Attentional mechanisms in the generation of sympathy

Judgment and Decision Making, Vol. 4, No. 4, June 2009, pp. 297–306
Attentional mechanisms in the generation of sympathy
Stephan Dickert?1 and Paul Slovic 2,3
1 Max Planck Institute for Research on Collective Goods, Bonn, Germany
2 Decision Research, Eugene, Oregon
3 Department of Psychology, University of Oregon
Abstract
Empathic responses, such as sympathy towards others, are a key ingredient in the decision to provide help to those
in need. The determinants of empathic responses are usually thought to be the vividness, similarity, and proximity of
the victim. However, recent research highlights the role that attention plays in the generation of feelings. We expanded
on this idea by investigating whether sympathy depends on cognitive mechanisms such as attention. In two studies we
found that sympathy responses were lower and reaction times were longer when targets were presented with distractors.
In addition, online sympathy judgments that allow attentional focusing on a target lead to greater affective responses
than judgments made from memory. We conclude that attention is an ingredient in the generation of sympathy, and
discuss implications for research on prosocial behaviour and the interaction between attention and emotions.
Keywords: emotions, attention, pro-social behavior.
1 Introduction
1994; Peters & Slovic, 2000; Slovic, Finucane, Peters, &
MacGregor, 2002). A direct consequence of this mecha-
Witnessing the suffering of others often invokes emo-
nism is that people are more likely to generate sympa-
tional reactions in the observers. The link between em-
thetic responses when they are able to mentally imag-
pathic responses and willingness to provide help to oth-
ine the victim. In fact, research on perspective taking
ers has been the subject of recent research on emotional
(e.g., Batson et al., 2007; Davis, 1994) supports this no-
responses and prosocial behaviour. Feelings such as em-
tion and shows that more empathic concern is generated
pathy, sympathy, compassion, distress, pity, and even an-
for victims that are similar to the perceiver (Loewenstein
ticipated regret are typically involved in decisions to pro-
& Small, 2007). Additionally, we seem to “feel” more
vide assistance or donate money to those in need (Batson,
for individual victims than for groups of victims because
1990; Batson, Eklund, Chermok, Hoyt, & Ortiz, 2007;
mental images of single, identi?ed victims are more vivid
Dickert, 2008; Kogut & Ritov, 2005a, 2005b; Loewen-
and concrete (Jenni & Loewenstein, 1997; Kogut & Ri-
stein & Small, 2007; Small, Loewenstein, & Slovic,
tov, 2005a; Slovic, 2007; Västfjäll, Peters, & Slovic, in
2007).
preparation).
Given the prominent role of emotions in prosocial be-
haviour, research has begun to tackle the important ques-
The predisposition to perceive groups of people as less
tion of what drives the generation of feelings relevant for
unitary is closely connected to Gestalt theories of percep-
helping others. Slovic (2007) suggests a model by which
tion (Hamilton & Sherman, 1996), an association which
mental images and attention are two vital precursors for
highlights that perceptual processes are closely related
emotional reactions towards others in distress. Mental
to other, more complex impression formation processes
images can contain affective tags that serve as a signal
(Glöckner & Betsch, 2008; Kahneman, 2003). The con-
for the selection of behavioural alternatives (Damasio,
nection between Gestalt principles of perception and im-
pression formation is of particular interest, as perceptual
?We kindly thank Tehila Kogut and Deborah Small for sharing their
processes impose limitations on our ability to process
picture stimuli, and Michael Posner, Ellen Peters, and Andreas Glöck-
ner for their valuable comments on earlier drafts. Support for this re-
large numbers of people in the same way that psycholog-
search was provided by the National Science Foundation under Grant
ical processes may restrict our ability to feel compassion
SES0649509 and by the William and Flora Hewlett Foundation. Cor-
for large numbers of victims. Additionally, it shows that
respondence concerning this paper should be addressed to: Stephan
Dickert, Max Planck Institute for Research on Collective Goods, Kurt-
perceptual and attentional processes can in?uence affec-
Schumacher-Str. 10, D-53113 Bonn. E-mail: dickert@coll.mpg.de.
tive reactions, as posited by Slovic (2007).
297

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
298
1.1 The interaction between emotions and
victims. If attentional focus is indeed facilitating the gen-
attention
eration of empathic emotions for individual victims, then
other members of a group can assume a distracting role
Research on the interplay between emotions and attention
for single individual members. Sympathy for individual
has often highlighted the selective effects of emotions
victims would then be dependent on the constraints that
on people’s attentional focus (e.g., Fox, 2002). Highly
the distraction places on the ability to attend to each indi-
relevant affective stimuli are processed faster and hold
vidual.
