2011. Vol.2, No.5, 486-491
Copyright © 2011 SciRes. DOI:10.4236/psych.2011.25075
Affective Modulation of the Startle Reflex Is an Ineffective
Methodology to Examine Depression-Linked Interpretative Biases
Leigh S. Goggin1,2, Mathew T. Martin-Iverson1,2, Paula R. Nathan3,4
1School of Medicine and Pharmacology, University of Western Australia, Perth, Australia;
2The Centre for Clinical Research in Neuropsychiatry, Perth, Australia;
3School of Psychiatry and Clinical Neurosciences, University of Western Australia, Perth, Australia;
4The Centre for Clinical Interventions, Perth, Australia.
Received May 17th, 2011; revised June 22nd, 2011; accepted July 25th, 2011.
Cognitive theory predicts that depression is associated with a bias to interpret ambiguous information in a
mood-congruent fashion. This negative interpretative bias may serve as a maintenance factor for the continua-
tion of a depressed mood state. The majority of studies investigating such interpretative biases suffer from a va-
riety of methodological problems. This research has utilized an objective physiological measure involving the
affective modulation of the human eye blink reflex in 25 depressed and 25 control subjects by depressive, de-
pressive-ambiguous, and distorted stimuli. Almost half of the depressed subjects suffered from a comorbid anxi-
ety disorder. In contrast to previous research utilizing the same methodology, depressed participants did not re-
act differently to non-depressed participants in terms of their blink reflex response to the various stimuli types.
This outcome is ascribed to the exclusion of anxiety-related stimuli in the current study. Depression-related
stimuli failed to augment blink amplitudes in both subject groups. Therefore, affective modulation of the startle
reflex is an ineffective methodology for the detection of depression-linked interpretative biases, as there is no
difference to how individuals react to depressive and neutral stimuli. In this study, patients diagnosed with social
anxiety disorder reacted to difficult-to-identify stimuli with augmented blink amplitudes, interpreted as an anxi-
ety response.
Keywords: Depression, Social Anxiety Disorder, Ambiguous, Interpretative Bias, Startle Reflex
The cognitive distortion known as interpretative bias—a
propensity to interpret ambiguous or innocuous events and
situations in a negative fashion—has long been associated with
depressive illnesses (Beck, 1967, 1976). It is possible that such
a phenomenon may act as a cognitive vulnerability to a depres-
sive disorder, and also as a maintenance mechanism for the
continuation of an existing depressed mood state. Previous
papers investigating the existence of negative interpretative
biases for ambiguous information in depression have relied
primarily upon subjective experimental methodologies such as
self-report (e.g., Watkins & Rush, 1983; Nunn, Mathews, &
Trower, 1997) or evaluative feedback (e.g., Gotlib, 1983;
Vestre & Caulfield, 1986) experiments. Unfortunately, this
body of research is plagued by mixed results and methodologi-
cal limitations such as experimenter demand effects, response
selection biases, and depression-linked autobiographical mem-
ory influences (Clark & Teasdale, 1985; Williams & Broadbent,
1986; MacLeod & Mathews, 1991).
The difficulties and criticisms associated with previous re-
search into depression-linked negative interpretative biases
prompted the development of a novel objective methodology
that was seemingly immune to such concerns. Lawson, MacLeod,
and Hammond (2002) utilised the phenomenon of human startle
reflex affective modulation in their investigation of interpreta-
tive biases in depression. The affective modulation of the hu-
man startle reflex was discovered by Vrana, Spence, and Lang
(1988) and since confirmed by a multitude of studies over a
diverse range of stimulus modalities and experimental condi-
tions. The startle reflex is a protective mechanism neurophysi-
ologically linked to the defensive motivational system’s re-
sponse to an aversive environmental cue. Therefore, the reflex
is facilitated when an organism is exposed to an aversive or
negative stimulus and inhibited when the appetitive motiva-
tional system is activated by a hedonic or positive stimulus
(Lang, Bradley, & Cuthbert, 1992).
