Anxiety sensitivity and auditory perception of heartbeat

ARTICLE IN PRESS

Behaviour Research and Therapy 44 (2006) 1739–1756 www.elsevier.com/locate/brat

Anxiety sensitivity and auditory perception of heartbeat$ R.A. Pollocka,, A.S. Carterb, N. Amirc, L.E. Marksd,e a

Psychiatric & Neurodevelopmental Genetics Unit, Center for Human Genetic Research, Massachusetts General Hospital & Harvard Medical School, 185 Cambridge Street, 6th Floor, Boston, MA 02114-2790, USA b Department of Psychology, University of Massachusetts, Boston, MA, USA c Department of Psychology, University of Georgia, Athens, GA, USA d John B. Pierce Laboratory, New Haven, CT, USA e Department of Epidemiology and Department of Psychology, Yale University, New Haven, CT, USA Received 7 January 2005; received in revised form 5 December 2005; accepted 20 December 2005

Abstract Anxiety sensitivity (AS) is the fear of sensations associated with autonomic arousal. AS has been associated with the development and maintenance of panic disorder. Given that panic patients often rate cardiac symptoms as the most fearprovoking feature of a panic attack, AS individuals may be especially responsive to cardiac stimuli. Consequently, we developed a signal-in-white-noise detection paradigm to examine the strategies that high and low AS individuals use to detect and discriminate normal and abnormal heartbeat sounds. Compared to low AS individuals, high AS individuals demonstrated a greater propensity to report the presence of normal, but not abnormal, heartbeat sounds. High and low AS individuals did not differ in their ability to perceive normal heartbeat sounds against a background of white noise; however, high AS individuals consistently demonstrated lower ability to discriminate abnormal heartbeats from background noise and between abnormal and normal heartbeats. AS was characterized by an elevated false alarm rate across all tasks. These results suggest that heartbeat sounds may be fear-relevant cues for AS individuals, and may affect their attention and perception in tasks involving threat signals. r 2006 Elsevier Ltd. All rights reserved. Keywords: Anxiety; Anxiety sensitivity; Panic disorder; Heartbeat perception

Introduction Anxiety sensitivity (AS), or the ‘‘fear of anxiety,’’ characterizes a tendency to interpret arousal sensations as having harmful consequences, which may be physiological (e.g., impending heart attack), psychological (e.g., going crazy), or social (e.g., losing control) (Reiss & McNally, 1985). AS is thought to represent a stable traitlike factor that is qualitatively different from general fear and anxiety (Reiss, 1987; Reiss, Peterson, Gursky, & McNally, 1986). Theoretical and empirical work over the past two decades has afforded AS and other ‘‘fearof-fear’’ constructs a central role in the etiology of anxiety disorders in general and panic disorder in particular $

Supported in part by NIH grant DC00271 to L.E.M.

Corresponding author. Tel.: +617 726 0956; fax: +617 726 0830.

E-mail address: [email protected] (R.A. Pollock). 0005-7967/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2005.12.013

ARTICLE IN PRESS 1740

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

(Barlow, 1988; Chambless & Gracely, 1989; McNally, 1990; Reiss & McNally, 1985). The misinterpretation of bodily cues may lead to a self-perpetuating fear-of-fear cycle whereby the fear of benign arousal sensations produces anxiety, which in turn increases the frequency and intensity of physiological sensations, and subsequently fuels apprehension regarding the significance of these sensations. Prospective studies suggest that AS predicts frequency and intensity of panic attacks (Maller & Reiss, 1992) as well as the spontaneous onset of panic attacks in a stressful situation (e.g., military training) (Schmidt, Lerew, & Jackson, 1997, 1999). If AS is characterized by particular cognitive states—both explicit and implicit beliefs about the catastrophic nature of certain physiological sensations—then these cognitions should in turn influence the perception and processing of these sensations (McNally, 1999a). Understanding the role of perceptual processes in AS may inform our understanding of emotional information processing in the pathogenesis of panic. To this end, the present research focuses on the role of AS in the detection and discrimination of auditory signals, such as heartbeat sounds, that may play a special and prominent role in AS and panic. If, for example, individuals high in AS find certain interoceptive signals, such as heartbeats, threatening, then these individuals may also exhibit differences in their approaches to making behavioral decisions about the naturally occurring (and potentially feared) bodily sensations that these signals represent. Although a great deal of research attention to date has attempted to understand patterns of cognitive bias for threat and interoceptive awareness in anxious individuals and individuals with panic (cf. McNally, 1999b), far less experimental work has focused on AS samples. Research suggests that anxious individuals preferentially process threat-relevant material (e.g., MacLeod, 1991; MacLeod, Mathews, & Tata, 1986; Mathews, 1990). Clinical and empirical evidence suggests that patients with panic disorder may interpret ambiguous situations (e.g., McNally & Foa, 1987; McNally, Foa, & Donnell, 1989) and internal stimuli, such as fluctuations in heart rate, as dangerous (e.g., Ehlers & Breuer, 1992; Ehlers, Margraf, Roth, Taylor, & Birbaumer, 1988), show favorable attention to threat cues (Asmundson, Sandler, Wilson, & Norton, 1993; Ehlers, Margraf, Davies, Roth, & Birmaunr, 1988; McNally, Riemann, & Kim, 1990), and preferentially remember threatening or fearful information (e.g., Amir, McNally, Riemann, & Clements, 1996). Although AS may confer risk for panic attacks (Schmidt et al., 1997, 1999), data suggesting a premorbid information processing bias are sparse. Researchers have examined the role of information processing bias in individuals with elevated AS who do not have a history of panic attacks. For example, McNally, Hornig, Hoffman, and Han (1999) investigated attention, memory, and interpretation bias in individuals with high and low AS who had not been diagnosed with any anxiety disorder. These researchers used a non-clinical sample, excluding participants with a history of panic, and found that individuals with elevated AS tended to interpret ambiguous panic relevant scenarios as threatening. No group differences were found for attentional bias or memory bias for threat. Similarly, McCabe (1999) examined memory bias in individuals with high and low AS without history of panic. McCabe found that individuals with elevated AS exhibited a memory bias for threat-related words, but not for anxiety related, positive, or neutral words. Because she did not find group differences in implicit memory (i.e., automatic, non-conscious; Schacter, 1992) and because individuals with panic disorder do show an implicit memory bias (Amir et al., 1996), McCabe proposed that this finding may delineate the differences between individuals with high AS and those with panic disorder. Finally, Amir and Beard (2004) hypothesized that individuals with elevated AS would show difficulty in inhibiting the threat relevant meaning of homographs (words with two meanings, e.g., faint). To test this hypothesis, the authors used a modified version of Gernsbacher’s inhibition paradigm in which participants were first presented with a sentence (e.g., ‘‘The car sounds were faint.’’) and then a probe word (e.g., ‘‘passout’’) and were asked whether or not the probe word was related to the meaning of the sentence (Gernsbacher, Verner, & Faust, 1999). One half of the sentences ended in a homograph, and one half ended in a nonhomograph word. Results revealed that individuals with elevated AS, but not those with low AS, showed difficulty in inhibiting the threat-relevant meaning of homographs when the interval between the presentation of the sentence and the probe word was short (i.e., 100 ms). Amir and colleagues (2004) concluded that difficulty inhibiting threat-relevant meanings may be a vulnerability factor involved in elevated AS, as well as the development of panic disorder.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1741