attention longer than affectively neutral stimuli (East-
In two experiments, we used a paradigm that placed
wood, Smilek, & Merikle, 2001). The apparent effects of
participants in a position to react empathically to victims
emotional stimuli (such as threatening or fearful stimuli)
in need of help and manipulated their ability to visually
on behavioural responses (e.g., orienting reaction times)
attend to a single target victim. It was expected that dis-
prompted researchers to suggest neural networks that al-
tracting attention away from individual targets decreases
low for attentional and emotional modulation of visual
emotional responses. We hypothesized that visual dis-
processes (Vuilleumier & Driver, 2007). Top-down mod-
tractors, in the form of other victims, negatively in?uence
ulation of emotionally signi?cant stimuli on attentional
the attention needed to generate sympathy towards a tar-
tasks indicates that neural networks exist that allow for
get victim, and that sympathy judgments are higher for
rapid communication between attentional and emotional
a single target victim presented alone vs. with distrac-
neural systems (Bush, Luu, & Posner, 2000). Whereas
tor victims. Additionally, we were interested in whether
emotions can direct attention to affectively salient objects
these emotional reactions are different when targets were
(Ochsner & Phelps, 2007; Vuilleumier, Armony, Driver,
evaluated online (i.e., while visually focusing on a pic-
& Dolan, 2003; Vuilleumier, 2005), attention can in?u-
ture of the target) vs. when these evaluations were made
ence emotional reactions by inhibiting as well as enhanc-
from memory. Hastie and Park (1986) propose that judg-
ing and generating emotional reactions to stimuli (Fenske
ments from memory are more effortful than spontaneous
& Raymond, 2006). The ability to shift attention helps
online judgments, and that attending to a target is easer
in regulating one’s own negative emotional state (Gross,
when judged online vs. from memory. Affective reac-
2002; Posner & Rothbart, 2007; Rueda, Posner, & Roth-
tions are thought to be stronger when mental representa-
bart, 2005), while focal (spatial) attention facilitates sub-
tions are attended to and more vivid (Pham, 2007; Slovic
sequent and more elaborate emotional processing and can
et al., 2002). Judgments based on memory retrieval, on
have a profound effect on the generation of emotions
the other hand, can lead to less vivid impression forma-
(Holmes, Vuilleumier, & Eimer, 2003).
tion (e.g., Reyna & Brainerd, 1995) and subsequently to
The effects of attention on the generation of emo-
weaker emotional responses.
tions were demonstrated in a series of insightful ex-
periments that highlight the emotional inhibitory conse-
quences of attending to speci?c locations in one’s visual
2 Experiment 1
?eld (Fenske et al., 2005; Fenske, Raymond, & Kunar,
2004; Raymond, Fenske, & Tavassoli, 2003). These au-
2.1 Method
thors consistently show that not attending to distractor
stimuli devalues these on affective as well as other di-
Participants. Fifty-eight participants (79% female) with
mensions, a phenomenon that they term the attentional
an average age of M = 21.9 (SD = 4.5) at the University
inhibition hypothesis (Fenske & Raymond, 2006). For
of Oregon completed Experiment 1. All participants had
example, smiling and neutral distractor faces were seen
normal or corrected-to-normal vision.
as less trustworthy than attended target faces, and abstract
Design and materials. Online vs. memory judgments
Mondrian stimuli were evaluated as less cheerful when
and presence vs. absence of distractors were manipulated
unattended.
in a fully factorial 2x2 within-subjects design. Partici-
In this article, we expand on the attentional inhibition
pants rated their sympathy for victims identi?ed by a spa-
hypothesis by examining the facilitating effects that at-
tial cue that appeared either before or after a target pic-
tention has on the generation of sympathy. Groups and
ture. Online judgments were realized by a spatial cue
individuals are processed differently on cognitive (Ariely,
presented before the target picture, while memory judg-
2001; Hamilton & Sherman, 1996) and emotional dimen-
ments featured a cue that appeared after the presentation
sions; affective reactions towards single identi?ed vic-
of the target picture. Thus, in the online judgment condi-
tims are often more intense compared to groups of vic-
tion, participants were able to focus on the speci?c loca-
tims (Slovic, 2007). A peculiar and inherent property of
tion where the target picture would appear and make an
a group of individuals is that attention to any single in-
online sympathy judgment while attending to the picture.
dividual can be decreased by the presence of the other
In the memory condition, the spatial cue appeared after

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
299
No distractors
Distractors
Fixation cross: 150 msec
Cue: 500 msec
Picture(s): 3000 msec
Sympathy rating: ~
Figure 1: Design schematic (online judgment pair).
the presentation of the target picture, and sympathy judg-
screen was interleaved. After seeing the target picture,
ments were based on a memory representation. Target
participants rated how much sympathy they felt on a ver-
pictures were either ?anked by seven distractor victims
tical sliding scale (300 pixels = 8.7° anchored by 0=“No
or presented alone. The primary dependent variable was
sympathy at all” to 300=“Very much sympathy”) with a
sympathy ratings for the target picture. Reaction times
vertically movable cursor corresponding to movements of
for these sympathy ratings were also recorded.
the mouse. The memory judgment condition was identi-
The target and distractor pictures were part of the same
cal to the online judgment condition, except that pictures
set of eight pictures (four female children and four male
were presented before the cues, such that participants ?rst
children, taken from Kogut & Ritov, 2005a,b). Partici-
saw one (or eight) pictures, but could identify the target
pants saw each picture an equal number of times in the
only after the pictures disappeared and the spatial cue
four conditions. In each condition the eight possible pic-
was presented. The order of pictures presented was de-
tures served as the target victim twice, resulting in 64 ex-
termined with a Latin-square to ensure that any effect of
perimental trials.
picture order on sympathy would be counterbalanced.