The affective modulation of the startle reflex allows for a
psychophysiological test of the existence of a negative inter-
pretative bias in depression that appears to be free from the
major limitations of the various subjective methodologies. By
presenting ambiguous stimuli to depressed individuals, it can be
inferred from their startle reflex response to a loud burst of
noise as to what particular interpretation they imposed upon the
stimuli. Accordingly, Lawson et al. (2002) developed a set of
negative, neutral, and ambiguous stimuli in order to test for an
interpretative bias in depressed individuals.
The stimuli of Lawson et al.’s (2002) study consisted of re-
corded spoken words of either a negative affective valence (e.g.,
alone, fright, lonely), neutral affective valence (e.g., dreamy,
prime, wheat), or an acoustically distorted hybrid word (e.g.,
cancer + dancer, scream + stream ) that served as the ambigu-
ous category of stimuli. Two groups of participants were re-
cruited based on their scores on the Beck Depression Inventory
(Beck, Steer, & Brown, 1996)—depressed and non-depressed
university students. All of the participants were presented with
one of the negative, neutral, or ambiguous words on each trial
of the experiment and instructed to imagine a situation involv-
ing themselves and that particular word. An acoustic startle
probe was generated during this imagery period on the majority
of the trials and the blink reflex of participants was recorded as
the measure of startle.
The results of Lawson et al.’s (2002) experiment revealed
that both groups exhibited increased blink amplitudes after
presentation of the negative words in comparison to the neutral
words, which is consistent with previous studies of blink reflex
affective modulation. However, the two groups differed in their
reaction to the ambiguous stimuli. Non-depressed participants
reacted to the ambiguous words in a neutral manner whereas
depressed participants reacted to them as if they were negative.
This finding suggests that the depressed group considered the
ambiguous stimuli unpleasant or negative, and therefore, gives
support to the hypothesis made by Beck’s (1967, 1976) cogni-
tive theory that depression is associated with the tendency to
impose a negative interpretation upon ambiguous information.
Although it appears that Lawson et al.’s (2002) study had
demonstrated that a negative interpretative bias for ambiguous
information exists in depressed individuals, three major limita-
tions can be identified with the research. Firstly, the partici-
pants consisted of a homogeneous student population and the
depressed group did not experience clinical levels of depressive
symptomatology. Secondly, a large number of the negative
words used in the study accessed emotional themes of anxiety
rather than depression. Anxiety and fear related words such as
rape, attack, and scream, for example, may have induced
greater levels of emotional arousal than depression related
words such as alone, gloom, and grief, for example. The am-
plitude of the human startle reflex is not only subject to the
affective valence of a stimulus, but also to its emotional arousal
(Sabatinelli, Bradley, & Lang, 2001), and previous research
indicates that sad stimuli fails to evoke augmented startle re-
sponses (Bradley, Codispoti, Cuthbert, & Lang, 2001). Finally,
the ambiguous stimuli differed from the non-ambiguous stimuli
on a property other than ambiguity of meaning; they were also
more difficult to understand. Participants with elevated levels
of anxiety may have worried that their inability to correctly
determine the ambiguous words would have embarrassing or
unpleasant consequences, thus increasing the negative affect
and emotional arousal of the stimuli, and subsequent blink am-
Several modifications to Lawson et al.’s (2002) methodology
were introduced in the present study to determine if their results
were representative of clinical depressive disorders, and de-
pendent on participant anxiety rather than depression. Firstly, a
clinically depressed sample was recruited for participation.
Secondly, only negative words of a primarily depressive theme
were retained in the auditory stimulus set. Finally, a fourth
category of stimuli was created in order to control for possible
anxiety effects stemming from indecision over the ambiguous
words. This category was similar to the ambiguous category,
but was composed of acoustically distorted neutral words. If
depressed individuals found both the distorted and ambiguous
categories of stimuli aversive, then it could be assumed that it
was the uncertainty over stimulus identity that they found un-
pleasant, rather than resorting to the notion of a depres-
sion-linked interpretative bias for the ambiguous stimuli.