Panic attacks are often accompanied by symptoms such as dizziness, faintness, heart palpitations or shortness of breath, and AS individuals report fear of these sensations. Researchers have described panic disorder patients as ‘‘interoceptive experts,’’ aware of and able to describe subtle changes throughout the body (Schands & Schor, 1982). However, studies examining interoceptive acuity among patients with panic disorder with respect to cardiac perception have yielded mixed results. It is important to differentiate accurate detection of heartbeat versus vigilance (or hypervigilance) to heartbeats in general. The former describes an objective state whereby an individual is able to correctly identify the actual heart rate or unique beating of his or her own heart. The latter characterizes a more subjective attention to or scanning for heartbeats, and may or may not reflect an accurate detection of these heartbeats. The heartbeat perception task employed in this study focuses specifically on subjective attention to and scanning for heartbeats. Some studies suggest greater cardiac awareness in panic disorder patients (e.g., Ehlers & Bruerer, 1992; King, Margraf, Ehlers, & Maddock, 1986), and others fail to replicate these findings reliably (e.g., Antony et al., 1995; Asmundson et al., 1993; Ehlers, Margraf, Roth et al., 1988; Knoll, Folten, & Hodapp, 1991; Pauli et al., 1991). More recently Stewart and colleagues (2001) found that high AS individuals reported higher heart rate estimates and greater accuracy in heart rate estimation than low AS individuals (Stewart, Buffett-Jerrott, & Kokaram, 2001). Researchers have also found that children with high AS and elevated panic/somatic symptoms showed enhanced heartbeat perception on a mental tracking task (Eley, Stirling, Ehlers, Gregory, & Clark, 2004). Methodological differences may contribute to the inconsistent findings regarding interoceptive acuity in both clinical and non-clinical populations. Furthermore, traditional methods for testing interoceptive acuity, such as heart rate tracking paradigms, may not offer information about attention and interpretation processes. These processes may, at least in part, drive the strategies that individuals use to monitor and respond to their bodily sensations. The present study uses signal detection theory (SDT) to examine auditory detection of heartbeat sounds against a background of white noise, and the discrimination between two different heartbeat sounds. Signal detection methodology allow us to quantify an individual’s perceptual and decisional components of the judgment process by comparing rates of correct detections to rates of false positive responses (Green & Swets, 1966; Macmillan & Creelman, 1991). In its simplest form, SDT assumes that individuals attempt to maximize the payoffs and minimize the risks associated with a decision, such as whether a signal was present or which of several possible signals was presented. These decisions will vary from very strict to very lenient according to the rules an individual uses in a given situation. Undoubtedly, misses and false alarms have different consequences. In a truly dangerous situation, the ability to selectively attend to threatening stimuli and react with anxiety is adaptive; rapid and automatic processing of information promotes escape. On the other hand, expending attentional resources towards non-dangerous threats is inefficient (Williams, Watts, MacLeod, & Mathews, 1988), and interferes with gaining an accurate perceptual representation of the environment (Pashler, 1998). In sum, SDT considers (1) the perception of a stimulus (sensitivity) and (2) the behavioral results of responding to that stimulus (criterion). Given this SDT framework, we designed an experiment to examine AS and the decisional processes individuals use when responding to heartbeat sounds. Specifically, we hypothesized that high and low AS listeners would use different response strategies (rules) in the face of different auditory signals (e.g., neutral tone, normal heartbeat, abnormal heartbeat). From a working theoretical perspective, high AS individuals, perhaps influenced by a biological predisposition to experience internal sensations with greater frequency (e.g., Kagan, 1994), a cognitive tendency to attend to and/or misinterpret internal sensations (e.g., Beck, Emery, & Greenberg, 1985; Clark, 1986, 1988; McNally, 1990), and/or learning from the environment (e.g., hearing misinformation from others, direct observation) that internal sensations are dangerous (McNally, 1999b), may exhibit different rules from non-AS individuals when responding to stimuli they may perceive as dangerous (e.g., heartbeat sounds). We had four explicit hypotheses: (1) High and low AS individuals will not differ in their perception or reporting of neutral auditory stimuli. (2) High and low AS individuals will differ in their reporting of normal heartbeat sounds, with high AS individuals showing a greater propensity, compared to low AS individuals, to report the presence of normal heartbeat sounds (i.e., high AS individuals will employ a lower criterion for target identification). Despite this hypothesized reporting bias, no group differences are expected in the ability to perceive normal heartbeat signals against a background of white noise (i.e., sensitivity). Specifically, normal

ARTICLE IN PRESS 1742

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

heartbeat sounds should not activate feelings of ‘‘threat’’ that would be expected to affect perception and sensitivity. (3) High and low AS individuals will differ in both their perception and reporting of abnormal heartbeat sounds. High AS individuals will have a lower criterion for reporting abnormal heartbeat sounds. Furthermore, if these sounds are deemed ‘‘threatening,’’ this perceived danger will interfere with the ability to selectively attend to abnormal heartbeat signals, and thereby reduce sensitivity. (4) High and low AS individuals will also differ in their ability to discriminate normal from abnormal heartbeat sounds. Because high AS individuals may have a tendency to ‘‘body scan’’ for cardiac fluctuations, they may be more likely to misinterpret normal bodily sensations as dangerous. This tendency would manifest as a bias to identify normal heartbeat sounds as abnormal, or in other words, to establish a lower criterion for reporting abnormal heartbeat sounds. Moreover, the impact of threat-related information will interfere with the ability of high AS individuals to gain a detailed perceptual representation of the unique qualities of the two heartbeat stimuli, thereby reducing sensitivity. Methods Participants Participants were college age males and females recruited in a two-stage process. First, 136 undergraduate volunteers completed the Anxiety Sensitivity Index (ASI: Reiss et al., 1986). After examining the distribution of ASI scores, we conducted a tertiary split of the ASI, assigning the top 33% to the high anxiety sensitive (high AS) group (ASIX24) and the bottom 33% (ASI p 14) to the low anxiety sensitive (Low AS) group. According to Peterson and Plehn (1999), ASI scores greater than or equal to 25 suggest problems with AS. Therefore, our ‘‘high AS’’ group may be considered within the range of clinical risk. The second stage of the study involved four computer-based auditory detection tasks. A minimum sample size of 24 per cell was chosen to attain power of .8 assuming a moderate effect size (approximately 0.3 at p ¼ :05, Rosenthal & Rosnow, 1991). Of eligible students, participants were 31 low AS and 34 high AS individuals, ranging in age from 18 to 30. Participants received either course credit or monetary compensation ($15.00) for their participation. Individuals with known hearing problems and benign heart conditions were excluded from the study. There were no statistically significant group differences on age, gender, or years of education. However, a larger percentage of the high AS group was female (20 females versus 11 males), and a higher percentage of the low AS group was male (20 males versus 11 females); this difference attained a trend level of significance, w2 ð1; N ¼ 65Þ ¼ 3:54, po:10. There were also no differences in the percentages of high and low AS individuals who were taking prescription medication, w2 ð1; N ¼ 65Þ ¼ 4:11, p4:10 (Table 1). Only three participants Table 1 Demographic and self-report data for high and low anxiety sensitivity participants Low ASI (n ¼ 31)

High ASI (n ¼ 34)

Age (M7SD) Years of education (M7SD) ASI (M7SD) STAI-S (M7SD) STAI-T (M7SD)

19.6872.15 14.0071.10 10.9773.47 29.3279.8 31.6079.1

19.5272.20 13.8271.00 28.8876.53c 34.30710.4a 41.25711.9b

Gender (%) Male Female Prescription Medication (%)

64.5 35.5 4.7 (n ¼ 3)

41.2 58.8 14.1 (n ¼ 9)

