Preattentive bias for snake words in snake phobia?

Behaviour Research and Therapy 42 (2004) 949–970 www.elsevier.com/locate/brat

Preattentive bias for snake words in snake phobia? Jenny Wikstro¨m a,
, Lars-Gunnar Lundh b, Joakim Westerlund a, Lennart Ho¨gman a a

Department of Psychology, Stockholm University, Stockholm, Sweden b Department of Psychology, University of Lund, Lund, Sweden

Received 26 October 2002; received in revised form ; accepted 23 July 2003

Abstract Stroop interference and skin conductance responses (SCRs) for words related to snakes, spiders, flowers, and mushrooms were studied in a group of women (n ¼ 40) with snake phobia who were randomised to a stress or no-stress condition. The 21 low-stress snake phobics showed Stroop interference for unmasked (but not for masked) snake words, compared with 21 age- and sex-matched controls. Stroop interference was not significantly different between high-stress and low-stress snake phobics. No support for stronger SCRs for masked snake words was found in snake phobics in a lexical decision task with masked presentations of the same words. The lack of a masked Stroop interference in snake phobics suggests a possible difference in cognitive–emotional mechanisms underlying specific phobia vs. other anxiety disorders that deserves further investigation. # 2003 Elsevier Ltd. All rights reserved. Keywords: Preattentive bias; Emotional Stroop task; Specific phobia; SCR; Masked words

1. Introduction Several lines of research exist concerning preattentive mechanisms involved in anxiety and anxiety disorders. Cognitive approaches in experimental psychopathology have mostly relied on indirect behavioural measures of information-processing as for example reaction time, whereas behaviourally oriented theorists have focused on measures of physiological arousal like skin ¨ hman, 1997; O ¨ hman & Soares, 1994). This study aimed at combining conductance responses (O


Corresponding author. Department of Clinical Neurosciences, Karolinska Institute, Psychology Section, Karolinska Hospital, SE-171 76 Stockholm, Sweden; Tel.: +46-8-517 750 740, fax: +46-8-30 72 98. E-mail address: [email protected] (J. Wikstro¨m). 0005-7967/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2003.07.002

950

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

these two approaches by studying information processing, as well as physical arousal for masked threat stimuli in individuals suffering from snake phobia. Theories propose that attentional, as well as preattentive biases for threatening information are contributing to and/or maintaining a range of anxiety disorders (Williams, Watts, MacLeod, & Mathews, 1997). One common method in this area is the emotional Stroop paradigm (for an overview, see Williams, Mathews, & MacLeod, 1996; Williams, Watts, MacLeod & Mathews, 1997). In the computer-based emotional Stroop task, the participant is required to report the colour of a word that appears on a computer screen as fast as possible, while ignoring the meaning of the word. Words with threatening content have been shown to increase colour-naming latencies in anxious individuals and this so called ‘Stroop interference’ is presumed to reflect an automatic tendency to attend to (or an inability to let go of) threatening information. Preattentive Stroop interference involving masked words has been shown in clinical anxiety patients with different diagnoses (Bradley, Mogg, Millar, & White, 1995; Harvey, Bryant, & Rapee, 1996; ¨ st, 1999; Mogg, Bradley, Millar, & White, 1995; Mogg, Lundh, Wikstro¨m, Westerlund, & O Bradley, Williams, & Mathews, 1993), as well as having associations with trait anxiety in nonclinical samples (MacLeod and Hagan, 1992; MacLeod and Rutherford, 1992; Mogg, Kentish, & Bradley, 1993; van Honk, Tuiten, van den Hout, Putman, de Haan & Stam, 2001). With regard to specific phobias, to our knowledge only one study has so far found Stroop interference for masked threat words. This was found with regard to spider words in spider phobics (van den Hout, Tenney, Huygens, & de Jong, 1997), whereas negative results have also been reported in one study (Thorpe & Salkovskis, 1997). Non-masked Stroop interference, on the other hand, has reliably been shown in spider phobic adults (Kindt & Brosschot, 1997; Kindt & Brosschot, 1998; Thorpe & Salkovskis, 1997), as well as in children suffering from spider phobia (Kindt, Bierman, & Brosschot, 1997; Kindt & Brosschot, 1999), although children’s Stroop interference failed to differentiate between phobic and non-phobic controls in these studies. In summary, the evidence of Stroop interference involving individuals suffering from specific phobia using unmasked presentation of words seems consistent, while the results with regard to masked Stroop interference for threat words in this population are still inconclusive and need further investigation. 1.1. Skin conductance responses to masked stimuli In another line of research, threatening pictorial stimuli that were shown at a preattentive level of awareness have been shown to elicit skin conductance responses in spider- and snake¨ hman and Soares, 1994; but see also Mayer, Merckelbach, & de Jong, 1999, fearful students (O ¨ hman, 1999). This preconscious fear response is thought of as genetically with comments by O ‘prepared’ through evolution in a way that enables fear conditioning for certain classes of aversive stimuli (angry faces, snakes, spiders) even in the absence of conscious awareness of the stimuli and being advantageous for surviving in life-threatening situations (for a review, see ¨ hman, 1997). O ¨ hman’s model concerning fear and anxiety implies that only a limited range of O threatening stimuli may be physiologically arousing when exposed at a level below conscious awareness. Because it can be argued that reading is a learned skill entering rather late in human development (both ontogenetically and phylogenetically), it might be expected that threatening linguistic stimuli are less likely to evoke autonomic responses at a preattentive level.

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

951

In opposition to this prediction, however, van den Hout, de Jong and Kindt (2000) found increased SCRs in spider phobics for masked spider words, as well as masked general threat words as compared to equivalent neutral stimuli. These authors interpret their finding as an ¨ hman’s model (O ¨ hman, 1993) indicating that conscious awareness is not required anomaly for O in order for threat words to produce physiological arousal and that the effect is a result of preattentive analysis of emotional valence. Some researchers have proposed that results from masked Stroop studies may reflect ‘a relatively superficial level of semantic analysis, for example, along a basic positive–negative valence dimension’ that interferes with reaction latencies (Mogg and Bradley, 1998: 826). Whether this ‘superficial semantic analysis’ of masked word stimuli in the emotional Stroop task actually is associated with physiological arousal as van den Hout et al. (2000) suggests, or reflects a strictly cognitive bias is not yet clear since these variables are seldom studied at the same time. A few studies, however, point towards Stroop interference for linguistic stimuli as a measure of cognitive emotional bias that seems to be unrelated to physical arousal or neuroendocrine changes both at conscious and preconscious levels (Chen, Lewin and Craske, 1996; van Honk, Tuiten, van den Hout et al., 2000). Using verbal equivalents to the pictorial stimulus categories (i.e. snakes, spiders, flowers, and ¨ hman and Soares (1994), one purpose of the present study was to invesmushrooms) used by O tigate masked Stroop interference in snake phobics, as well as physiological arousal due to subliminally presented snake words as measured by SCRs in an immediately following lexical decision task. This task involved masked exposure of the same categories of words as in the Stroop task and was supposed to reveal possible associations between Stroop interference and increased SCRs for masked snake words in snake phobics. 1.2. State-anxiety levels In two studies the induction of stress has been found to cause increased Stroop interference for spider words in spider fearful individuals (Chen, Lewin & Craske, 1996; Kindt & Brosschot, 1998). These findings may be contrasted with two other Stroop studies where increased stateanxiety resulted in a reversed Stroop effect (a ‘suppression’ effect) in spider fearful students (Mathews & Sebastian, 1993) and in social phobics (Amir et al., 1996). A second aim of the present study was therefore to investigate the effect of increased state anxiety on Stroop interference for snake words in snake phobics. This was done by randomly assigning half of the snake phobics to a high-stress condition, where they were informed before the Stroop task that they were going to do a behavioural approach test, approaching a real snake immediately afterwards, and the other half to a ‘low stress condition’ where no such information was given before the Stroop task. 1.3. Priming effects on the lexical decision task In two previous studies (Lundh, Wikstro¨m, & Westerlund, 2001; Lundh, Wikstro¨m, West¨ st, 1999) priming of unmasked words, but not masked words, was shown to improve erlund & O performance on a subsequent tachistoscopic identification task, where word stimuli were presented at the threshold for awareness. This effect was independent of word category, but is of theoretical interest because it suggests the possibility that the threshold for awareness may

