Biological Psychology 67 (2004) 343–358

Differences in startle modulation during instructed

threat and selective attention
Koen B.E. Böcker a,∗ , Johanna M.P. Baas b , J. Leon Kenemans b ,
Marinus N. Verbaten b
aDepartment of Psychopharmacology, Faculty of Pharmaceutical Sciences, Sorbonnelaan 16,
NL 3584 CA Utrecht, The Netherlands
b Department of Psychopharmacology, Utrecht Institute for Pharmacological Studies, Faculty of Pharmaceutical

Sciences, Rudolf Magnus Institute for Neurosciences, Utrecht University, Utrecht, The Netherlands

Received 10 December 2002; accepted 5 January 2004

Available online 11 March 2004

Abstract

This study investigated whether attentional processes contribute to fear-potentiated startle. Ten sub-
jects participated in a threat of shock experiment and an attentional control condition. In the threat of
shock experiment, visual cues indicated whether or not an aversive shock might occur. In the attentional
control, the shocks were replaced by faint vibrotactile stimuli that had to be counted. The P300 ampli-
tudes of the ERP evoked by the visual cues did not differ under threat and counting, which suggested
that both conditions engaged attention to the same extent. In contrast, startle potentiation in the threat
condition was an order of magnitude larger than the marginally significant attentional startle facilita-
tion in the counting condition. These results indicate that an attentional contribution to fear-potentiated
startle under the present experimental conditions is small. In addition, contextual effects of threat of
shock became manifest as baseline startle was facilitated relative to the attention condition. This may
reflect a more sustained state of anxiety on which cue-specific fear responses are superimposed.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Fear; Anxiety; Attention; Threat of shock; Startle; P300

1. Introduction

An increasing body of literature demonstrates that the startle reflex is enhanced during
aversive emotional states. This effect has proved especially robust with fear manipulations,

∗ Corresponding author.
E-mail address: K.B.E.Bocker@pharm.uu.nl (K.B.E. Böcker).

0301-0511/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.biopsycho.2004.01.001
344 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

and is generally referred to as fear-potentiated startle (FPS, e.g., Davis, 1992). In humans,
FPS has been demonstrated as a result of aversive conditioning (e.g., Hamm and Vaitl, 1996;
Lipp et al., 1994; Lipp et al., 1997; Lipp et al., 1998) or by instructed threat of shock (Grillon
et al., 1991; Grillon et al., 1993). Furthermore, Lang et al. (1995) developed a procedure
to induce negative affect by presenting intrinsically aversive lead stimuli, such as pictures
showing mutilated bodies (International Affective Picture System or IAPS). Presentation of
aversive pictures also leads to startle facilitation (see e.g., Lang et al., 1997, for a review).
However, startle magnitude can also be modulated by attention (e.g., Filion et al., 1998).
Initially, attentional startle modulation has been regarded as modality specific. Eye blink
startle is facilitated when the startle probes and the attended lead stimuli are in the same
modality, but is inhibited when they are in different modalities (e.g., Putnam, 1990). More
recently, it was demonstrated that attention can modulate startle in a modality-nonspecific
way. A number of studies reported larger auditory startle response during task-relevant
than during task-irrelevant lead stimuli, regardless of lead stimulus modality (Lipp et al.,
1997, 1998; Lipp, 2002; Böhmelt et al., 1999). This suggests the possibility that emotional
processing indirectly affects startle by engaging attention. In this paper, we address this
issue by comparing startle modulation during threat of shock with startle modulation in an
attentional task.
Behavioral research based on a cognitive-motivational approach to anxiety has indeed
demonstrated the existence of an attentional bias towards threatening stimuli (Mogg and
Bradley, 1998). Subjects who reported relatively high state anxiety in the experimental
situation were faster to react to dot probes at locations where a threatening word had been
shown just before (Mogg et al., 1997). In non-anxious subjects, event-related potentials
(ERPs) measured in an instructed threat of shock procedure indicated selective attention to
threat cues from 60 ms onwards (Baas et al., 2002a).
In previous studies that compared FPS and attentional startle modulation, occurrence of
the latter depended on the nature of the attention task. In the attention tasks, the lead stimuli
were associated with a non-aversive target stimulus, instead of the aversive stimulus used
in FPS (usually an electrical shock). When the target stimulus called for a motor response
findings were equivocal. In earlier studies, no startle modulation was observed (Hamm and
Vaitl, 1996; Lipp et al., 1994). In later studies, that used performance feedback combined
with monetary reward, attentional startle facilitation has been observed (Lipp, 2002; Lipp
et al., 2003b). Feedback has been reported to increase attentional startle modulation, at least
with easy tasks (Lipp, 2002; Lipp and Hardwick, 2003). In other studies, the attentional
task was to count occurrences of certain stimuli in a series of sequentially presented stimuli,
but only when presented with a slight increase in duration (first introduced by Filion et al.,
1993). This task requires participants to detect the occurrence of stimuli with the relevant
features, and to pay attention to the duration of those relevant stimuli to count the number of
longer-than-usual occurrences (discrimination and counting task). In these studies, that used
acoustic startle probes, attentional startle facilitation was observed with auditory and visual
lead stimuli (Lipp et al., 1997, 1998), but not with somatosensory lead stimuli (Lipp et al.,
1998). Because FPS was observed in all three modalities, it was concluded from the lack of
attentional facilitation in the somatosensory data that FPS could not be reduced to attentional
facilitation (Lipp et al., 1998; consult Lipp et al., 1997 for additional considerations). This
result is especially relevant for the present purposes because it involved relevant stimuli
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 345

