Content uploaded by Peter H Kenning
Author content
All content in this area was uploaded by Peter H Kenning on May 03, 2018
Content may be subject to copyright.
Brain Research Bulletin 67 (2005) 413–421
Evidence for a neural correlate of a framing effect: Bias-specific activity
in the ventromedial prefrontal cortex during credibility judgments
M. Deppea,∗,1, W. Schwindtb,1,J.Kr
¨
amera, H. Kugelb, H. Plassmannc,
P. Kenningc, E.B. Ringelsteina
aDepartment of Neurology, University of Muenster and University Hospital Muenster (UKM), Albert-Schweitzer-Str. 33,
D-48129 Muenster, Germany
bDepartment of Clinical Radiology University of Muenster and University Hospital Muenster (UKM), Germany
cDepartment of General Management, University of Muenster and University Hospital Muenster
(UKM), Germany
Available online 25 July 2005
Abstract
Background: Neural processes within the medial prefrontal cortex play a crucial role in assessing and integrating emotional and other implicit
information during decision-making. Phylogenetically, it was important for the individual to assess the relevance of all kinds of environmental
stimuli in order to adapt behavior in a flexible manner. Consequently, we can in principle not exclude that environmental information covertly
influences the evaluation of actually decision relevant facts (“framing effect”).
Objective: Totestthe hypothesis that the medial prefrontal cortexis involvedinto a framing effectwe employed functional magnetic resonance
imaging (fMRI) during a binary credibility judgment task.
Methods: Twenty-one subjects were asked to judge 30 normalized news magazine headlines by forced answers as “true” or “false”. To
confound the judgments by formally irrelevant framing information we presented each of the headlines in four different news magazines
characterized by varying credibility. For each subject the susceptibility to the judgment confounder (framing information) was assessed by
magazine-specific modifications of the answers given.
Results: We could show that individual activity changes of the ventromedial prefrontal cortex during the judgments correlate with the degree
of an individual’s susceptibility to the framing information.
Conclusion: We found (i) a neural correlate of a framing effect as postulated by behavioral decision theorists that (ii) reflects interindividual
differences in the degree of the susceptibility to framing information.
© 2005 Elsevier Inc. All rights reserved.
Keywords: Prejudgements; Framing-effect; Uncertainty; Branding; Ventromedial; Prefrontal cortex; Neuroeconomics; Decision-making; fMRI
1. Introduction
Decision-making is a process of generating, evaluating,
and selecting among a set of at least two choice alternatives
[13]. In real-life situations, the choices involve a variable
degree of uncertainty [21,40]. The current literature reflects
the growing interest in the investigation of the neural basis
and the mechanisms of decision-making and judgment, par-
∗Corresponding author. Tel.: +49 251 8348174; fax: +49 251 8352064.
E-mail address: mail@Michael-Deppe.de (M. Deppe).
1Authors contributed equally to this study.
ticularly in the context of interactions with emotions [36,38].
Very frequently, decisions and judgments are not only based
on the explicit information presented during a decision task,
but are also influenced by conscious and probably uncon-
scious implicit background knowledge, integrated into the
decision process [28]. Recently performed functional imag-
ing experiments, when targeting the neural correlates of
decision-making, game theory and economic behavior, were
mainly focused on explicit, decision relevant information
(for an overview, see Glimcher and Rustichini [18]). Inter-
actions between formally decision-relevant information and
environmental factors are not well understood [19]. Cortical
0361-9230/$ – see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.brainresbull.2005.06.017
414 M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421
regions predominantly relevant for human decision-making
include the orbitofrontal cortex (OFC) and the ventrome-
dial part of the prefrontal cortex (VMPFC) [12,26,39]. These
paralimbic regions play a central role in the integration of
information during decision-making. Based on connections
withassociation areas of allsensoryandexecutivemodalities,
limbic cortices and subcortical nuclei operate as integra-
tors of the physical and emotional attributes of objects. For
example, the VMPFC has been reported to represent intro-
spective,self-reflective information, stimulus-rewardsvalues
[25,31,33,45] or positive associations with visual stimuli like
the attractiveness of faces [32]. In addition to these find-
ings, deactivation in this region was observed, as soon as the
difficulty of a cognitive task increased [35]. The latter was
interpretedasanemotional gating aimed at inhibiting adverse
emotional signal to maximize the level of performance.
