Upload
james-stanley
View
213
Download
0
Embed Size (px)
Citation preview
Emotional specificity of startle potentiation during the
early stages of picture viewing
JAMES STANLEYand ROBERT G. KNIGHTDepartment of Psychology, University of Otago, Dunedin, New Zealand
Abstract
Potentiation of the startle blink reflex shortly after picture onset has only been reported in a sample of high animal-fear
participants during fear-evoking stimuli. The experiment tested the emotional specificity of startle reflex modification
at early (300 ms) and late (2–5 s) stages of picture viewing in an unselected sample (n5 55). Participants viewed two
negative picture categories, threat and disgust contents, in addition to neutral and positive pictures. Blink potentiation
occurred for threat contents at both probe times, relative to neutral responses. Disgust contents only potentiated blinks
relative to positive content responses. These results are inconsistent with a basic, prepulse interpretation of early startle
modification and suggest that affective modification can be observed shortly after picture onset, depending on the
specific emotional content of the picture.
Descriptors: Emotion, Startle, Reflex modification, Time course, Prepulse inhibition, Blink reflex
The startle reflex is reliably potentiated when elicited several
seconds after the onset of an emotionally negative picture stim-
ulus (e.g., Vrana, Spence, & Lang, 1988; for an in-depth review,
see Bradley, Cuthbert, & Lang, 1999). At shorter picture-to-
probe onset latencies (o500ms), startle blink modification dur-
ing negative-content pictures has shown either reflex inhibition
relative to neutral content startle (Bradley, Cuthbert, & Lang,
1993; Bradley & Lang, 2001; Levenston, Patrick, Bradley, &
Lang, 2000) or no difference between negative and neutral con-
tent responding (Codispoti, Bradley, & Lang, 2001). The results
of these studies testing the time course of emotional processing
are consistent with findings from the field of prepulse modifica-
tion of the startle reflex. The presentation of a prepulse (usually
an acoustic tone) immediately prior to elicitation of the startle
reflex sets in motion inhibitory mechanisms designed to protect
the processing of the initial stimulus from external influence
(Graham, 1992)Fin this case, protecting the prepulse stimulus
from the subsequent startle probe. The observed dampening of
the elicited reflex is referred to as prepulse inhibition (PPI).
Presentation of an acoustic tone prepulse between 30 and
500ms prior to startle elicitation leads to significant startle blink
inhibition, relative to no-prepulse startle instances (Anthony,
1985; Norris & Blumenthal, 1996). At longer latencies (e.g.,
800ms in Norris & Blumenthal, 1996), prepulse presentation has
no effect on blink magnitude; in experiments using affective pic-
tures as prepulse, startle reflex magnitude at probe latencies
longer than 800ms depends on the emotional valenceFpositive,
neutral, or negativeFof the picture (Bradley et al., 1993; Bradley
& Lang, 2001; Levenston et al., 2000). The reflex inhibition ob-
served for both negative and positive picture stimuli at probe
times earlier than 800ms has been interpreted as evidence for PPI
caused by the onset of a motivationally relevant stimulus (Brad-
ley et al., 1993).
In comparison, a sample of snake/spider aversive individuals
showed significant potentiation of the startle blink reflex as early
as 300ms after picture onset for their feared snake or spider
stimuli (Globisch, Hamm, Esteves, & Ohman, 1999). Low-fear
participants in this study showed no significant blink potentiat-
ion for these picture contents, regardless of the delay between
picture onset and startle probe presentation. These effects were
independent of the length of the picture presentation period (ei-
ther 6 s or 150ms, as a between-subjects manipulation), leading
the authors to conclude that ‘‘fear responses can be activated very
rapidly and with minimal stimulus input’’ (Globisch et al., 1999,
p. 73; Ohman, 1993).
Two other studies have reported blink potentiation at 250ms
for negative relative to positive pictures (Vanman, Boehmelt,
Dawson, & Schell, 1996, Experiment 2; Vanman, Dawson, &
Brennan, 1998). These experiments are somewhat confounded
by the absence of a neutral picture condition to test whether this
difference is due to startle potentiation for negative pictures, in-
hibition for positive pictures, or both. Participants also previewed
the pictures prior to the startle elicitation study, which the authors
concede may speed up subsequent processing of the pictures’
This researchwas completed in partial fulfillment of requirements for a
Doctor of Philosophy degree by James Stanley, under the supervision of
Robert G. Knight. The article was prepared with the aid of an Otago
University BridgingGrant to the first author. Aspects of this researchwere
previously presented at the Annual Meeting of the Society for Psycho-
physiological Research, October 2002, in Washington D.C. We acknowl-
edge the suggestions of three anonymous reviewers for improving this
article. James Stanley is now at the University of Birmingham, England.