in the same modality as that of the aversive shock. However, according to a more recent
study the absence of attentional startle facilitation in this modality was due to the specific
(pulsating) nature of lead stimulus (Lipp et al., 2003a).
Finally, the time-course of valence and attentional startle modulation has been compared
using a wide range of probe positions, from 120 ms to several seconds. Depending on
the exact manipulations valence effects either preceded (Vanman et al., 1996; Lipp et al.,
2001a) or lagged behind attentional modulation (Bradley et al., 1993; Robinson and Vrana,
2000), suggesting at least partially independent modes of startle modulation. Interpretation
of these findings is complicated by the fact that probe position per se also influences startle
magnitude. That is, at short (<1 s) cue–probe intervals startle is inhibited, while at longer
(>1 s) intervals, which are the main focus of the present paper, it is enhanced by attention
and negative affect (e.g., Filion et al., 1993, 1998). Yet, both Vanman et al. (1996) and Lipp
et al. (2001a) report an additive effect of affective startle modulation (negative affect cues
evoking larger startle) over the whole range of probe positions tested (250–4450 ms).
In the present study, the time-course and extent of differential stimulus processing were
assessed at the cortical level by recording ERPs. We present P300 data acquired with the
startle data in both the threat and attention tasks. The P300 amplitude reflects the relevance
of (Spencer et al., 2001a) or amount of attention devoted to (Kok, 2001) event categorization.
This interpretation extends towards motivational engagement and allocation of attentional
resources to affective pictorial stimuli (Schupp et al., 2000). Thus, the amplitude of the P300
component in the ERP to the visual cue stimuli is employed to assess possible quantitative
differences in selective attention between both conditions.
Startle modulation and P300 were compared during instructed threat of shock and an at-
tentional discrimination and counting task. The experimental conditions were designed such
that both modality-specific and modality-nonspecific selective attention were qualitatively
equal in all experimental conditions (compare Hamm and Vaitl, 1996, and the somatosen-
sory condition in Lipp et al., 1998). That is, the aversive stimulus in the threat condition
and the innocuous stimulus in the control condition were both somatosensory (which was
not always the case in previous studies that compared FPS to attentional startle modulation,
e.g., Lipp et al., 1998, visual and auditory conditions). These consisted of occasional shock
reinforcement and faint vibrotactile stimuli that had to be counted, respectively. Specific
visual cues indicated the possibility that a shock would occur as well as whether or not a
vibrotactile stimulus had to be counted. This is consistent with the majority of FPS studies
in which the cue stimuli are in a different modality than the aversive stimulus.
In order to maintain relatively high levels of selective attention, a fast rate of stimulus
presentation is commonly used in research into selective attention (e.g., Kenemans et al.,
1993). Therefore, cues were presented at a much faster rate than in previous studies of
attentional startle modulation, that is about 1 every 2 s (Baas et al., 1999). The startle probes
were delivered at exactly the same latencies as the shocks and task stimuli in the attention
conditions, i.e., at 1.2 and 1.4 s. In the past we observed a robust FPS effect under these
timing conditions (Baas et al., 1999, 2002a; Böcker et al., 2001a), which also allows for
ERP recordings with an acceptable signal-to-noise ratio. In contrast, the studies reviewed
above typically observed attentional startle facilitation at probe times of at least 3 s after lead
stimulus onset, and did not include the present probe times. Therefore the present results will
provide more insight in the temporal dynamics of FPS and attentional startle enhancement.
346 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

Finally, all participants were tested in both conditions, allowing a within-subjects design to
improve statistical power.
A second aim of the present study was to assess the contribution of context effects in
FPS experiments. Based on animal research a dissociation has been suggested between
contextual anxiety and cue-specific fear (Davis, 1992; Lang et al., 2000; Grillon, 2002). In
humans, startle potentiation by contextual anxiety has been demonstrated with several ex-
perimental procedures, such as re-exposure to the experimental room (Grillon and Morgan,
1999) and unpaired presentations of cues and shocks (Grillon and Davis, 1997; Ameli et al.,
2001). A context effect contributes to blink magnitudes of every startle evoked within the
threatening context, i.e., startle during threat and safe cues, intertrial intervals (Grillon and
Ameli, 1998a), and a baseline period. The present study allows within-subjects comparison
of startle evoked during irrelevant cues in the counting task with startle during safe cues, and
of baseline startle magnitude in the threatening context with that in the counting condition.
These comparisons will be used to assess the contribution of contextual anxiety to affective
startle modulation.
In sum, the present study addressed the contribution of attentional startle facilitation and
contextual anxiety to FPS. The amplitude of the P300 component of the ERP to the visual
cues was used to assess possible differences in selective attention between conditions.

2. Method

2.1. Participants

Ten healthy volunteers (seven females and three males) were recruited through bulletin
boards at Utrecht University (mean age: 21 ± 2 (S.D.) years old). None of them had partic-
ipated in a threat of shock experiment before. They received 22.50 for their participation.

2.2. Apparatus and stimuli

The eye blink component of the startle response was recorded with two tin EMG-electrodes
mounted over the orbicularis oculi muscle just below the right eye, with one electrode di-
rectly below the pupil and the other 1 cm lateral to it. This signal was amplified (13.3K times)
and filtered (3–200 Hz bandwidth) by a Contact Precision Instruments amplifier. The EEG
was recorded from 60 electrodes, mounted in an electrocap with tin electrodes, referenced
to the right mastoid. Vertical and horizontal EOGs were recorded from electrodes above
and below the left eye, and at the outer canthi of both eyes, respectively, for off-line ocular
artifact correction. These signals were amplified (10K times) and filtered (0.05–100 Hz)
by Ampligraph amplifiers. All signals were digitized at 900 Hz using NeuroScan Acquire
software (version 3).
Startle responses were evoked by auditory probes, consisting of 50 ms broadband noise,
presented binaurally at 115 dB(A) via Earlink 3A insert headphones. The visual lead stimuli
were high contrast black and white square wave gratings presented for 1.5 s on a gray
background. There were four possible conjunctions of two spatial frequencies (low 0.53
cycles per degree versus high 4.25 cycles per degree) and two orientations (horizontal and
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 347

