Event-related brain potentials and affective responses to threat in spider/snake-phobic and non-phobic subjects

Transcription

1 International Journal of Psychophysiology 57 (2005) Event-related brain potentials and affective responses to threat in spider/snake-phobic and non-phobic subjects Wolfgang H.R. Miltner*, Ralf H. Trippe, Silke Krieschel, Ingmar Gutberlet, Holger Hecht, Thomas Weiss Department of Biological and Clinical Psychology, Friedrich Schiller University, Am Steiger 3//1, D Jena, Germany Received 10 November 2004; received in revised form 25 January 2005; accepted 27 January 2005 Abstract We investigated cortical responses and valence/arousal ratings of spider phobic, snake phobic, and healthy subjects while they were processing feared, fear-relevant, emotional neutral, and pleasant stimuli. Results revealed significantly larger amplitudes of late ERP components (P3 and late positive complex, LPC) but not of early components (N1, P2, N2) in phobics when subjects were processing feared stimuli. This fear-associated increase of amplitudes of late ERP components in phobic subjects was maximal at centro-parietal and occipital brain sites. Furthermore, phobics but not controls rated feared stimuli to be more negative and arousing than fear-relevant, emotional neutral, and pleasant stimuli. Since late ERP components and valence/arousal ratings were only significantly increased when phobic subjects but not when healthy controls were processing feared stimuli, the present data suggest that P3 and LPC amplitudes represent useful neural correlates of the emotional significance and meaning of stimuli. D 2005 Elsevier B.V. All rights reserved. Keywords: ERP; Valence; Arousal; Animal phobia; Fear 1. Introduction Phobic fear in humans is more frequently elicited by objects like snakes, spiders, or rodents than by any other object of our every-day environment (Agras et al., 1969; Marks, 1969). Fear of these stimuli represents the core group of animal phobias with almost equal prevalence among different populations (9 14%) and a clear dominance in females. Whether this preference of phobic responses to a subgroup of animals is based on phylogenetic preparedness (Seligman, 1971; McNally, 1987; Öhman et al., 1995) or due to other evolutionary-based information processing systems in the human brain (LeDoux, 1996) or whether it is the result of conditioning or otherwise acquired cognitive schemata (Beck and Emery, 1985) and memory * Corresponding author. Tel.: ; fax: address: wolfgang.miltner@uni-jena.de (W.H.R. Miltner). functions (Bower, 1981; Williams et al., 1997; Lang et al., 2000) is still a matter of debate. However, there is general agreement that the presence of these stimuli in the environment of phobics grants these objects salient meaning by significantly pulling subjects attention towards them, increasing arousal and negative emotional feelings of threat and the induction of flight/fight responses (Williams et al., 1997; Lang et al., 2000). In the past, evidence for the salience of threatening stimuli was demonstrated by a number of studies indicating that the processing/encoding of such stimuli is significantly biased in clinically anxious subjects and subjects with hightrait anxiety scores (Williams et al., 1997). Researchers became particularly interested to disentangle individual differences of involuntary attention to threat and demonstrated that attention of anxious subjects is automatically pulled to threatening stimuli (MacLeod, 1991, 1999; MacLeod and Mathews, 1991; Logan and Goetsch, 1993; Williams et al., 1997). Results were based on different information processing paradigms including dichotic-listen /$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi: /j.ijpsycho

2 44 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) ing tasks (Mathews and MacLeod, 1986), the emotional Stroop test (Williams et al., 1996), the dot-probe-paradigm (MacLeod et al., 1986), and an emotional version of the Eriksen-flanker task (Streblow et al., 1985; Lavy et al., 1993). As a general result of these studies, it was shown consistently that the attention of subjects with anxiety or fear is automatically allocated to threatening stimuli at early stages of information processing. Furthermore, it was shown that this early allocation of attention happens particularly when stimuli are strongly related to subjects individual anxious concern (Logan and Goetsch, 1993; Williams et al., 1997) and when stimuli are presented outside of subjects attentional focus (Fox, 1996; Miltner et al., 2004). Additionally, studies by Öhman and coworkers (Öhman and Soares, 1993; Öhman et al., 1995) have shown that peripheral nervous system activity in animal phobics is significantly increased when subjects are exposed either to masked or unmasked feared stimuli as compared to neutral stimuli or stimuli related to other phobic fears (Öhman and Soares, 1998). From these and additional studies, Öhman and Mineka (2001) concluded that attention to feared stimuli is guided by latent dispositions and/or by memory networks and top-down controlled attentional sets that involuntarily and automatically allocate subjects attention towards threat. This priority processing finally lends these stimuli salient and significant emotional meaning. Despite the fact that cognitive paradigms and investigations on peripheral nervous system activities have greatly supported the presence of a selective encoding bias in phobic subjects, only little is known, how the brain of phobic subjects organizes this biased processing. Only few studies have tested the neural correlates of this biased, salient, and automatic processing of threatening stimuli by means of brain electrical event-related potentials (ERPs, i.e., the P1, N1, P2, P3, the late positive complex (LPC)). While the amplitude, latency, and topographic distribution of the P1 and N1 component in response to visual stimuli were demonstrated to represent correlates of different early selective attentional operations (Hillyard and Picton, 1979; Hansen and Hillyard, 1980), late components like the P3 and the LPC were demonstrated to be affected by the probability of an event s occurrence and its relevance to the observer s