How Arousal Modulates the Visual Contrast Sensitivity Function
Tae-Ho Lee
University of Southern California
Jongsoo Baek and Zhong-Lin Lu
Ohio State University
Mara Mather
University of Southern California
Recent evidence indicates that emotion enhances contrast thresholds in subsequent visual perception
(Phelps, Ling, & Carrasco, 2006) and perceptual sensitivity for low-spatial frequency but not high-spatial
frequency targets (Bocanegra & Zeelenberg, 2009b). However, these studies just report responses to
various frequencies at a fixed contrast level or responses to various contrasts at a fixed frequency. In the
current study, we measured the full contrast sensitivity function as a function of emotional arousal in
order to investigate potential interactions between spatial frequency and contrast. We used a Bayesian
adaptive inference with a trial-to-trial information gain strategy (Lesmes, Lu, Baek, & Albright, 2010)
and a fear-conditioned stimulus to manipulate arousal level. The spatial frequency at which people
showed peak contrast sensitivity shifted to lower spatial frequencies in the arousing condition compared
with the nonarousing condition and people had greater contrast sensitivity function bandwidth in the
arousing than in the nonarousing condition.
Keywords: emotion, early vision, perception, fear conditioning, contrast sensitivity, quick contrast
sensitivity function (qCSF)
Several pioneering psychophysical studies reveal that emotion
alters early visual perception. For instance, presenting an arousing
cue (e.g., a fearful face) can improve perception of subsequent
neutral gratings (e.g., Gabor patches; Phelps et al., 2006; see also
Woods, Philbeck, & Wirtz, 2013). In these studies, increasing
arousal lowered the contrast threshold, at least at the particular
spatial frequency tested. Furthermore, emotionally arousing stim-
uli can increase activation in early visual cortex regions such as V1
(Padmala & Pessoa, 2008).
However, a recent study suggests that arousal does not globally
enhance perception, but instead can either enhance or impair
contrast sensitivity depending on the spatial frequency of the
stimuli (Bocanegra & Zeelenberg, 2009b). In this study, spatial
frequency was determined by the number of sinusoidal luminance
cycles per degree (cpd) of visual angle in the Gabor patch (Figure
1A). Exposure to fearful faces decreased participants’ detection
sensitivity of subsequent high-spatial frequency gratings (e.g., 6
cpd), and increasing sensitivity of subsequent low-spatial fre-
quency gratings (e.g., 1.5 cpd). Similarly, in another study, expo-
sure to fearful faces impaired subsequent judgments about high-
spatial frequency characteristics of words but enhanced judgments
about low-spatial frequency characteristics of words (Borst &
Kosslyn, 2010). Emotions also seem to be more likely to be
elicited by low-spatial frequency stimuli than by high-spatial fre-
quency stimuli, although these emotion elicitation results are not
entirely consistent across studies (for a review see De Cesarei &
Codispoti, 2013).
Researchers have accounted for findings of different effects of
arousal on low- versus high-spatial frequency stimuli by arguing
that the amygdala responds to emotional arousal by potentiating
magnocellular-type channels in the visual system while suppress-
ing parvocellular-type channels (Bocanegra & Zeelenberg, 2009b,
2011a; Borst & Kosslyn, 2010). This bias favoring magnocellular-
type channels should preferentially enhance processing of low-
spatial frequency information because magnocellular cells within
the lateral geniculate nucleus of the thalamus have a lower spatial
resolution than parvocellular cells (Sincich & Horton, 2005).
Although the evidence to date indicates that emotion has a
strong influence in early visual perception, the results from the
previous studies described above each gave information about the
influence of only one dimension of early visual perception as a
function of emotional arousal. Phelps, Ling, and Carrasco (2006)
investigated the contrast psychometric function, that is, perfor-
mance as a function of stimulus contrast, at a fixed spatial fre-
quency, whereas Bocanegra and Zeelenberg (2009b) investigated
the orientation discrimination sensitivity for various spatial fre-
quencies at a fixed contrast level. Yet both spatial frequency and
contrast are critical for visual processing and they interact. For
This article was published Online First June 16, 2014.