Apparatus and procedure. Participants were seated
65cm in front of a 17” computer screen (resolution =
2.2 Results
1024 x 768) and were informed that they would see pic-
tures of children in need of a ?nancial contribution due to
Sympathy judgment. Sympathy ratings were averaged
a life-threatening disease. Participants were further told
across pictures for each condition. The results, depicted
that the child that elicited the highest average sympathy
in Figure 2, suggest that, regardless of whether judgments
would receive a donation from the experimenters on be-
were made online or from memory, target victims re-
half of the participants. Each trial started with a ?xation
ceived higher sympathy ratings when they were presented
cross presented for 150ms at the center of the screen, as
without distractor victims. Sympathy ratings were lowest
shown in Figure 1. In the online judgment conditions,
when the target victim was presented with distractors and
a red dot cue (size = .5°) appeared at one of eight pos-
judgments were made from memory.
sible locations (on a horizontal line) at the exact spatial
A repeated-measures analysis of variance (ANOVA)
location where the target picture (size = 4.5°) would ap-
with judgment mode and presence of distractors as
pear later. Dot cues were shown for 500ms, pictures for
within-subject factors revealed a signi?cant main effect
3000ms, and between dots and pictures a 150ms blank
for judgment mode, such that sympathy ratings were sig-

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
300
200
No distractors
2200
No distractors
Distractors
Distractors
90
1
2000
y
80
h
1
800
1
Sympat
1
70
eaction time
R
600
1
60
1
1
400
50
1
200
1
Online judgment
Memory judgment
Online judgment
Memory judgment
Figure 2: Mean sympathy ratings for Experiment 1.
Figure 3: Mean reaction times for Experiment 1.
ni?cantly higher when made online (M=179.6, SD=70.8)
ipants did not recall which picture to judge, we would ex-
vs. from memory (M=173.7; SD=72.5), F(1,57)=4.8, p
pect that their responses differ systematically compared
< .05, ? 2
p =.08.
Participants also gave higher sympa-
to the online condition. A 2 judgment condition (on-
thy ratings when target pictures were presented with-
line vs. memory) x 8 picture, repeated-measures ANOVA
out distractors (M=183.4, SD=70.1) vs. with distractors,
with sympathy ratings as the dependent variable revealed
(M=169.5; SD=73.2), F(1,57)=10.7, p < .01, ? 2
p =.16.
that some pictures evoked more sympathy than others,
Furthermore, a signi?cant interaction between judgment
F(7,399) = 56.4, p < .001, ?2=.50. However these ratings
mode and presence of distractors emerged from the data,
did not interact with the judgment condition, F(7,399) =
F(1,57)=10.8, p < .01, ? 2
p =.16. To further elucidate this
1.7, p = .10, ?2=.03, showing no evidence that the differ-
interaction, simple contrasts were conducted and revealed
ences in ratings of the individual pictures were dependent
that, for memory judgments, targets without distractors
on the judgment condition.
(M = 184.4; SD = 70.1) received signi?cantly higher sym-
Reaction time analysis.
A similar 2x2 repeated-
pathy ratings compared to targets with distractors (M =
measures ANOVA was conducted on participants’ av-
163.1; SD = 75.0), t(57)=3.8, p < .01, Cohen’s d=0.51.
erage reaction times for each condition. Results show
This difference was still marginally signi?cant for on-
that memory judgments took longer than online judg-
line judgments, with participants expressing higher sym-
ments, F(1,57)=9.8, p < .01, ? 2
p =.15, and participants’
pathy for targets without distractors (M = 183.3; SD =
reaction time was slower when distractors were present,
70.2) vs. with distractors present (M = 175.8; SD = 71.2),
F(1,57)=23.4, p < .001, ? 2
p =.29.
Additionally the in-
t(57)=1.8, p < .08, Cohen’s d = 0.24.
teraction between these two factors was also signi?cant,
It is possible that the observed decrease of sympa-
F(1,57)=15.7, p < .001, ? 2
p =.22. As can be seen in Fig-
thy judgments in the memory condition with distrac-
ure 3, reaction times were longer for memory judgments
tors was due to participants not being able to remember
when distractors were present. In this condition, partici-
which target picture they were rating and making an av-
pants had to ?rst recall which of the previously seen pic-
erage judgment instead. If this was indeed the case, we
tures corresponded to the cue before indicating their sym-
would expect lower variance in sympathy responses es-
pathy rating. Given that the reaction times are similar in
pecially in the memory judgment condition when distrac-
the other conditions, it is possible that the dif?culty and
tors were presented. However, we found no evidence for
compound effects of memory retrieval and distractors are
decreased variance in this condition relative to the other
responsible for the longer reaction times.
conditions. In fact, F-tests revealed that variances were
We further investigated whether participants’ sym-
similar in all four conditions, Fs<1.1, ps>.31. To fur-
pathy judgments correlated with reaction times.
