A baseline phase was included in which 90 trials of the exogenous cueing task were presented without differential conditioning of the cues: 48 validly cued trials, 24 invalidly cued trials, 12 catch trials, and 6 digit trials. Participants were informed that no UCS would be presented. Acquisition phase. Participants were informed that one of the cues would be followed sometimes by an UCS, and that no UCS would follow the other cue. The acquisition phase consisted of 180 trials of the exogenous cueing task: 96 validly cued trials, 48 invalidly cued trials, 24 catch trials, and 12 digit trials. Half of the CSC presentations was followed by an UCS. In order to facilitate the differential conditioning, the acquisition phase started with two buffer trials in which the CSC was followed by the UCS. Manipulation check. After the acquisition phase, participants rated the extent to which they expected that an UCS would follow the CSC and the CSK and how fearful they were during the presentation of the CSC and the CSK, on 11-point numerical rating scales (anchored 0Znot at all and 10Zvery strongly). The painfulness of the UCS was assessed using a similar rating scale. Furthermore, participants rated the (un)pleasantness of the UCS on 11-point numerical rating scale (anchored K5Zvery unpleasant; C 5Zvery pleasant). Finally, participants completed the Dutch version of the Pain Catastrophizing Scale (PCS; Crombez et al., 1998a; Sullivan et al., 1995). This is a 13item scale that measures the level of catastrophic thinking about pain in both non-clinical and clinical populations. Participants were asked to reflect on past painful experiences and to indicate the degree to which they experienced each of the 13 thoughts or feelings during pain (e.g. ‘I become afraid that the pain may get worse’) on a five-point scale from 0 (not at all) to 4 (all the time). The Dutch version of the PCS has been shown to be valid and reliable (Van Damme et al., 2002a). In this study, we used the median of the PCS (medianZ18) to differentiate between high-pain catastrophizers (nZ25, MZ23.84, SDZ5.27, range 19–38) and low-pain catastrophizers (nZ27, MZ 12.56, SDZ4.26, range 4–18). Available norms show that the distribution between high- and low-pain catastrophizers in our sample is representative for the population of undergraduate students (Van Damme et al., 2000). 2.5. Statistical analyses A 2 (pain catastrophizing: high, low)!2 (group: pain, control)!2 (cue validity: valid, invalid)!2 (signal: CSC,
S. Van Damme et al. / Pain 111 (2004) 392–399
395
Table 1 Means and standard deviations of self-reports as a function of pain catastrophizing and experiment group Pain group
UCS aversiveness UCS painfulness CSC expectancy CSK expectancy CSC fear CSK fear PCS
Control group
Total group, NZ26
High PCS, NZ13
Low PCS, NZ13
Total group, NZ26
High PCS, NZ12
Low PCS, NZ14
K1.77 (1.61) 2.65 (2.31) 7.15 (2.60) 1.34 (2.00) 4.54 (3.31) 1.12 (1.97) 17.85 (7.05)
K2.08 (1.75) 3.00 (2.58) 7.92 (2.33) 0.92 (1.44) 5.23 (3.65) 0.46 (1.13) 23.00 (5.52)
K1.46 (1.45) 2.31 (2.06) 6.38 (2.72) 1.77 (2.42) 3.85 (2.91) 1.77 (2.42) 12.69 (3.92)
0.00 (1.79) 0.31 (1.12) 7.46 (1.50) 1.38 (2.26) 2.12 (2.45) 0.42 (1.06) 18.11 (7.87)
K0.25 (2.34) 0.67 (1.61) 7.08 (1.51) 2.25 (2.77) 3.00 (2.83) 0.58 (1.44) 24.75 (5.05)
0.21 (1.19) 0.00 (0.00) 7.79 (1.48) 0.64 (1.45) 1.36 (1.86) 0.29 (0.61) 12.43 (4.70)
CSK) ANOVA with repeated measures was performed upon the mean response times in the acquisition phase.1 The trials in which the CSC was followed by an UCS were not analysed. In these trials, response times could be affected by both the CSC and the UCS, because there was a temporal overlap between the presentation of the UCS and the response to the target. As we were interested in the pure effect of the CSC, we omitted these trials from the analyses. Because we were interested in comparing attentional processes between the pain group and the control group, and between high- versus low-pain catastrophizers, group and pain catastrophizing were included as between-subject factors. Throughout the paper, Greenhouse–Geisser corrections (with corrected degrees of freedom) were presented when the sphericity assumption was violated (Mauchly’s Test of Sphericity; P!0.05). As an estimate of effect size, percentage of variance (PV) was reported for the hypothesized effects. Following Cohen’s (1988) guidelines, effect sizes of 0.01, 0.10, and 0.25 were used as thresholds to define small, medium and large effects, respectively.