Tae-Ho Lee, Department of Psychology, University of Southern Cali-
fornia; Jongsoo Baek and Zhong-Lin Lu, Department of Psychology, Ohio
State University; Mara Mather, Department of Psychology, Davis School
of Gerontology, and Neuroscience Graduate Programs, University of
Southern California.
This work was supported by NIH Grants RO1AG025340 and
K02AG032309. Zhong-Lin Lu has an intellectual rights (U.S. patent) for
the quick contrast sensitivity function (qCSF) toolbox (http://lobes.osu.edu/
qMethods.php).
Correspondence concerning this article should be addressed to Mara
Mather, Davis School of Gerontology, University of Southern California,
3715 McClintock Avenue, Los Angeles, CA 90089-0191. E-mail: mara
.mather@usc.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Emotion © 2014 American Psychological Association
2014, Vol. 14, No. 5, 978–984 1528-3542/14/$12.00 http://dx.doi.org/10.1037/a0037047
978
example, visual perception at a particular contrast level depends on
the spatial frequency of the target object (De Valois, Morgan, &
Snodderly, 1974; Georgeson & Sullivan, 1975; Keller, Stras-
burger, Cerutti, & Sabel, 2000; Owsley, 2003; Pasternak & Mer-
igan, 1981). Therefore, to globally assess the effects of emotion on
perception, investigations should probe a range of both spatial
frequencies and contrasts. Indeed, one recent neurophysiological
study considered these two factors in how emotion influences
perception (Song & Keil, 2013). In this study, the authors probed
early visual modulation, as indexed by steady-state visual potential
(ssVEPs), by presenting an emotionally arousing picture followed
by a contrast-varying Gabor patch. The Gabor patches were pre-
sented in one of two different spatial frequencies (2 cpd vs. 6 cpd).
For low-spatial frequency targets, viewing emotionally arousing
pictures just beforehand increased the accuracy of judging the
orientation of subsequent low-spatial frequency targets and in-
creased the ssVEP amplitude. In contrast, for high-spatial fre-
quency targets, preceding emotionally arousing pictures decreased
detection accuracy and ssVEP amplitude.
The current study was designed to provide a more complete
portrayal of the modulatory role of emotional arousal over both
contrast and spatial frequency in the early visual system. To
accomplish the current goal, we measured the contrast sensitivity
function (Cornsweet, 1970) as a function of emotional arousal
(Figure 1A). In visual psychophysics, threshold refers to the min-
imum physical strength of a stimulus that can be detected or
discriminated at some target accuracy level, such as 82% correct.
Sensitivity is defined as the reciprocal of threshold (1/threshold).
The contrast sensitivity function (CSF) characterizes sensitivity as
a function of the spatial frequency (the coarseness or fineness) of
visual stimuli. The fundamental advantage of measuring the con-
trast sensitivity function is that it reflects various contrast sensi-
tivity at different spatial frequency channels, each tuned to a
preferred spatial scale, and consequently it provides information
about the global tuning of the visual system in response to various
combination of contrast and spatial frequency of visual environ-
ment (Billock & Harding, 1996; Blakemore & Campbell, 1969;
Campbell & Robson, 1968; De Valois, Albrecht, & Thorell, 1982;
Hubel & Wiesel, 1968; Mallat, 1989; Pantle & Sekuler, 1968;
Sekuler, Wilson, & Owsley, 1984). Indeed, the contrast sensitivity
function has been extensively used as a model to investigate the
early visual system with various visual tasks (Campbell & Robson,
1968; Chung, Legge, & Tjan, 2002; Enroth-Cugell & Robson,
1966; Kwon & Legge, 2011; Legge & Foley, 1980; Movshon,
Thompson, & Tolhurst, 1978; Watson & Ahumada, 2005; Watson
& Solomon, 1997).