952

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

fluctuate as the result of priming at the supraliminal level. An interesting question is if the same kind of effect can be observed in the present study when word stimuli in an awareness check (a lexical decision task) are presented subliminally (i.e. at an exposure 17 ms below the threshold for awareness). If this is the case, the participants should perform better on ‘subliminal’ words previously shown unmasked than ‘subliminal’ words that were shown masked during the Stroop task. 1.4. Summary In all, the following hypotheses were stated: (1) Snake phobics would show Stroop interference for snake words as compared with the control group, both supraliminally and subliminally, and these Stroop interference effects should be specific for snake words; (2) the snake phobics would show stronger SCRs for masked snake words selectively, while a control group would show no such reaction in a subsequent lexical decision task; and (3) the participants would show a priming effect (i.e. better performance) on the lexical decision task for words that were unmasked during the Stroop task as compared to masked words. Due to the contradictory results (increased interference vs. ‘suppression’) of induced stress in earlier studies no hypothesis was stated concerning the induction of stress in the present study.

2. Method 2.1. Participants Fifty-two women were recruited to a one-session treatment study of snake phobia by advertisement in a daily newspaper. As a part of this project, they were interviewed by two research assistants using the Anxiety Disorders Interview Schedule, ADIS-IV (Brown, Di Nardo & Barlow, 1994) and diagnosed as snake phobics according to the criteria for snake phobia defined in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV, American Psychiatric Association, 1994). Twelve participants were excluded for various reasons (three because of concurrent spider fear, i.e. self-rated spider fear above three on a 0–8 rating scale, four due to dyslexia, four due to technical failure with the apparatus, and one due to more than 5% errors in colour naming responses). This resulted in a group of 40 snake phobics with normal colour vision and normal reading abilities, who scored a mean of 25.3 (SD ¼ 2:86) on the Snake Phobia Questionnaire (SNAQ) (Klorman, Weerts, Hastings, Melamed, & Lang, 1974). Twenty-one patients were randomised to the low-stress condition and 19 patients were randomised to a high-stress condition. These two subgroups did not differ significantly on SNAQ scores, mean age or the trait version of the State-Trait Anxiety Inventory (STAI-T). The 21 patients in the low-stress group were compared with 21 non-phobic control women matched for age (2 years), with normal colour vision and reading abilities. Exclusion criteria for controls were self-rated fear of spiders or snakes above three on a 0–8 scale. Three individuals were excluded due to snake or spider fear, and two due to colour naming errors above 5%. No individual had colour naming responses with an error rate above 5%. Controls were mainly psychology students, but some were also recruited through advertisement in central Stockholm

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

953

Table 1 Mean age, trait anxiety level, and state mood levels in snake phobic patients and controls. State mood is measured on a Visual Analogue Scale 0–100

Age STAIT-T VAS worried/ nerv. VAS sad VAS happy VAS annoyed a

Controls

Snake phobic patients

Low stress (n ¼ 21)

Low stress (n ¼ 21)

High stress (n ¼ 19)

M

SD

M

SD

M

SD

33.95 37.35a 19.62

11.17 8.68 16.54

34.19 40.33 33.71

10.77 9.78 25.04

32.26 37.58 62.63

9.31 9.74 26.35

8.76 59.43 5.42

10.72 19.56 8.67

17.67 50.43 13.14

17.82 21.28 20.03

20.42 46.05 16.21

24.44 29.81 23.21

Data missing from one participant (n ¼ 20).

and through personal contacts. All controls received payment (approximately 24 US$) or course credit for their participation. The mean age, trait anxiety, and the four state variables measured by Visual Analogue Scales (VAS) for the three groups are shown in Table 1. The control group and the low-stress phobic group were similar with regard to age, trait anxiety and VAS Happy and VAS Annoyed, but differed significantly on VAS Worried, tð40Þ ¼ 2:15; p < 0:05 and almost significantly on VAS Sad, tð40Þ ¼ 1:96; p < 0:06. 2.2. Psychometric measures 2.2.1. Trait version of Spielberger’s State-Trait Anxiety Inventory (STAI-T) This instrument contains 20 items that measure enduring symptoms of anxiety (Spielberger, Gorsuch, & Lushene, 1970). 2.2.2. Snake Phobia Questionnaire This questionnaire consists of 30 statements of self-reported fear toward snakes that have to be answered with ‘true’ or ‘false’. A total score (range 0 to 30) is calculated by summing all ‘true’ statements and serves as a measure of the degree of phobic fear (Klorman, Weerts, Hastings, Melamed & Lang, 1974). 2.2.3. Visual Analogue Scales All individuals were asked immediately before the Stroop task (directly after the mood induction manipulation) to fill out four visual analogue scales (’Worried/nervous’, ‘Happy’, ‘Sad’, ‘Annoyed’) in answer to the question: ‘How do you feel right now?’ The present degree of each specific mood state was indicated by a mark on a 100 mm long line by the participant with the left endpoint labelled as ‘not at all’ and the right endpoint labelled ‘maximum’.