task (Johnson, 1986). Additionally, P3 amplitude is significantly affected by the amount of attention being spent to a stimulus at later stages of information processing and is a significant correlate of different memory encoding operations (Johnson, 1986). Furthermore, P3 amplitude has been shown to be sensitive to the amount of subjects intentional engagement towards a stimulus and its demand characteristics (Johnson, 1986). Finally, and most important for the present study, it was shown that P3 amplitude is systematically affected by the individual meaning of a stimulus, i.e., by its emotional arousal and individual valence (Begleiter et al., 1967; Hömberg et al., 1981; Johnston et al., 1986; Farwell and Donchin, 1991; Naumann et al., 1997; Palomba et al., 1997; Schupp et al., 2000). These studies consistently indicate that P3 and LPC amplitudes are largest in response to those stimuli that subjects rate as emotionally meaningful and arousing. Furthermore, in a recent study by Pauli et al. (1997), P3 amplitude was used to investigate whether brain electrical activities of subjects with panic disorders are affected differently by stimuli whose meaning reflects critical concerns associated with emotional stress as compared to emotional neutral stimuli. Using body-related and nonsomatic words, presented tachistoscopically to 15 panic patients and 15 healthy controls at each subject s perceptual threshold for correctly identifying 50% of neutral words, it was demonstrated that panic patients recognized more bodyrelated words than non-body-related words. Furthermore, compared to words not related to individual concerns, bodyrelated words were associated with larger P3 amplitudes than non-body-related words. In healthy controls, no equivalent difference in word identification or in P3 amplitude was observed. When LPC activities to bodyrelated and non-body-related words were compared, panic patients additionally showed larger positive slow waves than healthy controls. With reference to the study by Pauli et al. (1997), the present study examined which ERP amplitude (N1, P2, N2, P3, LPC) is significantly increased when phobic subjects are processing visual stimuli related to their own specific phobic fear as compared to stimuli depicting objects of nonpersonal meaning or to objects with emotionally neutral or positive meaning. When P3 amplitude and the magnitude of LPC are signatures of salience and threat, we should obtain significant larger P3 and LPC amplitudes when phobic subjects are processing images related to their personal concerns than to stimuli evaluated as fearful by others or when processing neutral or positive stimuli. Control subjects, having been selected to be free of any phobia, should not show this enhancement of P3 amplitude in response to spider or snake stimuli. Responses to these stimuli should be like those observed for neutral or pleasant slides. In terms of the N1 component, which is considered to represent a signature of early processes related to attention (Hillyard and Picton, 1979; Hansen and Hillyard, 1980), we should obtain similar differences between stimulus categories and groups with larger N1 amplitudes in phobic subjects when processing objects of personal concern as compared to the stimuli from the other categories. Similar results are expected for the late positive complex. 2. Methods 2.1. Subjects Subjects were recruited from the student population of the University of Jena by means of handouts and posters.

3 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) Prior to the experiment, subjects received several questionnaires assessing either the presence of a specific phobic fear (Snake Phobia Questionnaire, SNQ, Spider Phobia Questionnaire, SPQ) (Klorman et al., 1974; Fredrikson, 1983) or unspecific anxiety symptomatology (German version of the Spielberger Trait Anxiety Inventory, STAI- X2; Laux et al., 1981). Subjects were included into the study when they scored above the 80th percentile on the SPQ or SNQ, but excluded, when these questionnaires simultaneously indicated the presence of both phobias. Subjects suffering from other anxiety disorders or high-trait anxiety (STAI-X2 > 70), or subjects with other physical or psychological complaints were also excluded. Subjects, who showed high scores on the SNQ (>20) and low scores on the SPQ (<9) and who met the criterion of the STAI-X2 questionnaire, became assigned to the snake phobia group and subjects who met the reverse criteria on the SNQ/SPQ (SNQ <9; SPQ >20) and fulfilled the STAI-X2 criterion were assigned to the spider phobic group. Control subjects were required to score below the 30th percentile on both phobia questionnaires and were not allowed to show STAI- X2 scores of above 60. With this procedure, a total number of 28 subjects was selected out of 612 subjects for participation in this study. Of these individuals, 14 subjects suffered from animal phobia (7 spider and 7 snake phobics) and 14 participants fulfilled the criteria for the control group. Twenty-three out of these 28 subjects were females. Subject s age ranged from 19 to 40 years (mean age: 22.6 years) with no significant differences between the three groups. Snake phobia test-scores (SNQ: F(2,25) = 58.6, p > ) and spider phobia test-scores (SPQ: F(2,25) = 47, p < ) differed significantly between groups with snake phobics expressing the strongest fear of snakes (SNQ: 22.2/1.2; SPQ: 13.6/8) and spider phobics the strongest fear of spiders (SPQ: 24.3/3.2; SNQ: 11.2/5.9) while controls subjects expressed no fear of both phobic objects (SPQ: 3.4/2.9; SNQ: 5.2/2.4). Before participation, subjects were informed about the experimental procedure and signed a consent form according to the declaration of Helsinki on experiments with human subjects. Subjects included into the experiment proper received 10 Euros for participation. Prior to the experiment, the local ethics committee approved the study Stimulus material The stimuli consisted of 240 animal slides, either belonging to the two phobic categories spiders (N = 60) and snakes (N =60), or to the category of emotional pleasant pictures (N = 60) depicting either puppies, bunnies, kittens, fawns, and kids or to a category of