vertical). They subtended 7.5◦ of visual angle. During threat of shock, the lead stimuli were
presented within a red frame that was present during the entire block. During the counting
task, the frame was colored green. Stimulus onset asynchrony (SOA) between lead stimuli
varied between 1.9 and 2.3 s (square distribution with 100 ms steps). The stimuli were
presented on a NEC Multisync PC monitor at 114 cm distance.
In the threat of shock condition aversive shocks were applied to the palmar side of the right
wrist by a pair of tin electrodes, placed about 3 cm apart at either side of the median nerve.
The shocks were produced by an Mk IV constant current unit, with an intensity of 1.8 mA
(Grillon et al., 1991; Baas et al., 2002a) and a duration of 100 ms. In the discrimination and
counting condition vibrotactile stimuli were applied to approximately the same site by a
small rotator device enclosed in a rubber cover. The duration of this stimulus was set just
above the individual detection threshold.
To record subjective responses to the experimental conditions two instruments were used.
The first was a 10 cm visual analogue scale (VAS) to rate the lead stimuli (both a priori and
at the end of every experimental block). Anchors at the extremes of the VAS were “very
unpleasant/very negative” at the left and “very positive/very pleasant” at the right. Second,
an abridged version of the state-part of the State-Trait Anxiety Inventory (STAI; Spielberger,
1983) was used, which consisted of the eight items that were found to be most discriminative
in previous studies (Böcker et al., 1999; Baas et al., 2002a). These were Dutch equivalents
of calm, tense, at ease, anxious, comfortable, nervous, relaxed and pleasant. These items
had a median item-total correlation of 0.60 in students and were suitable for forming an
abridged version, according to the manual for the Dutch edition of the STAI, (Van der Ploeg,
2000). The abridged STAI was filled out after reading the written instructions, prior to the
first block of each condition outside the booth. It was again filled out after the final block
of each condition inside the booth.

2.3. Set-up and procedure

The study had a within-subjects repeated measures design with two conditions. Half of
the participants started with the threat of shock condition and the other half with the count-
ing task. Each condition comprised four blocks. As is common in ERP research, in each
block a different conjunction of spatial frequency (SF) and orientation (e.g., high spatial
frequency; horizontal orientation) was associated with the shock or with to be counted vi-
brotactile stimuli. These lead stimuli served as threat and relevant cues respectively, and are
designated as ++ below. In addition, there were three safe or irrelevant cues in either con-
dition. One had the same spatial frequency (SF) as ++, but different orientation (e.g., high
SF; vertical; designated SF+), one had the same orientation but different spatial frequency
(e.g., low SF; horizontal; designated ORI+), and the third had different spatial frequency
and orientation (e.g., low SF; vertical; designated −−). With this set of stimuli effects of se-
lective attention to a certain stimulus feature (e.g., high SF, shared by SF+ and ++ stimuli)
can be separated from target detection and motor preparation, because the latter only affect
targets (++ stimuli). In fact, SF+ stimuli do evoke relatively early attention-related ERP
components in both simple RT conditions (e.g., Kenemans et al., 1993) and in a threat of
shock condition (Böcker et al., 2001b). Finally, previous research has shown strong FPS
with this set of stimuli (Böcker et al., 1999, 2001a).
348 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

The written instructions for either threat of shock or discrimination and counting were
presented outside of the experimental room. The threat instructions told the subjects that
one of the visual cues would occasionally be paired with an aversive electric shock that
was possibly slightly painful, but not harmful. It was explained that shocks would be pre-
sented in total about 15 times during this cue, and emphasized that it would not occur
during any of the other cues. The instructions for the counting task were to pay atten-
tion to the cues and to count occasionally occurring faint vibrations at their right wrists,
only when presented during the relevant cue and not during the others (catch trials). To
increase motivation for complying with the instructions, subjects were told that a block
would have to be repeated if the number they reported was wrong (none was repeated
in reality). Five subjects made one error (out of eight counts) and a sixth subject two
errors.
Upon arrival at the laboratory, participants read and signed an informed consent form
that explained the experimental procedures. Recording electrodes were attached. After
reading the instructions pertaining to the impending condition, participants filled out the
abridged STAI and were seated in a dimly lit, sound-attenuating booth in front of a PC
monitor. For participants who started in the threat condition, shock electrodes had been
attached outside the recording chamber and were connected to the shock unit inside the
chamber. This was explicitly noted to the subjects. For the other participants, the vibro-
tactile stimulator was attached to their wrist. After finishing the first task, participants
filled out the abridged STAI, were taken out of the booth for about 10 min in which
shock electrodes were attached or removed and the instructions for the second part were
presented.
Once in the cabin, the experiment started with the presentation of six startle probes (SOA
5–7 s) for the subject to get used to the probes and to induce some habituation. Subsequently
four blocks were run per condition. Each experimental block started with a visual display
that indicated which cue could be paired occasionally with an aversive shock or a to be
counted vibrotactile stimulus. This display was followed for 27 s by a gray screen with
a red or green fixation cross (for threat and counting, respectively) during which three
baseline startles were evoked (SOA 10, 11 or 12 s; average 11.3 s). This baseline startle
was used to assess context effects. Next a pseudo-random series of 216 cues (containing
54 cues of each type) was presented and a block finished with scoring each cue on a
VAS.
In each block, three to five threat cues were paired with a shock, and three to five pre-
sentations of each cue in the counting task were paired with a vibrotactile stimulus. The
experiment contained 15 paired cues of each type in total. Furthermore, in each block a
startle probe accompanied six cues of each type (11%). The resulting interval between
subsequent startle probes was 18.9 s on average. All shocks, vibrotactile stimuli and startle
probes were presented 1.2 s or 1.4 s following cue onset.
In the middle of each block of 9 min duration there was a 1 min break. During the
break subjects were looking at a blank screen, without any further instructions. In this
interval, three startles were elicited with a 19 and 22 s SOA, respectively. The experiment,
containing two conditions of four blocks each, lasted about 2 h in total, including breaks
(between blocks), application and removal of the recording electrodes and debriefing of the
participants.
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 349