Address reprint requests to: James Stanley, School of Psychology,
Birmingham University, Edgbaston, Birmingham, B15 2TT, United
Kingdom; e-mail: [email protected].
Psychophysiology, 41 (2004), 935–940. Blackwell Publishing Inc. Printed in the USA.Copyright r 2004 Society for Psychophysiological ResearchDOI: 10.1111/j.1469-8986.2004.00242.x
935
emotional content (Vanman et al., 1996). In the absence of such
previewing, no modification was observed at 250ms (Vanman et
al., 1996, Experiment 1). These observations of differential startle
for negative and positive contents at early probe times (Globisch et
al., 1999; Vanman et al., 1996, Experiment 2; Vanman et al., 1998)
appear inconsistent with a basic prepulse interpretation of early
picture processing and startle modification.
Hawk and Cook (2000) suggested that differential modifica-
tion for positive and negative pictures at short picture-to-probe
onset latencies may be due to PPI mechanisms being dampened
during viewing of negative pictures. This hypothesis was tested
by an experimental design combining the PPI and emotional
modification paradigms: While participants viewed a series of
affective pictures, an acoustic prepulse was presented 120ms be-
fore startle elicitation (between 3 and 5.5 s after picture onset).
Although the prepulse successfully inhibited startle, the results
indicated that the degree of PPI was consistent across positive,
neutral, and negative picture contents. Bradley and Lang (2001)
proposed that startle modification at the early stages of picture
viewing would depend on both reflex-inhibiting processes initi-
ated by the onset of a motivationally relevant picture, and emo-
tional reflex modification engendered by the picture’s valence
content. For negative picture contents, startle blink modification
will be influenced by both reflex-inhibiting attentional demands
and reflex-augmenting emotional processes, the latter being de-
pendent on the degree of stimulus aversiveness. It appears likely
from the above evidence that the differential startle modification
observed between positive and negative pictures at 250–300ms
(Globisch et al., 1999; Vanman et al., 1996, Experiment 2; Van-
man et al., 1998) is due to the activation of aversive startle po-
tentiation circuits rather than differences in PPI mechanisms
during negative and positive valence categories. The two-process
hypothesis thus provides a reasonable explanation of how the
highly aversive negative stimuli used in Globisch et al. (1999)
produced significant blink potentiation by 300ms.
There are two further important differences between Globisch
et al. (1999) and those experiments showing no blink potentiation
at 300ms (Bradley et al., 1993; Bradley & Lang, 2001; Codispoti
et al., 2001; Levenston et al., 2000). First, the use of a highly
fearful participant sample in Globisch et al. suggests that early
startle potentiation by negative emotional content may be specific
to high-fear individuals only, rather than specific to highly aversive
stimuli. Second, Globisch et al. used only threatening stimuli in
their negative picture category, whereas in the other studies cited
the negative condition consists of a variety of negative stimuli. The
following section considers whether startle modification differs
between certain categories of negative picture stimuli.
Emotional Specificity of Startle Potentiation
A central premise of the affective startlemodification paradigm is
that the systems responsible for these modification processes are
organized by the simple distinction of appetitive and aversive
drive systems (Lang, Bradley, & Cuthbert, 1998). This implies
that discrete emotions within each system (e.g., fear, anger, and
disgust as negative emotions) should produce similar effects on
the startle reflex, provided these foregrounds are of similar in-
tensity (level of emotional arousal).
Testing this hypothesis with negative emotional photographic
stimuli has revealed a more complex pattern of startle modifica-
tion. Balaban and Taussig (1994) included both fear- and dis-
gust-eliciting picture stimuli in their experimental design, but only
observed blink potentiation for fear contents. Lethbridge, Sim-
mons, and Allen (2002) reported no startle modification for pic-
ture contents depicting socially aversive (i.e., non-physical) events.
Levenston et al. (2000) tested startle modification in male
prison inmates, dividing negative contents into physical threat
and ‘‘victim’’ stimuli (photographs of injury or assault between
actors). Threat contents produced greater magnitude startle
blinks than victim contents (C. J. Patrick, personal communi-
cation, April 9, 2003). Preliminary results from undergraduate
female participants have also supported this finding (Bernat,
Patrick, Benning, Blonigen, & Hicks, 2002).