The conventional data collection method, the method of con-
stant stimuli, usually requires several hundred trials to estimate a
threshold at a single spatial frequency. To obtain a contrast sen-
sitivity function, researchers usually measure contrast thresholds at
several different spatial frequencies. The burden of data collection
is multiplied by the number of spatial frequencies tested. Even
Figure 1. (A) Schematic representation of the contrast sensitivity function, which was constructed using the
inverse of measured contrast thresholds, operationally defined as the point of inflection in the psychometric
function, at each spatial frequency. The contrast sensitivity function has a maximum value at intermediate
spatial frequencies, and decreases with both lower and higher spatial frequencies. In the quick contrast sensitivity
function method, the spatial contrast sensitivity function parameters (the peak gain, ␥
max
; the peak spatial
frequency f
max
; the bandwidth , which describes the function’s full width at half maximum; and the
low-frequency truncation level ␦) are directly estimated via Bayesian adaptive inference rather than measuring
multiple contrast thresholds at each spatial frequency based on psychometric functions. The dashed line indicates
predicted contrast sensitivities before truncation. (B) Conditioning phase trial sequence. (C) Orientation iden-
tification task trial sequence. Because the spatial frequency can be changed depending on viewing distance (i.e.,
defined by the number of sinusoidal cycles per one degree of visual angle), please note that the representative
spatial frequencies here were drawn for display purposes only.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
979
EMOTION AND THE CONTRAST SENSITIVITY FUNCTION
with the most advanced adaptive procedures for estimating a single
contrast threshold (e.g., QUEST or Psi method), measurement of a
contrast sensitivity function typically takes 500–1,000 trials. Dur-
ing such long sessions, visual performance is likely to change and
emotional responses are likely to habituate. Thus, measuring the
full contrast sensitivity function using the conventional method
faces obstacles in terms of testing time and data precision. To
overcome the disadvantage of the conventional method, we ad-
opted the quick CSF (qCSF) method (Lesmes et al., 2010), a
Bayesian adaptive procedure for efficient estimation of contrast
sensitivity functions. Instead of measuring contrast thresholds at
individual spatial frequencies (e.g., Figure 1A top), the qCSF
directly estimates the parameters of the contrast sensitivity func-
tion and computes contrast sensitivities over the full spatial-
frequency range based on the parameters values. By directly esti-
mating the model parameters, the qCSF method improves contrast
sensitivity function estimation in both testing time and precision
with a relatively small number of trials (e.g., less than 100 trials;
Hou et al., 2010).
Another important aspect of the current study is that we ex-
cluded confounds found in previous studies evaluating emotion’s
role in visual perception. Although one role of emotion is multi-
plicative amplification of transient covert attention’s influence on
the contrast threshold (Phelps et al., 2006) or spatiotemporal
resolution (Bocanegra & Zeelenberg, 2011b), other evidence im-
plies that emotional arousal can also influence perception irrespec-
tive of attention (Bocanegra, Huijding, & Zeelenberg, 2012; Bo-
canegra & Zeelenberg, 2009a, 2009b; Phelps et al., 2006;
Zeelenberg & Bocanegra, 2010). In the current study, we avoided
confounding attentional effects induced by transient covert atten-
tion by presenting just one target Gabor on the center of the screen.
By using a fixed location for target presentation, we also excluded
the effect of spatial uncertainty for the possible peripheral target’s
location; uncertainty influences overall visual perception as it
changes visual system resource allocation (Cameron, Tai, & Car-
rasco, 2002; Carrasco, Penpeci-Talgar, & Eckstein, 2000; Foley &
Schwarz, 1998; Pelli, 1985; Solomon, Lavie, & Morgan, 1997).
Although there are several accounts of how attention benefits early
visual perception, such as contrast gain (e.g., Sclar, Lennie, &
DePriest, 1989) or response gain (e.g., McAdams & Maunsell,
1999), the key point is that transient covert attention, induced by
the sudden appearance of a target stimulus at a peripheral location,
enhances visual perception. We wanted to avoid having this as a
potential mediator of our emotion effects (for more reviews, see
Carrasco, 2011; Carrasco, Ling, & Read, 2004; Carrasco et al.,
2000).