2.4. Data reduction and analysis

The VAS was scored from −5.0 (negative pole) to +5.0 (positive pole) and analyzed
with respect to cue-type (++, SF+, ORI+ and −−). The orbicularis oculi EMG was
epoched from 75 ms before to 125 ms after the onset of the startle probe. These epochs were
high-pass filtered (−3 dB at 28 Hz, −24 dB/octave roll-off, using NeuroScan Edit version
3.2) to remove the eye movement artifact (Van Boxtel et al., 1998). Next the signal was
converted to ␮V, baseline corrected for the pre-probe interval, rectified, smoothed with
a 40 ms wide boxcar filter, and baseline corrected once again. The startle amplitude was
scored as the peak amplitude in a 25–100 ms window following probe onset.
The following procedures were employed on the EMG trace before smoothing and the
final baseline correction to identify trials with zero-responses or corrupt baselines, for
example due to spontaneous blinks. The time-averaged amplitudes during the baseline
interval were transformed to z-values per block. When the baseline z-score on a trial was
larger than 2, the baseline was considered corrupt and the startle magnitude on that trial was
set as missing. In addition, trials on which the peak latency was below 45 ms were also set
to missing (spontaneous blinks early in the analysis window). Furthermore, on each trial the
S.D. of the amplitudes within the baseline period was compared to the S.D. in the analysis
window (i.e., 25–100 ms). If the increase in S.D. between baseline and analysis interval
was less than 55% or the startle magnitude was below zero then it was concluded that no
(discernable) startle response had occurred. In both cases the startle magnitude was set to
0. The reported cut-off scores have been cross-validated with visual inspection by the first
author. The procedure was semi-automatic; the decisions by the algorithm were confirmed
by visual inspection (blind to cue-type). Finally, startle magnitudes were averaged separately
for the first and second half of a block according to the type of cue during which they were
evoked. These averaged magnitudes, which each comprised three startles, were entered into
the statistical analyses.
Pure effects of context, not affected by residual effects from nearby threat cues or shock
delivery, were evaluated on baseline startle. A preplanned comparison between the baseline
startle magnitudes in the first block of each condition was performed as the purest test of
the context effect, as this measurement preceded presentation of threat cues and shocks.
The EEG was epoched from 250 ms before to 1200 ms after the onset of the cues. These
epochs were low-pass filtered (−3 dB at 30 Hz, −12 dB/octave roll-off) and reduced to
100 Hz sampling rate. Eye blink and movement artifacts were removed by regression of
the vertical and horizontal EOG (Kenemans et al., 1991). Artifact-free epochs (tested at
100 ␮V) were averaged per cue-type and condition. The P300 amplitude was measured at
POz (its maximum) as the average amplitude in a window from 300 to 400 ms.
All data were analyzed by univariate ANOVAs, i.e., all within-subject factors were treated
as repeated measures. These factors were context (threat, counting), cue (–, ORI+, SF+,
++; except for STAI data), block (1–4; for VAS and startle data) and half (before break,
after break; for startle only). Greenhouse–Geisser corrected degrees of freedom were used to
compensate for non-sphericity for within-subject factors with more than two levels (Vasey
and Thayer, 1987). The level of significance was set at 5%. Significant interactions were
broken down to simple effects. For factors with more than two levels the post-hoc contrasts
were polynomial for block and inverse Helmert for cue, respectively. The polynomials
350 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

described the time-course of startle magnitudes over time and the Helmert type evaluated
the effect of increased relevance of each cue over the former levels, that is, starting with
ORI+ versus --, ultimately ++ is tested against the mean of all other types. Because startle
level was expected to differ between conditions, the effect sizes of significant contrasts were
calculated (mean contrast value divided by its standard deviation over subjects).
To reduce the influence of individual differences in startle magnitudes on the ANOVA
results all analyses were performed on raw as well as subject-wise z-transformed startle
magnitudes. Furthermore, we expected a difference in startle level between conditions.
Differences in baseline startle can lead to differences in the extent of modulation (Grillon
and Baas, 2002). Therefore the data were also analyzed after z-transformation or expressed
as a percentage of startle magnitude during break, per subject per condition, to compare the
other factors with this main effect removed. Because patterns of significance were highly
similar for the within-subject factors in all four analyses, only the few exceptions will be
reported. In all analyses, a between-subjects factor order was included that indicated whether
subjects started with the threat or with the counting condition. This way the variance due
to the interaction between the order effect and that of the repeated measures factors was
removed from the effect and error terms for the tests of the repeated measures. Nevertheless,
significant interactions of the factor order with either cue or context will be reported.

3. Results

3.1. Questionnaire data

State anxiety was higher after the threat (17/1.2; M/SEM) than after the counting instruc-
tions (14/1.2) as measured by the abridged STAI (possible range 8–32) at the start of each
condition (context main effect F(1, 8) = 13.5, P < 0.01). The second STAI measurement,
after completion of each condition, indicated that state anxiety was higher during threat
(21/1.5) than during the three safe cues (16/1.0) and during the counting condition (15/0.9)
(Helmert contrast F(1, 7) = 22.9, P < 0.005; incomplete data for one participant; Con-
dition main effect (F(2, 14) = 15.8, P < 0.005, Greenhouse–Geisser ε = 0.81). Anxiety
during the safe cues did not differ significantly from that during counting (F(1, 7) = 1.7,
ns).
On the VAS (possible range −5 through 5) the visual ++ cues in the threat condition
were evaluated more negative (−2.3/0.5) than the safe ones (0.1/0.4); Helmert contrast
F(1, 8) = 16.6, P < 0.005). There were no significant differences in the appreciation of
the cues in the counting condition (overall mean, −0.3/0.2; F < 1, ns). Together, this gave
rise to a context by cue (F(3, 24) = 11.7, P < 0.005, ε = 0.50) interaction and a cue main
effect (F(3, 24) = 13.8, M, ε = 0.50).

3.2. Startle magnitudes: cue and context effects

As shown in Fig. 1 (left panel) startle elicited during threat-cues was larger than that
evoked during safe cues. In contrast, startle evoked during relevant cues was of almost the
same magnitude as that evoked during irrelevant ones. Furthermore, startle evoked during
safe cues was larger than that evoked during irrelevant ones. These observations were
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 351

Counting Count first

Startle Threat
Base-line startle Threat first
120 160

140
100
120
Magnitude (µV)

Magnitude (µV)

80
100

60 80

60
40
40
20
20

0 0
-- ORI+ SF+ ++ 1 2

Cue Condition

Fig. 1. Grand average startle magnitude for different cue types during both conditions (at the left). Baseline startle
evoked prior to those conditions, reflecting context effects, is shown per group as a function of condition (in
chronological order) and block (at the right).