954

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

2.3. Stimulus material In the conscious identification task, eight words rated as neutral in pilot tests were used: tva¨ttmedel (washing powder), cirkel (circle), definition (definition), frima¨rke (postage stamp), kalejdoskop (kaleidoscope), replik (reply), linjal (ruler), pendel (pendulum). Two additional words were included in the beginning of the test: ‘test’ (test) and ‘molekyl’ (molecule), serving as practice targets. Forty-eight words were used as target words in the Stroop and lexical decision tasks: 12 mushroom words, 12 flower words, 12 snake words and 12 spider words. A behaviour therapist with extensive experience in the treatment of specific phobias rated the words on their presumed threat-value for people suffering from snake phobia and spider phobia, respectively, in order to make the snake words and spider words equally threatening. Flower and mushroom words rated as neutral were matched to snake and spider words regarding relative frequency1 (Spra˚kbanken, 1999) and syllable length, and finally all words were divided into two equivalent word sets A and B (see Appendix). The general words denoting the categories themselves, that is, ‘ormar’ (snakes), ‘spindel’ (spider), ‘blomma’ (flower) and ‘svampar’ (mushrooms) were included in both word set A and B due to difficulties in finding matching equivalents. These words occurred alone or in compound words three times in each category in each word set. Fortyeight computer-generated non-word letter sequences were constructed in order to obtain the same word length as the corresponding target words. Four neutral words and four non-words were also used in a practice session prior to the lexical decision task. 2.4. Apparatus The stimuli were presented on a 17’’ NOKIA 447XS colour monitor (Espoo, Finland) attached to an AST Bravo MS/100 computer (AST Research, Irvine, CA). In order to register the reaction times in the Stroop task a voice-activated Hama condensator microphone (EM-01; HAMA, Manheim, Germany) was connected to the computer via a PST Serial response box (type #200; Psychology Software Tools, Pittsburgh, PA) and a combined mixer and compressor Fostex MN-50 (Fostex Corp., Tokyo, Japan). SCR was measured by an apparatus made by Biopac Systems Inc. (model mp 100A) linked to a Power Macintosh 7200/90 computer by means of computer software AcqKnowledge III for MP100WS (version 3.2). Electrodes of the EL204S Ag-AgCl type were used with isotonic paste as electrolyte (0.5% NaCl/100ml H2O). In order to obtain a starting point for each 5 s long SCR measurement, the AST Bravo computer was connected to the SCR apparatus. 2.4.1. Software The test was written in Microsoft Quick Basic and showed the words in SCREEN 12, with a resolution of 60 rows and 80 columns, and a viewable image size of 15.9’’. Stimulus presentations were synchronised with the vertical retrace of the monitor and the words were shown 1

‘Relative frequency’ is the mean estimate of a word’s occurrence if each source material (newspapers 1965; 1976; 1987; 1995–1998; novels 1976–1977 and 1980–81) would have consisted of one million words.

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

955

in lower-case letters 3.7 mm wide, on a black screen, within a fixation box consisting of fine white lines 9:5  6:5 cm, at a central screen location. 2.5. Procedure Both controls and patients filled in STAI-T questionnaires and then underwent an ADIS interview regarding fears and psychological health status; the controls, however, received a short version of the ADIS, whereas the patients received the full version. All participants were tested individually in accordance with a written manual. Patients were randomised to either a high-stress condition or a low-stress condition. In the high-stress condition the patients were informed of their participation in a behavioural approach test that was to take place immediately after the computer-based Stroop session in an adjacent room. In the low-stress condition this information was given after the Stroop task. Immediately before the test-session started, all participants filled in the Visual Analogue Scale of current mood. The program consisted of three parts: (1) a conscious identification task; (2) an emotional Stroop task; and (3) a lexical decision task, presented in the same order for each participant. The participants were seated about 100 cm from the computer monitor, with a 40 W desk lamp as the only light, turned against the wall away from the computer screen. 2.5.1. The conscious identification task The procedure in the conscious identification task was similar to the one in two previous ¨ st, 1999), studies (Lundh, Wikstro¨m & Westerlund, 2001; Lundh, Wikstro¨m, Westerlund & O although a forward mask was added in addition to the backward mask in order to improve the effectiveness of the masking procedure (Wikstro¨m, Lundh, & Westerlund, in press). Moreover, all words and masks were shown in yellow in order to obtain a strict criterion for awareness, because separate tests of the four Stroop colours had shown a non-significant advantage in visibility for yellow targets words (Wikstro¨m et al., in press). The two first target words served as practice trials and were not included in the analysis, but the remaining eight neutral words were randomised for each participant. The participants were instructed to try to identify the word even though it was presented very briefly at first. If they were unable to do this, the word was presented successively for longer and longer durations in steps of 17 ms until they were able to identify the word. The participants were not encouraged to make guesses, but neither were they explicitly discouraged from guessing. The conscious identification task consisted of 10 neutral words in all, ‘sandwiched’ between a forward and a backward mask consisting of a row of ‘X’s with the same word length as the target word, presented in the middle of a fixation box previously described in the software section. The first trial of a specific word consisted of (1) a forward mask for 17 ms; (2) a target word for 17 ms;2 and (3) a backward mask that remained on screen until the experimenter registered the response. The inter-trial interval was 2 s. If the participant reported not seeing the word or made an incorrect response, the experimenter registered this on the keyboard, whereupon the same word was shown again with an exposure 17 ms longer than the preceding trial (everything else being equal). This procedure was repeated until the experimenter registered a correct response by tapping the return key, initiating the first presen2

17 ms was the briefest presentation possible on the present equipment.

956

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

tation of the next word. The exposure duration for the masked stimuli in the following Stroop task and lexical decision task was automatically set 17 ms below the briefest duration at which the individual could identify any single word during the conscious identification task, following ¨ st, 1999, 2001). The exposure durthe same procedure as in Lundh, Wikstro¨m, Westerlund & O ation was measured from the onset of the target word to the onset of the backward mask. 2.5.2. The emotional Stroop task The emotional Stroop task was described as a kind of reaction test, where the task was to name the colours as fast as possible, regardless of the nature of the target. Testing began with a short practice session, in which the participants were introduced to the handheld microphone and were encouraged to practice on the colour naming task until they felt comfortable with the procedure. The stimuli in the practice session consisted of the words ‘yellow’, ‘blue’, ‘green’ and ‘red’ presented in the middle of a fixation box. The emotional Stroop task comprised two exposure conditions (masked and unmasked) and four word categories as described above. Word sets A and B were used in the unmasked and masked conditions alternately. Both word sets consisted of six words from each of the four categories spiders, snakes, flowers and mushrooms. The exposure duration for the masked condition was set 17 ms below the identification threshold established in the conscious identification task and the masking procedure was also similar to the one described above comprising a forward, as well as a backward mask in the following sequence: (1) a forward mask (17 ms); (2) a target word (duration determined by the conscious identification task); (3) a backward mask (displayed until response). The inter-trial interval was 2 s. The word stimuli were shown in either green, red, blue, or yellow, and the masks in each specific trial were shown in the same colour as the specific stimulus word. Exposure conditions, colours and word categories were completely randomised across the 192 trials for each individual. Colour naming latencies were recorded by the computer starting with the onset of the target word until the voice key detected the participant’s verbal response. The backward mask (masked condition) or the word (unmasked condition) remained on the screen until the response was registered by the software. 2.5.3. The lexical decision task All 48 words and 48 non-words from the Stroop task were included in the lexical decision task and presented in randomised fashion for each participant. As in the conscious identification task all masks and stimulus words were presented in yellow. The participants were instructed to respond verbally for each target and determine if the target was a Swedish word or a non-word. ‘Yes’ indicated a word and ‘No’ indicated a non-word. The participants were informed that an equal number of words and non-words were included in the task. A practice session with eight words and eight non-words was inserted prior to the main test. In order to obtain a starting point for each SCR measurement period, the computer was connected to the SCR apparatus sending a signal precisely one refresh cycle after the onset of each backward mask following each trial. Skin conductance responses for each stimulus was measured for a period of 5 s after this starting point. Pilot testing indicated that the participants had difficulties postponing their response for 5 s, and they were therefore allowed to respond at any point in time they found natural. The

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

957

experimenter recorded each response on the numeric keyboard not earlier than 5 s after the onset of the backward mask.