emotional neutral pictures (N = 60) showing adult toy fishes and birds (neutral category). Slide complexity was kept similar between all slides by displaying a single animal at the center of each slide whose size was about the same for all images. Furthermore, in all slides, the same colored background surrounded this animal. Also, slides of each category were selected from more than 600 slides depicting negative, neutral, and pleasant objects by means of a hierarchical cluster and subsequent discriminance analysis of emotional valence and arousal ratings for each single slide provided by different groups of subjects who participated in an earlier experiment. Slides of each category were composed of those slides whose position within the emotional space constituted by subjects valence and arousal ratings corresponded closest to each category s center within this space. Slides were presented in landscape format for a period of 4 s and in pseudo random order with the limitation that no more than two consecutive slides belonged to the same object category. Also, slides were arranged in blocks of 12 successive slides with equal numbers of neutral, pleasant, and negative slides in each block. After each single slide presentation, subjects were shown an additional slide for 4 s depicting a figural representation of the valence and arousal scales of the Self Assessment Manikin (SAM) (Lang et al., 1999) and requested to give separate verbal ratings (ranging from 1 to 9) about the emotional valence (1 = very unpleasant; 9 = very pleasant) and arousal (1 = very calm, 9 = highly aroused) induced by each slide. The interstimulus interval between two succeeding slides with objects varied between 8 and 12 s. Prior to the experiment, subjects filled in the German version of the STAI-X2 questionnaire (Laux et al., 1981) and then were prepared for physiological recordings. After attachment of electrodes, subjects became seated in an acoustically and electrically shielded EEG chamber (Industrial Acoustics Company, Germany) and were instructed and familiarized with the experimental and the SAM rating procedures. In order to facilitate the activation of an emotional action set, subjects were instructed to maximally attend to each slide and to focus on the emotional state and arousal induced by each slide for the following assessments by the SAM rating procedure Data recording and analysis The electroencephalogram was recorded from 9 electrode sites mounted onto each individual s scalp according to the international system (Jasper, 1958) using a DCamplifier operated in DC-mode. Ag/AgCl-electrodes were placed at locations F3, FZ, F4, C3, CZ, C4, P3, PZ, and P4 and fixed by an electrode cap. Prior to application of the cap, all electrode sites were cleansed with alcohol (70% isopropanol) and electrodes were injected with an electrolyte jelly by means of a syringe. Additionally, skin resistance at each electrode site was reduced to below 5 kv by removing the upper skin layers with the tip of the syringe. For off-line control of eye-movement and blink artifacts, the vertical electroencephalogram (VEOG) was recorded from electrodes placed 1 cm above and 1 cm below the midline of the left eye and electrodes for horizontal eye

4 46 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) movement recordings (electrooculogram, EOG) were fixed to the outer canti of both eyes. All active electrodes were referenced to linked mastoids. The amplifier s high frequency filter was set to a cut-off frequency of 30 Hz and data of all channels were digitized online and sampled at a rate of 250 Hz. EEG epochs were recorded for 2048 ms starting 500 ms before the onset of visual stimuli. This 500 ms period served as baseline. In order to account for possible electrode drifts between epochs caused by DC_shifts, amplifiers were reset at 2500 ms pre_stimulus before each trial. Recording epochs then were corrected for excessive EEG artifacts such as muscle contractions by excluding all epochs from further analyses whose EEG activity exceeded T150 AV. Additionally, trials with horizontal eye movements of more than 50 AV were discarded from all further analysis. Correction of vertical eye movements and blinks was performed offline on single trial epochs using the procedure proposed by (Gratton et al., 1983). Since in the present study we were mainly interested to assess whether phobic and non-phobic subjects show significant differences in ERP components when processing stimuli with feared contents as compared to stimuli with fear-relevant (i.e., negative), neutral, or pleasant contents, the experimental groups and stimulus categories were rearranged according to the following procedure: spider and snake phobic subjects were collapsed into one group of animal phobic subjects (N =14). Second, trials where stimuli from the snake and spider categories were presented were split into two new experimental categories termed the Ffeared_ category and the Ffear-relevant, non-phobic_ category. The feared category now includes all stimulus presentations where spider phobics were exposed to spider images and snake phobics were exposed to snake pictures. Similarly, the fear-relevant, non-phobic category now includes all stimulations where spider phobics were exposed to snake images and snake phobics were exposed to spider images. The arrangement and denotation of the other stimulus conditions remained unchanged. In order to keep this rearrangement of snake and spider images into the two new categories similar for control subjects, snake and spider images were also rearranged for these subjects by randomly reassigning half of the snake pictures and half of the spider pictures to a dummy category of Ffeared_ pictures and the other half of spider and snake pictures to a dummy category of Ffear-relevant_ pictures. Artifact-corrected waveforms were averaged separately for each electrode, each condition, and each individual. Magnitudes and latencies for the ERP components N1, P2, N2, and P3 were extracted from these average waveforms at the maximal base-to-peak amplitude measure of each component following stimulus onset after waveforms were baseline-corrected by subtracting each individual s average