Twenty-five clinically depressed psychiatric outpatients and
25 healthy control subjects matched with the patient group for
age and gender participated in the study. All of the subjects
reported English to be their primary language. Participants in
the clinically depressed group were outpatients recruited from
the Centre for Clinical Interventions in Perth, Western Australia,
and all were about to commence cognitive-behavioural therapy
for a depressive episode. Psychologists at the clinic selected
those patients they deemed appropriate for the study based on
diagnosis, current symptomatology, and capability to give in-
formed consent. Exclusion criteria for the study included the
presence of brain damage, neurological disorder, psychosis,
mania, or substance abuse disorder. Seventeen (68%) of the
patients were diagnosed with Major Depressive Disorder, six
(24%) were diagnosed with Bipolar Disorder (currently in a
depressive phase lasting at least two weeks), and two patients
(8%) were diagnosed with a Major Depressive Episode super-
imposed on Dysthymic Disorder.
Twelve (48%) of the depressed outpatients suffered from a
comorbid anxiety disorder (social anxiety disorder = 5, gener-
alised anxiety disorder = 3, agoraphobia with panic = 2, agora-
phobia without panic = 1, and panic disorder = 1). Fourteen
(56%) of the patients were currently prescribed antidepressant
medication (selective serotonin re-uptake inhibitors). Comorbid
anxiety was considered a factor for covariate analysis.
Healthy control subjects were recruited from the general
community through the use of advertisements and information
sheets. Mean ages of the patient and control group were 42.2
(range 24 - 69) and 37.7 years (range 19 - 59) respectively. Mean
years of education was 13.8 (range 9 - 22) and 15.8 (range 10 -
24) respectively. Both groups consisted of fifteen (60%) females.
Target Stimuli
The auditory stimulus sets used in this experiment were gen-
erously provided by Lawson et al. (2002) and were therefore
identical to those employed in their own study. Eighteen
healthy volunteers rated words from the negative category on
two five-point Likert measures, the first assessing the degree to
which each individual word accessed emotional themes of
anxiety and/or depression, and the second assessing the emo-
tional arousal level of the word. Twenty of the negative words
that were deemed to be the most closely related to sad and de-
pressed affect were selected for use in the experiment. Negative
words such as attack, battle, cancer, explode, fear, fright, rape,
and threat, for example, were discarded, leaving only words
related primarily to depressive themes for use in the experiment.
The remaining twenty depression-linked negative words in-
cluded alone, gloom, grief, guilt, inferior, lonely, loser, and
sorrow, for example. In contrast to the anxiety related words
(mean = 3.28; t(38) = 11.3, p < 0.05), the depressive words
(mean = 1.23; t(38) = 1.7, p > 0.05) were not rated significantly
higher than the neutral words (mean = 0.73) in terms of emo-
tional arousal.
Ten of the depressive words were left intact and the remain-
ing half were used in their ambiguous form as created by Law-
son et al. (2002), for example, lonely + lovely = lo*ely. Ten
intact neutral words were selected at random from Lawson et
al.’s (2002) neutral category set to act as the control stimuli.
The distorted neutral category of words was developed by mix-
ing the neutral words with their reversals. For example, the
word foot was fused together with its reversal teef to produce a
distorted neutral word that was quite difficult to hear and un-
derstand correctly.
In summary, the auditory stimuli for this study consisted of
40 single spoken words divided into four separate categories:
depression-linked negative words (e.g., alone, grief), depres-
sion/neutral ambiguous words (e.g., gloom/bloom, guilt/built),
neutral words (e.g., dancer, river), and distorted neutral words
(e.g., caring/gnirac, silent/tnelis).
Experimental Hardware
An Acer TravelMate 212 TXV notebook was used to present
the visual and auditory stimuli involved in the experiment, and
to record subject eye blink responses. These responses, defined
as increases in electromyographic (EMG) activity from the
orbicularis oculi muscle of the left eye, were essentially electric
signals produced by the motor structures that elicit muscle con-
tractions in that region. The signals were recorded using two
tin-cup surface electrodes 6-mm in diameter and 2-mm in depth
(Electro-cap International, model E-21, USA). The electrodes
were filled with conductive paste (Ten 20 conductive EEG
paste) in order to provide some consistency to the hydration and
conductivity of the skin beneath the electrodes. One electrode
was placed approximately 1-cm below the pupil while the other
was positioned 1-cm apart from the first, at the lateral canthus.