3. Results 3.1. Conscious identification task In the phobic group, 35% or 14 of the individuals could identify at least one word at an exposure duration of 51 ms, 30% (12) had an identification threshold of 68 ms, 20% (8) had 85 ms, 2.5% (1) had 102 ms, and the remaining 12.5% (5) could identify at least one word at 119 ms. In the control group, about 48% or 10 individuals had an identification threshold of 51 ms, 33% (7) identified a word at 68 ms, 14% (3) at 85 ms and 5% (1) at 102 ms. The mean duration of the masked presentation was 55.7 ms (SD 15.1) for the phobic group and 47.0 ms (SD 22.4) for the controls. This difference was not statistically significant (p ¼ 0:115). 3.2. Stroop interference effects The mean colour-naming latencies calculated from all the participants’ median values for each word type and presentation condition are shown in Table 2. Due to technical failure one participant in the control group had to be excluded, and the matched participant in the Table 2 Means and standard deviations for colour naming latencies and Stroop Interference for spider words and snake words in ms for snake-phobics in both stress conditions and non-phobic controls Controls

Snake phobic patients

Low-stress (n ¼ 20)a

Low-stress (n ¼ 21)

High-stress (n ¼ 19)

M

SD

M

SD

M

SD

Supraliminal Spider Snake Flower Mushroom Spider-interf. Snake-interf.

649.3 646.0 638.1 634.8 12.9 9.5

100.8 106.6 94.6 91.2 39.2 43.2

755.9 878.6 739.8 747.2 14.5 135.2

140.7 184.4 114.2 115.0 55.1 107.5

744.2 849.8 701.5 719.5 33.7 139.3

142.1 170.6 117.5 138.5 67.6 100.3

Subliminal Spider Snake Flower Mushroom Spider-interf. Snake-interf.

566.1 567.5 564.3 570.8 1.4 0.0

81.7 75.0 79.9 77.3 37.6 31.7

612.4 611.3 610.2 614.0 1.5 2.6

71.3 72.4 72.6 76.0 35.5 34.2

601.1 613.0 613.8 604.2 7.9 3.9

77.1 89.0 88.3 96.3 40.1 25.5

a

Data missing from one participant due to technical problems.

958

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

low-stress phobic group was therefore also excluded. The number of participants included in the control vs. phobic comparisons was consequently 20 individuals in each group. To test the hypothesis of a Stroop interference effect for snake words relative to spider words specifically in the phobic group, a Stroop interference index was calculated for each individual by subtracting the mean of the two median colour naming latencies for flower words and mushroom words (neutral baseline) from the median colour naming latencies of spider words and snake words respectively, in each exposure condition separately. The interference indices for unmasked snake words and unmasked spider words were significantly correlated, rð60Þ ¼ 0:37; p ¼ 0:004, but these indices showed no significant correlations with Stroop interference for masked snake or spider words, and neither did the two latter (all r < 0:19). The SNAQ scores among the phobic patients did not correlate significantly with the unmasked snake interference, rð40Þ ¼ 0:26; p ¼ 0:11, or with the masked snake words, rð40Þ ¼ 0:04; p ¼ 0:80. No linear or curvilinear relationships were found between trait anxiety and Stroop interference indices in any group. 3.2.1. Effects of snake phobia A mixed analysis of variance (ANOVA) was first conducted for the unmasked exposure condition with group (2: low-stress snake phobics; controls) as between subjects-variable and interference type (2: spider, snake) as within-subjects variable. The results showed a significant main effect of group, F ð1; 38Þ ¼ 12:81; p < 0:01, with snake phobics showing greater Stroop interference in general than controls, and a main effect of interference type, F ð1; 38Þ ¼ 25:65; p < :001, which indicates that all participants showed a greater Stroop interference for snake words. More importantly, there was also a significant interaction between interference type and group, F ð1; 38Þ ¼ 28:67; p < 0:001. In order to explore this interaction we analysed simple main effects of interference type showing a significant difference between interference type in the low-stressed phobics, 135 ms for snake words vs. 14.5 ms for spider words, tð19Þ ¼ 5:98; p < 0:001, but not in the control group (p ¼ 0:77). A similar ANOVA was then performed for the masked condition, but showed no significant main effect for group, interference type, or interaction effect involving interference type and group (all Fð1; 38Þ < 1). Thus hypothesis 1 was confirmed only with regard to unmasked words: low-stress snake phobic individuals showed a Stroop interference effect for snake words selectively as compared to the control group in the supraliminal condition, but failed to show any effects in the subliminal condition. 3.2.2. Effects of the stress induction Twenty-one snake phobics in the low-stress group were compared with 19 snake phobics in the high-stress condition. The efficacy of the mood induction manipulation was confirmed by the fact that the high-stress phobics scored significantly higher on the VAS scale ‘worry/nervousness’, tð38Þ ¼ 3:56; p ¼ 0:001. This indicates that the information prior to the Stroop task that a behavioural approach test would follow immediately after the test session was effective in increasing state anxiety in the high-stress phobic group, even above the elevated stress level that was shown by the low-stress phobics as compared to the normal controls. To explore differences in snake-phobics’ Stroop interference for snake words as a function of current stress levels a mixed analysis of variance was conducted with group (2: high-stress

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

959

phobics; low-stress phobics) as between-subjects variable, and interference type (2: snake; spider) as within-subjects variable in the masked and the unmasked conditions separately. The analysis in the unmasked condition resulted in a significant main effect of interference type, F ð1; 38Þ ¼ 57:89; p < 0:001, although no main effect or interaction involving stress-level approached significance (all F ð1; 38Þ < 1). A similar ANOVA for the masked condition showed no significant main effects or interactions of neither interference type, nor stress level, (all F ð1; 38Þ < 1). This means that no empirical support was found for increased stress as a mediator in Stroop interference for snake words in snake phobics.

3.3. Skin conductance responses in the lexical decision task SCRs were measured by the apparatus during a time period of 5.0 s starting one refresh cycle after the onset of each backward mask following each target stimulus for each individual. For each 5 s measurement period electrodermal activity was measured by mean skin conductance level, and as measures of skin conductance variability served standard deviation and range (defined as the distance between the lowest and the highest skin conductance level during this period). Standard deviation and range may be closer to the more common SCR magnitude (mean amplitude of the event related SCRs computed across all stimulus presentations, including those with a null-response according to criteria specified beforehand, see Dawson, Schell & Filion, 1990: 304) than the former mean variable. For each participant mean values of these three measures (mean; range; standard deviation) were calculated for 12 words of each category (mushroom, flower, snake, spider) and 48 non-words. Since the range variables and standard deviation variables for controls and phobics were highly correlated within each word type (spider, snake, flower, mushroom and non-words), r(61) ranging from 0.996 to 0.997, all p < 0:001, only the former variable (range) was retained in the following analyses. Consistent with previous research regarding correlations among EDA components (Dawson, Schell & Filion, 1990), range values were only moderately correlated with the mean SCR levels for all word types rð61Þ ¼ 0:63 to 0.79, all p<.001 (with the exception of mushroom words, rð61Þ ¼ 0:33; p < 0:05) indicating that these two variables (mean and range) may reflect partly different processes. Group means for mean and range values of SCRs for each word type are shown in Table 3. To explore differences in SCRs for different word types as well as non-words, mixed ANOVAs were calculated with group (3: controls; low-stress snake phobics, high-stress snake phobics) as between-subjects variable and word type (5: spider; snake; flower; mushroom; nonwords) as within-subjects variable, first using the mean, and then the range, as dependent variables. With regard to the mean level of skin conductance, the analysis revealed no significant effects for group, F ð2; 58Þ ¼ 1:73; p ¼ 0:19, or word type, F ð4; 232Þ ¼ 2:0; p ¼ 0:10, and no significant interaction between word type and group, F ð8; 232Þ < 1. Similarly, with regard to SCR range values, there was no significant effect for group, F ð2; 58Þ < 1, or word type, F ð4; 232Þ ¼ 1:57; p ¼ 0:18, and no significant interaction between word type and group, F ð8; 232Þ < 1. These results do not support the existence of differences in SCRs between groups due to word type.