confirmed by the results of the ANOVA. A main effect of cue (F(3, 24) = 9.0, P < 0.05,
ε = 0.37) was found, due to the startle potentiation for the ++ stimuli (overall Helmert
contrast (F(1, 8) = 9.7, P < 0.05) both in the threat (F(1, 8) = 9.8, P < 0.05) and
(marginally) in the counting condition (F(1, 8) = 5.0, P = 0.06; after z-transformation
F(1, 8) = 5.8, P < 0.05). However, the latter was much smaller (4±6.0 ␮V, M±S.D., effect
size 0.71) than the former (30 ± 28 ␮V, effect size 1.0 increasing to 1.8 after subject-wise
z-transformation; context by cue Helmert contrast F(1, 8) = 9.3, P < 0.05; context by cue
in the main analysis F(3, 24) = 8.9, P < 0.05, ε = 0.36). After removal of the context effect
(see below) by condition-wise z-transformation, similar effects of cue and its interaction
with context remained (F(3, 24) = 19.1, P < 0.005, ε = 0.46 and F(3, 24) = 17.2,
P < 0.0005, ε = 0.59, respectively).
A main effect of context (F(1, 8) = 17.2, P < 0.005) reflected larger startle magnitudes
in the threat than the counting condition. The context effect was significant for all types
of cues (all F(1, 8) > 13.5, P < 0.01). Therefore, the main effect does not just reflect
the potentiation during the threat cue, and can be interpreted as a more general effect.
However, the context effect was larger in the group starting with the threat condition (38 ±
16 ␮V; F(1, 4) = 26.9, P < 0.01) than in the group starting with the counting condition
(13 ± 22 ␮V; F(1, 4) = 1.7, ns). This apparent interaction only became significant after a
subject-wise z-transformation (F(1, 8) = 7.7, P < 0.05). Finally, the effects of decreasing
magnitudes over and within blocks showed up as significant effects of block (F(3, 24) =
21.9, P < 0.0005, ε = 0.50), half (F(1, 8) = 41.0, P < 0.0005) and their interaction
(F(3, 24) = 6.3, P < 0.05, ε = 0.58).
352 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

3.3. Baseline startle: context effects

As shown by Fig. 1 (right panel), baseline startle magnitudes were larger in the threat
than in the counting condition for the threat-first group (F(1, 4) = 13.9, P < 0.05). In
contrast, there were no significant context effects in the counting-first group, if anything
the context effect went in the opposite direction in this group (larger amplitude during
counting; F(1, 4) = 1.3, ns; context by order F(1, 8) = 11.5, P < 0.01). The baseline
preceding the first block of each condition showed the same pattern of results (context
by order F(1, 8) = 13.3, P < 0.01). Furthermore, baseline startle decreased over blocks
(F(3, 24) = 13.1, P < 0.005, ε = 0.44; see Fig. 1, right panel). This gradual decrease
over trials added to the context effect (larger baseline startle in the threat condition) in the
threat-first group, and counteracted this effect in the counting-first group. Therefore we
compared the baseline startle of the final block of the first task with those of the first block
of the second task. For the threat-first group baseline startle increased by 3 ± 10 ␮V (F < 1,
ns), while for the counting-first group the increase measured 51 ± 15 ␮V (F(1, 4) = 11.5,
P < 0.05; block by context F(1, 8) = 7.1, P < 0.05).

3.3.1. P300
P300 amplitudes at POz (Fig. 2) were larger for threat and relevant cues than for other
types of cues (Helmert contrast, F(1, 8) = 23.6, P < 0.005). Furthermore, P300 was larger
for SF+ than for other safe or irrelevant cues (F(1, 8) = 18.3, P < 0.005). P300 amplitudes
and modulations were nearly identical in both contexts. Accordingly, the ANOVA showed

Counting - POz Threat - POz

7.5 7.5
--
ORI+
SF+
++
5.0 5.0
Amplitude (µV)

2.5 2.5

0.0 0.0

-2.5 -2.5
-250 0 250 500 750 -250 0 250 500 750
Latency (ms)

Fig. 2. Grand average ERP for different cues in both contexts at electrode POz. The largest differences are in the
P300 range (analysis window indicated by vertical lines).
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 353

a cue main effect (F(3, 24) = 20.6, P < 0.005, ε = 0.38) in the absence of a context main
effect (F(1, 8) = 2.1, ns) or a context by cue interaction (F < 1, ns; for threat against
relevant cues F(1, 8) = 1.3, ns).

4. Discussion

The present study addressed the question whether, and if so to what extent, startle mod-
ulation in a fear-potentiated startle experiment reflects the engagement of attention. Fear
potentiated startle, induced by a threat of shock procedure, was compared to attentional
startle facilitation, induced by a discrimination and counting task. State anxiety measures
and subjective evaluations of the cues confirmed that threat cues induced negative affect. In
fact, state anxiety measures in the threat condition rose to levels at which an attentional bias
towards threat can be anticipated (Mogg et al., 1997). As expected, the threat cues induced
fear potentiation of startle magnitude. Attentional startle facilitation in the counting task
was an order of magnitude smaller than fear potentiation in the threat condition (4 ␮V ver-
sus 30 ␮V). In contrast, P300 amplitudes were about equal for threat and for attention cues,
indicating that comparable amounts of attention-related cortical activation were elicited by
the two kinds of cue.
With respect to the main issue of this paper, we conclude that the contribution of atten-
tional startle facilitation to the observed potentiation during threat of shock was small, if
not negligible. Yet, P300 amplitudes suggested that the relevance of categorization of the
cues in the threat of shock and counting conditions, and the amount of attention devoted to
this process (Spencer et al., 2001a; Kok, 2001) did not differ. Moreover, skin conductance
recordings from previous studies suggested that arousal and orienting did not differ between
threat and attention conditions either (e.g., Hamm and Vaitl, 1996; Lipp et al., 1997, 1998).
Therefore, the extrapolation from the small attentional startle facilitation in the counting
condition to a presumably small contribution of attentional facilitation of startle in the threat
condition seems to be justified.
Similar conclusions have been drawn based on comparisons between fear potentiation of
startle and modality-specific (Lipp et al., 1994; Siddle et al., 1997) or modality nonspecific
attentional startle facilitation, using either a reaction time (Hamm and Vaitl, 1996) or a
counting task (Lipp et al., 1998, somatosensory modality). The latter study reports a robust
FPS effect and, under comparable experimental conditions, the absence of attentional startle
facilitation. The outcome of the present study replicates this result (although reporting a
small attentional enhancement) using a different methodology (stimulus rate, probe times,
within versus between-subject comparisons).
However, recently Lipp and colleagues have observed reasonably large attentional startle
facilitation with somatosensory targets in both an RT task with visual cues (Lipp, 2002;
Lipp et al., 2003b) and a counting task (Lipp et al., 2003a). The latter study revealed that
the previous failure to observe startle facilitation in such an experiment was due to the
use of a particular (continuously pulsating) somatosensory stimulus. One factor that might
explain the difference between the results reported by Lipp (2002) and Lipp et al. (2003b)
and those obtained in the present study is the absence of performance feedback combined
with monetary reinforcement. In Lipp (2002), a trend towards attentional facilitation was
354 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