activity at each electrode and of each condition within the baseline period from all succeeding voltages. Scoring of N1, P2, and N2 amplitudes was achieved within windows ranging from 100 ms to 200 ms for N1 (Cz), 190 ms to 300 ms for P2 (Cz), and 250 ms to 380 ms for N2 at Cz. P3 amplitudes at all electrodes were scored within 340 ms and 520 ms post-stimulus with reference to Pz. For determination of the LPC, an area measure within the time window between 450 and 700 ms post-stimulus was used. ERP peak amplitudes of each component were submitted separately to a 2433 repeated measures analysis of variance (ANOVA) using the between-subject factor Group (phobics, controls) and three within-subject factors Stimulus Category (feared, fear-relevant, neutral, pleasant), Sagittal (electrodes at the left hemisphere (i.e., electrodes F3, C3, and P3), the midline (Fz, Cz, and Pz), and at the right hemisphere (F4, C4, and P4) and Coronal (electrodes at frontal (F3, Fz, and F4), central (C3, Cz, and C4), parietal sites (P3, Pz, and P4). Each component s topographical distribution was scaled separately for each condition and subject by its corresponding norm vector, i.e., by the square root of the sum of squared voltages over all electrode locations (see McCarthy and Wood, 1985). The normalized data then were submitted separately for each component to a second 2433 repeated measures ANOVA and topographical differences between conditions and groups were accepted when the interactions between factors Category/ Group and one of the topographical factors Coronal or Sagittal remained significant. Results of the ANOVA with non-normalized data will only be detailed, when the ANOVA with normalized data became significant. Data on valence and arousal ratings (SAM) of each stimulus condition were averaged for each individual across 5 blocks of 12 succeeding stimuli each and submitted separately to a 245 repeated measures ANOVA using the between-subjects factor Group and the within-subject factors Stimulus Category and Blocks (1 to 5). When appropriate, Greenhouse-Geisser epsilon factors (indexed as Fe=_ throughout the text) were used to correct the F-values of all ANOVAs in order to minimize violations of sphericity (Greenhouse and Geisser, 1959). Significance level was pre-set to.05 for all analyses. Significant effects involving any of the repeated measures factors were followed up by post-hoc tests between all levels of the respective factor. SPQ, SNQ, and STAI-X2 data were analyzed by t-tests for independent groups (phobic subjects vs. control subjects). 3. Results 3.1. STAI Average STAI-Trait scores of subjects were not significantly different between groups (STAI-X2: snake phobics (Mean/Standard Deviation) 33.6/6.3, spider phobics 38/ 7.3, controls 37.8/7.1).

5 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) SAM Average valence ratings and arousal ratings in response to feared, fear-relevant, pleasant, and neutral stimuli of phobic and control subjects are shown in Fig. 1. For valence, there were significant main effects of factors Stimulus Category ( F(3,78)=120.1, p <0.0001; e =0.66) and Group ( F(1,25)=12.3, p <0.002) and a significant interaction between both factors ( F(3,78) =21.6, p <0.0001). Phobic subjects rated objects associated with their personal fear as most unpleasant while control subjects rated these stimuli only slightly negative ( p < 0.001). No significant valence differences between groups were obtained for neutral or pleasant stimuli. A similar picture emerged for arousal ratings where stimuli of the four stimulus categories ( F(3,78) =32.5, p <0.0001, e =0.54) were rated significantly different with feared stimuli being rated as most arousing followed by fear relevant, neutral, and pleasant stimuli. A significant main effect of factor Group F(1,25)=6.2, p =0.02) revealed that phobics became overall more aroused than controls while watching the slides. According to a significant interaction of stimulus Category by Group ( F(3,78) = 7.79, p = ), phobic subjects rated slides depicting feared objects as most arousing ( p <0.001) while no significant difference between groups was obtained for fear-relevant, neutral, or pleasant stimuli ERP-data Grand average waveforms of control and phobic subjects in response to neutral, pleasant, fear-relevant, and feared stimuli at all electrode sites are shown in Fig. 2. ANOVA results for N1 amplitude and P2 amplitudes showed no significant main effect for factors Group or Stimulus Category and no significant interaction between both factors. Also the ANOVA of N2 amplitudes did not reveal significant main effects for factors Group or Stimulus Category but a significant interaction Category by Group ( F(3,78) =3.8, p =0.017). While phobics showed most negative amplitudes in response to pleasant stimuli, followed by fear-relevant, and neutral stimuli and least negative amplitudes when processing feared stimuli, control subjects showed most negative amplitudes in response to feared stimuli followed by fear-relevant and pleasant stimuli. Least negative amplitudes were observed when subjects were processing pleasant stimuli (all comparisons p < 0.03). Average Valence Ratings A Controls B Phobics Blocks Blocks Average Arousal Ratings A Blocks Neutral Pleasant Fear-relevant Feared B Blocks Fig. 1. Average valence (A) and arousal ratings (B) of experiment one to 60 feared (red lines), 60 fear-relevant (green), 60 positive (blue), and 60 neutral (black) visual stimuli of snake and spider phobic subjects (left column) and non-phobic subjects (right column). Y-axes are scaled according to the Self-Assessment Manikin (SAM) procedure ranging from 1 (very unpleasant/very calm) to 9 (very pleasant/very aroused). X-axes represent average ratings of 5 consecutive blocks of 12 stimuli of each stimulus category. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