Note: AS ¼ Anxiety Sensitivity, STAI-S ¼ State Trait Anxiety Inventory- State Version, STAI-T ¼ State Trait Anxiety Inventory- Trait Version. a po:05. b po:01. c po:0001.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1743

reported taking psychiatric medications (i.e., Effexor, Wellbutrin, and Dexedrine); two were in the high AS group and one in the low AS group. Self-report measures Participants completed the Anxiety Sensitivity Index (ASI, Reiss et al., 1986), a 16-item questionnaire designed to measure fear of anxiety sensations, which shows both discriminant and convergent validity in clinical and nonclinical anxious populations (cf. Taylor, 1995). Participants also completed the state and trait versions of the State-Trait Anxiety Inventory (STAI, Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983), which measures an individual’s current and general level of anxiety. Equipment We used the SoundForger sound-editing program to generate white noise, mix white noise with heartbeat sounds and tones, and calibrate signals in noise. Intensity of stimuli was measured and calibrated with a General Radio 1987 Sound Level Meter with a 9A-type coupler. An IBM compatible Pentium class laptop computer presented digitized files of heartbeats and pure tones through Sony MDR/CD30 headphones to the subject. Cardiac sounds One normal heartbeat and one abnormal heartbeat were selected from a CD-Rom (Criley, Criley, & Zalace, 1997) used for training in cardiac auscultation, and recorded as .wav files. Sounds were selected to maximize similarity in spectral content and temporal envelope of heartbeat type and were piloted in a sample of 74 undergraduates (Pollock, Amir, Freshman, & Marks, 2000). Twelve different heartbeat sounds, categorized as normal, gallop, and murmur heartbeats, were presented to participants with and without a history of panic attacks. Participants rated the heartbeats on familiarity (representative of a typical heartbeat) (1 ¼ familiar, 10 ¼ unfamiliar) and normality (1 ¼ normal, 10 ¼ abnormal). Regardless of group membership, participants rated heartbeats as familiar (gallop) and unfamiliar (murmur) similarly. However, individuals with a history of panic rated unfamiliar heartbeat sounds as significantly more abnormal than did non-panic participants. In the current study, a ‘‘gallop’’ heartbeat was chosen to represent the abnormal heartbeat category so that the spectral characteristics of the normal and abnormal heartbeats could be matched as closely as possible.1 Each stimulus included two complete heartbeat evolutions, which for normal heartbeats contained 4 spikes, and for abnormal ‘‘gallop’’ heartbeats contained 6 spikes. The waveform characteristics of the heartbeat sounds are depicted in Fig. 1. Detection tasks Participants completed three signal-in-white-noise detection tasks. Each task consisted of 72 two-second trials; half of the trials were noise alone and half of the trials contained a signal masked with one of two levels of white noise, 70.5 and 66.5 db. Stimuli were presented randomly at high and low noise levels. Signals were one second in duration and began 0.5 s after the onset of the white noise. In the first experiment, the signal was either a high frequency (2000 Hz) or low frequency (200 Hz) tone, presented randomly and with equal probability. In the second experiment, the signal was a normal heartbeat. In the third experiment, the signal was an abnormal heartbeat. 1 Different cardiopathies will manifest in heartbeat sounds of variable familiarity to the medically untrained ear. For example, valvular stenosis (narrowing at the level of the valve) or regurgitation/insufficiency (incomplete valvular closure producing a backwards flow of blood across the valve) produces ‘‘whistle and murmur’’ sounds. These ‘‘whooshing’’ sounds provide stark contrast to the ‘‘lub-dub’’ heard in normal heartbeats. Along with normal heartbeat, other cardiac pathologies (e.g., gallops) exist where the number or rate of ‘‘lub-dub’’ beats is altered. Since gallop cardiopathies are explicitly more similar in their auditory characteristics to the familiar ‘‘lub-dub’’ pattern of a normal heartbeat, they may be considered by most listeners as more descriptive of ‘‘normal heartbeat.’’ (E. J. Pollock, personal communication, June 14, 1999).

ARTICLE IN PRESS 1744

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

Fig. 1. Spectral characteristics of normal and abnormal heartbeat. Note: (a) Normal heartbeat, (b) abnormal heartbeat, x-axis ¼ time, y-axis ¼ amplitude.

Discrimination task Participants completed one discrimination task involving procedures identical to those described above, with the following exceptions: (1) the experiment consisted of 84 two-second trials; (2) every trial contained a normal or abnormal heartbeat in noise presented randomly and with equal probability. Procedure All participants read and signed an informed consent form prior to study entrance. Participants completed the STAI-S and STAI-T, prior to beginning the computer tasks. Participants sat alone in a sound-attenuated chamber. They could not see the experimenter, but could signal the experimenter through an intercom if they had a question, or to indicate that they had completed a task. Participants listened to each stimulus three times prior to beginning the computer task to gain familiarity with the noise and signals in noise. A 2-min break was provided between each experiment. The task order was constant; participants first completed the neutral tone detection tasks, followed by the normal heartbeat detection task, abnormal heartbeat detection task, and finally the heartbeat discrimination task. A constant order was selected to maximize statistical power. The normal heartbeat tasks preceded all tasks that included an abnormal heartbeat to minimize the potential for carry-over effects. For each trial, participants received one type of stimulus, noise alone or a signal plus noise. Responses to stimuli were prompted by a computer-generated message (‘‘Did you hear a heartbeat?’’), and were made by pressing the appropriate key (e.g., ‘‘yes’’ or ‘‘no’’). A second message subsequently appeared asking participants, ‘‘How confident are you in your decision, (1–6)?’’ The scale was anchored by 1 ¼ ‘‘I am 100% uncertain of my decision’’, 3 ¼ I am somewhat sure of my decision, and 6 ¼ ‘‘I am 100% certain of my decision.’’ The procedure was identical for the discrimination task with the exception that students were made aware that each trial contained either a normal or abnormal heartbeat and were prompted to determine whether the heartbeat signal they heard was normal or abnormal. Following the experiments, all participants were fully debriefed. Analytic procedures SDT provides a model and set of analytic procedures for distinguishing an individual’s behavioral criteria (e.g., response to stimuli) and his or her sensitivity to a particular signal in noise. SDT holds that the internal representation on each trial is a scalar quantity reflecting the effect of noise alone or signal plus noise. According to the theory, in a yes–no decision task, an individual establishes a criterion to respond ‘‘yes.’’

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1745

Payoffs for different outcomes (e.g., correct detection, false alarm) will contribute to the location of the response threshold (Macmillan & Creelman, 1991; Gescheider, 1997). To this end, SDT allows for the quantification and comparison of decision strategies used by individuals high or low on a measure of AS. Detection rates of heartbeat and neutral tones were calculated for each subject based on the accuracy of distinguishing signal plus noise from noise alone. For each participant, we calculated the following response variables (Macmillan & Creelman, 1991). Response

Signal present

Signal absent

‘‘Yes’’ ‘‘No’’

A C

B D

H ¼ PðAÞ=A þ C;

(1)

F ¼ PðBÞ=B þ D;

(2)

d 0 ¼ zH2zF ,

(3)

c ¼ ½
:05ðzHÞ þ ðzF Þ.