A third electrode, the ground electrode, was a silver/silver-
chloride disposable adhesive patch placed on the participant’s
left hand.
The EMG signal detected by the electrodes was conveyed to
a high performance AC preamplifier (Grass Instruments, model
CP511, West Warwick, USA) where it was amplified and
broadly filtered using a low pass cut-off of 1-kHz and a high
pass cut-off of 30-Hz, with an additional 50-Hz notch filter.
The filtered signal was relayed to a connector box (National
Instruments, San Diego, USA), and then transmitted to a 12-bit
analog-to-digital converter (DAQ card, model 6062E, National
Instruments, San Diego, USA) where it was digitised and sam-
pled at a rate of 1-kHz. The system recorded the signal for a
time period of 500-ms for each individual trial (100-ms prior to
onset of the blink-eliciting stimulus and 400-ms after its occur-
rence). The recorded EMG activity was then stored on the
notebook’s hard drive for subsequent offline scoring purposes.
The auditory stimuli and the blink-eliciting stimuli were both
presented by the notebook. The former consisted of the re-
corded spoken words and the latter consisted of a 30-ms burst
of white noise with a near-instantaneous rise time created by a
LabVIEW program (National Instruments, San Diego, USA). A
white noise generator (Goldline, USA) provided background
white noise, and this input was combined with that of the note-
book by a 2-channel stereo preamplifier performance mixer
(Stanton Magnetics, model ESM-11, China). This ensured the
background white noise was presented at a level of 60-dB (A),
the auditory stimuli were presented at 62-dB (A), and the
blink-eliciting stimuli were presented at 97-dB (A). All of these
sounds were conveyed to participants through a pair of head-
phones (Optimus Pro, model XB-100, Intertran, Australia),
which also attenuated ambient noises from the environment. A
sound level meter (Bruel & Kjaer, model 2237A, Australia)
was used to calibrate the levels of these auditory stimuli prior to
each session of testing.
Experimental Software
A LabVIEW version 6.1 (National Instruments, San Diego,
USA) program developed by two of the authors controlled the
presentation and timing of all the visual and auditory stimuli,
with the exception of the background white noise, which was
on continuously during the startle testing procedure. In addition
to experimental control, the program also collected the physio-
logical data from participants and stored it in raw form on the
notebook’s hard drive for later analysis. The LabVIEW pro-
gram rectified the data and digitally filtered the rectified data
using a finite impulse response band-pass filter, with a low
pass filter set at 250-Hz and a high pass filter set at 78-Hz.
Finally, a second-order low-pass Chebyshev filter smoothed
the data.
A second LabVIEW program was devised for the purposes
of scoring the signal. The first function completed by the pro-
gram was to convert the filtered, rectified, and smoothed EMG
signal from digital to metric units (μV). The second function of
the program was to score the data. The scoring program de-
tected the onset amplitude threshold of each blink (defined as
the mean plus three standard deviations of value of the baseline
amplitude of the 100-ms of activity occurring before the onset
of the startle stimulus), the peak amplitude of each blink (the
maximum point of the signal exceeding onset amplitude
threshold), and the end of response of each blink (defined as the
first time that the signal fell under the onset amplitude threshold
since the beginning of the blink).
The program applied three criteria to the scoring of blink
traces to ensure that only those blinks included in the analysis
could be confidently linked to the effect of the blink-eliciting
stimuli and not to randomly occurring endogenous blinks.
Firstly, blink responses were excluded if the onset amplitude
threshold did not transpire within 17-ms to 150-ms of the
blink-eliciting stimulus. Blinks occurring outside of this win-
dow were unlikely to be time-locked to the stimulus. Secondly,
blink responses were discarded if a blink occurred in the
100-ms prior to the blink-eliciting stimulus, because such
blinks could produce paired-pulse inhibition. Finally, if the
program was unable to detect any blink at all (i.e., the onset
amplitude threshold was not reached and pre-startle baseline
activity was normal), then the trace was rejected from the study.
Approximately 10% of blink traces were excluded from the
study on the basis of these three criteria.