3. Results 3.1. Self-report data All self-report data are shown in Table 1. The experimental manipulation was successful. First, participants in the pain group rated the UCS as more aversive and painful than participants in the control group (aversiveness: F(1,51)Z14.07, P!0.001; painfulness: F(1,51)Z21.63, P!0.001). Second, in both the pain group and the control group, differential conditioning effects emerged as a result of the acquisition phase. Participants expected the UCS 1
For reasons of clarity, baseline RTs were not included in this ANOVA. As there were no UCS presentations in this phase, inclusion of these RTs analysis would result in very complex and useless interactions. A separate analysis of the baseline phase showed a significant main effect of cue validity, indicating that responses were faster to validly cued targets compared to invalidly cued targets. No other interpretable effects were found.
significantly more after the presentation of the CSC than after the presentation of the CSK (pain group: t(25)Z9.94, P!0.001; control group: t(25)Z11.13, P!0.001). Furthermore, participants reported more fear during the presentation of the CSC than during the presentation of the CSK (pain group: t(25)Z5.29, P!0.001; t(25)Z4.20, P! 0.001). The pain group had no stronger expectations than the control group that the CSC would be followed by the UCS (F!1). However, the pain group reported significantly more fear than the control group during the presentation of the CSC, F(1,51)Z8.98, P!0.01. Finally, it should be noted that there were no significant differences in PCS scores between the pain group and the control group, neither for the total groups, nor for high- and low-pain catastrophizers separately (all F!1) (see also Table 1). 3.2. Response time data After visual inspection, RTs smaller than 150 ms and larger than 750 ms were considered as outliers and omitted (1.80%). Trials with errors were also excluded from the RT analyses (2.00%). Two participants were excluded from the analyses. The first frequently responded to catch trials (66.67%), had a large number of outliers (41.67%), and made a large number of errors (16.67%). The second frequently responded to catch trials (45.83%). A 2 (pain catastrophizing: high, low)!2 (group: pain, control)!2 (cue validity: valid, invalid)!2 (signal: CSC, CSK) ANOVA with repeated measures was performed. Mean response times are shown in Table 2. We were interested in the cue validity!signal interaction, indicating effects of differential conditioning upon attentional engagement and disengagement. More particularly, we wanted to investigate the effect of the UCS type (pain versus control) upon this interaction. Surprisingly, the three-way interaction of group!cue validity!signal failed to reach significance, F(1,48)Z 0.27, MSEZ133.92, n.s. However, the four-way interaction effect of pain catastrophizing!group!cue validity!signal, proved to be significant, F(1,48)Z5.57, MSEZ133.92, P!0.05, indicating that the effect of group is dependent upon the level of pain catastrophizing. In order to disentangle this interaction, we used a three-step approach. First, we performed a 2 (group: pain, control)!2 (cue validity: valid,
396
S. Van Damme et al. / Pain 111 (2004) 392–399
Table 2 Mean RTs (and standard errors) as a function of cue validity, signal, catastrophic thinking, and experiment group Total group
VC IC CSC CSK VC CSC VC CSK IC CSC IC CSK
Pain group
Control group
High PCS
Low PCS
High PCS
Low PCS
High PCS
Low PCS
319 (8) 355 (8) 334 (7) 339 (8) 313 (8) 324 (8) 356 (7) 355 (9)
327 (7) 367 (7) 344 (7) 350 (8) 321 (8) 333 (8) 367 (7) 367 (8)
324 (11) 359 (10) 341 (10) 342 (11) 318 (11) 330 (11) 364 (10) 354 (12)
326 (11) 374 (10) 347 (10) 353 (11) 321 (11) 330 (11) 373 (10) 376 (12)
313 (11) 351 (11) 328 (10) 337 (12) 308 (11) 319 (11) 347 (10) 355 (12)
329 (10) 360 (10) 341 (9) 347 (11) 321 (11) 336 (10) 361 (9) 358 (11)
VCZvalidly cued; ICZinvalidly cued.