Spider Snake Flower Mushroom Non-words

Word type

0.298 0.299 0.298 0.296 0.298

(0.004) (0.005) (0.004) (0.003) (0.003)

0.019 0.021 0.018 0.016 0.019

(0.018) (0.020) (0.016) (0.015) (0.017)

SD 0.299 0.300 0.301 0.299 0.299

M (0.006) (0.011) (0.009) (0.007) (0.004)

SD 0.022 0.026 0.025 0.024 0.025

M (0.027) (0.037) (0.035) (0.030) (0.034)

SD

M

M

SD

Mean

Range

Mean

Range

High-stress snake phobic (n ¼ 19)

Low-stress snake phobic (n ¼ 21)

0.297 0.298 0.298 0.295 0.298

M

Mean

(0.007) (0.005) (0.006) (0.006) (0.003)

SD

Control (n ¼ 21)

0.022 0.025 0.024 0.024 0.026

M

Range

(0.020) (0.021) (0.022) (0.020) (0.021)

SD

Table 3 SCRs: mean level and mean range values (in microSiemens) for control group and low-stress/high-stress snake phobics over 5 s after offset of each specific stimulus during the lexical-decision task

960 J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

961

3.4. Performance and priming effects on the lexical decision task Fig. 1 summarizes the participants’ performance on the lexical decision task, for each word category. Two things may be noted in this figure. First, the participants tended to perform better than chance (i.e. 50% correct) on both words and non-words; this raises the question of whether the words and non-words during the lexical decision task were truly ‘subliminal’. Second, the participants performed better on words that they had already seen unmasked during the Stroop task, than on masked words that they had been exposed to during the Stroop task, and than unprimed non-words; further, this improved performance is particularly apparent among the phobic patients. This suggests that words that were seen unmasked during the Stroop task causes a priming effect, in the sense of lowering the threshold for word/non-word discrimination of these words. 3.4.1. Were the masked words ‘subliminal’? In research with masked stimuli, it is common to distinguish between a subjective and an objective threshold (Merikle, 1992). The term ‘objective threshold’ refers to an exposure duration below which individuals do not perform better than chance on forced discrimination tasks (as the lexical decision task), and is different from the ‘subjective threshold’, defined by Merikle, Smilek and Eastwood (2001) as the exposure duration below which individuals report an inability to consciously identify the stimulus (as in the conscious identification task in the present study). To test if the words during the lexical decision task were exposed under the ‘objective threshold’, two-high threshold analysis (Snodgrass & Corwin, 1988) was used, separately for the words that had been unmasked and masked during the Stroop task. A sensitivity

Fig. 1. Proportion correct responses for each stimulus type on the lexical decision task for all individuals: controls (n ¼ 21), snake phobics in the low-stress condition (n ¼ 21), and snake phobics in the high-stress condition (n ¼ 19).

962

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

measure Pr that indicates the ability to discriminate words and non-words was computed, defined as Pr ¼ HIT  FA, where HIT ¼ the probability of a ‘word’ response when a word is presented, and FA ¼ the probability of a ‘word’ response when a non-word is presented. The discrimination ability with regard to unmasked Stroop words for all 61 participants was Pr ¼ 0:197, which was significantly better than chance, tð60Þ ¼ 7:37; p < 0:0001. The discrimination ability with regard to masked Stroop words was also significantly better than chance, Pr ¼ 0:134; tð60Þ ¼ 5:75; p < 0:0001. 3.4.2. Priming effects To test if the improved performance on unmasked vs. masked Stroop words was significant, and to explore the possibility of selective effects on lexical decision of snake words for snake phobics, a mixed ANOVA was conducted with group (3: low-stress snake phobics; high-stress snake phobics, controls) as between subjects-variable, and priming type (2: unmasked, masked) and word category (4: snake, spider, flower, mushroom) as within-subjects variables. This analysis showed a significant main effect of priming type, Fð1; 58Þ ¼ 12:22; p < 0:001, that is, the participants performed better on Stroop words previously shown in the unmasked condition (63.6% correct) than in the masked condition (57.3% correct). There was a trend towards a significant main effect for group, Fð1; 58Þ ¼ 2:74; p ¼ 0:07, and also a significant interaction effect between group and word category, Fð6; 174Þ ¼ 2:95; p < 0:01. Analysis of simple main effects showed that the groups differed significantly in their lexical decision performance on snake words, F ð2; 58Þ ¼ 5:59; p ¼ 0:006, and flower words, Fð2; 58Þ ¼ 3:37; p ¼ :041, but not on spider words, F ð2; 58Þ ¼ 0:98, or mushroom words, F ð2; 58Þ ¼ 0:68. Tukey post-hoc tests showed that the high-stressed snake phobics performed significantly better than the controls on both snake words and flower words. Analysis of simple main effects of word category showed that the high-stressed snake phobics performed differently on the four word categories, Fð3; 54Þ ¼ 4:28; p < 0:01. Tukey post-hoc tests showed that they performed significantly better on snake words (69.8% correct) and flower words (70.2% correct) than on mushroom words (55.7% correct). 3.5. The role of the exposure duration for the effects It should be noted that the participants were assigned individual exposure durations for all masked stimuli, both during the Stroop task and the lexical decision task, on the basis of their performance on the conscious identification task. This raises the question whether the results that have so far been reported (i.e. lexical decision performance above chance on masked words, priming effects on the lexical decision task, and the absence of Stroop interference and SCR effects of masked snake words in snake phobics) are independent of exposure duration. 3.5.1. The role of exposure duration for performance on the lexical decision task All participants were split into three groups, as a function of the exposure duration that they had been assigned on the basis of the conscious identification task: one group with an exposure duration of 34 ms (n ¼ 24), one with an exposure duration of 51 ms (n ¼ 19), and one with an exposure duration of 68 ms or more (n ¼ 18). The discrimination ability with regard to unmasked Stroop words was significantly better than chance for all three groups: Pr ¼ 0:056; tð23Þ ¼2.22, p<0.04; Pr=0.224, t(18)=5.77, p<0.0001; and Pr=0.358, t(17)=7.00,