We also avoided adaptation effects potentially induced by using
the same sensory modality to evoke emotional arousal and to
measure perception (e.g., inducing arousal via a visual cue and
testing perception via a visual target). Arousing and nonarousing
visual stimuli can differ in low-level visual features such as the
energy distribution of frequencies (e.g., Delplanque, N=diaye,
Scherer, & Grandjean, 2007); this is important as spatial frequency
adaptations can change subsequent perception and global tuning of
the visual system (see De Valois, 1977; Maffei, Fiorentini, & Bisti,
1973; Wilson & Humanski, 1993). To eliminate these concerns,
we used fear-conditioned auditory tones as the arousing and non-
arousing cues. Classical fear conditioning is a process in which
affective significance is endowed to a previously neutral stimulus
(conditioned stimulus; conditional stimulus, CS) by pairing it with
an aversive stimulus such as an electric shock (unconditioned
stimulus, US). After repeated pairings with US, the CS becomes
capable of increasing emotional arousal (CS⫹) and triggers a
series of autonomic response such as increased skin conductance
response (conditioned response, CR). A previous study showed
that fear conditioned auditory cues modulated activity in early
visual cortex regions such as V1 (see the Conditioning 2 section of
Padmala & Pessoa, 2008), thus arousal induced by auditory stimuli
can modulate basic visual processes. In the current study, we used
two different tones (500 Hz vs. 1,500 Hz) as the CS’s. The pitch
of the CS⫹ tone was counterbalanced across observers; for exam-
ple, some observers received the electric shock after hearing the
high-pitched tone, and some after hearing the low-pitched tone
(see Method for more details). By adopting classical fear condi-
tioning, we avoided all possible confounding sources in terms of
low-level visual features of experimental stimuli.
Materials and Method
Observers
Twenty-eight observers (16 male)
1
with corrected-to-normal
vision volunteered for this study and gave informed consent.
Observers were naïve to the purpose of the experiment. During the
experiment, observers performed a fear-conditioning session fol-
lowed by a perceptual detection session using the conditioned
tones as modulators of arousal in each trial.
Stimuli and Apparatus
Target displays contained single Gabor patches (5.2° ⫻ 5.2° of
visual angle; 1.9° sinusoidal gratings enveloped by a Gaussian).
For the current study, the 16 possible grating spatial frequencies
were spaced log linearly from 0.25 cpd to 36 cpd; the 60 possible
grating contrasts were spaced log linearly from 0.1% to 100%. All
the visual stimuli were generated in real time using the qCSF
toolbox (http://lobes.osu.edu/qMethods.php) and the PsychTool-
box extensions (Brainard, 1997; Pelli, 1997) based on Matlab
2010b (The MathWorks Corp., Natrick, MA). The visual stimuli
were presented at 85 Hz on a 19-in. CRT monitor (NEC AccuSync
90; NEC Corp. Tokyo, Japan) with a resolution of 1,280 ⫻ 960
pixels. Using a display attenuator that combines two 8-bit output
channels of the graphics card, the display system produced 14-bit
gray-level resolution (Li, Lu, Xu, Jin, & Zhou, 2003), and then was
gamma-corrected. Observers viewed the stimuli in a soundproof
and dimly lit room at a viewing distance of 2.5 m with their head
position stabilized by a chin rest. The mild electric shock used as
an unconditioned stimulus (US) was delivered to the third and
fourth fingers of the left hand via a shock stimulator for humans
(E13–22; Coulbourn Instruments, Allentown, PA). Two tones (500
Hz and 1,500 Hz) were adopted as conditioned stimuli (i.e., CSs)
1
One observer had extremely high max gains in both the arousing and
nonarousing conditions (982.60 and 282.31). These values were far greater
than three standard deviations above the peak gain mean (56.69 and 59.90
for arousing and nonarousing conditions, respectively). We excluded this
observer from the final analysis but note that all significant effects re-
mained significant when he was included.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
980
LEE, BAEK, LU, AND MATHER
to avoid possible confounding effects of using stimuli in the same
sensory modality to induce emotional arousal and to measure
perceptual processing. To confirm the success of the emotional
arousal manipulation, skin conductance responses as a function of
CSs (i.e., fear conditioned-arousing tone vs. nonarousing tone)
were measured using a BIOPAC MP-150 system (250 Hz sam-
pling rate; BIOPAC Systems, Goleta, CA) during the experiment.