Al-
ther exclude the possibility that participants were mak-
though faster reactions times were generally accompa-
ing average judgments in the memory condition, we also
nied by higher sympathy ratings in each of the four con-
examined whether signi?cant differences in sympathy
ditions (rs ranged from ?.24 to ?.01), the correlation
judgments existed for the individual pictures. If partic-
approached conventional signi?cance levels only in the

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
301
online judgment/no-distractor condition, r(57) = ?.24,
ting to test whether the effects of attention on the genera-
p<.08. However, across all observations, the correlation
tion of empathic feelings generalize to a different culture.
between sympathy ratings and reaction times was statis-
tically signi?cant, r(230) = ?.13, p = .037.
3.1 Method
2.3 Discussion
Participants. Forty-eight participants (53% female) from
the University of Bonn, Germany, and community mem-
Experiment 1 was designed to investigate the role of
bers with an average age of 25.7 (SD = 7.4) took part in
attention in the generation of sympathy. Results sup-
this study and were paid an average of 12C as compen-
ported the hypothesis that a single target victim evokes
sation for their time in a test battery that included other
more sympathy when presented alone vs. ?anked by dis-
experiments unrelated to this study.
tractor victims, which is in line with ?ndings that emo-
Design and materials. The design and materials were
tional responses decrease as the number of victims in-
similar to those used in Experiment 1. Participants saw a
crease (Slovic, 2007). This effect was present when judg-
total of 64 experimental trials, in which they rated their
ments were made online as well as from memory, how-
sympathy for one of eight possible children suffering
ever it was especially pronounced when affective judg-
from an unspeci?ed disease. The two variables of interest
ments were made based on memory. It seems likely that
(judgment mode and presence of distractors) were manip-
online processing enabled more vivid images and, in turn,
ulated in the same fashion as in Experiment 1. However,
stronger empathic responses than memory processing, an
unlike Experiment 1, here we reduced the number of dis-
explanation that dovetails nicely with research on person
tractor pictures to three and used a block-design where
perception (Hamilton & Sherman, 1996) and affect (e.g.,
trials were blocked by judgment mode. Half of the par-
Slovic, et al., 2002; Loewenstein, Weber, Hsee, & Welch,
ticipants made online judgments for the ?rst 32 trials and
2001).
memory judgments for the second 32 trials, and this or-
A possible alternative explanation of the results in Ex-
der was reversed for the other half. Within each block,
periment 1 is that participants gave lower sympathy rat-
target pictures were presented randomly with and with-
ings in the memory judgment/distractor condition simply
out distractors. The target and distractor pictures could
because they were unable to clearly identify which target
appear in four locations: above, below, right, or left at an
they were supposed to rate. However, similar results (al-
equal distance from a central ?xation cross. As in Experi-
beit less pronounced) were obtained for the online judg-
ment 1, the distractor variable was crossed with judgment
ment condition, which cannot be explained by an account
mode. On half of the trials a spatial cue appeared before
that focuses solely on unsuccessful retrieval of the target
the picture(s), and followed the pictures on the other half
picture. Moreover, if participants were unable to identify
of the trials. The spatial cue was presented for 500ms and
the individual pictures in the memory judgment/distractor
the pictures for 4000ms. At the end of each trial, partic-
condition and instead gave average ratings, we would ex-
ipants rated their respective sympathy level of the target
pect little difference in how the individual pictures were
picture with a sliding scale (500 pixels = 13.8°, anchored
rated in this condition. This was not the case, and dif-
by 0=“No sympathy at all” to 500=“Very much sympa-
ferences in sympathy ratings for individual pictures did
thy”).
not depend on the judgment condition. Nonetheless, we
After completion of the 64 experimental trials, 24 ma-
addressed this concern directly in Experiment 2 to clar-
nipulation check trials were added in which participants
ify the role that correct identi?cation plays in empathic
had to correctly identify a target picture. The correct
responses.
identi?cation of the target picture was a concern only for
trials in the memory condition with distractors. Thus, the
manipulation check trials had a similar structure as the
3 Experiment 2
memory judgment condition: In each of these trials, four
pictures were presented after the ?xation cross, followed
Experiment 2 sought to replicate the general ?ndings in
by a cue. Participants were then asked to judge whether
Experiment 1 and rule out the possibility that lower sym-
a test picture corresponded to the target picture identi?ed
pathy ratings were mainly a product of unsuccessful re-
by the cue. Three types of manipulation check trials were
trieval of the target picture. The number of distractors
used, each presented a total of eight times: The test pic-
was reduced and the viewing time for the pictures was ex-
ture was (1) identical to the target picture, (2) not iden-
tended to facilitate better encoding. We also added a ma-
tical to the target picture but part of the picture set used
nipulation check to verify that participants could indeed
in the study, and (3) not identical to the target picture and
identify the target retrospectively. Moreover, we were in
belonged to a completely different set of pictures not used
a position to replicate the basic ?ndings in a different set-
in the 64 experimental trials.

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
302
3.2 Results
260
No distractors
Manipulation check. Overall, every participant answered
Distractors
more than 87% of the manipulation check trials correctly,
240
and none performed below 75% in any of the three ma-
nipulation check trial types. We took this as evidence
that participants were quite able to correctly identify the
y
220
h
target picture and base their sympathy judgment on the
correct mental representation when target pictures were
Sympat
200
presented with distractors in the memory condition.