(4)

(1) Hit rate gives the probability of correctly reporting a signal when it is present. (Eq. (1)), (2) False alarm rate gives the probability of incorrectly reporting noise as a signal (Eq. (2)), (3) Sensitivity (d0 ) is a measure of the individual’s ability to discriminate between signal and noise (Eq. (3)), and (4) Criterion/Response threshold (c) is a measure of the likelihood of responding ‘‘yes’’ or ‘‘no’’ given a decision task (Eq. (4)). ROC curves Participants’ decision-making strategies were plotted via a Receiver Operating Characteristic (ROC) (Gescheider, 1997; Swets, Dawes, & Monahan, 2000). We calculated ROC curves by accumulating probabilities of hits and false alarms across confidence ratings for signal and noise trials, respectively. The ROC plots the z-transformed probability that the person responds ‘yes’ (or gives a particular confidence rating) when the signal is present, P(hit), against the corresponding probability of the same response when the signal is absent, or negative, P(false alarm) (Swets et al., 2000). Statistical analyses Prior to testing formal hypotheses, preliminary analyses were conducted to consider appropriate demographic covariates (e.g., gender and age). The associations between AS group,2 and self-reported state and trait anxiety, were examined with Pearson correlations. As noted above, there was a trend for females to be higher on the ASI than males; therefore, gender was included as a covariate in initial models. However, since across heartbeat tasks there were no significant main effects for gender on the outcome variables hit rate, hit rate, false alarm rate, sensitivity, and criterion, analyses are presented without gender included. As expected, high AS individuals reported higher levels of state anxiety and trait anxiety than Low AS participants (Table 1). The primary focus of the following four experiments was to evaluate differences between individuals high and low on AS in their auditory perception of neutral tones, normal heartbeats, and abnormal heartbeats, and in their ability to discriminate between normal and abnormal heartbeat sounds. For each of the experiments, we submitted response variables (hit rate, false alarm rate, sensitivity and response bias) to a 2 (Group: high AS, low AS)  2 (Noise Level: high, low) factorial ANOVA with noise as a repeated factor. 2

A categorical AS grouping variable (high and low) was used since a tertiary split had been conducted on the original continuous data.

ARTICLE IN PRESS 1746

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

Results Experiment 1: detection of neutral tones In this and subsequent experiments, participants were excluded from the final analyses if their data were judged to be unreliable (e.g., hit rate o 20%, false alarm rate480%). In Experiment 1, this led to the exclusion of two participants. Five participants indicated time constraints and did not complete the first experiment. They did, however, provide data for Experiments 2–4. Twenty-seven low AS and 32 high AS participants were included in the analyses to follow. Table 2 depicts the means and standard deviations for high and low AS participants for the outcome variables in the first experiment. Hit rate and false alarm rate: High and low AS participants performed comparably on neutral tone tasks, as evidenced by similar correct detection and correct rejection rates. Hit rates were significantly higher in low versus high noise levels (2000 Hz [F ð1; 57Þ ¼ 105:83, po:0001, Z2 ¼ :65], and 200 Hz [F ð1; 57Þ ¼ 91:9, po:0001, Z2 ¼ :62]), but false alarm rates did not differ across noise levels [F ð1; 57Þ ¼ 1:50, p ¼ :22, Z2 ¼ :02]. Sensitivity and criterion: Similarly, there were no significant differences between high and low AS listeners in terms of their sensitivity or criterion, with all values of F o:86, p4:40, Z2 o:02. There were significant main effects of noise for sensitivity, 2000 Hz [F ð1; 57Þ ¼ 101:45, po:0001, Z2 ¼ :64], 200 Hz [F ð1; 57Þ ¼ 97:00, po:0001, Z2 ¼ :63] and criterion, 2000 Hz [F ð1; 57Þ ¼ 116:6, po:0001, Z2 ¼ :67], 200 Hz: [F ð1; 57Þ ¼ 69:76, po:0001, Z2 ¼ :53]. None of the two-way interactions between AS and noise level was statistically reliable, with all values of F o:50, p4:50, Z2 o:02. Experiment 2: detection of normal heartbeat By the exclusion criteria described earlier, five participants’ data were deemed invalid and were not considered in the analyses that follow. The study included 27 low AS and 33 high AS participants. Results for Experiment 2 are presented below in Figs. 2 and 3. Hit rate and false alarm rate: A main effect of noise level emerged for hit rate [F ð1; 58Þ ¼ 54:54, po:0001, Z2 ¼ :49] but not for false alarm rate [F ð1; 58Þ ¼ 0:01, p ¼ :90, Z2 o:0001], indicating that respondents were more likely to correctly detect normal heartbeats in lower levels of noise, but were equally likely to report having heard a normal heartbeat in noise alone trials regardless of background noise levels. High AS individuals had slightly higher mean hit rates for normal heartbeats than did low AS individuals; however, these differences constituted only a statistical trend [F ð1; 58Þ ¼ 2:81, po:10, Z2 ¼ :05]. Consistent with our predictions, the high AS group exhibited a significantly greater false alarm rate than the low AS group

Table 2 Means and standard deviations of response variables for high and low AS in neutral tone detection tasks Low AS (n ¼ 27) High noise

High AS (n ¼ 32) Low noise

High noise

Low noise

Hit rate Low tone (200 Hz) High tone (2000 Hz)

.517.28 .547.29

.907.26 .967.13

.507.28 .567.28

.897.23 .977.09

False alarm rate

.227.22

.207.19

.237.19

.207.21

Sensitivity Low tone (200 Hz) High tone (2000 Hz)

1.3171.09 1.3471.36

3.2671.17 3.5071.24

0.9971.35 1.2871.56

2.9271.68 3.4871.14

Criterion Low tone (200 Hz) High tone (2000 Hz)

.577.94 .567.91

Note: AS ¼ Anxiety Sensitivity.


.677.49
.597.54

.597.74 .447.75


.297.72
.567.61

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1

0.5

0.9

0.45 False Alarm Rate

0.8 Hit Rate

0.7 0.6 0.5 0.4 0.3

0.4 0.35 0.3 0.25 0.2 0.15

0.2

0.1

0.1

0.05

0

1747

0 High

Noise Level Low AS High AS

(a)

High

Low

Low Noise Level Low AS High AS

(b)

3.5 1.2

3

1 2.5 0.8 2 c

d'

0.6 1.5

0.4

1

0.2

0.5

0

0

-0.2 High

Low

Low Noise Level

(c)

High

Low AS

High AS

Noise Level (d)

Low AS

High AS

Fig. 2. Response variables for high and low AS in normal heartbeat detection task. (a) Hit rate: normal heart beat; (b) False alarm rate: normal heart beat; (c) Sensitivity: normal heartbeat; (d) Criterion: normal heart beat. Note: AS ¼ Anxiety Sensitivity, (a) low noise4high noise, po.0001, (b) high AS4low AS, po.01, (c) low noise4high noise, po.0001, (d) low noise4high noise, po.0001; low AS4high AS, po.05.

[F ð1; 58Þ ¼ 9:47, po:01, Z2 ¼ :14]. Neither of the two-way interactions between AS and noise level was significant, with all values of F o:50, p4:50, Z2 o:02. Sensitivity and criterion: Participants evidenced lower d0 in high versus low noise levels [F ð1; 58Þ ¼ 60:77, po:0001, Z2 ¼ :51],] and a more lax threshold of response (criterion) [F ð1; 58Þ ¼ 95:27, po:0001, Z2 ¼ :62]. AS groups did not differ in their sensitivity (d0 ) to normal heartbeats [F ð1; 58Þ ¼ 0:94, Z2 ¼ :02]. Finally, and consistent with our expectations, high AS respondents showed a significantly lower threshold of response than low AS individuals, [F ð1; 58Þ ¼ 4:24, po:05, Z2 ¼ :07], suggesting a response bias for normal heartbeats. ROC analysis: Fig. 3 shows ROCs based on the confidence ratings of high AS and low AS listeners. In the low noise condition, there was an evident difference in sensitivity when participants used strict criteria; however, this difference was no longer present when the criteria were lax. For the high noise trials, high AS and low AS participants performed comparably, and relatively close to chance. This pattern of sensitivity indicates that signal trials and noise trials were difficult to discriminate in the high noise condition regardless of level of AS.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1748

Experiment 2: Low Noise condition

Experiment 2: High Noise condition 3

2

2

1

1 Z-hit

Z-hit

3

0

-1

0

-1 Low AS

-2

Low AS

-2

High AS

-3

High AS

-3 -3

-2

-1

0 1 Z-false alarm

2

3

-3

-2

-1

0 1 Z-false alarm

2

3

Fig. 3. ROC curves for normal heartbeat detection task. Note: AS ¼ Anxiety Sensitivity.