Procedure
An initial fear-conditioning session established the emotionally
arousing nature of the CS⫹ tone. During the session, one of the
tones was paired with electric shock (CS⫹) and the other tone was
not paired with shock (CS⫺). Fourteen observers were conditioned
with the high-pitched tone and 13 with the low-pitched tone as the
CS⫹. Each trial began with a fixation cross jittered to appear for
7 s to 10 s. Then one of the CS tones played for 0.7 s. If the tone
was designated as the CS⫹, a shock was delivered right after the
offset of the CS⫹ tone, and terminated after 0.2 s (Figure 1B). On
each trial, observers were asked to press a button to indicate
whether the tone pitch was low or high. A total of 30 trials were
presented in a random order: 10 CS⫹ with shock; 10 CS⫹ without
shock; 10 CS⫺ tones. Therefore, CS⫹ tones were followed by a
shock with a 50% partial reinforcement schedule. Prior to the
experiment, observers were informed which tone predicts the US,
but they were not informed about the probability of US delivery.
The intensity of “highly unpleasant but not painful” electric shock
was determined individually for each observer before the condi-
tioning session (mean intensity ⫽ 1.77 mA, range 1.1–2.3 mA).
The trials that included shocks were excluded in subsequent anal-
yses.
Following the conditioning session, the orientation identifica-
tion task was administered. The trials began with a fixation point
jittered to last between7sto10s;then either the CS⫹ or CS⫺
was played for 0.7 s to manipulate arousal level. Following a 1-s
blank screen, one target Gabor grating was presented in the center
of the screen for 0.05 s. Observers indicated the target orientation
(counterclockwise or clockwise; ⫾ 45°) via a button press during
the response period (Figure 1C). A small black square (0.6°) was
presented in the center of the target grating to reduce stimulus
uncertainty, which could affect the contrast sensitivity function
measurement, especially in the high-frequency region (Woods,
1996). We put a 1-s blank interval between the arousing cue and
target onsets to avoid possible temporal competition in processing.
Note that each trial’s stimulus type (i.e., contrast and spatial
frequency) was determined via the qCSF algorithm for the two CS
conditions separately based on the observer’s responses in previ-
ous trials (see section below for more details), and therefore each
observer was tested with a different combination of target proper-
ties. To minimize extinction of conditioned responses, booster
trials consisting of a shock after a CS ⫹ tone without any target
stimulus on that trial were randomly intermixed with the other
trials. Booster trials were excluded from subsequent analyses. The
task consisted of 100 CS⫹ and 100 CS⫺ trials randomly ordered.
In addition to these 200 trials, there were approximately 12%
additional “booster” CS⫹ tone trials involving shock to prevent
extinction.
qCSF Method Implementation
In the current study, we estimated individuals’ contrast sensi-
tivity functions in both CS⫹ and CS⫺ conditions using the qCSF
method (Lesmes et al., 2010). Contrast sensitivity is defined as the
reciprocal of contrast threshold (1/threshold) at 82% accuracy in
the orientation discrimination task. For example, a threshold of
0.02 translates into a sensitivity of 50; and a sensitivity of 50
means that the observer’s threshold is 1/50 ⫽ .02 (or 2% contrast).