Here, we investigated the role of the medial prefrontal
cortex in human subjects during decision-making. The sub-
jectshadtogive binary credibility judgments (“true”,“false”)
of messages under the influence of confounding framing
information. Our study was based on Kahneman’ and Tver-
sky’sconcept, that information processed bythe human brain
contributingtoa decision can be divided into (I) explicit deci-
sion relevant information and (II) formally, but not actually,
decision irrelevant, implicit information [22,23]. Decisions
and judgments based on type I information, could be eval-
uated by situation-related facts maintained in the working
memory (WM) (“task”) and by usually incomplete, con-
scious, explicit episodic memory contents (“knowledge”).
Decisions based on this type of information can be eval-
uated corresponding to a certain decision strategy, or by
explicit rules allowing to formalize and quantify the deci-
sion relevant information by scalar (one-dimensional) vari-
ables reflecting measures like “probability”, “likelihood”,
“expected utility”, “costs”, or “use” of multiple decision or
judgment options. This type of information is usually the
basis for a formal approach of decision-making [13]. The
type II information represents completely different infor-
mation integrated into the neuronal decision process, like
conscious or unconscious experiences with a similar task in
the past. Further examples for type II information are asso-
ciations with intrinsic scales like “beauty” or “faithfulness”,
perceived pain, rewards, punishments, sensorial information,
emotions and other signaled body information [27,37]. The
intuitive integration of implicit information can be triggered
by associations evoked by the decision object or the decision
situation. In the literature, the latter process is described as
“framing” [20,24,41]. According to Volz et al., the two vari-
ants of information and uncertainty have neural correlates,
which can be assessed by means of functional brain imaging
during judgment tasks [42–44]. The objective of the present
study was to investigate the neural basis of framing processes
and their influence on internal uncertainty. We hypothesized
that the ventromedial prefrontal cortex is involved in inte-
grating implicit type II information into behavioral-relevant
brain processes.
2. Materials and methods
2.1. Functional MRI experiment—subjects
Eleven male (mean age 25.5, range 20–33) and 10 female
(mean age 25, range 22–29) healthy subjects participated in
the present study. Standard exclusion criteria for MR exami-
nations, like metal implants, were applied. Because we used
visual stimuli, subjects with strong myopia or other rele-
vant constraints of vision were also excluded. All subjects
provided written informed consent prior to the scanning ses-
sions. The subjects were also informed that the examination
could potentially reveal medically significant findings and
were asked if they would like to be notified in this case. The
study was approved by the local ethic committee.
2.2. Behavioral protocol: the judgment task
For the judgment task, we developed 45 completely fic-
titious, but potentially conceivable statements in the form
of magazine headlines, with topics related to current news.
The statements were designed targeting a variable credibil-
ity. They had first been rated by 100 different subjects on
a score ranging from 1, “absolutely incredible” to 7, “abso-
lutely credible” (see Fig. 1). The 10 most credible (group
“+”), the 10 most incredible (group “−”) and 10 ambigu-
ous headlines (group “0”) were selected for the present fMRI
experiment (factor credibility C, levels “+”, “−”, “0”). In the
following, three illustrative statements are given from group
“+”, group “0” and group “−”, respectively: “since 1995 the
number of ice bears decreased more than 5%”, “American
research institutes proved that coffee can promote the gen-
eration of cancer”, “train ticket prices will decrease strongly
for short-distance travels in autumn” (for further details see
Fig. 1). During fMRI, the 30 statements were presented four
times in combination with each of four different magazine
logos of well known German news magazines, here denoted
as A, B, C and D. For the data analysis these logos were
used as the four levels of the judgment confounder M. The
inter-stimulus interval between all 4×30 = 120 judgments
was standardized to 10s each. During repetitive presenta-
tion, the headlines were semantically, but not syntactically
identical. Details about the design of the visual presentation
are illustrated in Fig. 2. The subjects were requested (forced
choice) to judge the presented headlines as “true” or “false”
by pressing corresponding buttons on a magnetic resonance
compatible response box. The responses were recorded by
thestimulationsoftware ShowPics, which wasusedforvisual
presentation [10]. Separately for each magazine M, we cal-
culated the response bias
BM=Ntrue −Nfalse
Ntrue +Nfalse (1)
with Ntrue,Nfalse: number of ratings judged as “true” or
“false”, respectively (−1<BM<1). Additionally, the vari-
ability of the response bias was determined by the standard
M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421 415
Fig. 1. Mean ratings and standard deviation of 45 totally fictitious news magazine statements from 100 subjects assessed by questionnaires. The ratings were
graded into seven steps (1, totally uncredible, ..., 7, totally credible). The 10 most credible (group “+”), the 10 most incredible (group “−”) and 10 indifferent
statements (group “0”) were selected for our experiment. By this kind of “normalization” it was expected, that statements of the three groups would mainly be
judged as “false”, “ambiguous” and “true”, respectively.