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

963

p<0.0001; respectively. The discrimination ability with regard to masked Stroop words was significantly better than chance for the 51 ms-group, Pr ¼ 0:127; tð18Þ ¼ 3:54; p ¼ 0:002, and for the 68 ms-group, Pr ¼ 0:284; tð18Þ ¼ 6:22; p < 0:0001, but not with regard to the 34 ms-group, Pr ¼ 0:028; tð23Þ ¼ 1:33; p ¼ 0:196. This suggests that the only words that could be said to be exposed below the ‘objective threshold’ in the lexical decision task were the masked Stroop words that were exposed at an exposure duration of 34 ms. The same pattern of results were found when the control group was excluded from the analysis. 3.5.2. The role of exposure duration for the priming effects on the lexical decision task The main effect of priming type remained significant for the groups with an exposure duration of 51 ms or more; the 51-ms group performed significantly better on the supraliminally primed words (66.9% correct) than on the subliminally primed words (57.2% correct), F ð1; 18Þ ¼ 7:22; p < 0:015, and the same was true for the 68 or more ms-group (71.8 vs. 64.4% correct), F ð1; 17Þ ¼ 6:49; p < 0:022. However, the participants who had an exposure duration of 34 ms showed no significant effect (54.9 vs. 52.1% correct), F ð1; 23Þ ¼ 0:84. That is, those participants who had an exposure duration of 34 ms in the lexical decision task showed no evidence of improved performance on words that had been exposed supraliminally vs. subliminally during the Stroop task. Excluding the controls from the analysis did not alter the pattern of results. 3.5.3. The role of exposure duration for Stroop interference of masked snake words The above-mentioned results suggest that only in the 34-ms group were the masked words ‘subliminal’ in the sense of being exposed safely under what is generally referred to as the ‘objective threshold’. But does this mean that the participants who had a longer exposure duration than 34 ms may have been able to consciously read the masked words in the Stroop task? If so, the snake phobic participants with these exposure durations may be expected to show some degree of Stroop interference for masked snake words (because they showed a strong interference effect for unmasked snake words, which they could consciously read). In order to test this possibility, the results on the masked Stroop task was reanalysed for the 37 participants (13 low-stressed phobics, 13 high-stressed phobics, and 11 controls) who were assigned an exposure duration of 51 ms or more. In a mixed ANOVA with group (3: low-stress snake phobics; high-stress snake phobics, controls) as between subjects-variable, and Stroop interference (2: snake words, spider words) as within-subjects variable, however, no main effects or interaction effects even remotely approached significance (all p > 0:60).3 Interestingly, the 26 snake phobics with an exposure duration of 51 ms or more showed no Stroop interference at all for masked snake words (interference index ¼ 0:2 ms), which may be compared with the very substantial interference effect of 142.1 ms that these 26 snake phobics showed for unmasked snake words. 3

Due to the improved performance on snake words, as well as flower words during the lexical decision task the same analysis was performed with word type (4: spider; snake; flower; mushroom) as within-subjects variable, in order to guard against the possibility of using an inappropriate baseline in calculating the Stroop interference index. This ANOVA showed no main effect of word type, p ¼ 0:77, or interaction between group and word type, p ¼ 0:95, (Greenhouse-Geisser corrected df due to violation of the sphericity assumption). The main effect of group showed a weak trend toward significance, F ð1; 34Þ ¼ 2:17; p ¼ 0:13.

964

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

3.5.4. The role of exposure duration for the SCR effects of masked snake words The SCR results were also reanalysed separately for the 37 participants (13 low-stressed phobics, 13 high-stressed phobics, and 11 controls) who had an exposure duration of 51 ms or more. In a mixed ANOVA with group (3: low-stress snake phobics, high-stress snake phobics, controls) as between subjects-variable, and word type (5: spider, snake, flower, mushroom, nonwords) as within-subjects variable, and the mean SCR level as the dependent variable, there was a significant main effect of word type, F ð4; 136Þ ¼ 3:52; p ¼ 0:009, but no main effect of group, F ð2; 34Þ ¼ 1:35; p ¼ 0:23, and no significant interaction, F ð8; 136Þ ¼ 1:02; p ¼ 0:43. In a similar ANOVA with SCR range values as the dependent variable, there was also a significant main effect of word type, Fð4; 136Þ ¼ 2:96; p ¼ 0:022, but no main effect of group, F ð2; 34Þ ¼ 0:75; p ¼ 0:48, and no significant interaction, Fð8; 136Þ ¼ 0:58; p ¼ 0:79. Tukey post-hoc tests revealed that in this analysis, the participants showed higher mean skin conductance level to the snake words than to mushroom words, and an identical analysis regarding SCRs (range values) showed a similar trend at the border of being significant.

4. Discussion The present study investigates cognitive bias, as well as autonomic arousal for linguistic stimuli in snake phobics at different levels of awareness and state anxiety. The main findings are a strong Stroop interference for unmasked snake words in snake phobics, in combination with a complete absence of an equivalent cognitive bias in the masked word condition. No significant effects of the stress induction were found among the snake phobics. Interestingly, however, the participants showed a priming effect (i.e. better performance) on the lexical decision task for words that were unmasked during the Stroop task, as compared to masked words; that is, they were significantly better than chance in their lexical decisions of these words, although they were exposed ‘subliminally’. Finally, there were no significant differences in SCRs between snake phobics and controls, although all participants with an exposure duration of at least 51 ms (i.e. both snake phobics and controls) responded more to snake words than to mushroom words. These various aspects of the results are discussed in turn below. 4.1. Stroop effects on unmasked, but not on masked words The attentional bias at a conscious level corroborates previous findings in spider phobics (Kindt and Brosschot, 1997; Kindt and Brosschot, 1998; Lavy, van den Hout & Arntz, 1993; Thorpe & Salkovskis, 1997). Although the null results on masked words are consistent with the results in spider phobia found by Thorpe and Salkovskis (1997), they contradict the positive results reported by van den Hout et al. (1997). However, the report by van den Hout et al. showed a very small masked Stroop effect (much smaller than in the unmasked condition) that unexpectedly varied in direction (interference vs. facilitation) for high vs. low severity of phobia, and consequently should be almost absent had the sample not been divided (see van den Hout et al., Fig. 1, p. 31). Further, their design did not include a control group, and they did not report the results of the awareness checks, which makes the results even harder to interpret. A similar comparison between high vs. low degree of phobic severity was not done in the present