Sympathy judgments.
A preliminary analysis con-
80
1
?rmed that no signi?cant difference existed between the
orders in which the blocked online and memory judg-
ments were made, which justi?ed simplifying further
60
1
analyses to a 2 (online vs. memory judgment) x 2 (dis-
tractors vs. no distractors) within-subjects design. An
Online judgment
Memory judgment
ANOVA with judgment mode and presence of distractors
revealed a signi?cant main effect for distractors, F(1,47)
Figure 4: Mean sympathy ratings for Experiment 2.
= 4.19, p < .05, ? 2
p = .08. Participants gave higher sym-
pathy judgments when pictures were presented without
distractors (M = 230.5, SD = 69.4) than when distractors
p= .01, ? 2
p =.19. As expected, sympathy judgments were
were present (M = 218.8, SD = 90.6). The main effect
faster when made online than when made from memory.
for judgment mode was not signi?cant, F(1,47) = 1.01,
Additionally, sympathy judgments were faster when tar-
p = .32, ? 2
p = .02, however the mean difference was in
gets were presented without distractors. Apart from the
the predicted direction such that participants gave higher
main effects, the interaction between judgment mode and
sympathy judgments when making their judgment online
distractors was also signi?cant, F(1,47) = 11.37, p= .01,
(M = 227.3, SD = 86.6) vs. from memory (M = 222.0, SD
? 2
p =.20. Figure 5 illustrates that judgments were gener-
= 73.8). Finally, although the interaction between judg-
ally slower when distractors were present, but that this ef-
ment mode and presence of distractors was not signi?-
fect was particularly present when judgments were made
cant, F(1,47) < 1, p = .72, ? 2
p = .01, simple contrasts
online, t(47)=5.5, p <.001, Cohen’s d = 0.79. In the mem-
revealed that participants gave signi?cantly higher sym-
ory condition, the effect of distractors on participants’ re-
pathy judgments without distractors (M = 228.5, SD =
action times was minimal, t(47)=.1, p =.93, Cohen’s d =
71.0) vs. with distractors (M = 215.5, SD = 76.5) only
0.01. Unlike in Experiment 1, we did not ?nd that reac-
in the memory judgment condition, t(47) = 4.06, p <.001,
tion times correlated with sympathy ratings, (rs ranged
Cohen’s d = 0.59. This effect was also present but not sig-
from .14 to -.02), ps >.35.
ni?cant when participants made online judgments (M =
222.1, SD = 105.3 and M = 232.4, SD = 67.8 for with and
3.3 Discussion
without distractors, respectively), t(47) = 1.13, p =.26,
Cohen’s d = 0.16. See Figure 4 for details.
Experiment 2 was designed to replicate and extend the
As in Experiment 1, we tested whether the decrease in
?nding that presenting distractors reduces emotional re-
sympathy judgments in the memory condition with dis-
sponses to individual targets while controlling whether
tractors was based on participants’ not being able to iden-
participants could successfully recall the target picture in
tify the target picture retrospectively and instead making
the memory condition. As predicted, participants gave
an average judgment. A 2 judgment condition (online
lower sympathy ratings towards individual victims when
vs. memory) x 8 picture repeated-measures ANOVA indi-
distractor victims were presented simultaneously. Ad-
cated that although sympathy ratings differed for the pic-
ditionally, in Experiment 2 sympathy judgments from
tures, F(7,329) = 13.5, p < .001, ?2=.22, this did not de-
memory were lower than online judgments, as was ex-
pend on whether pictures were seen in the memory or on-
pected, albeit not signi?cantly so.1 It is of note that reduc-
line judgment condition as indicated by a non-signi?cant
1In order to increase power, we reanalyzed our sympathy data com-
interaction, F(7,329) = 0.4, p = .87, ?2=.01.
bining both experiments. The results showed that participants gave
Reaction time results.
A similar 2 x 2 factorial
higher sympathy ratings when target pictures were shown without dis-
ANOVA for reaction times showed a signi?cant main
tractors (M = 205, SD = 73.1) compared to when they were presented
with distractors (M = 191.8, SD = 85.5), F(1,105) = 14.0, p < .001,
effect for judgment mode, F(1,47)=26.52, p < .001,
?2=.12. Sympathy ratings were also higher when participants made
? 2
p =.36, and presence of distractors, F(1,47) = 10.75,
their judgments online (M = 201.2, SD = 82.4) vs. from memory (M =

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
303
4 General discussion
2200
No distractors
Distractors
Two experiments were conducted to test the hypothesis
that distractors and judgment mode in?uence the gener-
2000
ation of emotions. Speci?cally, these studies were de-
signed to test whether attention is a precursor for em-
800
1
pathic feelings, as proposed by Slovic (2007). Across
both experiments, sympathy for others was lower when
distractor victims were present. Additionally, we found
eaction time
R
600
1
evidence that sympathy judgments were higher when
made online vs. from memory. Both of these results in-
dicate that attention to a target can intensify an emotional
1
400
response. Reaction time results further show that judg-
ments from memory take longer, which is indicative of a
more dif?cult retrieval process compared to online judg-
200
1
ments. We found partial evidence that longer reaction
Online judgment
Memory judgment
times coincide with lower sympathy ratings indicating
Figure 5: Mean reaction times for Experiment 2.