6 48 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) A F3 Fz F4 B F3 Fz F4 C3 Cz C4 C3 Cz C4 0 P3 Pz P4 P3 Pz P4 µv +25 [ms] Neutral Pleasant... Fear-relevant Feared Fig. 2. Grand averages of event-related potentials at 3 frontal, 3 central, and 3 parietal electrode sites of snake and spider phobic subjects (right column B) and non-phobic subjects (left column A) to feared (red lines), fear-relevant (green), positive (blue), and neutral (black) visual stimuli. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 3.4. P3 Analysis of P3 amplitudes showed no significant main effect for factors Stimulus Category and Group but a significant interaction between both factors ( F(3,78) = 3.9, p = 0.02). P3 amplitudes of phobic subjects were most pleasant in response to feared stimuli (7.3 AV) and almost 75% larger than in control subjects (4.2 AV) whereas in response to the other stimulus categories both groups showed similar P3 magnitudes (phobics fear-relevant: 5.5 AV, neutral: 5.1 AV, pleasant: 4.3 AV; control subjects fear-relevant: 5.5 AV, neutral: 5.2 AV, pleasant: 5.1 AV). Significant main effects of factors Sagittal ( F(2,52) = 3.4, p =0.048, e =0.9) and Coronal ( F(2,52)=138.5, p < , e = 0.61) indicated that P3 amplitudes were larger at midline (5.3 AV) and at the right hemisphere (6 AV, both did not differ significantly) as compared to the left hemisphere (4.5 AV) ( p < 0.03). P3 amplitude was most pleasant at parietal sites (15 AV) followed by vertex (3.5 AV) and negative at frontal sites ( 3.1 AV, all p <0.03). A significant interaction between factors Stimulus Category and Coronal ( F(6,156)=6, p =0.0008, e =0.54) and a significant three way interaction Stimulus Category by Coronal by Group ( F(6,156) = 3.4, p = 0.018) were not confirmed by normalized data. All other interactions involving the factors Group or Stimulus Category did not reveal significant effects. Overall P3 latencies were not significantly different between both groups, and Stimulus Categories, however, there was a trend-wise interaction Stimulus Category by Group ( F(3,78) = 2.9, p = 0.053). While phobic subjects showed no significant difference of latencies between stimulus categories in response to feared (419 ms), fearrelevant (425 ms), and pleasant (427 ms) stimuli, latency of P3 amplitude of this group was fastest to neutral stimuli (412 ms). Conversely, P3 latencies of control subjects to fear-relevant (418 ms), pleasant (433 ms), and neutral stimuli (424 ms) were not significantly different, however, when processing feared stimuli (468 ms) P3 latency became significantly delayed (all sig. comparisons p < 0.024) LPC ANOVA of the LPC revealed no significant differences between Groups and Stimulus Category. There was a trendwise interaction between Stimulus Category and Group ( F(3,78) =2.7, p =0.055). A highly significant main effect of factor Coronal ( F(2,52) = 115.7, p < , e = 0.63; frontal: 0.2 AV; central: 7.7 AV; parietal: 16.7 AV) and a significant interaction between Stimulus Category and Coronal ( F(6,156) = 4.1, p = 0.008, e = 0.54; normalized: F(6,156) = 2.7, p = 0.04) indicated largest LPC activities at posterior sites when feared stimuli were processed (17.9 AV), followed by fear-relevant (16.7 AV), and pleasant stimuli (16 AV). Smallest amplitudes at parietal sites were obtained for neutral stimuli (15.8 AV) (all p < 0.03). In response to all stimulus categories, amplitudes at anterior sites were small and negative during most stimulus conditions (feared: 0.11 AV; fear-relevant: 0.7AV; pleasant 1.6 AV; neutral 0.2 AV). At central sites, amplitudes were pleasant for all stimulus categories (feared: 8.5 AV; fearrelevant: 7.3 AV; pleasant 8.2 AV; neutral 6.6 AV) but only

7 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) about half the magnitude than at parietal sites (feared: 17.9 AV; fear-relevant: 16.8 AV; pleasant 16.3 AV; neutral 15.8 AV) (all p < 0.04). A significant three way interaction between factors Group, Stimulus Category and Coronal ( F(4,104) =5.7, p =0.009) was not confirmed by normalized data. All other interactions involving factors Group or Category were not significant. 4. Discussion Valence and arousal ratings of spider/snake phobic and control subjects in response to neutral and pleasant stimuli of the present study confirmed earlier observations in a German population (Hamm and Vaitl, 1993) that images of young animals are commonly judged to be of positive valence while ornament fishes and birds are rated as emotionally neutral. Like in the earlier studies by Hamm and Vaitl (1993) and Lang et al. (1999), subjects of the present study also judged these stimuli to be of only minimal arousing quality. A similar correspondence to these earlier findings was observed in control subjects for spider and snake images. Like in the Hamm et al., and Lang et al. studies, control subjects considered spider and snake stimuli to be significantly more negative and more arousing than neutral and pleasant stimuli. However, compared to subjects of the present phobic group, they were judged by control subjects to be much less negative and much less arousing. When spider phobics were processing spider images and snake phobics were processing snake images, i.e., when phobic subjects were processing feared stimuli, valence and arousal ratings of these stimuli significantly reflected the threatening quality of these stimuli for these individuals and confirmed their fear of spiders and snakes assessed by the Snake Phobia and Spider Phobia Questionnaires (Klorman et al., 1974; Fredrikson, 1983). Although, results of both phobia questionnaires indicated that spider phobics considered themselves not to be anxious of snakes and snake phobics reported not being fearful of spiders, arousal and valence ratings to these stimuli indicate that both subgroups of phobic subjects were also more negatively affected by these fear-relevant stimuli than control subjects. Rating values on arousal and valence in response to these fear-relevant stimuli were in between responses to neutral/pleasant and feared stimuli. Finally, across blocks of stimuli, results demonstrate that these patterns of ratings were fairly consistent in both groups of subjects for all stimulus categories throughout the course of stimulation. Results from the present studies confirm a number of earlier findings (Begleiter et al., 1967; Hömberg et al., 1981; Johnson, 1986; Farwell and Donchin, 1991) demonstrating that visual images are helpful tools to probe individual differences in the processing of stimuli with