Experiment 3: detection of abnormal heartbeat Results for Experiment 3 are shown in Figs. 4 and 5. Data of four participants were excluded by the criteria described earlier. Thus, analyses were based on data of 29 low AS and 32 high AS participants. Hit rate and false alarm rate: Once again, hit rates were significantly greater in low levels of background noise [F ð1; 59Þ ¼ 14:71, po:0001, Z2 ¼ :20], and, as in Experiment 2, there was no main effect of noise level for false alarm [F ð1; 59Þ ¼ 0:37, p4:50, Z2 ¼ :01]. In contrast to the findings in Experiment 2, high AS individuals had a significantly lower hit rate [F ð1; 59Þ ¼ 5:40, po:05, Z2 ¼ :08] than did low AS individuals. Consistent with our prediction, the groups differed in their false alarm rate [F ð1; 59Þ ¼ 4:41, po:05, Z2 ¼ :07], with high AS participants showing a greater likelihood than low AS participants to report an abnormal heartbeat in noise alone trials. No two-way interactions were significant. Sensitivity and criterion: There were significant main effects of noise level on sensitivity [F ð1; 59Þ ¼ 54:37, po:0001, Z2 ¼ :48], and criterion [F ð1; 59Þ ¼ 31:36, po:0001, Z2 ¼ :07]; participants had both a greater sensitivity and a lower response threshold for abnormal heartbeats in low levels of noise. As Fig. 4c reveals, and consistent with our prediction, low AS participants showed greater sensitivity [F ð1; 59Þ ¼ 12:21, po:001, Z2 ¼ :17] to abnormal heartbeats than did high AS individuals, a function of their greater correct detection and lower false alarm rates. Contrary to our hypothesis, AS was not associated with differences in criterion [F ð1; 58Þ ¼ :44, p4:50, Z2 o:01]. Furthermore, there were no significant two-way interactions. ROC analysis: Fig. 5 shows ROCs based on the confidence ratings of high AS and low AS listeners. The high AS group showed lower d0 across noise conditions, with group differences becoming more pronounced in the high noise condition. Experiment 4: discrimination between normal and abnormal heartbeat Results for Experiment 4 are shown in Figs. 6 and 7. Participants included 30 low AS and 34 high AS students. One student was excluded due to unreliable data. Hit rate and false alarm rate: A significant main effect of noise level emerged for hit rate [F ð1; 62Þ ¼ 22:6, po:0001 Z2 ¼ :27]. Participants had a higher rate of correct detection in low levels of noise. There was no main effect of false alarm rate [F ð1; 62Þ ¼ 1:40, p ¼ :24 Z2 ¼ :02]. Similar to the findings in Experiment 3, low AS participants correctly identified normal and abnormal heartbeats more often than did high AS participants

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1749

0.3

1

0.25 False Alarm Rate

Hit Rate

0.8 0.6 0.4

0.2 0.15 0.1

0.2

0.05

0

0 High

High

Low

Low Noise Level

Noise Level Low AS

(a)

High AS

Low AS

(b)

High AS

0.8

5

0.7 4

0.6 0.5

d'

3 c

0.4 2

0.3 0.2

1

0.1 0 High

0

Low

High

Low

-0.1 Noise Level (c)

Low AS

High AS

Noise Level (d)

Low AS

High AS

Fig. 4. Response variables for high and low AS in abnormal heartbeat detection task. (a) Hit rate: abnormal heart beat; (b) False alarm rate: abnormal heart beat; (c) Sensitivity: abnormal heartbeat; (d) Criterion: abnormal heart beat. Note: AS ¼ Anxiety Sensitivity, (a) low noise4high noise, po.0001, low AS4high AS, po.05, (b) high AS4low AS, po.05, (c) low noise4high noise, po.0001; low AS4high AS, po.001, (d) low noise4high noise, po.0001.

(hit rate) [F ð1; 62Þ ¼ 5:65, po:05 Z2 ¼ :08]. As expected, high AS listeners showed a greater tendency than low AS listeners to report that they heard an abnormal heartbeat when a normal heartbeat was presented (false alarm) [F ð1; 62Þ ¼ 4:95, po:05 Z2 ¼ :07]. There were no significant two-way interactions [F ð1; 62Þo1:00, p4:50 Z2 o:01]. Sensitivity and criterion: As expected, participants showed greater sensitivity on the heartbeat discrimination task in low levels of noise [F ð1; 62Þ ¼ 36:59, po:0001, Z2 ¼ :37]. Furthermore, listeners maintained a significantly more lax response threshold in lower levels of noise [F ð1; 62Þ ¼ 15:01, po:0001 Z2 ¼ :20]. Consistent with their performance on the abnormal heartbeat detection task, low AS individuals were better able to discriminate between normal and abnormal heartbeats than were high AS individuals [F ð1; 62Þ ¼ 6:62, po:01, Z2 ¼ :10]. AS groups did not differ on criterion [F ð1; 62Þo1:0, p4:45 Z2 o:01], and none of the twoway interactions was significant [F ð1; 62Þo1:5, p4:40 Z2 o:01]. ROC analysis: Fig. 7 shows ROCs. In both the high and low noise conditions, low AS participants showed greater sensitivity than high AS participants, especially when listeners used more stringent criteria, i.e., d0 when zFo0.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1750

Experiment 3: High Noise condition 3

2

2

1

1 Z-hit

Z-hit

Experiment 3: Low Noise condition 3

0

0

-1

-1

Low AS

Low AS -2

-2

High AS

High AS

-3

-3 -3

-2

-1

0 1 Z-false alarm

2

3

-3

-2

-1

0 1 Z-false alarm

2

3

Fig. 5. ROC curves for abnormal heartbeat detection task. Note: AS ¼ Anxiety Sensitivity.

Discussion Consistent with information processing models of anxiety, high AS and low AS listeners showed different response patterns in tasks involving fear-relevant auditory stimuli (heartbeats) but not neutral tones. These results provide preliminary evidence that individuals high in AS do not manifest global perceptual differences in sensitivity to or response biases for processing general auditory stimuli. The implication of these findings suggests that emergent variance in response to heartbeat stimuli may be attributable to fear-specific differences in perceptual processing. Specifically, individuals high and low on AS did not differ in their detection of neutral signals, whether these signals involved high- or low-frequency tones. In contrast, on a normal heartbeat detection task, high AS individuals showed an elevated false alarm rate and lower threshold for reporting normal heartbeats (criterion) when compared to their low AS counterparts. This pattern was consistent across conditions of high and low background white noise. Moreover, high AS was associated with decreased performance (sensitivity) on both an abnormal heartbeat detection task as well as a heartbeat discrimination task. High AS participants were consistently less able to discriminate abnormal heartbeats from white noise and less able to discriminate abnormal heartbeats from normal heartbeats than were low AS participants. On tasks involving abnormal heartbeats, high AS respondents made more false alarms than low AS respondents by (1) reporting having heard a heartbeat when no heartbeat was presented in a detection task, and (2) labeling a normal heartbeat abnormal in a discrimination task. Taken together, this pattern of results suggests that AS may be associated with differences in response strategy with respect to both sensitivity and criterion, and depending on the type of heartbeat sound presented. Specific findings from this study are discussed below. Although little evidence is available regarding information processing patterns in individuals with high AS, empirical studies of panic may inform our understanding of AS. For example, McNally (1999a, b) has commented on the equivocal association between panic and interoceptive acuity, stating, ‘‘merely because a person fears interoceptive cues does not mean that he or she will be especially good at detecting them’’ (McNally, 1999b, p. 193). In this sense, the term ‘‘sensitivity,’’ or (d0 ) as defined by SDT, may be comparable to the term ‘‘acuity’’ used in other studies of cardiac perception. Consistent with our expectations, as well as with previous findings suggesting that perceptual acuity for cardiac sensations is not enhanced by history of panic attacks (e.g., Antony et al., 1995; Asmundson et al., 1993), high AS was neither associated with better perception of normal or abnormal heartbeat signals in noise, nor associated with improved discrimination between normal and abnormal heartbeat sounds.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1