observed without performance feedback (16% increase), similar to the trend that was ob-
served in the present counting condition (9% increase). A possible drawback of the use of
performance feedback to enhance the salience of non-aversive task stimuli is that it might
induce affective anticipation. It has been established by different research groups that the
anticipation of affective stimuli, such as feedback stimuli, leads to facilitation of startle
magnitude (Sabatinelli et al., 2001; Lipp et al., 2001b; Dichter et al., 2002; Nitschke et al.,
2002). A second issue is task difficulty. Difficult auditory and visual reaction time tasks
induced more acoustic startle facilitation than an easy task (Lipp et al., 2000; Lipp and
Hardwick, 2003). However, this was reversed in a task requiring more continuous perfor-
mance (Neumann, 2002). A visual continuous performance task might even lead to overall
blink inhibition (see Hazlett et al., 2001 also). This effect was ascribed to modality-specific
attention. In the present study, with comparable timing characteristics, the somatosensory
detection task was around the threshold for detection and might have called for the alloca-
tion of a lot of attentional resources to this modality. A final difference between the studies
by Lipp et al. and the present one is the percentage of targets, which was 25% versus 7%,
respectively. This was equated to the percentage of shocks during the threat condition, but
it might have affected the extent of startle modification. In sum, the extent of attentional
startle modulation to be expected in the present counting condition is difficult to predict
from previous studies. Yet, P300 amplitudes indicate that attentional resources allocated in
the counting and threat conditions did not differ in the present experiment, as was expected
due to the equivalence of task characteristics. Nevertheless, startle modulation was much
larger during threat than during counting, which we attribute to genuine fear potentiation in
the former condition. The small attentional startle facilitation for relevant compared to irrel-
evant cues in the counting condition might reflect a mixture of different, modality-specific
and nonspecific, attentional effects.
A further difficulty in comparing results across studies is that startle modulation is ex-
pressed differently. Nevertheless, it could be argued that the presently used cue-probe inter-
vals were too short to allow for larger attentional startle facilitation (as observed by Vanman
et al., 1996; Lipp et al., 1997, 1998, 2001a, 2003a,b). However, with these same intervals
we did find sizable affective modulation, indicating that the startle system as such can be
modulated within these intervals. Furthermore, it should be noted that probe positions in
all studies, including the present one, were at or near the time that the target (vibrotactile
stimulus) or target feature (longer-than-usual duration) was present. Finally, P300 results
indicated comparable cortical processing in the two cue conditions long before the end
of the interval. So, at best, a putative mechanism for attentional startle modulation and the
mechanism of the presently observed affective modulation may have very different temporal
characteristics. This may well reflect that they also differ in anatomical substrates. Perhaps
most importantly, these arguments again imply that the present affective modulation cannot
be reduced to attentional modulation.
In future studies, it would be interesting to see at what latencies robust attentional startle
facilitation appears. Preferably, such studies should also involve anticipation of positively
valenced stimuli, as well as skin conductance recordings. Generalizing from the available
evidence, their outcome is likely to show a dissociation in time between cued anticipation
of aversive and non-aversive stimuli. That would permit the postulation of two mechanisms
of startle modulation (cf. Lipp, 2002), one in the attentional and another in the affective
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 355

domain. These mechanisms might interact in a task-dependent way, to create diverse patterns
of emotional and attentional startle facilitation, depending on the exact task conditions (e.g.,
Vanman et al., 1996, experiment 1 versus experiment 2).
In the present study, P300 amplitude was used to index attentional processing of both
kinds of cues. Modulations of late positive potentials similar to the present P300 effects have
been observed for positive and negative IAPS pictures compared to neutral pictures with a
similar fast presentation rate (Schupp et al., 2000). These modulations were interpreted to
reflect motivational engagement and commitment of attentional resources. In the present
study, P300 and startle amplitudes showed different patterns of modulation. Unlike startle,
P300 amplitude did not differentiate between relevant and threat cues. Furthermore, P300
was increased for cues that matched the relevant and threat cues on spatial frequency relative
to the other irrelevant and safe cues (Fig. 2), whereas startle was not (Fig. 1). Irrelevant
grating stimuli with the same spatial frequency as target gratings have been shown to engage
selective attention (e.g., Kenemans et al., 1993). The fact that startle modulation does not
differentiate between different types of irrelevant cues again leads to the conclusion that in
the present experiment startle modulation does not reflect selective attention.
A further aim of this study concerned effects of context, rather than of specific cues.
Context effects were indeed observed; startle during the safe cues in the threat context was
larger than that during irrelevant cues in the counting context. A first alternative explana-
tion is that the startle magnitude increase on safe cues resulted from the temporal proximity
of threat cues, although such a residual effect is expected to disappear within 5 s at most
(Grillon et al., 1993). A second possibility is that this increase might have resulted from
shock sensitization in the threat condition (Hamm and Vaitl, 1996). However, neither of
these possibilities can explain the difference in baseline startle between the task conditions.
Baseline startles directly following the instructions were larger in the threat than in the
attention condition, even though the conditions were presented in the same experimental
room. This increase is paralleled by an increase in state anxiety following the threat com-
pared to following the counting instructions. This finding corroborates the interpretation
that at least part of the increase in startle magnitude during safe compared to irrelevant cues
is due to the context of the threat condition.
Other authors also reported similar effects of affective context on startle magnitude
(Grillon and Davis, 1997; Ameli et al., 2001). The presence of shock electrodes attached
to the wrist probably contributed to these effects (Grillon and Ameli, 1998b1 ; Baas et al.,
2002b). Cue and context effects of these types have been interpreted as manifestations of
fear (cue specific) and anxiety (more sustained and generalized), respectively (Davis, 1992;
Lang et al., 2000; Baas et al., 2002b; Grillon, 2002. Fear and anxiety have different but over-
lapping and probably interacting neuronal substrates and have been associated with different
anxiety disorders. With threat of shock, both fear and contextual anxiety determine startle
magnitude during threat cues. Both effects can be observed independently by making the
appropriate comparisons. That is startle during threat cues should be compared with startle
during safe cues for fear, and baseline startle in a threat condition with baseline startle in a

1 The paper does not report tests for the effect of shock electrode placement for safe cues only. Additional

analyses revealed that this effect is indeed highly significant (C. Grillon, personal communication, January 9,
2002).
356 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

neutral task condition for anxiety. The possibility of studying both fear and anxiety in the
same subjects and in the same experimental session is advantageous for pharmacological
and patient research on fear and anxiety in human subjects.