invalid)!2 (signal: CSC, CSK) ANOVA separately in high- and low-pain catastrophizers. Second, we calculated the effect of pain signals and vibrotactile signals on attentional engagement and disengagement. The engagement component relates to the validly cued trials. We tested whether the presentation of the CSC as a valid cue facilitated target detection compared to the presentation of the CSK. The disengagement components relates to the invalidly cued trials. We tested whether the presentation of the CSC as an invalid cue impaired target detection compared to the presentation of the CSK. Third, we compared attentional engagement and disengagement between the pain group and the control group. High-pain catastrophizers. All effects are illustrated in Table 2. There was a significant main effect of cue validity, F(1,23)Z158.04, MSEZ211.72, P!0.001. Responses were significantly faster to validly cued targets compared to invalidly cued targets. Also the main effect of signal was significant, F(1,23)Z5.14, MSEZ120.64, P!0.05. Responses were significantly faster to targets preceded by the CSC compared to targets preceded by the CSK. There were two significant two-way interactions (group!signal, F(1,23)Z4.30, MSEZ120.64, P!0.05; cue validity! signal, F(1,23)Z7.16, MSEZ132.98, P!0.05), but these were subsumed under the hypothesized three-way interaction of group!cue validity!signal, F(1,23)Z4.03, MSEZ132.98, PZ0.057. This interaction indicates a group-dependent modulation of spatial attention by the anticipation of somatosensory stimulation. More particularly, we expected that the effects of somatosensory signals on attentional engagement and disengagement would be more pronounced in the pain group than in the control group. In order to test this, we calculated the effects of signals of somatosensory stimulation on engagement (RT to validly cued CSK targets minus RT to validly cued CSC targets) and disengagement (RT to invalidly cued CSC targets minus RT to invalidly cued CSK targets), and compared these effects between the pain and control group, using t-tests. The effects are illustrated in Fig. 2. Attentional engagement was facilitated both by signals of impending pain (MZ11.23, SDZ19.37; t(12)Z2.09, PZ0.05) and by signals of impending vibrotactile stimulation (MZ11.08,
SDZ11.58; t(11)Z3.32, P!0.01). However, engagement was equally facilitated by somatosensory signals in the pain group and in the control group, t(23)!1 (PVZ0.00). Attentional disengagement was significantly retarded by signals of impending pain (MZ10.38, SDZ14.79; t(12)Z 2.53, P!0.05), but not by signals of impending vibrotactile stimulation (MZK8.00, SDZ16.68; t(11)Z1.66, PZ 0.125). Disengagement was significantly more retarded by somatosensory signals in the pain group than in the control group, t(23)Z2.92, P!0.01. This effect had a large effect size (PVZ0.27). Low-pain catastrophizers. All effects are illustrated in Table 2. There was a significant main effect of cue validity,
Fig. 2. Indices of attentional engagement (ENG) and disengagement (DISENG) as a function of experiment group and level of catastrophic thinking about pain. The engagement index was calculated by subtracting the mean RT to validly cued CSC targets from the mean RT to validly cued CSK targets. Positive difference scores indicate stronger attentional engagement to the CSC compared to the CSK. The disengagement index was calculated by subtracting the mean RT to invalidly cued CSKtargets from the mean RT to invalidly cued CSC targets. Positive difference scores indicate slower disengagement of attention from the CSC compared to the CSK.