1751

0.3

0.9 0.25 False Alarm Rate

0.8

Hit Rate

0.7 0.6 0.5 0.4 0.3 0.2

0.2 0.15 0.1 0.05

0.1 0

0 High

Low

High

Noise Level (a)

Low AS

High AS

Low Noise Level

(b)

Low AS

3.5

1.2

3

1

2.5

High AS

0.8 c

d'

2 0.6

1.5 0.4 1 0.2

0.5

0

0 High

High

Low

(c)

Low AS

High AS

Low Noise Level

Noise Level (d)

Low AS

High AS

Fig. 6. Response variables for high and low AS in heartbeat discrimination task. (a) Hit rate: discrimination task; (b) False alarm rate: discrimination task; (c) Sensitivity: discrimination task; (d) Criterion: discrimination task. Note: AS ¼ Anxiety Sensitivity, (a) low noise4high noise, po.0001; low AS4high AS, po.05, (b) high AS4low AS, po.05, (c) low noise4high noise, po.0001; low AS4high AS, po.01, (d) low noise4high noise, po.0001.

We predicted that an abnormal heartbeat sound would be a fear-relevant stimulus that would be associated with reduced sensitivity in high AS individuals for detection and discrimination tasks. Research suggests that anxious individuals are characterized by decreased attentional capacity, high levels of distractibility, and compromised performance in the face of threat (e.g., Mathews, 1990). Inevitably, these factors will affect sensitivity, evidenced by false alarm rates that exceed the tendency for correct detection (i.e., hit rates). Because AS individuals tend to associate abnormal heartbeats with anxious arousal, their presence may be sufficient to increase fear in individuals who associate such sensations with danger (Asmundson et al., 1993; Ehlers, Breuer, Dohn, & Figenbaum, 1992, 1995). That high AS individuals had lower hit rates and higher false alarm rates in the abnormal but not normal heartbeat condition is fully consistent with abnormal heartbeats functioning as a fear-relevant stimulus. To this end, listeners’ response styles may have been influenced by the verbal instructions prior to the two tasks which included abnormal heartbeat signals (i.e., Experiments 3 and 4) and explicitly conveyed the following information: ‘‘you will hear an abnormal heartbeat’’. Unambiguous information about the presence of abnormal heartbeats may have increased state anxiety, arousal, and distractibility, and therefore decreased ability to perform on a task requiring selective attention to multiple stimuli. A limitation of this study is that

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1752

Experiment 4: High Noise condition 3

2

2

1

1 Z-hit

Z-hit

Experiment 4: Low Noise condition 3

0

0

-1

-1 Low AS -2

Low AS

-2

High AS

High AS

-3

-3 -3

-2

-1

0 1 Z-false alarm

2

3

-3

-2

-1

0 1 Z-false alarm

2

3

Fig. 7. ROC curves for heartbeat discrimination task. Note: AS ¼ Anxiety Sensitivity.

changes in state anxiety, arousal, and distractibility were not systematically assessed. Interestingly, this pattern of decreased sensitivity emerged only in tasks involving abnormal heartbeats. If the normal heartbeat did not evoke the same fear response as the abnormal heartbeat, then a difference in d0 for the normal heartbeat task would not have been expected. This fear process may interfere with an individual’s ability to create a detailed and accurate perceptual representation of contextual stimuli. It follows that if AS individuals do not maintain well-elaborated representations of the abnormal heartbeat stimuli, then they will also have a more difficult time detecting them against a background of white noise and discriminating among them. Recently, researchers have applied statistical Decision Theory to a variety of fields where diagnostic questions exist, and yes–no type decisions are prevalent (Swets et al., 2000). In the field of radiology, for example, studies have used methods of SDT to investigate sensitivity to difficult to perceive mammographic masses (e.g., Chan et al., 1999; Getty, Pickett, D’Orsi, & Swets, 1988). In the presence of true pathology, the consequences of reporting an abnormality that does not exist (false alarm) may ultimately be less costly than failing to detect a cancerous growth (miss). Similarly, high AS individuals may consider it less costly to falsely detect a heartbeat than to miss a heartbeat. Our results suggest that AS was reliably associated with an elevated false alarm rate across heartbeat tasks. Along with a strategy for preventing ‘‘misses,’’ we would also expect a response bias for heartbeat. In this light, and as depicted in the ROC curve for the normal heartbeat detection task (Fig. 3), high AS listeners may employ different strategies depending on the perceived ‘‘risk’’ of their decision. For example, when background noise was low and participants used strict criteria for responding, the ROC curve in the normal heartbeat detection task was similar in shape to the ROC curves for the abnormal and discrimination heartbeat detection tasks. The shape of these curves shows high AS individuals performing less well on the heartbeat tasks relative to low AS individuals. Our findings are consistent with a study by Ehlers and Breuer (1992), who attempted to replicate findings suggesting enhanced perceptual acuity for heartbeats in panic patients. The authors employed a heart rate estimation procedure, which included strict (‘‘count only those heartbeats about which you are sure’’) and lax (‘‘count all of the heartbeats you feel in your body’’) instructions for participants. Panic disorder patients showed a more accurate estimation of heartbeat in the lax condition only. The ‘‘strict’’ condition may have evoked additional anxiety and uncertainty, thereby interfering with attention perceptual acuity. It is important to note methodological differences between the Ehlers and Breuer (1992) study and the current design; in the former, participants were asked to estimate their actual heart rates, whereas the latter employed a simulated heartbeat detection task involving heartbeat