that sympathy judgments may be sensitive to the timing
of the emotional response. Greater temporal distance be-
tween encountering and emotionally reacting to a victim
might be related to and contribute to other determinants
ing the number of distractors from seven to three across
of sympathy, such as vividness and newness (Loewen-
experiments did not eliminate the effect of distractors on
stein & Small, 2007). Our ?ndings also suggest that em-
sympathy judgments. However, compared to Experiment
pathic concern for others and the often observed reduc-
1, participants in Experiment 2 gave lower sympathy rat-
tion in empathy for multiple victims is, at least partly,
ings in general.
a result of divided attention.
This result contributes
Reaction times in Experiment 2 were slightly differ-
to recent advances made in the exploration of how at-
ent compared to Experiment 1. Memory judgments took
tentional mechanisms in?uence social-emotional evalu-
longer than online judgments in both Experiments, but
ations, and extends research by Fenske and colleagues
the effect of distractors was most prominent in the mem-
(e.g., Fenske & Raymond, 2006), who examined the in-
ory condition in Experiment 1, whereas in Experiment
hibitory effects of attention on judgments of trustworthi-
2 the effects of distractors was most visible in the on-
ness of faces (Fenske et al., 2005; Raymond, Fenske, &
line condition. It would be expected that seven distrac-
Westoby, 2005) and cheerfulness (Fenske, Raymond, &
tors are more attention diverting than three, resulting in
Kunar, 2004).
slower judgments in the memory condition in Experi-
Our results are best viewed from the perspective that
ment 1 compared to Experiment 2. However, it is unclear
the effects of attention are not unilateral, and can inhibit
why participants were faster in the online judgment/no-
as well as facilitate emotional reactions. Thus, we sug-
distractor condition compared to the distractor condition
gest that sympathy is not generated to a similar degree
in Experiment 2 and could be due to the change in dis-
when target victims are ?anked by distractor victims due
tance between distractors and target. Additionally, it is
to attentional constraints of the perceptual system (e.g.,
possible that having fewer distractors leads to a slower,
Posner & Raichle, 1994). Furthermore, our results also
more comparative process underlying the sympathy judg-
shed new light on the identi?ed-victim and singularity
ment of the target. Hyde and Spelke (2009) show that
effects reported in research on prosocial behaviour (e.g.,
a fundamental difference exists in how small and large
Kogut & Ritov, 2005a,b; Small, Loewenstein, & Slovic,
numbers are processed, and that adults treat small num-
2007) by suggesting that a precursor to affective reactions
bers (i.e., 1–3) as separate objects that are individually
is attentional focus.
compared.
In our studies we used pictures of other victims as dis-
tractors to target victims. It is possible that the observed
attention effects on sympathy ratings are not limited to
195.6, SD = 77.0), F(1,105) = 4.1, p < .05, ?2=.04. The interaction be-
the use of other victims as distractors and could be visi-
tween presence of distractors and judgment mode was also signi?cant,
ble with other classes of stimuli. However, evidence ex-
F(1,105) = 4.6, p < .001, ?2=.04. Of note is that sympathy judgments
ists that human faces tend to be processed differently and
were higher for targets shown without distractors in both the online and
memory condition, t(105) = 1.86, p = .065 and t(105) = 5.2, p <.01,
are more attention grabbing than other, non-face stim-
respectively.
uli (Downing, Chan, Peelen, Dodds, & Kanwisher, 2006;

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
304
Lavie, Ro, & Russell, 2003; Ro, Russell, & Lavie, 2001;
sibly a higher willingness to help others, it might be bet-
Theeuwes & Van der Stigchel, 2006). In fact, Ro et al.
ter to use presentation formats that take advantage of at-
demonstrate that faces are preferentially attended to in
tentional processes (e.g., single presentation). While we
comparison to other common objects, and Theeuwes and
have used a research task speci?c to prosocial behavior,
Van der Stigchel argue that the discrimination of human
the attentional mechanisms discussed in the current pa-
faces from other objects is based on pre-attentive and un-
per are important for other tasks that capitalize on the re-
conscious processing, which automatically draws focal
lationship between emotions and decision making (such
attention to the faces.
as the endowment effect). Future research should address
the extent to which attention is a precursor to affective
4.1 Alternative accounts and limitations
reactions related to the construction of preferences and
valuations.