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

965

study due to the limited range of SNAQ scores. The range restriction in phobic severity in the present study, which involved phobics with a severe degree of phobia, might also explain the lack of a significant association between the unmasked Stroop interference and phobia severity scores of the kind that was obtained by van den Hout et al. in their study. Although preattentive bias for threat words previously has been demonstrated for other anxiety disorders, these disorders have mostly been accompanied by increased levels of trait anxiety (Bradley et al., 1995; Mogg, Bradley et al., 1993; Lundh et al., 1999; Mogg et al., 1995; van den Hout, Tenney, Huygens, Merckelbach, & Kindt, 1995), indicating that a broader construct of ‘negative affect’ may be associated with the masked Stroop interference for threat words. It may be speculated that worry in the form of rumination over threat-related concerns is needed to obtain preattentive Stroop interference for threat relevant words. Snake phobics, however, usually do not suffer from elevated trait anxiety, and if they tend to avoid everything that might elicit the conditioned fear response (including verbal rumination over the feared subject) very little internal priming of verbal constructs is likely to occur—an internal priming that may perhaps be a necessary prerequisite for masked Stroop interference to take place. Further research, however, is needed to clarify if cognitive biases evoked by masked words are less common in specific phobia than in anxiety disorders that are accompanied by high trait anxiety, such as GAD. The optimal design here would be to directly compare, in one study, patients with specific phobia and other anxiety disorders. For example, we do not know if the masked Stroop effect for threat-related words would materialise in GAD patients with the particular masking procedure used in the present study (i.e. with both a forward and a backward mask). On the other hand, it seems unlikely that the complete absence of preattentive bias is due to the particular methodology in the present study, because an anxiety-related bias for masked threat words has been found in a non-clinical sample with the same kind of procedure in a previous study (Wikstro¨m et al., in press). It could also be argued that the results with regard to the masked words in the present study ¨ hman’s model of preattentive bias for threat in specific phobia as being are consistent with O ¨ hman, 1997). Again, specific to a relatively limited range of pictorial, non-symbolic stimuli (O however, the experimental design used in the present study does not allow for any conclusion on this topic; the optimal design here would be to compare word stimuli and pictorial stimuli as part of the same study. 4.2. Effects of the stress induction Although the stress induction in the present study was associated with an increased state anxiety, it did not cause any significant effect on the Stroop task, neither in the form of a suppression effect (as was the case in some earlier studies concerning spider fear and social phobia; Amir et al., 1996; Mathews & Sebastian, 1993), nor in the form of an increased Stroop interference (as previously has been reported by Chen et al., 1996, and Kindt & Brosschot, 1998). These contradictory results are hard to explain and the matter needs further investigation. However, one possible explanation may be that even the phobics who did not undergo stress induction were at a high level of current stress as compared to the normal controls. That is, even if the phobic groups in the two stress conditions differed on state anxiety, the low-stress snake phobic group also experienced a relatively high level of anticipatory anxiety due to the test

966

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

situation. Actively helping this group to relax by a mood induction procedure or by more information might have improved the differentiation between the phobic participants in the two stress conditions. 4.3. Priming caused better discrimination of ‘subliminal’ words on the lexical decision task A significant difference in discrimination performance on the lexical decision task was shown among all participants, demonstrating a priming effect stemming from the unmasked exposure of words during the Stroop task, as compared with the masked Stroop words, lasting at least up to 15 min after the Stroop task. Although this priming effect may indicate that some individuals performed at a level above the criterion for an objective threshold of awareness on the lexical decision task, it is notable that this state of affairs was not associated with any differential colour naming latencies for masked words during the preceding Stroop task. This lack of a preattentive Stroop interference is even more striking as the size of the unmasked Stroop effect is very large compared to previous similar studies concerning specific phobia (39 ms, Lavy et al., 1993; 10-20 ms, Chen et al., 1996; 84 ms, Kindt and Brosschot, 1997; 27–43.8 ms, Kindt & Brosschot, 1998), and only a small degree of awareness would be sufficient in order to interfere with colour naming behaviour. It may be argued that the divergent Stroop results in the two exposure conditions supports the use of a subjective criterion of conscious awareness (i.e. the conscious identification task) in the Stroop paradigm. These results also suggest that the forced choice lexical decision task is overly conservative, that is, a task that may also be sensitive to preattentive cognitive processes, besides conscious processes (Merikle et al., 2001). 4.4. Effects on electrodermal responses Finally, the present study did not reveal any increased SCRs for masked snake words in snake phobics than for non-phobic controls, as previously have been shown in one study (van den Hout et al., 2000). It is noteworthy, however, that increased SCRs to masked snake words were seen in participants with longer exposure durations (0.51 ms), although there was no evidence of Stroop interference for snake words at the same exposure durations. This dissociation indicates that physical arousal and cognitive bias for threat may be due to separate mechanisms. The fact that stronger SCRs due to snake words were seen in all participants (and not only phobics) in the longer exposure conditions is hard to explain. It must be noted, however, that because the strength and characteristics (its amplitude, latency, rise time, and habituation) of a typical SCR to a masked word was not known beforehand, the analyses of SCR effects were not based on specific SCRs as defined by specific criteria, but on mean values of skin conductance levels and variability for a beforehand specified period of time after each stimulus—a procedure that may have resulted in a loss of power (for example, this method might not be sensitive to SCRs with long latencies and/or long rise times). Another criticism may be that verbal responses also may have confounded the SCR variance in various ways, and the SCR results should therefore be considered with caution. Possible response interference effects in SCRs due to short inter-trial intervals may further have resulted in lack of power in the SCR measurements (Dawson, Schell & Filion, 1990), although no

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

967

systematic error-variance should have been introduced because presentation of stimuli in the lexical decision task were totally randomized. A reason for the relatively short inter-trial interval in the lexical decision task was a wish to keep this task similar to the Stroop task, to justify comparisons of cognitive and physiological effects from these two different tasks, as well as to avoid fatigue in the participants.

5. Conclusion The most suggestive results of the present study are: (a) the large discrepancy between strong Stroop effects for unmasked snake words and no effect on masked snake words in patients with severe snake phobia; and (b) the increased ability in the lexical decision task to discriminate ‘subliminal’ words that had been shown unmasked as part of a previously performed Stroop task. The former result suggests that specific phobia may differ from other anxiety disorders, like GAD, in terms of preattentive processing of word stimuli—to test this possibility, however, these two different kinds of patient groups should be compared directly within the same study.

Acknowledgements We wish to thank Else Waaler for help with the data collection, and Professor Lars-Go¨ran ¨ Ost for help with the generation of snake and spider words. Appendix. Word set A

Word set B

Spider: Korsspindel (Cross-spider) Spindelva¨v (cobweb) Harkrank (crane-fly) Skorpion (scorpion) Spindel (spider) Ha˚rig (hairy)

Fa˚gelspindel (tarantula) Spindelna¨t (spider’s web) Insektsben (insect’s leg) Tarantel (tarantula) Spindel (spider) Kryp (creepy thing)

Snake: Anakonda (anaconda) Boaorm (boa-constrictor) Huggorm (viper) Slingra (coil) Ormar (snakes) Reptil (reptile)

Pytonorm (python) Skallerorm (rattlesnake) Huggta¨nder (fangs) Ringla (twine) Ormar (snakes) Kra¨ldjur (reptile)

968

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

Flower: Fo¨rga¨tmigej (forget-me-not) Smo¨rblomma (buttercup) Mandelblom (meadow saxifrage) Kronblad (petal) Blomma (flower) Planta (plant)

Liljekonvalj (lily of the valley) ¨ ngsblomma (meadow flower) A Va˚rblomma (spring flower) Groblad (plantain) Blomma (flower) Stja¨lk (stem)

Mushroom: Va˚rmusseron (St George’s mushroom) Kantarell (chanterelle) Skogssvamp (wildcrafted mushrooms) Flugsvamp (fly agaric) Svampar (mushrooms) Murkla (morel)

Fja¨llskivling (parasol mushroom) Champinjon (champignon) Karljohan (cep) Bla¨cksvamp (ink cap) Svampar (mushrooms) Ro¨ksvamp (puff-ball)