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1753

sounds. Furthermore, because panic history was not assessed in the current study, the relationship between panic disorder, AS and heartbeat perception cannot be assumed. Relevant to the current findings, Thayer, Friedman, Borkovec, Johnsen, and Molina (2000, p. 366) found that GAD patients showed a precognitive defensive cardiac response to threat words over repeated trials. The authors concluded that ‘‘increased sensitivity to threatening information is accompanied by a defensive response to evade that information when detected (Thayer et al., 2000). ‘‘Consistent with the idea of strategic inhibition, the GAD participants in this study showed decreased habituation over time or attentional dysregulation for threat. In a similar light, the high AS group in the current study showed decreased sensitivity to abnormal heartbeat, which may be at least partially explained by motivated inattention or cognitive avoidance in the face of threatening information, and ‘‘an attempt to shield against the impact of threat’’ (Thayer et al., 2000, p. 366). With respect to response threshold (criterion), we hypothesized that if AS individuals pay attention to and furthermore fear the consequences of internal physiological sensations, they would also maintain a lower threshold of responding to heartbeat sounds than low AS individuals. High AS individuals did in fact evidence a lower criterion than low AS individuals when responding to normal heartbeats. Contrary to our expectation, this response strategy shifted in response to abnormal heartbeat sounds, where there were no differences between high and low AS respondents. Several plausible explanations exist. Although high AS individuals may evidence pre-attentive or automatic biases for threat relevant information (e.g., symptoms of autonomic arousal), they may also use more controlled attentional processes when making decisions about threat information. For example, high AS individuals may be quite practiced at noticing, labeling, and reporting the sensations they interpret fearfully (e.g., heart palpitations). For this reason, in a signal detection paradigm involving a condition where AS listeners are plainly informed that they will hear a ‘‘threat’’ signal (i.e., abnormal heartbeat), they may strategically inhibit a known pattern of anxious responding to threat information (Mogg, Bradley, de Bono, & Painter, 1997; Mogg, Mathews, & Weinman, 1987). Similarly, Getty, Swets, Pickett, and Gonthier (1995) provide an example whereby warning system operators deliberately change their strategies of responding to ‘‘danger’’ according to experiences with the signals. The authors report that in many instances, warning system operators have responded slowly or not at all to warnings of danger; the reason given for this lack of response is that ‘‘the warnings have cried wolf too often to be credible’’ (Getty et al., 1995, p. 19). Specifically, in the abnormal heartbeat tasks, high AS individuals may have shifted their response criterion in an effort to compensate for a generally known tendency to ‘‘cry wolf.’’ Moreover, Getty et al. (1995) state that faulty firings of ‘‘warning’’ have not only proven inaccurate responses to danger, but they have also proven serious distractions and interruptions of performance (e.g., decreased sensitivity). The results we see for high AS individuals may indicate not only a system that is attempting not to respond to a perceived danger, but also a system that is distracted at the same time by automatic tendencies to respond to a perceived danger. In other words, patterns of sensitivity and criterion may be influenced by competing processes of anxiety-induced distraction and strategic inhibition during abnormal heartbeat tasks. In contrast, for the normal heartbeat task—in the absence of both explicit threat and the additional attentional demands required by ‘‘threat’’—high AS individuals may have allowed themselves to maintain a more lax criterion, and exercised their more automatic response bias to heartbeats in general. In sum, state anxiety and arousal may interfere with the high AS group’s performance and decision rules. In our sample, high AS participants indicated significantly higher levels of state anxiety (as measured by the STAI-S) than low AS participants prior to performing the auditory detection tasks (Table 3). It may be informative in future studies to obtain physiological measures of state anxiety (e.g., heart rate, GSR) as well as subjective ratings (e.g., normality, danger) about the different heartbeat stimuli throughout the experiments in order to assess changes in response to perceptual demands. It is important to note that it is not clear whether the differences between high and low AS groups on the heartbeat detection tasks are a function of pre-attentive, automatic threat detection or more controlled attentional processes in anxiety maintenance (Mathews & MacLeod, 1986). Although in general, signal detection tasks rely primarily on controlled processing, the role of automatic processing bias cannot be disentangled from controlled processing strategies within this set of experiments. Moreover, there may be

ARTICLE IN PRESS 1754

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

reciprocal influences between controlled and pre-attentive or automatic processing. For example, pre-attentive cognitive biases, such as the tendency to catastrophize interoceptive cues, may influence purposeful decisionmaking strategies used by listeners in these heartbeat detection tasks. A difficult issue in any study of interoceptive sensitivity is the use of external stimuli as an analogue to internally perceived events. Thus, one inherent limitation in this study is related to the stimuli used to assess perceptual bias. Though individual differences in cardiac perception may play an etiological role in the development of panic, cardiac perception is neither an essential component of AS nor the fear-of-fear model of panic disorder. Furthermore, although both normal heartbeats and gallop cardiopathies are sounds produced by the human heart, they differ on multiple acoustic dimensions. Qualitative differences in duration and amplitude of the heart sounds may effect perception. In future experiments, it will be important to maximize the extent to which the temporal envelope and spectral characteristics of the heartbeats are matched without altering the qualities unique to the cardiac pathologies. Furthermore, this study did not obtain baseline anxiety ratings about normal and abnormal heartbeat stimuli. The inclusion of such ratings will aid in the interpretation of replication studies. Another limitation of the study is the lack of reliable measurement of panic disorder or panic history to characterize the potential relationship between panic history and the associations between AS and heartbeat perception identified in this study. It will also be important to assess depression as a potential confounding variable in understanding the relationship between AS and heartbeat perception in future studies. Taken as a whole, the ability to perceive changes in heart rate may not differentiate high AS from low AS participants. Rather, specific individual differences may lie in a tendency to misperceive fear-relevant information. This finding supports cognitive models proposing that panic is associated with the catastrophic misinterpretation of physiological sensations and inaccurate beliefs about the harmfulness of these sensations (e.g., Clark, 1988). If, in fact, a panic attack is a clinically relevant ‘‘false alarm’’ (possibly cued by the misperception of a benign physiological experience), and AS is considered a cognitively relevant vulnerability for panic disorder, then the current findings may offer a perceptual analogue for the tendency to misinterpret internal sensations. Results from this study, therefore, may be useful in elucidating mechanisms that may play a role in the development of anxiety and specifically fears of autonomic arousal associated with panic disorder. Conclusions The current pilot study used methods of SDT to study the strategies that high and low AS individuals apply when making choices about weakly perceived normal and abnormal heartbeat sounds. Results from the current study provide preliminary support for the hypothesis that different patterns of auditory perception and judgment of normal and abnormal heartbeat sounds may distinguish individuals who may be vulnerable to develop panic disorder. Furthermore, results offer preliminary evidence for variance within AS that may be characterized by different perceptual patterns for fear-relevant information. This methodology builds upon information-processing methods by incorporating auditory threat-relevant stimuli, and builds upon interoceptive acuity paradigms by assessing preferential processing of threatening material. Moreover, this approach offers wide applicability for both clinically anxious and vulnerable populations. An auditory perception task offers a means to assess information-processing bias in younger children who may not be able to (1) read and/or (2) comprehend lexical tasks involving words specific to the experience of physiological arousal or panic. Delineating perceptual patterns in individuals considered vulnerable for anxiety disorders may be particularly useful in early identification, early intervention, and prevention.

References Amir, N., & Beard, C. (2004). Inhibitory difficulties and anxiety sensitivity. Cognitive Therapy and Research, 28, 283–292. Amir, N., McNally, R. J., Riemann, B. C., & Clements, C. (1996). Implicit memory bias for threat in panic disorder: Application of the white noise paradigm. Behaviour Research and Therapy, 34, 157–162. Antony, M. M., Brown, T. A., Craske, M. G., Barlow, D. H., Mitchell, W. B., & Meadows, E. A. (1995). Accuracy of heartbeat perception in panic disorder, social phobia, and nonanxious subjects. Journal of Anxiety Disorders, 9, 355–371.