different emotional quality on the subjective level, the peripheral, and the central level of the nervous system. The significance of threat was additionally emphasized by large magnitudes of the P3 and the late positive complex of brain electrical potentials when phobic subjects were processing feared stimuli. This observation suggests that the P3 amplitude is affected positively by stimuli of emotional significant arousal and meaning. Like in the Pauli et al. (1997) study, where panic patients demonstrated larger P3 and LPC amplitudes when processing body-related words associated to panic as compared to words not associated to panic, P3 and LPC amplitude of phobic subjects of the present studies were significantly increased in response to stimuli whose meaning reflects critical fear concerns of these subjects. While in the Pauli et al. study P3 amplitude was demonstrated to be largest at centro-parietal brain regions of panic patients in response to panic-related stimuli, but not affected differentially at other brain areas, data of the present studies further revealed that P3 amplitude of phobic subjects is also larger at parietal areas when feared stimuli but not when stimuli of the non-feared categories are processed. This larger parietal activation to threat was not observed in controls. Our results confirm recent findings by Junghöfer et al. (2001) that the processing of highly arousing emotional stimuli significantly activates posterior brain areas involved in the analysis of visual stimuli. It is important to note that the increased magnitudes of the P3 and LPC did not purely reflect negative valence and/ or increased arousal induced by stimuli rated as emotionally negative. While spider stimuli were rated significantly more negatively and arousing by snake phobics and snake objects were rated more negatively and arousing by spider phobics, the magnitudes of phobic subjects late components in response to these fear-relevant stimuli did not differ significantly from those observed to neutral and pleasant stimuli. Furthermore, P3 and LPC amplitudes of phobic subjects to neutral, pleasant, and fear-relevant stimuli were almost similar to those observed in control subjects. Phobic subjects only showed increased magnitudes of the P3 and LPC when they were processing fearful stimuli of personal concern. Our present data do not confirm results from studies based on EEG-spectral analysis suggesting that the processing of threat is significantly associated with increased right hemispheric activation, whereas the processing of positive stimuli is associated with increased left hemispheric activation (Davidson, 1992, 2001, 2002). In both groups of subjects, the processing of emotional stimuli (pleasant or negative) was not accompanied by differential activation of one of the both hemispheres nor were there significant differences in hemispheric activation when phobic subjects were processing feared stimuli. Reasons for missing differential hemispheric activations in the present studies might be that we used normalized ERPs measures whereas previous studies either did not use normalized measures for ERP analyses or had based their findings on emotion-related hemispheric differences on spectral measures of the ongoing EEG. Data from the present experiment further suggest that earlier components of the ERP, i.e., the N1, P2, and N2 components, might not be as closely related to the emotional

8 50 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) valence and threatening meaning of a stimulus as the P3 component. This was surprising to us. Behavioral studies on the priority processing of threat (MacLeod, 1991) and on the attentional bias to threat in different groups of anxious subjects (Logan and Goetsch, 1993; McNally, 1996; Williams et al., 1997; MacLeod, 1999; McNally et al., 1999) have demonstrated that threat is capturing the attention of anxious subjects. From these observations, one would expect that threatening stimuli should significantly increase the attention of phobic subjects towards them and that this should result in a significant larger amplitude of at least the N1 component of ERPs (Hillyard and Picton, 1979; Hansen and Hillyard, 1980) in phobic subjects as compared to control subjects or to non-feared stimuli. One possible explanation for the missing differences between N1 amplitudes to threatening and non-threatening stimuli might be related to the stimulus material used in the present studies. As outlined by Mathews and MacLeod (2002), positive findings on the attentional bias in anxious subjects were mainly based on paradigms where stimuli were composed of ambiguous information. In most of the earlier studies on the attentional bias neutral and threatening information were presented simultaneously, whereas no attentional bias was observed in anxious subjects when non-ambiguous stimuli were presented (Mathews and Milroy, 1994). Finally, from a recent study by our group (Miltner et al., 2004) using a visual search paradigm where subjects were requested to search for a threatening or a neutral object depicted in an array of neutral images, it became clear that threatening and non-threatening information did not differentially affect early perceptual states of stimulus processing but later attentional selection processes involved in the organization of speeded behavioral responses to threat. This suggestion is in part supported by the P3-results of the present study. As outlined above, P3 is considered as a correlate of later attentional selection processes. According to this interpretation of P3 amplitude, phobic subjects showed significantly increased P3 amplitudes when processing threatening information as compared to negative, pleasant, or neutral information or as compared to control subjects. Our suggestion that the increased P3 amplitudes to phobic stimuli in phobic subjects represent a neural correlate of the threatening meaning of stimuli might be questioned by the fact that our results might have been contaminated by different stimulus probabilities of the four stimulus categories for both groups of subjects. Phobic subjects might have categorized neutral, fear-related, and pleasant stimuli into one group of