References American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders, (DSM-IV), 4th edition. Amir, N., McNally, R. J., Rieman, B. C., Burns, J., Lorenz, M., & Mullen, J. T. (1996). Suppression of the emotional Stroop effect by increased anxiety in patients with social phobia. Behaviour Research and Therapy, 34, 945–948. Bradley, B. P., Mogg, K., Millar, N., & White, J. (1995). Selective processing of negative information: Effects of clinical anxiety, concurrent depression, and awareness. Journal of Abnormal Psychology, 104, 532–536. Brown, T. A., Di Nardo, P. A., & Barlow, D. H. (1994). Anxiety Disorders Interview Schedule for DSM-IV. Albany: University at Albany, New York, Center for Stress and Anxiety Disorders. Chen, E., Lewin, M. R., & Craske, M. G. (1996). Effects of state anxiety on selective processing of threatening information. Cognition and Emotion, 10, 225–240. Dawson, M. E., Schell, A. M., & Filion, D. L. (1990). The electrodermal system. In J. T. Cacioppo, & L. G. Tassi(Eds.), Principles of psychophysiology: Physical, social, and inferential elements. Cambridge: Cambridge University Press. Harvey, A. G., Bryant, R. A., & Rapee, R. M. (1996). Preconscius processing of threat in posttraumatic stress disorder. Cognitive Therapy and Research, 20, 613–623. Kindt, M., Bierman, D., & Brosschot, J. (1997). Cognitive bias in spider fear and control children: Assessment of emotional interference by a card format and a single-trial format of the Stroop task. Journal of Experimental Child Psychology, 66, 163–179. Kindt, M., & Brosschot, J. (1997). Phobia-related bias for pictorial and linguistic stimuli. Journal of Abnormal Psychology, 106, 644–648. Kindt, M., & Brosschot, J. F. (1998). Stability of cognitive bias for threat cues in phobia. Journal of Psychopathology and Behavioral Assessment, 20, 351–367. Kindt, M., & Brosschot, J. F. (1999). Cognitive bias in spider-phobic children: Comparison of a pictorial and a linguistic spider Stroop. Journal of Psychopathology and Behavioral Assessment, 21, 207–220. Klorman, R., Weerts, T. C., Hastings, J. C., Melamed, B. G., & Lang, P. (1974). Psychometric description of some specific-fear questionnaires. Behavior Therapy, 5, 401–409. Lavy, E. H., van den Hout, M., & Arntz, A. (1993). Attentional bias and spider phobia: Conceptual and clinical issues. Behaviour Research and Therapy, 31, 17–24.

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

969

Lundh, L.-G., Wikstro¨m, J., & Westerlund, J. (2001). Cognitive bias, emotion, and somatic complaints in a normal sample. Cognition and Emotion, 15, 249–277. ¨ st, L.-G. (1999). Preattentive bias for emotional information in Lundh, L.-G., Wikstro¨m, J., Westerlund, J., & O panic disorder with agoraphobia. Journal of Abnormal Psychology, 108, 222–232. MacLeod, C., & Hagan, R. (1992). Individual differences in the selective processing of threatening information, and emotional responses to a stressful life event. Behaviour Research and Therapy, 30, 151–161. MacLeod, C., & Rutherford, E. M. (1992). Anxiety and the selective processing of emotional information: mediating roles of awareness, trait and state varables, and personal relevance of stimulus material. Behaviour Research and Therapy, 30, 479–491. Mathews, A., & Sebastian, S. (1993). Suppression of emotional Stroop effects by fear-arousal. Cognition and Emotion, 7, 517–530. Mayer, B., Merckelbach, H., & de Jong, P. (1999). How to handle contradictory results? Discount the data or refine ¨ hman. Journal of Psychophysiology, 257–259. the theories? A reply to O Merikle, P. M. (1992). Perception without awareness. Critical issues. American Psychologist, 47, 792–795. Merikle, M., Smilek, D., & Eastwood, J. D. (2001). Perception without awareness: perspectives from cognitive psychology. Cognition, 79, 115–134. Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational analysis of anxiety. Behaviour Research and Therapy, 36, 809–848. Mogg, K., Bradley, B. P., Millar, N., & White, J. (1995). A follow up study of cognitive bias in generalized anxiety disorder. Behaviour Research and Therapy, 33, 927–935. Mogg, K., Bradley, B. P., Williams, R., & Mathews, A. (1993). Subliminal processing of emotional information in anxiety and depression. Journal of Abnormal Psychology, 102, 304–311. Mogg, K., Kentish, J., & Bradley, B. P. (1993). Effects of anxiety and awareness on colour-identification latencies for emotional words. Behavour Research and Therapy, 31, 559–567. ¨ hman, A. (1993). Fear and anxiety as emotional phenomena: Clinical phenomenology, evolutionary perspectives, O and information-processing mechanisms. In M. Lewis, & J. M. Haviland (Eds.), Handbook of emotions (pp. 511– 536). New York: The Guilford Press. ¨ hman, A. (1997). Unconscious pre-attentive mechanisms in the activation of phobic fear. In G. C. L. Davey (Ed.), O Phobias—a handbook of theory, research and treatment (pp. 349–374). John Wiley & Sons Ltd. ¨ hman, A. (1999). Contributing to confusion rather than knowledge: A commentary on Mayer et al. Journal of PsyO chophysiology, 13, 160–162. ¨ hman, A., & Soares, J. J. F. (1994). ‘Unconscious anxiety’: Phobic responses to masked stimuli. Journal of AbnorO mal Psychology, 103(2), 321–340. Snodgrass, J. G., & Corwin, J. (1988). Pragmatics of measuring recognition memory: Applications to dementia and amnesia. Journal of Experimental Psychology: General, 117, 34–50. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Spra˚kbanken. (1999). Concordances of the Bank of Swedish on the Net, [online database]. University of Gothenburg. Available: http://spraakbanken.gu.se/. [1999]. Thorpe, S. J., & Salkovskis, P. M. (1997). Information processing in spider phobics: the Stroop colour naming task may indicate strategic but not automatic attentional bias. Behaviour Research and Therapy, 35, 131–144. van den Hout, M., de Jong, P., & Kindt, M. (2000). Masked fear words produce increased SCRs: An anomaly for ¨ hman’s theory of pre-attentive processing in anxiety. Psychophysiology, 37, 283–288. O van den Hout, M., Tenney, N., Huygens, K., & de Jong, P. (1997). Preconscious processing bias in specific phobia. Behaviour Research and Therapy, 35, 29–34. van den Hout, M., Tenney, N., Huygens, K., Merckelbach, H., & Kindt, M. (1995). Responding to subliminal threat cues is related to trait anxiety and emotional vulnerability: a successful replication of MacLeod and Hagan (1992). Behaviour Research and Therapy, 33, 451–454. van Honk, J., Tuiten, A., van den Hout, M., Koppeschaar, H., Thijssen, J., de Haan, E., & Verbaten, R. (2000). Conscious and preconscious selective attention to social threat: different neuroendocrine response patterns. Psychoneuroendocrinology, 25, 577–591.

970

J. Wikstro¨m et al. / Behaviour Research and Therapy 42 (2004) 949–970

van Honk, J., Tuiten, A., van den Hout, M., Putman, P., de Haan, E., & Stam, H. (2001). Selective attention to unmasked and masked threatening words: relationships to trait anger and anxiety. Personality and Individual Differences, 30, 711–720. Wikstro¨m, J., Lundh, L.-G., & Westerlund, J. (in press). Stroop effects for masked threat words: Preattentive bias or selective awareness? Cognition and Emotion. Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3–24. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional disorders. (2nd edn). Chichester: John Wiley & Sons Ltd.