ARTICLE IN PRESS R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

1755

Asmundson, G. J. G., Sandler, L. S., Wilson, K. G., & Norton, G. R. (1993). Panic attacks and interoceptive acuity for cardiac sensations. Behaviour Research and Therapy, 31, 193–197. Barlow, D. (1988). Anxiety and its disorders: The nature and treatment of anxiety and panic. New York: Guilford Press. Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety disorders and phobias. New York: Basic Books. Chambless, D. L., & Gracely, E. J. (1989). Fear of fear and the anxiety disorders. Cognitive Therapy and Research, 13, 9–20. Chan, H-P., Sahiner, B., Helvie, M. A., Petrick, N., et al. (1999). Improvement of radiologists’ characterization of mammographic masses by using computer-aided diagnosis: An ROC study. Radiology, 21, 817–827. Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24, 461–470. Clark, D. M. (1988). A cognitive model of panic attacks. In S. Rachman, & J. D. Maser (Eds.), Panic: Psychological perspectives (pp. 71–89). Hillsdale, NJ: Erlbaum. Criley, J. M., Criley, D., & Zalace, C. (1997). The physiological origins of heart sounds and murmurs: The unique interactive guide to cardiac diagnosis. Boston: Blaufuss Medical Multimedia. Ehlers, A., & Breuer, P. (1992). Increased cardiac awareness in panic disorder. Journal of Abnormal Psychology, 101, 371–382. Ehlers, A., Breuer, P., Dohn, D., & Figenbaum, W. (1995). Heartbeat perception and panic disorder: Possible explanations for discrepant findings. Behaviour Research and Therapy, 33, 69–76. Ehlers, A., Margraf, J., Davies, S., Roth, W. T., & Birmaunr, N. (1988). Selective processing of threat cues in subjects with panic attacks. Cognition and Emotion, 2, 201–219. Ehlers, A., Margraf, J., Roth, W. T., Taylor, CB., & Birbaumer, N. (1988). Anxiety induced by false heart rate feedback in patients with panic disorder. Behaviour Research and Therapy, 26, 1–11. Eley, T. C., Stirling, L., Ehlers, A., Gregory, A. M., & Clark, D. M. (2004). Heart-beat perception, panic/somatic symptoms and anxiety sensitivity in children. Behaviour Research and Therapy, 42, 439–448. Gernsbacher, M. A., Verner, K. R., & Faust, M. E. (1999). Investigating differences in general comprehension skill. Journal of Experimental Psychology—Learning Memory and Cognition, 16, 430–445. Gescheider, G. A. (1997b). Psychophysics: The fundamentals. Mahwah, New Jersey: Lawrence Ehrlbaum Associates. Getty, D. J., Swets, J. A., Pickett, R. M., & Gonthier, D. (1995). System operator response to warnings of danger: A laboratory investigation of the effects of the predictive value of a warning on human response time. Journal of Experimental Psychology: Applied, 1, 19–33. Getty, D. M., Pickett, R. M., D’Orsi, C. J., & Swets, J. A. (1988). Enhanced interpretation of diagnostic images. Investigative Radiology, 23, 240–252. Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics. New York: Wiley. Kagan, J. (1994). Galen’s prophecy: Temperament in human nature. New York: Basic Books. King, R., Margraf, J., Ehlers, A., & Maddock, R. (1986). Panic disorder—Overlap with symptoms of somatization disorder. In I. Hand, & H. U. Wittchen (Eds.), Panic and phobias (pp. 72–77). Berlin, FRG: Springer. Knoll, J. F., Folten, A., & Hodapp, V. (1991). Heart rate perception ability: A comparison of two methods. Journal of Psychophysiology, 5, 113. Maller, R. G., & Reiss, S. (1992). Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety Disorders, 6, 241–247. MacLeod, C. M. (1991). Half a century of research on the stroop effect: An integrative review. Psychological Bulletin, 109, 163–203. MacLeod, C., Mathews, A., & Tata, P. (1986). Attentional bias in emotional disorders. Journal of Abnormal Psychology, 95, 15–20. Macmillan, N. A., & Creelman, D. C. (1991). Detection theory: A user’s guide. Cambridge: Cambridge University Press. Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research and Therapy, 28, 455–468. Mathews, A., & MacLeod, C. (1986). Discrimination of threat cues without awareness in anxiety states. Journal of Abnormal Psychology, 95, 131–138. McCabe, R. E. (1999). Implicit and explicit memory for threat words in high- and low-anxiety-sensitive participants. Cognitive Therapy and Research, 23, 21–38. McNally, R. J. (1990). Psychological approaches to panic disorder: A review. Psychological Bulletin, 108, 403–419. McNally, R. J. (1999a). Theoretical approaches to the fear of anxiety. In S. Taylor (Ed.), Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety (pp. 3–16). Mahwah, NJ: Lawrence Erlbaum Associates. McNally, R. J. (1999b). Anxiety sensitivity and information-processing biases for threat. In S. Taylor (Ed.), Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety (pp. 183–198). Mahwah, NJ: Lawrence Erlbaum Associates. McNally, R. J., & Foa, E. B. (1987). Cognition and agoraphobia: Bias in the interpretation of threat. Cognitive Therapy and Research, 11, 567–581. McNally, R. J., Foa, E. B., & Donnell, C. D. (1989). Memory bias for anxiety information in patients with panic disorder. Cognition and Emotion, 3, 27–44. McNally, R. J., Riemann, B. C., & Kim, E. (1990). Selective processing of threat cues in panic disorder. Behaviour Research and Therapy, 28, 407–412. McNally, R. J., Hornig, C. D., Hoffman, E. C., & Han, E. M. (1999). Anxiety sensitivity and cognitive biases for threat. Behavior Therapy, 30, 51–61. Mogg, K., Bradley, B. P., de Bono, J. D., & Painter, M. (1997). Time course of attentional bias for threat information in non-clinical anxiety. Behaviour Research and Therapy, 35, 297–303. Mogg, K., Mathews, A., & Weinman, J. (1987). Memory bias in clinical anxiety. Journal of Abnormal Psychology, 96, 94–98. Pashler, H. E. (1998). The psychology of attention. Cambridge: MIT Press.

ARTICLE IN PRESS 1756

R.A. Pollock et al. / Behaviour Research and Therapy 44 (2006) 1739–1756

Pauli, P., Marquardt, C., Hartl, L., Nutzinger, D. O., Holzl, R., & Strain, F. (1991). Anxiety induced by cardiac perceptions in patients with panic attacks: A field study. Behaviour Research and Therapy, 29, 137–145. Pollock, R.A., Amir, N., Freshman, M., Marks, L.E. (2000). The emotional perception of auditory heartbeat. Poster, presented at the 34th meeting of the association for the advancement of behavior therapy. New Orleans. Reiss, S., & McNally, R. J. (1985). Expectancy model of fear. In S. Reiss, & R. R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp. 107–121). San Diego, CA: Academic Press. Reiss, S., Peterson, R., Gursky, D., & McNally, R. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behavioural Research and Therapy, 24, 1–8. Reiss, S. (1987). Theoretical perspectives on the fear of anxiety. Clinical Psychology Review, 7, 585–596. Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods and data analysis. New York: McGraw-Hill. Schacter, D. L. (1992). Understanding implicit memory. A cognitive neuroscience approach. American Psychologist, 47, 559–569. Schands, H. C., & Schor, N. (1982). The modern syndrome of phobophobia and its management. In R. L. Dupont (Ed.), Phobia: A comprehensive summary of modern treatments (pp. 107–116). New York: Brunner/Mazel. Schmidt, N., Lerew, D., & Jackson, R. (1997). The role of anxiety sensitivity in the pathogenesis of panic: Prospective evaluation of spontaneous panic attacks during acute stress. Journal of Abnormal Psychology, 106, 355–364. Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1999). Prospective evaluation of anxiety sensitivity in the pathogenesis of panic: Replication and extension. Journal of Abnormal Psychology, 108, 532–537. Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the state-trait anxiety inventory. Palo Alto, CA: Consulting Psychologist Press. Stewart, S. H., Buffett-Jerrott, S. W., & Kokaram, R. (2001). Heartbeat awareness and heart rate reactivity in anxiety sensitivity: A further investigation. Anxiety Disorders, 15, 535–553. Swets, J. A., Dawes, R. M., & Monahan, J. (2000). Psychological science can improve diagnostic decisions. Psychological Science in the Public Interest, 1, 1–24. Taylor, S. (1995). Anxiety sensitivity: Theoretical perspectives and recent findings. Behaviour Research and Therapy, 33, 243–258. Thayer, J. F., Friedman, B. H., Borkovec, T. D., Johnsen, B. H., & Molina, S. (2000). Phasic heart period reactions to cued threat and nonthreat stimuli in generalized anxiety disorder. Psychophysiology, 37, 361–368. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1988). Cognitive psychology and the emotional disorders. Chichester, UK: Wiley.

Further reading Amir, N., Coles, M. E., Brigidi, B., & Foa, E. B. (2001). The effect of practice on recall of emotional information in individuals with generalized social phobia. Journal of Abnormal Psychology, 110, 76–82. Pollock, R.A. (2001). Anxiety sensitivity and the auditory perception of heartbeat. Doctoral dissertation, Yale University, 2001. Dissertation Abstracts International, 62, 1593.