Prior to the arrival of the participant, the equipment was cali-
brated for internal voltage readings and for the appropriate
sound levels. Calibration was conducted in the same fashion for
every individual testing session. After the participants had
completed the information and consent procedure, they were
seated before the computer and had electrodes attached to the
orbicularis oculi muscle under their left eye and to the area of
skin on top of their left hand. The subjects next received verbal
instructions as to what the task entailed and what was required
of them. They were then given a set of headphones to wear for
the purposes of auditory stimuli presentation. Participants were
informed that the screen would display three colored lights,
which would guide them through each trial of the experiment.
The illumination of the green light signaled the beginning of a
trial, and it remained lit for the duration of the entire trial. After
a two-second interval, an auditory target stimulus (a neutral,
negative, ambiguous, or distorted word) was played. An amber
light was then illuminated 100-ms after the presentation of the
auditory stimulus, indicating that the participant should begin to
imagine a situation involving themselves that was evoked by
that particular word. To encourage compliance, participants
were told that the experimenter would question them as to what
imagery they created for specific words at the conclusion of the
Eight-seconds after the amber light illumination, both of the
lights were terminated and a red light was illuminated to signal
the end of that particular trial. Participants were asked to stop
imagining and to relax when the red light was turned on. An
intertrial interval of thirteen-seconds elapsed before the com-
mencement of the next trial, which was signaled by the illumi-
nation of the green light. Blink-eliciting stimuli were presented
on 71.4% of the trials, always occurring 5.5-seconds into the
imagery period. This was to ensure that the startle stimulus was
relatively unpredictable. The startle stimulus was also played
randomly on approximately 16.7% of the intertrial intervals,
always occurring 5-seconds into the intertrial interval. The
purpose of this intertrial interval startle was also to make the
blink-eliciting stimulus less predictable to participants. The
subjects were then informed that the first three trials were prac-
tice trials, and that after successful completion of these trials,
they would embark on the actual experiment, which consisted
of the 56 trials.
Three control subjects and one patient were excluded from
the results for failure to blink in response to the blink-eliciting
stimuli. The criterion for exclusion was if a participant blinked
on less than 50% of the trials in any one of the four auditory
stimulus categories. The study therefore analysed the data from
22 controls and 24 patients. The recorded blink amplitude EMG
data was analysed in a mixed-design General Linear Model of
analysis of variance (ANOVA). Pair-wise comparisons were
conducted using the Sidak adjustment for multiple comparisons.
Unless stated otherwise, sphericity of the data was violated as
determined by Mauchly’s test and the Greenhouse-Geisser
adjustment was used in the analysis of results.
Figure 1 shows that the overall size of the blink amplitudes of
depressed participants (n = 24) was not significantly different
from that of the control subjects (n = 22; F(1,44) = 1.52, p > 0.05,
η2 = 0.03, power = 0.23). There was no significant difference in
the amplitude of blink responses to the various stimuli types, with
all participants displaying approximately equivalent responses to
each stimulus category (F(2.4,107.2) = 0.17, p > 0.05, η2 = 0.00,
power = 0.08, ε = 0.81). There was also no significant interac-
tion involving participant mood and stimuli type, indicating that
there were no fundamental differences in how depressed and
control subjects reacted to each of the stimulus types (F(2.4,107.2)
= 0.28, p > 0.05, η2 = 0.01, power = 0.10, ε = 0.81).
Controls Depressed
Neg ative
Neutr al
Su bject ty p e
amplitud e
Figure 1.
Blink amplitude data with standard error bars for each stimulus type as
a function of participant mood.