Alternative accounts for how perceptual experience
translates into affective evaluations (such as the mere-
exposure effect and perceptual ?uency) could potentially
References
explain our results. However, the fact that participants
saw each victim an equal amount of times makes explana-
Ariely, D. (2001). Seeing sets: Representation by statis-
tions based on theories that capitalize on the mere expo-
tical properties. Psychological Science, 12, 157–162.
sure effect less likely (Murphy & Zajonc, 1993; Zajonc,
Batson, C. D. (1990). How social an animal? The human
1968). The result that individual victims received higher
capacity for caring. American Psychologist, 45, 336–
sympathy ratings when presented alone seems to be better
346.
explained by the in?uence that distractors have on partic-
Batson, C. D., Eklund, J. H., Chermok, V. L., Hoyt, J.
ipants’ attention to the individual target when other vic-
L., & Ortiz, B. G. (2007). An additional antecedent of
tims were present. Research on perceptual ?uency (e.g.,
empathic concern: Valuing the welfare of the person
Winkielman, Schwarz, Fazendeiro, & Reber, 2003) posits
in need. Journal of Personality and Social Psychology
that easier cognitive processing elicits differential emo-
93, 65–74
tional reactions. Recall that sympathy judgments were
Bush, G., Luu, P. & Posner, M. I. (2000). Cognitive and
lowest in the distractor/memory condition, which is also
emotional in?uences in the anterior cingulate cortex.
cognitively most demanding. However, the effect of dis-
Trends in Cognitive Science, 4, 215–222.
tractors was also present in the cognitively less demand-
Damasio, A. R. (1994). Descartes’ error: Emotion, rea-
ing online judgment condition. Furthermore, the percep-
son, and the human brain. New York: Avon.
tual ?uency account suggests that ease of processing re-
Davis, M. H. (1994). Empathy: A social psychological
sults in distinctly positive affect. However, empathic con-
approach. Madison, WI: Brown & Benchmark.
cern is usually classi?ed as an arousal state with distinctly
Dickert, S. (2008). Two Routes to the Perception of Need:
negative valence that motivates prosocial behaviour in or-
The role of affective and deliberative information pro-
der to reduce this negative feeling (Batson, 1990).
cessing in pro-social behavior. Doctoral Dissertation,
A possible limitation in our design was that we did not
University of Oregon, OR.
collect sympathy judgments for distractors, which would
Downing, P., Chan, A., Peelen, M., Dodds, C., & Kan-
have allowed us to directly compare whether attended tar-
wisher, N.(2006). Domain speci?city in visual cortex.
gets receive more sympathy than unattended distractors.
Cerebral Cortex, 16, 1453–1461.
Additionally, because we were interested in the connec-
Eastwood, J. D.,Smilek, D., & Merikle, P. M. (2001) Dif-
tion between perceptual systems and emotions we specif-
ferential attentional guidance by unattended faces ex-
ically investigated the role of focal visual attention, which
pressing positive and negative emotion. Perception &
does not allow de?nite conclusions for other forms of at-
Psychophysics 63, 1004–1013.
tention (Posner & Rothbart, 2007).
Fenske, M. J. & Raymond, J. E. (2006). Affective in-
?uences of selective attention. Current Directions In
4.2 Implications and future research
Psychological Science, 15, 312–316.
Fenske, M. J., Raymond, J. E., Kessler, K., Westoby,
Our results point to exciting relationships between atten-
N., & Tipper, S. P. (2005). Attentional inhibition has
tional and affective systems, which are of importance in
social-emotional consequences for unfamiliar faces.
understanding the generation of feelings and its conse-
Psychological Science, 16, 753–758.
quences for behavior. Presenting a group of people in
Fenske, M. J., Raymond, J. E., & Kunar, M. A. (2004).
need of help can increase the dif?culty of attending to
The affective consequences of visual attention in pre-
any single individual, leading to lower sympathy. Conse-
view search. Psychonomic Bulletin and Review, 11,
quently, in order to elicit more empathic concern and pos-
1055–1061.

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
305
Fox, E. (2002) Processing of emotional facial expres-
Peters, E. & Slovic, P. (2007). Affective asynchrony and
sions: The role of anxiety and awareness. Cognitive,
the measurement of the affective attitude component.
Affective & Behavioral Neuroscience, 2, 52–63.
Cognition and Emotion, 21, 300–329.
Glöckner, A., & Betsch, T. (2008). Modeling option and
Peters, E., & Slovic, P. (2000).
The springs of ac-
strategy choices with connectionist networks: Towards
tion: Affective and analytical information processing
an integrative model of automatic and deliberate deci-
in choice. Personality and Social Psychology Bulletin,
sion making. Judgment and Decision Making, 3, 215–
26, 1465–1475.
228.
Pham, M. T. (2007). Emotion and rationality: A critical
Gross, J. (2002). Emotion regulation: Affective, cogni-
review and interpretation of empirical evidence. Re-
tive, and social consequences. Psychophysiology, 39,
view of General Psychology, 11, 155–178.
281–291.
Posner, M. I. & Raichle, M. E. (1994). Images of Mind.
Hamilton, D. L. & Sherman, S. J. (1996). Perceiving per-
Scienti?c American Books.
sons and groups. Psychological Review, 103, 336–355.
Posner, M. I. & Rothbart, M. K. (2007). Research on at-
Hastie, R., & Park, B. (1986). The relationship between
tention networks as a model for the integration of psy-
memory and judgment depends on whether the judg-
chological sciences. Annual Review of Psychology, 58,
ment task is memory-based or on-line. Psychological
1–23.