Using a well-known functional form of the contrast sensitivity
function, the qCSF method applies Bayesian inference to directly
estimate the parameters of the contrast sensitivity function by
optimal placements of test stimuli based on trial-by-trial response
from the observer. The functional form imposed by the qCSF
method is a truncated log-parabola, which is specified by four
parameters: (a) the peak gain (optimal maximum contrast sensi-
tivity) ␥
max
; (b) the spatial frequency of peak sensitivity ƒ
max
; (c)
the bandwidth , which describes the function’s full width at half
maximum (in octaves); and (d) the low-frequency truncation level
␦ (Figure 1A). With this functional form, the observer’s response
on each trial is used to update the posterior distributions of the
parameters (␥
max
,ƒ
max
, , ␦). The contrast sensitivity (or thresh
-
old) over the full range of spatial-frequencies is computed based
on the estimated parameter values in the end of the procedure.
Simultaneously, the trial outcomes (i.e., the updated contrast sen-
sitivity function) are used to select the particular grating frequency
and contrast level that maximizes the expected information gain
for the next trial. In sum, the goal of the qCSF is to efficiently
search the stimulus space to gain information in the parameter
space.
Results
On each trial, skin conductance responses were calculated by
subtracting a baseline (average signal between 0 s and 1 s) from
the peak amplitude during the 1 s–7 s time window following the
CS onset. CS⫹ cues yielded stronger skin conductance responses
than CS⫺ cues during both the initial conditioning session, t(26) ⫽
5.67, p ⬍ .001, and the orientation identification task, t(26) ⫽
3.83, p ⬍ .001, confirming the success of the arousal manipulation
via fear conditioning in the current study (Figure 2A).
The qCSF method estimated parameters for each CS (i.e., CS⫹
and CS⫺) were compared. Early visual perception differed signif-
icantly depending on whether the CS was arousing or nonarousing
(Figures 2B & 2C). A maximum contrast sensitivity (␥
max
)of
51.00 was found for CS⫹ trials at a spatial frequency (f
max
) of 1.53
cpd. In contrast, a ␥
max
of 56.56 was found at f
max
of 1.70 cpd for
CS⫺ trials. That is, the average contrast to reach 82% accuracy
was 1.96% at 1.53 cpd in the CS⫹ condition, but 1.77% at 1.70
cpd in the CS⫺ condition. Paired t tests revealed that there was a
significant difference in f
max
, t(26) ⫽⫺2.26, p ⫽ .03, and a
marginal difference in ␥
max
, t(26) ⫽⫺1.83, p ⫽ .079. Also, the
bandwidth () of the contrast sensitivity function was wider for
CS⫹ trials (3.11 octave) than for CS⫺ trials (2.84 octave), t(26) ⫽
2.79, p ⫽ .01. However, there was no significant difference in
truncation, ␦, of the contrast sensitivity function (p ⫽ .44). The
area under the curve (AUC) for the contrast sensitivity function
curve constructed by estimating these four parameters did not
differ by CS (p ⫽ .82).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
981
EMOTION AND THE CONTRAST SENSITIVITY FUNCTION
Discussion
To provide a more comprehensive understanding of how emo-
tional arousal influences early visual perception, the current study
used conditioned stimuli to manipulate arousal on a trial-by-trial
basis while quantifying observers’ contrast sensitivity function, a
fundamental measure of early visual system. To our knowledge,
this is the first study to measure how arousal influences the
contrast sensitivity function. Observers’ arousal levels were ma-
nipulated by a fear conditioning procedure that yielded greater skin
conductance responses for CS⫹ cues than for CS⫺ cues in both
phases of the experiment. Playing a tone previously associated
with shock influenced subsequent visual perception, as reflected in
significant changes in the contrast sensitivity function as a function
of CS (Figure 2B). First, there was an overall shift of the center of
the contrast sensitivity function (f
max
) toward lower spatial fre
-
quencies during CS⫹ trials relative to CS⫺ trials (1.53 cpd vs.
1.70 cpd), indicating that arousal led to greater contrast sensitivity
for low-spatial frequencies. Second, the bandwidth ()ofthe
contrast sensitivity function—the range of frequencies over which
the observer can detect a given level of contrast—increased during
CS⫹ trials compared with CS⫺ trials (3.11 vs. 2.84 octave). In
addition, we found a marginally significant arousal-induced con-
trast sensitivity loss at the peak spatial frequency (␥
max;
p ⫽ .078).