non-threatening stimuli that opposed the category of feared stimuli (i.e., probability of threat to nonthreat = 1:3), whereas control subjects might have categorized these stimuli into four different categories with equal probability. Since stimulus probability is known to be a critical source of variance for P3 amplitude (Johnson, 1986), this would imply that the lower probability of threatening stimuli would have induced the larger P3 amplitudes in phobic subjects without the meaning of the stimuli would have played any role for the increase of P3 amplitude. While the present studies cannot clear up such an objection, results from a recent study of our group (Miltner et al., 2002), using the classical Oddball paradigm, do not support this assumption. In this study, four different runs of the paradigm were used with the probability of targets and non-targets (20:80 or 80:20) and the task relevancy of stimuli (targets and non targets had to be counted in different runs) being permutated across the four runs of the paradigm. By subsequent collapsing all data from the four runs into one data set for each individual, effects of probability and task relevancy across stimulus categories became cancelled out with the result that the different meaning of stimuli belonging to the two stimulus categories remained the only differential source of variance for the individual ERP waveforms. The results of this study fully supported the P3 results of the present experiments, demonstrating significantly larger P3 amplitudes in phobic subjects than in control subjects when spider stimuli were processed. P3 amplitudes induced by threat were almost of the same magnitude than those of the present studies. However, no difference of P3 amplitude between groups was obtained for neutral stimuli. Finally, it might be argued that the increased amplitudes of P3 in phobic subjects in response to feared stimuli do not reflect the significant emotional valence of these stimuli but first of all its significant arousing qualities (Schupp et al., 2000). Unfortunately, this argument is hardly testable since valence and arousal of (at least visual) stimuli are correlated quadratically, i.e., when arousal of a visual stimulus is rated to be high by subjects, also stimulus valence is rated to be strong, independently of the direction of valence (i.e., negative vs. positive; see Hamm and Vaitl, 1993; Lang et al., 1999). This squared relation between valence and arousal holds no matter whether stimuli are negative or positive. To test the exclusive impact of arousal on the P3 and LPC amplitude, it is not sufficient to demonstrate that very arousing negative and very positive stimuli are associated with larger P3 amplitudes but neutral, low arousing stimuli are not (Schupp et al., 2000). It would require the application of highly arousing stimuli with neutral valence. If P3 to such stimuli would not be different from P3 amplitudes to highly arousing stimuli with strong emotional valence (either positive or negative), one could correctly talk of P3 amplitude as of primarily reflecting the arousing quality of a stimulus. However, according to the IAPS, such stimuli are virtually not available. Also, the demonstration that phobic subjects show similarly large P3 amplitudes to stimuli whose valence is significantly negative and arousing for non-phobic subjects would not conflict with our suggestion that P300 represents a correlate of significant emotional stimulus valence or meaning. There is no reason why phobics should not consider pictures of mutilation and other negative images of strong arousal as being emotionally very negative like non-phobics do. Even when phobics show large P3 amplitudes to very positive and

9 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) arousing stimuli like erotica (what they do), this would not abrogate the concept that P3 amplitude is a correlate of stimuli with significant emotional valence. We do not state that it is an exclusive correlate of threat processing. However, the present study confirms that it is strongly correlated with the processing of emotionally very meaningful unpleasant stimuli. In summary, the present study indicates that the processing of feared stimuli of personal concern significantly induces negative affect and increased arousal in phobics. This processing of threat is accompanied by a significant increase of P3 amplitude in phobics. Our present experiments, therefore, suggest, that among the many cognitive and arousal functions assigned to the P3 amplitude during the last three decades, P3 amplitude represents a significant neural correlate of the meaning and salience of emotionally significant stimuli. The future inclusion of source analysis methods might additionally help to reveal in more detail what brain areas are involved when some individuals become threatened by specific stimuli and at what period within the stream of information processing this individual difference of stimulus impact is generated. Acknowledgements Research was supported by grants from the Deutsche Forschungsgemeinschaft (DFG) to WHRM (MI 265/6-1, MI 265/6-2) and in part by joint grants from the Federal Ministry of Education and Science and the Ministry of Science, Research, and Art of Thueringen to WHRM (Klinische Neurowissenschaften FS 1, ZP 1.4). We would like to thank two earlier students, N. Tillmann and S. Sierich, for their help during data acquisition. References Agras, W.S., Sylvester, D., Oliveau, D., The epidemiology of common fear and phobia. Compr. Psychiatry 10, Beck, A.T., Emery, G., Anxiety Disorders and Phobias: A Cognitive Perspective. Basic Books, New York. Begleiter, H., Gross, M.M., Kissin, B., Evoked cortical responses to affective visual stimuli. Psychophysiology 3, Bower, G.H., Mood and memory. Am. Psychol. 36, Davidson, R.J., Emotion and affective style: hemispheric substrates. Psychol. Sci. 3, Davidson, R.J., The neural circuitry of emotion and affective style: prefrontal cortex and amygdala contributions. Soc. Sci. Inf. Stud. 40, Davidson, R.J., Anxiety and affective style: role of prefrontal cortex and amygdala. Biol. Psychiatry 51, Farwell, L.A., Donchin, E., The truth will out: interrogative polygraphy ( lie detection ) with event-related brain potentials. Psychophysiology 28, Fox, E., Selective processing of threatening words in anxiety: the role of awareness. Cogn. Emot. 10, Fredrikson, M., Reliability and validity of some specific fear questionnaires. Scand. J. Psychol. 