In some experiments, differences in stimulus intensity between
emotional categories can better explain startle modification ef-
fects. Two studies (Bradley, Codispoti, Cuthbert, & Lang, 2001;
Bradley, Codispoti, Sabatinelli, & Lang, 2001) reported increas-
ing blink magnitude across eight specific negative categories, oc-
curring in parallel with increases in the subjective arousal ratings
for these categories (in ascending order of arousal: pollution,
loss, illness, contamination, accidents, mutilation, animal attack,
and human attack). Yartz and Hawk (2002) tested two disgust
picture categories (one with pictures including blood or injury,
the other non-blood disgusting images) alongside positive, neu-
tral, and fear stimuli. For female participants, blink magnitude
during the disgust categories (considered together) was greater
than blink magnitude during fear stimuli (Yartz & Hawk, 2002),
although the disgust-blood category was rated as more arousing
than the other negative categories.
One observation across all of these studies is that physical
threat and fear stimuli potentiate startle more consistently than
other negative contents, and this effect cannot be entirely attrib-
uted to differences in stimulus intensity.
Experimental Design and Hypotheses
The following study explores the possibility that between-study
differences in early startle modification can be attributed to the
types of emotional stimuli included in the negative picture cat-
egory. In addition to positive and neutral pictures, participants in
the current study viewed two categories of negative stimuli, dis-
gust and threat contents, while the startle reflexwas elicited either
300ms after picture onset (early probe time condition) or be-
tween 2 and 5 s after picture onset (late probe time).
Through the testing of an unselected sample of participants,
we hypothesized that early startle potentiation will be observed
for the threat content stimuli only. At the late probe time we
predicted that both negative picture contents would potentiate
startle relative to neutral and positive content trials.
Preliminary evidence from an unselected sample (Stanley,
2003) suggested that startle potentiation occurs for some threat
contents by 300ms. The threat contents used here include human
and animal threat content pictures. Pictures of spiders and snakes
were excluded from the threat condition (a) to allow generali-
zation of the findings of Globisch et al. (1999) to a wider range of
threat stimuli and (b) because previously collected subjective
ratings of pleasantness and arousal indicated that snake and spi-
der pictures were more sensitive than other negative stimuli to
individual differences in fearfulness (Stanley, 2003).
Method
Participants
An unselected sample of 55 University of Otago students
(32 female) completed the experiment, including 30 first-year
936 J. Stanley and R.G. Knight
psychology students who received course credit for participating.
An on-campus student job placement center provided the re-
maining 25 participants, who received NZ$10 for taking part.
Participant age ranged from 18 to 44 years, with a median of 19
years and a mean of 20.95 years.
Four participants did not complete the experiment, due to
computer failure (n5 1), absence of blink responding during the
habituation session (n5 2), and withdrawing from the study
halfway through the experiment (n5 1). The data analysis sec-
tion includes the exclusion criteria used for analyses of blink and
SCR magnitude. From the pool of those who completed the
experiment, 48 participants (30 female) contributed data to the
blink magnitude analyses and 43 participants (25 female) con-
tributed to the SCR analyses.
Picture Stimuli
Twenty-four pictures were selected from the International Af-
fective Picture System (IAPS; Lang, Bradley, & Cuthbert, 1999)
to comprise positive, neutral, and negative emotional conditions.
The negative picture condition consisted of two distinct catego-
ries of pictures: The ‘‘threat’’ category included pictures of hu-
man and animal (excluding snakes and spiders) threat directed
toward the viewer, such as a gun pointed at the screen. The
‘‘disgust’’ category primarily included pictures of mutilated bod-
ies, as well as other pictures chosen to elicit feelings of disgust
(e.g., human feces, dead animals). Picture selection for the two
negative categories included identifying pictures that met the
above descriptions, and then matching stimuli between these
categories on the basis of valence and arousal ratings, collected
using the Self-Assessment Manikin (Hodes, Cook, & Lang,
1985), from an independent sample of 88 participants (64 female)
from the University of Otago. Pictures were selected for the
neutral and positive categories on the basis of having been used in
previously published studies.1 Table 1 reports the means of va-
lence and arousal ratings for the different picture categories from
the sample of 88 participants, as well as the range ofmean ratings
for pictures in each category. A further 36 pictures (14 positive,
14 neutral, 8 negative) were presented as filler material; startle
probes were never presented on these trials. Participants there-
fore viewed a total of 20 pictures from each valence group during
the experiment.