Acknowledgements

The authors want to thank Prof. Dr. Mark Geyer for the useful discussions during his
visiting professorship and Gert Camfferman for his technical assistance. The Netherlands
Organization for Scientific Research NWO is gratefully acknowledged for funding this
project. This research was conducted while the first author was supported by a grant of the
Foundation for Behavioural and Educational Sciences of this organization (#575-25-015) to
Dr. Hilde Huizenga and Prof. Dr. Peter Molenaar in collaboration with Dr. Chantal Kemner,
Dr. J. Leon Kenemans and Dr. Jan de Munck.

References

Ameli, R., Ip, C., Grillon, C., 2001. Contextual fear-potentiated startle conditioning in humans: replication and
extension. Psychophysiology 38, 383–390.
Baas, J., Böcker, K.B.E., Mijnhardt, F., Kenemans, J.L., Verbaten, M.N., 1999. Modulation of startle and P300
amplitudes in an experimental model for anxiety. Psychophysiology 36, S29 (abstract).
Baas, J.M.P., Grillon, C., Böcker, K.B.E., Brack, A.A., Morgan, C.A., Kenemans, J.L., Verbaten, M.N., 2002b.
Benzodiazepines have no effect on fear-potentiated startle in humans. Psychopharmacology 161, 233–247.
Baas, J.M.P., Kenemans, J.L., Böcker, K.B.E., Verbaten, M.N., 2002a. Threat-induced cortical processing and
startle potentiation. Neuroreport 13, 133–137.
Böcker, K.B.E., Baas, J.M.P., Dekker, D.X., Kenemans, J.L., Verbaten, M.N., 1999. An ERP compatible
fear-potentiated startle paradigm. Psychophysiology 36, S34 (abstract).
Böcker, K.B.E., Baas, J.M.P., Kenemans, J.L., Verbaten, M.N., 2001a. Stimulus-preceding negativity induced by
fear: a manifestation of affective anticipation. International Journal of Psychophysiology 43, 77–90.
Böcker, K.B.E., Baas, J.M.P., Kenemans, J.L., Verbaten, M.N., Huizenga, H.M., 2001b. Electrophysiological
manifestations of fear-induced selective attention. Biomedizinische Technik 46 (2), 239–241.
Böhmelt, A.H., Schell, A.M., Dawson, M.E., 1999. Attentional modulation of short-and long-lead-interval mod-
ification of the acoustic startle eyeblink response: comparing auditory and visual prestimuli. International
Journal of Psychophysiology 32, 239–250.
Bradley, M.M., Cuthbert, B.N., Lang, P.J., 1993. Pictures as prepulse-attention and emotion in startle modification.
Psychophysiology 30 (5), 541–545.
Davis, M., 1992. The role of the amygdala in fear and anxiety. Annual Review of Neuroscience 15, 353–375.
Dichter, G.S., Tomarken, A.J., Baucom, B.R., 2002. Startle modulation before, during and after exposure to
emotional stimuli. International Journal of Psychophysiology 43 (2), 191–196.
Filion, D.L., Dawson, M.E., Schell, A.M., 1993. Modification of the acoustic startle-reflex eyeblink: a tool for
investigating early and late attentional processes. Biological Psychology 35, 185–200.
Filion, D.L., Dawson, M.E., Schell, A.M., 1998. The psychological significance of human startle eyeblink modi-
fication: a review. Biological Psychology 47 (1), 1–43.
Grillon, C., 2002. Startle reactivity and anxiety disorders: aversive conditioning, context, and neurobiology. Bio-
logical Psychiatry 52 (10), 958–975.
Grillon, C., Ameli, R., 1998a. Effects of threat and safe signals on startle during anticipation of aversive shocks,
sounds, or airblasts. Journal of Psychophysiology 12, 329–337.
Grillon, C., Ameli, R., 1998b. Effects of threat of shock, shock electrode placement and darkness on startle.
International Journal of Psychophysiology 28, 223–231.
 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358 357