deviation
σB=1
4
M=A–D
(B−BM)2(2)
using the mean bias
B=(BA+BB+BC+BD)
4(3)
2.3. Image presentation
A dedicated shielded fMRI projection system (Covilex,
Magdeburg, Germany) provided high quality image presen-
Fig. 2. Graphical design of the visual presentation. Additionally to the text,
which represented the explicit decision relevant information, one of the four
different logos A, B, C, or D and corresponding style elements of four Ger-
man news magazines were presented.
tation. Controlled by a personal computer in the MR con-
trol room, images were projected onto an approximately
50cm×50cm field on a screen fixed at the rear opening
of the MR bore. A subject lying in the bore could view the
screen via a 45◦mirror fixed at the top of the head coil. As
the images covered about 50% of the subject’s whole field
of view, even small details of the displayed logos could be
recognized easily. Care was taken to present the different
news headlines and logos equal in size, position, background
and luminance in order to prevent confounding visual stim-
ulation. Additionally, the position of the logos was changed
randomly from trial to trial to focus the attention to the logos
and to prevent habituation effects.
2.4. Preparation of volunteers
Shortly before the scanning session, the volunteers were
informedaboutthe planned scanning procedure and the judg-
ment task they would have to make: after the acquisition
of structural brain images the images sensitive to changes
in brain function would be taken, and their participation
would be required. Every 10s a news magazine headline
would be projected. The volunteers were asked to decide
after reading the text, whether the statement is true or false
by pressing the corresponding button on the response box.
They were informed that two runs of 120 decisions each
would be presented. The subjects were advised to avoid head
movements. Subsequently, they were positioned in the scan-
ner. Head fixation was performed by foam pads and a soft
headband. Earplugs and headset were employed together to
416 M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421
protectagainst scanner noise and to allow for communication
with the volunteer, e.g. to announce the commencement of
the decision task after finishing the preceding structural scan.
Theheadsetwasadditionally fixed to the coil to improve head
immobilization.
2.5. MR image acquisition
All data were acquired on a 3.0T whole body scanner
(Intera T30, Philips, Best, NL) equipped with Master gra-
dients (nominal gradient strength 30mT/m, maximal slew
rate150 mT/(m ms)). Forspinexcitation andresonancesignal
acquisition, a circularly polarized transmit/receive birdcage
head coil with a HF reflecting screen at the cranial end was
used. Coil diameter was 275mm, coil length 230 mm.
Following a survey, a 3D isotropic T1w dataset of
the whole head with a measured voxel size of 1.0mm
edge length was acquired for anatomical identification and
coregistration into the Talairach space [11] using a Turbo
Field-echo-technique in sagittal slice orientation with 3D-
acquisition,i.e.phase encoding in two directions (apandslice
encoding direction lr); FOV 256mm ×205 mm ×160 mm
(frequency encoding×phase encoding ×slice encoding in
fh/ap/lr direction), measured matrix 256 ×205×160, recon-
structed after zero filling to 512×410×320, i.e. recon-
structed edge length 0.5mm; contrast was defined by
TR=7.4mm, TE=3.4ms, FA=9◦, an inversion recov-
ery pre-pulse every 805ms = every 102 acquisitions, one
saturation slab caudal to the acquired volume. Acquisi-
tion bandwidth (BW) per pixel was 217.1Hz, total BW
55.578kHz, with two signal averages total acquisition time
was 11:01min.
For functional images blood oxygenation level dependent
(BOLD)contrast images were acquired using a T2*weighted
single shot gradient echo-planar imaging (EPI) sequence
which covered nearly the whole brain. The data set con-
sisted of 36 transversal slices of 3.6mm thickness without
gap, FOV 230mm ×230 mm, acquired matrix 63×64 (ap/lr
direction), reconstructed matrix 64×64, i.e. reconstructed
isotropic voxels with 3.6mm edge length, with phase encod-
ing in ap direction. Slices were oriented parallel to the ac–pc-
line.
Contrast parameters were TR=3000ms, TE=50ms,
FW=90◦, EPI-Factor (Echo train length) 63, frequency
selective fat suppression by a preceding inversion pulse
(SPIR-technique),BWper pixel in frequencyencodingdirec-
tion 2452.8 Hz, total BW 156.98kHz, BW per pixel in phase
encoding direction 23.3 Hz, the total acquisition time per set
of 25 slices was 3 s. Prior to each fMRI run 10 dummy scans
were acquired to allow for equilibration of magnetization.