Stimulus Presentation
A custom-designed program (Jones, 2000) controlled both pic-
ture and probe presentation through a Pentium III computer.
The program displayed each picture for 6 s on a 40-cm monitor
(60-Hz refresh rate) positioned 1.2m from the participant; in-
tertrial intervals (ITIs) were randomized, ranging from 15 to 21 s
duration. Startle probe presentation consisted of a 50-ms burst of
white noise at 95 dB (A) with near-instantaneous rise time, pre-
sented binaurally over headphones.
Experimental Design
Startle probe presentation occurred during picture viewing on all
six pictures in each emotional category. Half of these startle
probes occurred 300ms after picture onset (early probe time),
with the other half occurring at random time points between 2
and 5 s after picture onset (late probe time). Means (and SE) of
timings for the late probe instances in each emotional category
were as follows: positive, 3407.9ms (56.8); neutral, 3547.2ms
(71.4); disgust, 3430.4ms (86); threat, 3557.6ms (67.6).
Each emotional category/probe time condition thus consisted
of data from three pictures. Startle was also elicited during one
intertrial interval in each block. The picture set was divided into
five blocks of 12 pictures each, with the provisions that within
each block no two adjacent pictures contained the same valence
content (positive, neutral, negative), and that in each block the
number of pictures from each emotional category (threat, dis-
gust, neutral, positive) and number of early/late probe instances
were equal. There were two possible assignments of probes to
specific pictures in each block, so that a picture was probed at the
early time for one version of that block and in the late interval for
the second version of that block. Block presentation order also
varied, so that there were a total of eight groups for probe as-
signment/block order, counterbalanced across participants.
Physiological Recordings
The disposable electrodes used for recording blink EMG activity
fromorbicularis oculi had a 1.5 � 1.5 cm conductive surface area
(Blue Sensor BRS-50-K), preventing placement with an inter-
electrode distance of 1 cm (as recommended by Fridlund & Ca-
cioppo, 1986). The experimenter consistently used a slightly
wider interelectrode distance of approximately 2 cm (center to
center), with the negative lead electrode centered 1.5 cm below
the eyelid in line with the pupil of the left eye and the positive lead
electrode placed lateral and superior to the first electrode; the
ground electrode was placed on the forehead, below the hairline.
A PowerLab signal processor controlled by the MacLab
Chart program recorded both EMG and SCR signals (recording
software and MacLab hardware designed by ADInstruments).
The raw EMG signal was recorded at 1,000 samples per second,
and filtered off-line with a 28–500Hz bandpass (van Boxtel,
Boelhouwer, & Bos, 1998) prior to rectification and smoothing
with a 10-ms time constant. For the skin conductance recordings,
bright-plated, dry-operation electrodes (i.e., requiring no con-
ductive paste; ADInstruments) were attached to the medial
phalanges of the participant’s index and ring fingers on his or her
nondominant hand, and this signal was digitized at 400 samples
per second. The stimulus presentation computer produced sig-
nalsmarking onset and offset of the picture stimulus and onset of
the startle probe stimulus. Data sampling began 2 s prior to pic-
ture onset and ended 4 s after picture offset.
Calculation of blink response magnitude consisted of iden-
tifying the peak rectified EMGamplitude in a 20–150mswindow
Emotional specificity of early startle potentiation 937
Table 1. Mean Valence and Arousal Ratings for Startled Pictures
by Emotional Category
Dimension Positive Neutral Disgust Threat
ValenceMean 7.69 5.45 2.25 3.08Range 7.1–8.2 5.1–5.9 1.6–3.4 2.7–3.7
ArousalMean 3.81 1.95 5.4 5.34Range 2.4–6.1 1.2–3 3.6–6.5 3.6–6.5
Note: Reported ratings were collected during an earlier study.
1The IAPS numbers for the pictures on which startle was elicited areas follows: threat condition: 1300, 1301, 6244, 6250, 6260, and 6300;disgust condition: 3140, 3150, 3170, 9008, 9140, and 9400; neutral con-dition: 2840, 5740, 7006, 7182, 7503, and 7830; positive condition: 1460,2040, 2530, 7350, and 8030; additional pleasant pictures were 4210 formen and 4520 for women.
following startle probe presentation, and then subtracting the
mean rectified EMG level for the 20-ms period immediately prior
to probe presentation. The SCRmagnitude calculation identified
the largest change in SC amplitude occurring between 1 and 4.5 s
after picture onset. SCR magnitudes were only calculated for
pictures on which startle was elicited, due to the limited selection
of pictures and the specific experimental hypotheses.