The results of this study, did not replicate Lawson et al.’s
(2002) finding that depressed individuals exhibit an interpreta-
tive bias for the negative plausible alternative of the ambiguous
words. The depressed patients did not exhibit increased blink
reflex amplitudes in response to the ambiguous stimuli. In addi-
tion, neither the control group nor the depressed group responded
to the negative-depressive category of stimuli with augmented
blink amplitudes with anxiety-related negative words excluded
from the study. Thus, the larger startle responses displayed by
both the control subjects and the depressed subjects in Lawson
et al.’s (2002) study may be attributed to the presence of anxi-
ety-related negative words such as rape and attack in that ex-
periment. The failure of the depression-related negative words
to evoke larger blink amplitudes in response to a startling
stimulus supports previous affective modulation research
(Bradley et al., 2001; Kreibig, Wilhelm, Roth, & Gross, 2011)
This may be due to the effects of the low arousal level of the
depressive words or to the inherent difference between the
emotions of anxiety and depression with regard to startle reflex
Bradley et al. (2001) discovered that in contrast to anxi-
ety-provoking visual images (e.g., animal attack and mutilation),
pictures of a depressive nature (e.g., loss and pollution) failed
to evoke a heightened blink response in their sample of healthy
university students. In addition to the probable low arousal
effect, Bradley et al. (2001) hypothesized that negative stimuli
can vary in its theoretical “relevance” to life and immediate
survival. Anxiety-related negative stimuli are perceived to be
most threatening to an individual’s existence, and thus activate
the defensive motivational system, while depression-related
negative stimuli are considered to be of far less immediate
consequence and result in minimal activation of the system.
The result of the present study replicates that of Bradley et al.
(2001) with regard to the effects of depressive stimuli.
Covariate analysis controlling for comorbid anxiety in the
depressed group revealed no significant effects. However, an
interesting result was discovered in the five patients with a
comorbid diagnosis of social anxiety disorder. They exhibited
ControlsOther PatientsSxD Patients
N egative
Su bj e ct ty p e
Figure 2.
Mean blink amplitude with standard error bars for each stimulus type
as a function of participant mood and social anxiety disorder diagnosis
(asterisks denote a significant difference [p < 0.05] from the neutral
category within each group).
significantly larger overall blink amplitudes compared to the
control group and the other depressed patients (F(2,43) = 3.67, p
< 0.05, η2 = 0.15, power = 0.64). A significant main effect was
discovered for stimulus type (F(2.6,111.8) = 3.10, p < 0.05, η2 =
0.07, power = 0.67, ε = 0.87), and there was also a significant
interaction involving participant type and stimuli category,
indicating that a diagnosis of social anxiety disorder had an
influence upon the amplitude of blink responses to the various
categories of stimuli (F(5.2,111.8) = 4.35, p < 0.01, η2 = 0.17,
power = 0.96, ε = 0.87). Pairwise comparisons revealed that in
contrast to the control group and the other depressed patients,
socially anxious patients exhibited significantly larger blink
amplitudes to the negative, ambiguous, and distorted stimuli in
comparison to the neutral stimuli (Figure 2).
It can be inferred from this increased amplitude of the startle
response that the socially anxious patients were in a highly
anxious state throughout the experiment. This could be attrib-
uted to the experimental context itself, which involved aspects
of social interaction, observation, and perceived evaluation. The
DSM-IV describes social anxiety disorder as a “marked and
persistent fear of one or more social or performance situations
in which the person is exposed to unfamiliar people or possible
scrutiny by others” (APA, 1994). Therefore, it is likely that the
experimental situation evoked significant anxiety in socially
phobic individuals due to fact that it involved observation and
scrutiny by an unfamiliar person (the experimenter).
It is possible that patients with social anxiety disorder may
have felt that they were ‘failing’ the experiment by not cor-
rectly identifying the various words, and this would result in
them being evaluated negatively by the experimenter and ap-
pearing foolish in the eyes of others. Such a perception would
render the ambiguous and distorted stimuli aversive and threat-
ening in this patient group, leading to further worry and anxiety,
and a subsequently augmented startle response. The results
from the social anxiety subgroup primarily indicate that the
procedure has sufficient power to detect a real effect. However,
due to the small sample size of the social anxiety disorder sub-
group, replication with a larger number of patients with social
anxiety disorder is required.
The current finding that neither words with depressing con-
tent nor ambiguous stimuli that can be plausibly interpreted as
either a neutral or depressive word are sufficiently aversive to
affect the startle reflex has a major implication for the utility of
Lawson et al.’s (2002) procedure. The use of depressive stimuli
for the affective modulation of the eye blink reflex indicates
that, while considered unpleasant in ratings of affective valence,
depressive stimuli do not increase blink amplitudes relative to
neutral stimuli in healthy controls and depressed patients with-
out a diagnosis of social anxiety disorder. Therefore, Lawson et
al.’s (2002) methodology is not capable of detecting the pres-
ence of a depression-linked interpretative bias, as there seems
to be little difference in how individuals react to depressive and
neutral stimuli (Bradley et al., 2001; Kreibig et al., 2011).