S. Van Damme et al. / Pain 111 (2004) 392–399
F(1,25)Z186.47, MSEZ231.45, P!0.001. Responses were significantly faster to validly cued targets compared to invalidly cued targets. The significant group!cue validity interaction shows that this effect was stronger in the pain group compared with the control group. Also the main effect of signal was significant, F(1,25)Z4.65, MSEZ 216.61, P!0.05. Responses were significantly faster to targets preceded by the CSC compared to targets preceded by the CSK. There was a significant cue validity!signal interaction effect, F(1,25)Z6.81, MSEZ134.78, P!0.05. Decomposition of this effect shows that signals of impending somatosensory stimulation affect attentional engagement but not disengagement. Responses to validly cued targets were significantly faster in trials with the CSC as a cue compared to trials with the CSK as a cue, t(26)Z3.27, P!0.01, reflecting facilitated engagement. Responses to invalidly cued targets were not slower in trials with the CSC as a cue compared to trials with the CSK as a cue, t(26)!1, indicating no retarded disengagement. The threeway interaction of group!cue validity!signal failed to reach significance, F(1,25)Z1.74, MSEZ134.78, n.s., indicating that the effects of signals of somatosensory stimulation on spatial cueing were not group-dependent. In order to test this, we compared the effects of signals of somatosensory stimulation on attentional engagement and disengagement between the pain and control group, using a similar procedure as in high-pain catastrophizers. The effects are illustrated in Fig. 2. Attentional engagement was significantly facilitated by signals of impending vibrotactile stimulation (MZ14.43, SDZ17.78; t(13)Z 3.04, PZ0.01), but not by signals of impending pain (MZ9.46, SDZ20.94; t(12)Z1.63, PZ0.129). However, the difference between the pain group and the control group was not significant, t(25)!1 (PVZ0.02). Attentional disengagement was not retarded by signals of impending pain (MZK3.69, SDZ20.41; t(12)!1), and by signals of impending vibrotactile stimulation (MZ3.14, SDZ15.69; t(13)!1). Furthermore, there was no difference between the pain group and the control group, t(25)!1 (PVZ0.04).
4. Discussion The main objective of this study was to investigate whether pain anticipation modulates spatial attention, and to examine the pain-specificity of this modulation. The results partially supported the hypotheses, and can be readily summarised: spatial attention was more strongly modulated by the anticipation of pain than by the anticipation of vibrotactile stimulation. However, this was only the case for participants high in catastrophic thinking about pain. First, there were no differences in attentional engagement when comparing signals predicting pain with signals predicting vibrotactile stimulation. Second, disengagement was more retarded by signals predicting pain than by signals predicting vibrotactile stimulation for high-pain
397
catastrophizers. This effect was not found for low-pain catastrophizers. We will now discuss the findings more in detail. The experimental manipulation proved to be successful. In line with Van Damme et al. (2004b), somatosensory stimulation was rated as more aversive and painful in the pain group than in the control group. Furthermore, selfreported expectation and fear indicated conditioning effects in both groups. There was no difference in the expectation of somatosensory stimulation after the conditioned signal between the pain group and the control group. However, in the pain group, more fear was reported during the conditioned signal than in the control group. Finally, reaction time data showed an overall faster detection of targets preceded by valid cues compared to targets preceded by invalid cues. This strong cue validity effect is in line with classic spatial cueing studies (Posner et al., 1987). Our results replicate and extend the findings of previous studies. First, in line with Van Damme et al. (2004c), we found that the anticipation of pain facilitated attentional engagement. However, this effect was not pain-specific, because engagement was equally facilitated by signals of impending pain as by signals predicting vibrotactile stimulation. This finding is in line with a recent crossmodal cueing study by Van Damme et al. (2004b), who found that the detection of pain stimuli and vibrotactile stimuli was equally facilitated by a cue correctly predicting the sensory modality of the stimuli. No differences were found between high- and low-pain catastrophizers. Second, we found that the anticipation of pain retarded disengagement of attention in high-pain catastrophizers. This effect proved to be painspecific, because disengagement