24, Gratton, G., Coles, M.G., Donchin, E., A new method for off-line removal of ocular artifact. Electroencephalogr. Clin. Neurophysiol. 55, Greenhouse, S.W., Geisser, S., On methods in the analysis of profile data. Psychometrika 24, Hamm, A.O., Vaitl, D., Emotionsinduktion durch visuelle Reize: validierung einer Stimulationsmethode auf drei Reaktionsebenen. Psychol. Rundsch. 44, Hansen, J.C., Hillyard, S.A., Endogenous brain potentials associated with selective auditory attention. Electroencephalogr. Clin. Neurophysiol. 49, Hillyard, S.A., Picton, T.W., Event-related brain potentials and selective information processing in man. Prog. Clin. Neurophysiol. 6, Hömberg, V., Grünewald, G., Grünewald-Zuberbier, E., The variation of p300 amplitude in a money-winning paradigm in children. Psychophysiology 18, Jasper, H., The ten twenty electrode system of the international federation. Electroencephalogr. Clin. Neurophysiol. 10, Johnson, R.E., A triarchic model of P300 amplitude. Psychophysiology 23, Johnston, V.S., Miller, D.R., Burleson, M.H., Multiple P3s to emotional stimuli and their theoretical significance. Psychophysiology 23, Junghöfer, M., Bradley, M.M., Elbert, T.R., Lang, P.J., Fleeting images: a new look at early emotion discrimination. Psychophysiology 38, Klorman, R., Weerts, T.C., Hastings, J.E., Melamed, B.G., Lang, P.J., Psychometric description of some specific-fear questionnaires. Behav. Ther. 5, Lang, P.J., Bradley, M.M., Cuthbert, B.N., International Affective Picture System (IAPS). Technical Manual and Affective Ratings. The Center for Research in Psychophysiology, University of Florida, Gainsville, FL. Lang, P.J., Davis, M., Öhman, A., Fear and anxiety: animal models and human cognitive psychophysiology. J. Affect. Disord. 61, Laux, L., Glanzmann, P., Schaffner, P., Spielberger, C.D., Das State- Trait-Angstinventar. Theoretische Grundlagen und Handanweisung. Beltz Testgesellschaft, Weinheim. Lavy, E., Van den Hout, M., Arntz, A., Attentional bias and facilitated escape: a pictorial test. Adv. Behav. Res. Ther. 15, LeDoux, J., The Emotional Brain. The Mysterious Underpinnings of Emotional Life. Simon and Schuster, New York. Logan, A.C., Goetsch, V.L., Attention to external threat cues in anxiety states. Clin. Psychol. Rev. 13, MacLeod, C., Clinical anxiety and the selective encoding of threatening information. Int. Rev. Psychiatry 3, MacLeod, C., Anxiety and anxiety disorders. In: Dalgleish, T., Power, M.J. (Eds.), Handbook of Cognition and Emotion. John Wiley Sons Ltd., Chichester, England, pp MacLeod, C., Mathews, A., Biased cognitive operations in anxiety: accessibility of information or assignment of processing priorities. Behav. Res. Ther. 29, MacLeod, C., Mathews, A., Tata, P., Attentional bias in emotional disorders. J. Abnorm. Psychology 95, Marks, J.M., Fear and Phobias. Basic Books, New York. Mathews, A., MacLeod, C., Discrimination of threat cues without awareness in anxiety-states. J. Abnorm. Psychology 95, Mathews, A., MacLeod, C., Induced processing biases have causal effects on anxiety. Cogn. Emot. 16, Mathews, A., Milroy, R., Processing of emotional meaning in anxiety. Cogn. Emot. 8, McCarthy, G., Wood, C.C., Scalp distributions event-related potentials: an ambiguity associated with analysis of variance models. Electroencephalogr. Clin. Neurophysiol. 62, McNally, R.J., Preparedness and phobias: a review. Psychol. Bull. 101,

10 52 W.H.R. Miltner et al. / International Journal of Psychophysiology 57 (2005) McNally, R.J., Cognitive bias in the anxiety disorders. Neb. Symp. Motiv. 43, McNally, R.J., Hornig, C.D., Hoffman, E.C., Han, E.M., Anxiety sensitivity and cognitive biases for threat. Behav. Ther. 30, Miltner, W.H.R., Krieschel, S., Hecht, H., Trippe, R., Weiss, T., Attention bias in phobics: cortical and behavioral correlates. Psychophysiology 39, S60. Miltner, W.H.R., Krieschel, S., Hecht, H., Trippe, R., Weiss, T., Eyemovements and behavioral responses to threatening and non-threatening stimuli during visual search in phobic and non-phobic subjects. Emotion 4, Naumann, E., Maier, S., Diedrich, O., Becker, G., Bartussek, D., Structural, semantic, and emotion-focused processing of neutral and negative nouns: event-related potential correlates. J. Psychophysiol. 11, Öhman, A., Mineka, S., Fears, phobias, and preparedness: toward an evolved module of fear and fear learning. Psychol. Rev. 108, Öhman, A., Soares, J.F., On the automatic nature of phobic fear: conditioned electrodermal responses to masked fear-relevant stimuli. J. Abnorm. Psychology 102, Öhman, A., Soares, J.F., Emotional conditioning to masked stimuli: expectancies for aversive outcomes following nonrecognized fearrelevant stimuli. J. Exp. Psychol. Gen. 127, Öhman, A., Esteves, F., Soares, J., Preparedness and preattentive associative learning electrodermal conditioning to masked stimuli. J. Psychophysiol. 9, Palomba, D., Angrilli, A., Mini, A., Visual evoked potentials, heart rate responses and memory to emotional pictorial stimuli. Int. J. Psychophysiol. 27, Pauli, P., Dengler, W., Wiedemann, G., Montoya, P., Birbaumer, N., Flor, H., Buchkremer, G., Behavioral and neurophysiological evidence for altered processing of anxiety-related words in panic disorder. J. Abnorm. Psychology 106, Schupp, H.T., Cuthbert, B.N., Bradley, M.M., Cacioppo, J.T., Ito, T., Lang, P.J., Affective picture processing: the late positive potential is modulated by motivational relevance. Psychophysiology 37, Seligman, M.E.P., Phobias and preparedness. Behav. Ther. 2, Streblow, H., Hoffmann, J., Kasielke, E., Experimentalpsychologische analyse von Gedaechtnisprozessen bei Phobikern experimental psychological analysis of memory processes in phobics. Z. Psychol. 193, Williams, J.M.G., Mathews, A., MacLeod, C., The emotional stroop task and psychopathology. Psychol. Bull. 120, Williams, J.M.G., Watts, F.N., MacLeod, C., Mathews, A., Cognitive Psychology and Emotional Disorders. Wiley, Chichester.