These SCR conditions therefore roughly correspond to the
early and late probe time categories: In the ‘‘probe’’ condition,
startle was elicited almost simultaneously with picture onset (i.e.,
the early probe time); for the ‘‘no-probe’’ condition, responses
were only included for late probe time instances when startle was
elicited at least 3.5 s after picture onset (and thus not likely to
influence measurement of SCRs to the picture stimulus in the 1–
4.5 s time window for SCR measurement).
Transformations of blink and SCRdatawere performed prior
to analysis to render the data more suitable for assessing differ-
ences in responding to the within-subjects factors.2 The blink
magnitude transformation converted the uncorrected data into a
T score distribution (mean of 50, standard deviation of 10). The
blink standardization procedure included data from all startle
instances (during picture and ITI trials). Range correction of the
SCR magnitude data (Lykken & Venables, 1971) transformed
each picture’s SCR magnitude to a proportion of the range of
SCRmagnitudes elicited in that participant (range being equal to
the maximal SCR shown minus the minimal SCR shown).
Procedure
Once the participant entered the laboratory, the experimenter
explained that the procedure involved viewing a series of neg-
ative, neutral, and positive emotional pictures, and that the par-
ticipant was to view each picture for its entire time on-screen.
They werewarned that loud noises could be presented at any time
during picture presentation and occasionally during ITIs, but
that these stimuli could be ignored. Participants provided written
informed consent at this stage.
The participants then washed their hands with soap to ensure
accurate SCR measurement. After cleaning the EMG electrode
sites with an ethanol pad and applying OmniPrep skin prepa-
ration paste (D. O.Weaver), the experimenter attached the EMG
and SCR electrodes (details listed above). At this stage the par-
ticipant received three startle probe presentations (15-s ITI) to
ensure that blink responses were observable and relatively noise
free on the EMG record. Presentation of the five picture blocks
followed, and in the short break between blocks (approximately
1min) the experimenter checked that the participant was still
comfortable and wished to continue the experiment. Following
presentation of the five picture blocks, the experimenter removed
the electrodes, debriefed the participant on the experimental hy-
potheses, and thanked them for their assistance. The entire pro-
cedure lasted approximately 40min.
Data Analysis
Exclusion criteria involved identifying those participants who
showed infrequent or no responding to the startle probe. For
blink magnitude analyses, data from a participant were excluded
if they showed an absence of blink responding on all three pos-
sible startle instances in any emotional category/probe time con-
dition (smaller than 10 mVmagnitude in the raw data; this criteria
correctly identified non-blinks with this recording setup). Data
from a participant were included in the SCR analyses provided at
least one nonzero response was available in each emotional cat-
egory/probe presentation condition.
For the analysis of blink magnitude, emotional category
and probe time were within-subjects factors, with gender as
a between-subjects factor. Greenhouse–Geisser corrections
were applied to main effects or interactions involving emotion-
al category, and the correction epsilon (e) values are re-
ported where appropriate. Planned contrasts for significant
effects involving emotional category compared each negative
category separately against neutral and positive responses. To
control for possible Type 1 errors, the alpha level for each test
was set to .0125 (.05 overall alpha, divided by four contrasts).
Contrasts between the two negative categories were not
performed, as the primary experimental hypothesis related to
how these contents modified startle relative to neutral/positive
contents.
A similar strategy was used for SCR magnitude, with emo-
tional category and probe presentation aswithin-subjects factors,
and gender also included. Planned contrasts again followed up
significant effects involving emotional category (with a reduced
alpha level of .0125). The nature of these contrasts differed
slightly from those used for the blink analysis.
Results
Blink Magnitude
In the analysis of blink magnitude by emotional category,
gender was not significant as a main effect, F(1,46)5 0.56,
p5 .46, nor was it involved in significant interactions with
emotional category, Fs(3,138)o1; subsequent analyses excluded
this factor.