The parametric changes revealed in our study indicates that
increasing emotional arousal leads the visual system to more
effectively detect coarse visual input information (e.g., scene gist
or the overall shape of an object), while losing fine details. In
addition, our results help integrate previous psychophysical stud-
ies. As can be seen in our contrast sensitivity function (Figure 2C),
emotional arousal increases contrast sensitivity at relatively low-
spatial frequency ranges but reduces it at high ranges. Thus, the
current findings offers further support not only for Phelps et al.’s
(2006) study that showed emotion-induced enhancement in con-
trast perception with (fixed) low-spatial frequency target (2 cpd)
across the experiment, but also for Bocanegra and Zeelenberg’s
(2009b) study that found emotion enhanced sensitivity for low-
spatial frequency stimuli but impaired it for high-spatial frequency
stimuli (see also Borst & Kosslyn, 2010; Song & Keil, 2013).
The contrast sensitivity function shift during the emotional trials
of our study indicates that momentary fluctuations in arousal
stimulate dynamic adaptation of neuronal receptive fields (i.e.,
neuronal tuning) for visual system optimization (Billock & Hard-
ing, 1996; Blakemore & Campbell, 1969; Campbell & Robson,
1968; De Valois et al., 1982; Hubel & Wiesel, 1968; Mallat, 1989;
Pantle & Sekuler, 1968; Sekuler et al., 1984). Attentional manip-
ulations can also lead to dynamic adaption of contrast sensitivity
(Carrasco, 2011). However, attention leads to a different pattern of
contrast sensitivity modulation than seen in the emotion condition
of our study; exogenous covert attention enhances overall contrast
sensitivity across spatial frequencies (Carrasco et al., 2004; Car-
rasco et al., 2000). In contrast, in our study, emotion shifted peak
sensitivity at the cost of maximal sensitivity without changing the
area under the curve. As noted earlier in the article and in the
Method section, in the current study, covert attention to the pe-
riphery was unnecessary and the location of the targets was always
the same. Thus, transient covert attention was not a critical factor
in determining detection of the target in the current study, allowing
us to evaluate the effect of emotional arousal without this con-
founding factor.
Another aspect of our results worth noting is that, across the two
conditions, the contrast sensitivity function was shifted toward
lower spatial frequency levels than expected under normal circum-
stances (e.g., Owsley, 2003; Sekuler, Hutman, & Owsley, 1980).
Although we found that the emotionally arousing condition in-
duced significantly greater f
max
shifting than the nonarousing
condition did, the contrast sensitivity function in the nonarousing
condition was also lower than the normal range of f
max
(2 cpd–4
cpd). This might be due to the current design. In the main task, we
randomly presented either CS⫹ or CS⫺ tones prior to each target.
Thus, as CS⫺ trials were intermixed with CS⫹ trials, observers
may have had a baseline level of arousal throughout the whole
session that was higher than normal. However, the random inter-
mixing of trials was useful in that it revealed that a brief increase
in arousal induced by a CS⫹ tone can alter perception. Future
research should compare sessions with fear-conditioned tones with
sessions without any fear-conditioned tones to examine whether
there are more global effects of arousal (e.g., Lee, Itti, & Mather,
2012) in addition to the transient trial-by-trial effects seen here.
Presenting a fear-conditioned stimulus activates the amygdala
(e.g., Lim, Padmala, & Pessoa, 2009), and it has been argued that
Figure 2. (A) Skin conductance responses during the study (B) Estimated parameters; peak gain (␥
max
), peak
spatial frequency (f
max
), and bandwidth (). (C) Contrast sensitivity function as a function of spatial frequency
for CS⫹ and CS⫺ conditions. Error denotes the standard within-subject error term (Loftus & Masson, 1994).
ⴱ
p ⬍ .05.
ⴱⴱ
p ⬍ .01.
ⴱⴱⴱ
p ⬍ .001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
982
LEE, BAEK, LU, AND MATHER