was significantly more retarded by signals of impending pain than by signals predicting vibrotactile stimulation. This finding is in line with the crossmodal cueing study of Van Damme et al. (2004b), in which the detection of auditory stimuli was more retarded by cues predicting pain than by cues predicting vibrotactile stimulation. Our findings have a number of theoretical implications. Facilitated engagement by the anticipation of pain is not a pain-specific process. This indicates that there is no threatrelated bias in the initial processing of signals predicting somatosensory stimulation. This suggests that the amount of attention initially allocated to signals is probably determined by their salience and not by their threat value. In contrast to this, retarded disengagement by the anticipation of pain is a pain-specific process. This indicates that the amount of attention held on signals is determined by their threat value. Once a signal is perceived as threatening, attention is kept on the threat location, in order to allow the initiation of protective behaviour and escape strategies. When the signal is tagged as non-threatening, further processing of the signal is inhibited, and attention is directed to other relevant locations. Furthermore, this would explain the differential effects of high- and low-pain catastrophizers. High-pain catastrophizers probably perceive the signal of
398
S. Van Damme et al. / Pain 111 (2004) 392–399
impending pain as highly threatening. As pain responses of high catastrophizers are activated very swiftly (Sullivan et al., 2001), these participants will worry more about, and be more fearful of, anticipated pain compared with low-pain catastrophizers (Aldrich et al., 2000). Consequently, they will process the signal predicting pain thoroughly, and hold attention to its location. In contrast, low-pain catastrophizers, for whom the anticipated pain is not threatening, will inhibit further processing of the signal, and direct attention to other relevant locations. Our findings may have some clinical relevance. Most patients suffering from chronic pain will perceive their pain and signals of impending pain as highly threatening. Consequently, the initial interruption by and processing of such pain signals should be considered a normal process. However, based on our findings, it could be expected that these patients will have pronounced difficulties disengaging attention from the anticipated pain location and directing it to other environmental stimuli. Such fixation of attention by the anticipation of pain might become dysfunctional, resulting in hypervigilance, fear, worry, and avoidance behaviour (Aldrich et al., 2000; Crombez et al., 1998b; Peters et al., 2002). However, our ideas about the relation between these concepts and difficulty disengaging are speculative, and require further investigation with patient populations. This study has a number of limitations, which require further consideration. First, this study was conducted with pain-free volunteers, using experimental pain stimuli. Therefore, one must be cautious in generalizing the results to chronic pain patients. Our findings need further confirmation, both in non-clinical and clinical populations. Second, the majority of the participants in our study were female, limiting the generalizability of the results. Third, although we were able to demonstrate that the pattern of results for high-pain catastrophizers is robust (Van Damme et al., 2002b, 2004a), our conclusions were drawn from relatively small sample sizes. Therefore, it is possible that the statistical power in our study was low, resulting in the detection of large effects but leaving undetected differences with small effect size. In future research, work with a priori selected high- and low-pain catastrophizers could be considered, thereby requiring smaller sample sizes. In sum, evidence was found that spatial attention is modulated by the anticipation of pain. Furthermore, only the modulation of attentional disengagement and not engagement proved to be pain-specific. Finally, painspecificity was only found in individuals who perceived pain as highly threatening. One challenge for future research might be the combination of behavioural and neurophysiological measures, to foster our understanding of the differential representation of attentional engagement and disengagement in the brain (Legrain et al., 2002; Lorenz et al., 2003).
Acknowledgements This study was supported by a research grant (G.0107.00) to Geert Crombez by the Fund for Scientific Research Flanders. The authors would like to thank Charlotte Nollet for her help with the data collection, and Ernst Koster for his always inspiring comments.