Late probe time blinks (M5 51.35, SE5 0.43T units) were
greater in magnitude than early probe time blinks (M5 48.05,
SE5 0.44T units), indicated by a significant main effect
for probe time, F(1,47)5 15.96, po.001. Blink magnitude also
varied on the basis of foreground emotional category,
F(3,141)5 9.76, po.001, e5 .96. Emotional category did not
interact with probe time, F(3,141)5 0.73, p5 .526, e5 .94. Fig-
ure 1 presents means and standard errors of blink magnitude by
emotional category at each probe time. Table 2 presents raw blink
magnitude data for each emotional category and ITI blinks.
Averaged over both probe times, blinks during threat con-
tents were of greater magnitude than neutral or positive blinks,
Fs(1,47)5 17.92 and 23.02, both pso.001.
Disgust content blinks were significantly larger than positive
content blinks, F(1,47)5 7.26, po.001, but did not differ from
neutral content blinks, F(1,47)5 2.72, p5 .106.
SCR Magnitude
To check that arousal was equated in the threat and disgust cat-
egories, a planned comparison was performed on the SCR data
from no-probe trials (see Table 3), and this indicated that there
was no significant difference in SCR magnitude between these
938 J. Stanley and R.G. Knight
2Blink magnitude data were transformed to correct for the fact thatsome participants had greater rawmagnitude blink responses (on averageover emotional categories) than other participants. Thirty-two of the 48participants included in the analysis had a mean blink magnitude of lessthan 60mV; 10 participants had mean magnitudes of between 70 and160mV. Because response variance increases with average magnitude(r5 .78), the decision was made to standardize these data. The maineffect of emotional category was still significant with the raw data,F(3,141)5 7.73, po.001, e5 .86, and the ordering of the cell means forthe Emotional Category � Probe Time conditions was the same with thetwo data sets.
two negative categories, F(1,41)5 .42, p5 .521. The affective
condition SCRs were of greater magnitude than neutral SCRs,
F(1,41)5 8.9, p5 .005.
The main effect of gender on SCR magnitude approached
significance, F(1,41)5 3.04, p5 .089. On average over emo-
tional category, SCRs elicited on trials with a 300-ms probe
were greater in magnitude (M5 0.41, SE5 0.026 of partici-
pant response range) than SCRs on trials with no probe
(M5 0.1, SE5 0.014 of participant response range), F(1,41)5
160.07, po.001. The magnitude of SCRs also depended on pic-
ture emotional category, F(3,123)5 5.65, p5 .002, e5 .85.
Emotional category did not interact with gender, F(3,123)5
1.09, p5 .35, nor with probe presentation, F(3,123)5 1.32,
p5 .273, e5 .9.
Averaged across the two probe presentation conditions, SCR
magnitudes did not differ between threat and positive contents,
F(1,41)5 2.36, p5 .132, or between disgust and positive con-
tents, F(1,41)5 0.11, p5 .739. The contrast between threat and
disgust content SCRs fell just outside of the corrected alpha level,
F(1,41)5 5.86, p5 .02. As the planned comparison reported
above indicated that there was no significant difference between
these categories for the no-probe trials, an exploratory compar-
ison was performed on the probe data, which indicated that
SCRs were of greater magnitude during threat relative to disgust
pictures (see Table 3) when startle was elicited 300ms after pic-
ture onset, F(1,41)5 7.77, p5 .008.
A final test indicated greater SCR magnitude for the three
affective conditions (threat, disgust, positive) compared to neu-
tral responses, F(1,41)5 17.98, po.001.
Discussion
Blink responses for threat contents were potentiated relative to
neutral and positive content blinks. For disgust contents, blinks
were potentiated relative to positive contents only. These effects
were consistent across the 300-ms and late-interval responses.
The early potentiation observed for threat contents replicates
the findings of Globisch et al. (1999), extending their result to an
unselected sample with the use of a different selection of pictures.
Disgust content blinks were not significantly different from
neutral at either early or late probe times (in a manner similar to
the negative condition in Codispoti et al., 2001).
These results are inconsistent with the prediction, based on
simple tone prepulse studies, that blink responses to interesting
picture stimuli will be inhibited at early probe times. Instead, the
results are consistent with the prediction of Globisch et al. (1999)
that threatening picture contents will activate defensive mecha-
nisms early in the picture viewing period. This could occur as an
outcome of Bradley and Lang’s (2001) two-process account of
early startle modification, where the net effect of prepulse inhi-
bition and affective potentiation will determine the degree of
startle modification for negative contents. This interpretation
suggests that the highly arousing threat contents in this study and
in Globisch et al. (1999) activated blink potentiation circuits to a
degree surpassing the effects of prepulse inhibition. Disgust con-
tent blinks did not differ from neutral at the early probe time,
which could potentially indicate (a) some aversive activation
from the picture content, but not enough to surpass attentional
inhibition or (b) a lack of any aversive/attentional modifica-
tionFin other words, the pictures failing to produce emotional
processing that differed from neutral. Both possibilities can be
addressed through the stimulus arousal/intensity dimension.