Review, 93, 258–268.
Raymond, J. E., Fenske, M. J., & Tavassoli, N. T. (2003).
Holmes, A., Vuilleumier, P., & Eimer, M. (2003). The
Selective attention determines emotional responses to
processing of emotional facial expression is gated by
novel visual stimuli. Psychological Science, 14, 537–
spatial attention: Evidence from event-related brain
542.
potentials. Cognitive Brain Research, 16, 174–184.
Raymond, J. E., Fenske, M. J., & Westoby, N. (2005).
Hyde, D., & Spelke, E. (2009). All numbers are not
Emotional devaluation of distracting patterns and
equal: An electrophysiological investigation of small
faces: A consequence of attentional inhibition during
and large number representations. Journal of Cogni-
visual search? Journal of Experimental Psychology:
tive Neuroscience, 21, 1039–1053.
Human Perception and Performance, 31, 1404–1415.
Jenni, K. E. & Loewenstein, G. (1997). Explaining the
Reyna, V. F., & Brainerd, C. J. (1995). Fuzzy-trace the-
“identi?able victim effect.” Journal of Risk and Un-
ory: An interim synthesis. Learning and Individual
certainty, 14, 235–257.
Differences, 7, 1–75.
Kahneman, D. (2003). A perspective on judgment and
choice: Mapping bounded rationality. American Psy-
Ro, T., Russell, C, & Lavie, N. (2001). Changing faces:
chologist, 58, 697–720.
A detection advantage in the ?icker paradigm. Psycho-
logical Science, 12, 94–99.
Kogut, T. & Ritov, I. (2005a). The “Identi?ed Victim”
effect: An identi?ed group, or just a single individual?
Rueda, M. R., Posner, M. I., & Rothbart, M. K. (2005).
Journal of Behavioral Decision Making, 18, 157–167.
The development of executive attention: Contributions
Kogut, T., & Ritov, I. (2005b). The singularity of identi-
to the emergence of self-regulation. Developmental
?ed victims in separate and joint evaluations. Organi-
Neuropsychology, 28, 573–594.
zational Behavior and Human Decision Processes, 97,
Slovic, P. (2007). “If I look at the mass I will never act”:
106–116.
Psychic numbing and genocide. Judgment and Deci-
Lavie, N., Ro, T., & Russell, C. (2003). The role of per-
sion Making, 2, 79–95.
ceptual load in processing distractor faces. Psycholog-
Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D.
ical Science, 14, 510–515.
G. (2002). The affect heuristic. In T. Gilovich & D.
Loewenstein, G., & Small, D. (2007). The scarecrow and
Grif?n & D. Kahneman (Eds.), Heuristics and biases:
the tin man: The vicissitudes of human sympathy and
The psychology of intuitive judgment (pp. 397–420).
caring. Review of General Psychology , 11, 112–126.
New York: Cambridge University Press.
Loewenstein, G., Weber, E. U., Hsee, C. K., & Welch, N.
Small, D. A., Loewenstein, G., & Slovic, P. (2007). Sym-
(2001). Risk as feelings. Psychological Bulletin, 127,
pathy and callousness: Affect and deliberations in do-
267–286.
nation decisions. Organizational Behavior and Human
Murphy, S. T., & Zajonc, R. B. (1993). Affect, cogni-
Decision Processes, 102, 143–153.
tion, and awareness: Affective priming with optimal
Theeuwes, J., & Van der Stigchel, S. (2006). Faces cap-
and suboptimal stimulus exposures. Journal of Per-
ture attention: Evidence from inhibition of return. Vi-
sonality and Social Psychology, 64, 723–739.
sual Cognition, 13 , 657–665.
Ochsner, K. & Phelps, E. (2007). Emerging perspectives
Västfjäll, D., Peters, E., & Slovic, P. (In preparation).
on emotion–cognition interactions. Trends in Cogni-
Representation, affect, and willingness-to-donate to
tive Sciences, 11, 317–318.
children in need. Unpublished manuscript.

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Attentional mechanisms in the generation of sympathy
306
Vuilleumier, P (2005). How brains beware: neural mech-
Winkielman, P., Schwarz, N., Fazendeiro, T., & Reber,
anisms of emotional attention. Trends in Cognitive Sci-
R. (2003). Hedonic marking of processing ?uency:
ences, 9, 586–594.
Implications for evaluative judgment. In J. Musch &
Vuilleumier, P., Armony, J. L., Driver, J., & Dolan, R. J.
K.C. Klauer (Eds.), Psychology of evaluation: Affec-
(2003). Distinct spatial frequency sensitivities for pro-
tive processes in cognition and emotion (pp. 189– 217).
cessing faces and emotional expressions. Nature Neu-
Mahwah, NJ: Erlbaum.
roscience, 6, 624–631.
Zajonc, R. B. (1968). Attitudinal effects of mere expo-
Vuilleumier, P., & Driver, J. (2007). Modulation of vi-
sure. Journal of Personality and Social Psychology, 9,
sual processing by attention and emotion: windows
1–27.
on causal interactions between human brain regions.
Philosphical Transactions of The Royal Society, 362,
837–855.

---