References Aldrich S, Eccleston C, Crombez G. Worrying about chronic pain: vigilance to threat and misdirected problem solving. Behav Res Ther 2000;38:457–70. Cohen J. Statistical power analysis for the behavioural sciences. San Diego, CA: McGraw-Hill; 1988. Crombez G, Eccleston C, Baeyens F, Eelen P. When somatic information threatens, catastrophic thinking about pain enhances attentional interference. Pain 1998a;75:187–98. Crombez G, Vervaet L, Lysens R, Baeyens F, Eelen P. Avoidance and confrontation of painful back straining movements in chronic back pain patients. Behav Modif 1998b;22:62–77. De Clercq A, Crombez G, Roeyers H, Buysse A. A simple and sensitive method to measure timing accuracy. Behav Res Methods Instrum Comput 2003;35:109–15. Eccleston C, Crombez G. Pain demands attention: a cognitive-affective model on the interruptive function of pain. Psychol Bull 1999;125: 356–66. Honore´ J, He´non H, Naveteur J. Influence of eye orientation on pain as a function of anxiety. Pain 1995;63:213–8. Jensen J, McIntosh AR, Crawley AP, Mikulis DJ, Remington G, Kapur S. Direct activation of the ventral striatum in anticipation of aversive stimuli. Neuron 2003;40:1251–7. Legrain V, Gue´rit J, Bruyer R, Plaghki L. Attentional modulation of the nociceptive processing into the human brain: selective spatial attention, probability of stimulus occurrence, and target detection effects on laser evoked potentials. Pain 2002;99:21–39. Lorenz J, Minoshima S, Casey KL. Keeping pain out of mind: the role of the dorsolateral prefrontal cortex in pain modulation. Brain 2003;126: 1079–91. Peters ML, Vlaeyen JWS, Kunnen AMW. Is pain-related fear a predictor of somatosensory hypervigilance in chronic low back pain patients? Behav Res Ther 2002;40:85–103. Pincus T, Morley S. Cognitive processing bias in chronic pain: a review and integration. Psychol Bull 2001;127:599–617. Ploghaus A, Tracey I, Gati JS, Clare S, Menon RS, Mathews PM, Rawlins JNP. Dissociating pain from its anticipation in the human brain. Science 1999;284:1979–81. Ploghaus A, Becerra L, Borras C, Borsook D. Neural circuitry underlying pain modulation: expectation, hypnosis, placebo. Trends Cogn Sci 2003;7:197–200. Porro CA, Baraldi P, Pagnoni G, Serafini M, Facchin P, Maieron M, Nichelli P. Does anticipation of pain affect cortical nociceptive systems? J Neurosci 2002;22:3206–14. Porro CA, Cettolo V, Francescato MP, Baraldi P. Functional activity mapping of the mesial hemispheric wall during anticipation of pain. NeuroImage 2003;19:1738–47. Posner MI, Inhoff A, Friedrich FJ, Cohen A. Isolating attentional systems: a cognitive-anatomical analysis. Psychobiology 1987;15:107–21. Spence C, Nicholls MER, Driver J. The costs of expecting events in the wrong sensory modality. Percept Psychophys 2001;63:330–6. Spence C, Bentley DE, Phillips N, McGlone FP, Jones AKP. Selective attention to pain: a psychophysical investigation. Exp Brain Res 2002; 145:395–402.
S. Van Damme et al. / Pain 111 (2004) 392–399 Sullivan MJL, Bishop SR, Pivik J. The Pain Catastrophizing Scale: development and validation. Psychol Assess 1995;7:524–32. Sullivan MJL, Thorn B, Haythornthwaite JA, Keefe F, Martin M, Bradley LA, Lefebvre JC. Theoretical perspectives on the relation between catastrophizing and pain. Clin J Pain 2001;17: 52–64. Van Damme S, Crombez G, Vlaeyen JWS, Goubert L, Van den Broeck A, Van Houdenhove B. De Pain Catastrophizing Scale: psychometrische karakteristieken en normering [The Pain Catastrophizing Scale: psychometric characteristics and norms]. Gedragstherapie 2000;33: 209–22. Van Damme S, Crombez G, Bijttebier P, Goubert L, Van Houdenhove B. A confirmatory factor analysis of the Pain Catastrophizing Scale: invariant factor structure across clinical and non-clinical populations. Pain 2002a;96:319–24.
399
Van Damme S, Crombez G, Eccleston C. Retarded disengagement from pain cues: the effects of pain catastrophizing and pain expectancy. Pain 2002b;100:111–8. Van Damme S, Crombez G, Eccleston C. Disengagement from pain: the role of catastrophic thinking about pain. Pain 2004a;107:70–6. Van Damme S, Crombez G, Eccleston C, Goubert L. Impaired disengagement from threatening cues of impending pain in a crossmodal cueing paradigm. Eur J Pain 2004b;8:227–36. Van Damme S, Lorenz J, Eccleston C, Koster EHW, De Clercq A, Crombez G. Fear-conditioned cues of impending pain facilitate attentional engagement in an emotional modification of the exogenous cueing paradigm. Neurophysiol Clin 2004c;34:33–9. Vanderiet K, Adriaensen H, Carton H, Vertommen H. The McGill Pain Questionnaire constructed for the Dutch language (MPQ-DV): preliminary data concerning reliability and validity. Pain 1987;30:395–408.