Participants did not provide ratings of the pictures in this exper-
iment, leaving the SCR magnitude data as the sole measure of
stimulus intensity. For trials in which SCRs were not influenced
by the startle probe, these data suggested that arousal levels were
equivalent in the two negative conditions, and augmented rela-
tive to neutral content SCRs. For trials on which startle was
elicited almost simultaneously with picture onset, SCR magni-
tude was much greater for the threat compared to the disgust
condition. From the no-probe results, it appears unlikely that the
disgust contents produced no emotional activation, and thematch-
ing of arousal between the two negative categories suggests that a
factor other than stimulus intensity is responsible for the differ-
ences in blink magnitude between threat and disgust contents.
It is possible that the differences in blink modification reflect
other stimulus characteristics unrelated to emotional content
(e.g., visual complexity). One limitation to this study was the
relatively small number of pictures used for each of the categories
(six pictures per category), which leaves the possibility open that
these differences in stimulus characteristics may have influenced
Emotional specificity of early startle potentiation 939
Table 2. Raw Blink Magnitude (in Microvolts) Means and
Standard Errors for Each Emotional Category and ITI Instances
Positive Neutral Disgust Threat ITI
Mean 50.01 52.27 55.85 58.43 57.89(SE ) (4.81) (5.44) (5.87) (5.55) (5.88)
Table 3. Standardized SCR Magnitude (Proportion of Range)
Means and Standard Errors for Each Emotional Category, Divided
by Probe Presentation
Probe condition Positive Neutral Disgust Threat ITI
No probe 0.12 0.06 0.1 0.12 N/A(SE ) (0.02) (0.01) (0.02) (0.03)Probe 0.40 0.36 0.41 0.48 0.35(SE ) (0.03) (0.03) (0.03) (0.03) (0.03)
Early Late40
45
50
55
60
Positive Neutral Disgust Threat
Probe Time
Blin
k M
agni
tude
(T
-sco
re)
Figure 1. Mean standardized blink magnitude by emotional category, at
the early and late probe times. Error bars indicate one standard error.
Dotted line represents mean level of ITI responses.
the pattern of responses. The small picture sets were largely due
to the experimental hypotheses regarding the emotional specifi-
city of each category, and replication with a larger number of
exemplars and/or a different picture set will strengthen the va-
lidity of these conclusions.
This experiment showed that early startle potentiation is not
limited to preselected, highly fearful participants (cf. Globisch
et al., 1999), with threat content pictures successfully potentia-
ting startle within 300ms of picture onset. Subsequent studies
should assess whether the observed early startle modification
differences between threat and disgust content stimuli reflect
arousal differences between these categories in the present study,
or a more qualitative difference in the processing of these two
picture contents.
REFERENCES
Anthony, B. J. (1985). In the blink of an eye: Implications of reflexmodification for information processing. In P. K. Ackles, J. R.Jennings, & M. G. H. Coles (Eds.), Advances in psychophysiology(Vol. 1, pp. 167–218). Greenwich, CT: JAI Press.
Balaban, M. T., & Taussig, H. N. (1994). Salience of fear/threat in theaffective modulation of the human startle blink. Biological Psychol-ogy, 38, 117–131.
Bernat, E., Patrick, C., Benning, S. D., Blonigen, D., &Hicks, B. (2002).Parsing effects of affective stimulus content on startle reflex modu-lation in men [Abstract]. Psychophysiology, 39(Suppl. 1), S21.
Bradley, M. M., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (2001).Emotion and motivation I: Defensive and appetitive responding inpicture processing. Emotion, 1, 276–298.
Bradley, M. M., Codispoti, M., Sabatinelli, D., & Lang, P. J. (2001).Emotion and motivation II: Sex differences in picture processing.Emotion, 1, 300–319.
Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1993). Pictures as pre-pulse: Attention and emotion in startle modification. Psychophysiol-ogy, 30, 541–545.
Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1999). Affect and thestartle reflex. In M. E. Dawson, A. M. Schell, & A. H. Bohmelt(Eds.), Startle modification: Implications for neuroscience, cognitivescience, and clinical science (pp. 157–183). Cambridge, England:Cambridge Univeristy Press.
Bradley, M. M., & Lang, P. J. (2001). Orienting and emotion: Startlemodulation during picture perception. Manuscript submitted forpublication.
Codispoti, M., Bradley, M. M., & Lang, P. J. (2001). Affective reactionsto briefly presented pictures. Psychophysiology, 38, 474–478.
Fridlund, A. J., & Cacioppo, J. T. (1986). Guidelines for human elect-romyographic research. Psychophysiology, 23, 567–589.
Globisch, J., Hamm, A. O., Esteves, F., & Ohman, A. (1999). Fearappears fast: Temporal course of startle reflex potentiation in animalfearful subjects. Psychophysiology, 36, 66–75.
Graham, F.K. (1992). Attention: The heartbeat, the blink, and the brain.In B. A. Campbell, H. Hayne, & R. Richardson (Eds.), Attention andinformation processing in infants and adults: Perspectives from humanand animal research (pp. 3–29). Hillsdale, NJ: Erlbaum.
Hawk, L. W., & Cook, E. W., III. (2000). Independence of valencemodulation and prepulse inhibition of startle. Psychophysiology, 37,5–12.
Hodes, R. L., Cook, E.W., &Lang, P. J. (1985). Individual differences inautonomic response: Conditioned association or conditioned fear?Psychophysiology, 22, 545–560.
Jones, P. (2000). Computer programme for presenting photographic andacoustic startle stimuli [Computer program]. Dunedin, New Zealand:Department of Psychology, Univeristy of Otago.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1998). Emotion,motivation, and anxiety: Brain mechanisms and psychophysiology.Biological Psychiatry, 44, 1248–1263.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1999). Internationalaffective picture system (IAPS) [photographic slides]. Gainseville, FL:Center for Research in Psychophysiology, University of Florida.
Lethbridge, R., Simmons, J. G., & Allen, N. B. (2002). All things unpleas-ant are not equal: Startle reflex modification while processing socialand physical threat [Abstract]. Psychophysiology, 39(Suppl. 1), S51.
Levenston, G. K., Patrick, C. J., Bradley, M. M., & Lang, P. J. (2000).The psychopath as observer: Emotion and attention in pictureprocessing. Journal of Abnormal Psychology, 109, 373–385.
Lykken, D. T., & Venables, P. H. (1971). Direct measurement of skinconductance: A proposal for standardization. Psychophysiology, 8,656–672.
Norris, C. M., & Blumenthal, T. D. (1996). A relationship between in-hibition of the acoustic startle response and the protection of prepulseprocessing. Psychobiology, 24, 160–168.
Ohman, A. (1993). Fear and anxiety as emotional phenomena: Clinicalphenomenology, evolutionary perspectives, and information-process-ing mechanisms. In M. Lewis & J. M. Haviland (Eds.), Handbook ofemotions (pp. 511–536). New York: Guilford Press.
Stanley, J. (2003). Emotional specificity of early startle potentiation.Unpublished doctoral dissertation, University of Otago, Dunedin,New Zealand.
van Boxtel, A., Boelhouwer, A. J. W., & Bos, A. R. (1998). OptimalEMG signal bandwidth and interelectrode distance for the recordingof acoustic, electrocutaneous, and photic blink reflexes. Psychophys-iology, 35, 690–697.
Vanman, E. J., Boehmelt, A. H., Dawson, M. E., & Schell, A. M. (1996).The varying time courses of attentional and affective modulation ofthe startle eyeblink reflex. Psychophysiology, 33, 691–697.
Vanman, E. J., Dawson, M. E., & Brennan, P. A. (1998). Affective re-actions in the blink of an eye: Individual differences in subjectiveexperience and physiological responses to emotional stimuli. Person-ality and Social Psychology Bulletin, 24, 994–1005.
Vrana, S. R., Spence, E. L., & Lang, P. J. (1988). The startle proberesponse: A new measure of emotion. Journal of Abnormal Psychol-ogy, 97, 487–491.
Yartz, A. R., & Hawk, L. W., Jr. (2002). Addressing the specificity ofaffective startle modulation: Fear versus disgust. Biological Psychol-ogy, 59, 55–68.
(Received August 7, 2003; Accepted July 1, 2004)
940 J. Stanley and R.G. Knight