Based on this observation, it would be impossible to determine
whether an individual interpreted an ambiguous stimulus in a
depressive fashion from their startle response to that stimulus.
The results of Lawson et al.’s (2002) experiment may there-
fore be attributed to the presence of highly arousing anxi-
ety-related words. Their anxious subjects may have exhibited
an interpretative bias for threat-related words in the ambiguous
stimuli, or alternatively, simply found the ambiguous stimuli
aversive for some other reason. An important implication from
the results of the current study is that participant social anxiety
levels may act as a major confounding factor in experimental
research designs investigating depressive illnesses or symptoms,
particularly those employing physiological assessments such as
the startle response.
American Psychological Association (1994). Diagnostic and statistical
manual of mental disorders. Washington, DC: American Psycho-
logical Association.
Beck, A. T. (1967). Depression: Clinical, experimental and theoretical
aspects. New York, NY: Harper & Row.
Beck, A. T. (1976). Cognitive therapy and the emotional disorders.
New York, NY: International Universities Press.
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Depression Inven-
tory-Manual (2nd ed.). San Antonio: The Psychological Corporation.
Bradley, M. M., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (2001).
Emotion and motivation I: Defensive and appetitive reactions in pic-
ture processing. Emotion, 1, 276-298.
Clark, D. M., & Teasdale, J. D. (1985). Diurnal variation in clinical
depression and accessibility of memories of positive and negative
experiences. Journal of Abnormal Psychology, 91, 87-95.
Gotlib, I. H. (1983). Perception and recall of interpersonal feedback:
Negative bias in depression. Cognitive Therapy and Research, 7,
399-412. doi:10.1007/BF01187168
Kreibig, S. D., Wilhelm, F. H., Roth, W. T., & Gross, J. J. (2011).
Affective modulation of the acoustic startle: Does sadness engage the
defensive system? Biological Psychiatry, 87, 161-163.
Lang, P. J., Bradley, M. M., & Cuthbert, B. C. (1992). A motivational
analysis of emotion: Reflex-cortex connections. Psychological Sci-
ence, 3, 44-49. doi:10.1111/j.1467-9280.1992.tb00255.x
Lawson, C., MacLeod, C., & Hammond, G. (2002). Interpretation
revealed in the blink of an eye: Depressive bias in the resolution of
ambiguity. Journal of Abnormal Psychology, 111, 321-328.
MacLeod, C., & Mathews, A. (1991). Cognitive-experimental app-
roaches to the emotional disorders. In P. R. Martin (Ed.), Handbook
of behaviour therapy and psychological science: An integrative ap-
proach (pp. 116-150). New York, NY: Pergamon.
Nunn, J. D., Mathews, A., & Trower, P. (1997). Selective processing of
concern-related information in depression. British Journal of Clinical
Psychology, 36, 489-503.
Sabatinelli, D., Bradley, M. M., & Lang, P. J. (2001). Affective startle
modulation in anticipation and perception. Psychophysiology, 38,
719-722. doi:10.1111/1469-8986.3840719
Vestre, N. D., & Caulfield, B. P. (1986). Perception of neutral person
ality descriptions by depressed and nondepressed subjects. Cognitive
Therapy and Research, 10, 31-36. doi:10.1007/BF01173380
Vrana, S. R., Spence, E. L., & Lang, P .J. (1988). The startle probe
response: A new measure of emotion? Journal of Abnormal Psy-
chology, 97, 487-491. doi:10.1037/0021-843X.97.4.487
Watkins, J. T., & Rush, A. J. (1983). Cognitive response test. Cognitive
Therapy and Research, 7, 425-436. doi:10.1007/BF01187170
Williams, J. M. G., & Broadbent, K. (1986). Autobiographical memory
in suicide attempters. Journal of Abnormal Psychology, 95, 144-149.