Grillon, C., Ameli, R., Merikangas, K., Woods, S.W., Davis, M., 1993. Measuring the time-course of anticipatory
anxiety using the fear-potentiated startle reflex. Psychophysiology 30, 340–346.
Grillon, C., Ameli, R., Woods, S.W., Merikangas, K., Davis, M., 1991. Fear-potentiated startle in humans: effects
of anticipatory anxiety on the acoustic blink reflex. Psychophysiology 28 (5), 588–595.
Grillon, C., Baas, J.M.P., 2002. Comments on the use of the startle reflex in psychopharmacological challenges:
impact of baseline startle on measurement of fear-potentiated startle. Psychopharmacology 164 (2), 236–238.
Grillon, C., Davis, M., 1997. Fear-potentiated startle conditioning in humans: explicit and contextual cue condi-
tioning following paired versus unpaired training. Psychophysiology 34 (4), 451–458.
Grillon, C., Morgan, C.A., 1999. Fear-potentiated startle conditioning to explicit and contextual cues in Gulf War
veterans with posttraumatic stress disorder. Journal of Abnormal Psychology 108 (1), 134–142.
Hamm, A.O., Vaitl, D., 1996. Affective learning: awareness and aversion. Psychophysiology 33 (6), 698–710.
Hazlett, E.A., Dawson, M.E., Schell, A.M., Nuechterlein, K.H., 2001. Attentional stages of information processing
during a continuous performance test: a startle modification analysis. Psychophysiology 38 (4), 669–677.
Kenemans, J.L., Kok, A., Smulders, F.T., 1993. Event-related potentials to conjunctions of spatial frequency
and orientation as a function of stimulus parameters and response requirements. Electroencephalography and
clinical Neurophysiology 88, 51–63.
Kenemans, J.L., Molenaar, P.C., Verbaten, M.N., Slangen, J.L., 1991. Removal of the ocular artifact from the
EEG: a comparison of time and frequency domain methods with simulated and real data. Psychophysiology
28, 114–121.
Kok, A., 2001. On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology 38, 557–577.
Lang, P.J., Bradley, M.M., Cuthbert, B.N., 1995. International Affective Picture System (IAPS): Technical Manual
and Affective Ratings. The center for research in psychophysiology, Gainesville.
Lang, P.J., Bradley, M.M., Cuthbert, B.N., 1997. Motivated attention: Affect, activation, and action. In: P.J. Lang,
R.F. Simons, M.T. Balaban (Eds.), Attention and Orienting: Sensory and Motivational processes. Lawrence
Erlbaum Associates, Mahwah, NJ, pp. 97–135.
Lang, P.J., Davis, M., Öhman, A., 2000. Fear and anxiety: animal models and human cognitive psychophysiology.
Journal of Abnormal Psychology 61 (3), 137–159.
Lipp, O.V., 2002. Anticipation of a non-aversive reaction time task facilitates the blink startle reflex. Biological
Psychology 59 (2), 147–162.
Lipp, O.V., Cox, D., Siddle, D.A.T., 2001b. Blink startle modulation during anticipation of pleasant and unpleasant
stimuli. International Journal of Psychophysiology 15 (3), 155–162.
Lipp, O.V., Hardwick, S.A., 2003. Attentional blink modulation in a reaction time task: performance feedback,
warning stimulus modality, and task difficulty. Biological Psychology 62 (2), 115–132.
Lipp, O.V., Neumann, D.L., Pretorius, N.R., McHugh, M.J., 2003a. Attentional blink modulation during sustained
and after discrete lead stimuli presented in three sensory modalitiesm. Psychophysiology 40 (2), 285–290.
Lipp, O.V., Neumann, D.L., Siddle, D.A.T., Dall, P.J., 2001a. Assessing the effects of attention and emotion on
startle eyeblink modulation. International Journal of Psychophysiology 15, 173–182.
Lipp, O.V., Sheridan, J., Siddle, D.A.T., 1994. Human blink startle during aversive and nonaversive Pavlovian
conditioning. Journal of Experimental Psychology: Animal Behavior Processes 20 (4), 380–389.
Lipp, O.V., Siddle, D.A.T., Dall, P.J., 1997. The effect of emotional and attentional processes on blink startle
modulation and on electrodermal responses. Psychophysiology 34 (3), 340–347.
Lipp, O.V., Siddle, D.A.T., Dall, P.J., 1998. Effects of stimulus modality and task condition on blink startle
modification and on electrodermal responses. Psychophysiology 35 (4), 452–461.
Lipp, O.V., Siddle, D.A.T., Dall, P.J., 2000. The effect of warning stimulus modality on blink startle modification
in reaction time tasks. Psychophysiology 37 (1), 55–64.
Lipp, O.V., Siddle, D.A.T., Dall, P.J., 2003b. The effects of unconditional stimulus valence and conditioning
paradigm on verbal, skeleto-motor, and autonomic indices of human Pavlovian conditioning. Learning and
Motivation 34, 32–51.
Mogg, K., Bradley, B.P., 1998. A cognitive-motivational analysis of anxiety. Behaviour Research and Therapy 36,
809–848.
Mogg, K., Bradley, B.P., DeBono, J., Painter, M., 1997. Time-course of attentional bias for threat information in
non-clinical anxiety. Behaviour Research and Therapy 35 (4), 297–303.
Neumann, D.L., 2002. Effect of varying levels of mental workload on startle eyeblink modulation. Ergonomics
45 (8), 583–602.
358 K.B.E. Böcker et al. / Biological Psychology 67 (2004) 343–358

Nitschke, J.B., Larson, C.L., Smoller, M.J., Navin, S.D., Pederson, A.J.C., Ruffalo, D., Mackiewicz, K.L., Gray,
S.M., Victor, E., Davidson, R.J., 2002. Startle potentiation in aversive anticipation: evidence for state but not
trait effects. Psychophysiology 39 (2), 254–258.
Putnam, L.E., 1990. Great expectations: anticipatory responses of the heart and brain. In: Rohrbaugh, J., Para-
suraman, R., Johnson Jr., R. (Eds.), Event-Related Brain Potentials. Basic Issues and Applications. Oxford
University Press, New York, pp. 109–129.
Robinson, J.D., Vrana, S.R., 2000. The time-course of emotional and attentional modulation of the startle eyeblink
reflex during imagery. International Journal of Psychophysiology 37 (3), 275–289.
Sabatinelli, D., Bradley, M.M., Lang, P.J., 2001. Affective startle modulation in anticipation and perception.
Psychophysiology 38 (4), 719–722.
Schupp, H.T., Cuthbert, B.N., Bradley, M.M., Cacioppo, J.T., Ito, T., Lang, P.J., 2000. Affective picture processing:
the late positive potential is modulated by motivational relevance. Psychophysiology 37 (2), 257–261.
Siddle, D.A.T., Lipp, O.V., Dall, P.J., 1997. The effect of unconditional stimulus modality and intensity on blink
startle and electrodermal responses. Psychophysiology 34 (4), 406–413.
Spencer, K.D., Dien, J., Donchin, E., 2001a. Spatiotemporal analysis of the late ERP responses to deviant stimuli.
Psychophysiology 38, 343–358.
Spielberger, C.D., 1983. Manual for the State-Trait Anxiety Inventory (STAI-Form Y). Consulting Psychologists
Press, Palo Alto, CA.
Van Boxtel, A., Boelhouwer, A.J., Bos, A.R., 1998. Optimal EMG signal bandwidth and interelectrode distance
for the recording of acoustic, electrocutaneous, and photic blink reflexes. Psychophysiology 35 (6), 690–697.
Van der Ploeg, H.M., 2000. Handleiding bij de Zelf-Beoordelings Vragenlijst (ZBV). Een Nederladstalige bew-
erking van de Spielberger State-Trait Anxiety-Inventory. STAI-DY. Swets Test Publishers, Lisse.
Vanman, E.J., Böhmelt, A.H., Dawson, M.E., Schell, A.M., 1996. The varying time-courses of attentional and
affective modulation of the startle eyeblink reflex. Psychophysiology 33 (6), 691–697.
Vasey, M.W., Thayer, J.F., 1987. The continuing problem of false positives in repeated measures ANOVA in
psychophysiology: a multivariate solution. Psychophysiology 24 (4), 479–486.