2.6. Credibility ratings of the news magazines
About two weeks after the fMRI-measurements subjects
were asked about their opinions about the four different news
magazines A, B, C and D. This rating was performed short
enough after the scans to avoid shifts in preferences and long
enough to prevent a bias by the presentation itself. The 21
subjectswererequestedtorate the credibility of the four news
magazines according to a credibility score CMranging from
1, “absolute credible” to 7, “absolute incredible”. Again, the
index “M” represents one of the four different magazines,
respectively (M=A, B, C and D).
2.7. Data analysis
Data analysis was performed using Statistical Paramet-
ric Mapping (SPM2; Wellcome Department of Cognitive
Neurology, London, UK; http://www.fil.ion.ucl.ac.uk/spm)
[14–16] to correct for head movements and allow func-
tional data sets to be entered into group analyses. All
EPI volumes were spatially normalized and re-sampled to
2mm×2mm×2mm resolution to the MNI EPI standard
template of 152 averaged brains [1]. All normalized func-
tional volumes were smoothed with an isotropic Gaussian
kernel (4mm FWHM). Global changes in fMRI response
from scan to scan were removed by proportionally scaling
to have a common global mean voxel value. To correct for
long-termeffects,we applied high pass filtering with a cut-off
frequency of 0.008Hz. The hemodynamic responses with-
out temporal derivatives were modeled into an event-related
statistical design based on the General Linear Model. The
temporal events were defined by the time points when the
subjects pressed the “true” or “false” buttons to account for
systematic, effort depended durations of the decision pro-
cesses. The events were categorized according to the four
different levels of the implicit stimulation, i.e. the occurrence
of the statements in one of the four different news magazine
brands A, B, C and D. According to these four levels of the
factor magazine type M, a one-factorial ANOVA was cal-
culated for each examined subject (single subject analysis).
Additionally, we calculated t-contrasts for each subject by
employing weighting factors for the four conditions A, B,
C and D that linearly represented the individual judgment
biases BM. For this analysis, all calculations were performed
using an error probability of p<0.001, corrected for multi-
ple comparisons. For the group analysis we employed the
F-maps from the one-factorial single subject ANOVA as the
dependent variable and the subjects’ individual suggestibil-
ity as assessed by σB(standard deviation of mean bias ¯
B, Eq.
(2)) as the independent variable.
2.8. Coordinate assignment
Because the MNI (SPM2) space uses a coordinate sys-
tem, which is not exactly congruent with the one intro-
duced by Talairach and Tournoux, all coordinates cal-
culated by SPM2 were automatically transformed to the
Talairachand Tournouxspace andassignedto cortical regions
with the T2T-database Java applet (http://neurologie.uni-
muenster.de/T2T/)[11].
M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421 417
3. Results
3.1. Credibility ratings of the journals
The credibility ratings CM(scale ranging from 1, “abso-
lute credible” to 7, “absolute incredible”, see above) of
the four news magazines by the 21 examined subjects
revealed that journal A was regarded as relatively credi-
ble (mean rating CA=1.9±0.73), journal B and C showed
no significant differences and were rated close to 3 (mean
CB=2.9±1.26;C: mean CC= 3.0 ±1.00),andjournal D was
consistently marked as the most incredible magazine (mean
CD=5.5±1.44). For a graphical illustration of the ratings
given by the 21 subjects see Fig. 3.
3.2. Judgments
According to the “normalization” of the headlines’ cred-
ibility (Fig. 1), the actual judgments averaged over the four
journals were mainly “true” for the “+” group, “false” for
the “−“group and nearly equally distributed within the “0”
group (Fig. 4). According to Eq. (1) this leads to an overall
mean bias for all 21 examined subjects close to zero, repre-
senting balanced decisions in total (Fig. 5). Reaction times
did not differ significantly between expected responses for
unequivocal judgments like “true” for group “+” statements
and “false” for of group “−” statements. The higher deci-
sion uncertainty of the statements belonging to group “0”
resulted in systematically delayed responses as compared to
the groups with lower ambiguity (“+” and “−”), irrespective
of their evaluation as “true” or “false” (see Fig. 6). “Abnor-
mal” judgments, i.e. T-judgments for “−” statements and
F-judgments for “+” statements, occurred relatively seldom
Fig.3. Mean credibilityratings CMof the four differentmagazines M=A,B,
C and D by the 21 subjects who underwent fMRI during the news headline
judgment task. The logos of these journals were employed to confound
judgments about the credibility of news headlines. The score ranges from
1, “absolutely credible magazine” to 7, “absolutely incredible magazine”.
The ratings were assessed by a questionnaire two weeks after the fMRI
experiment.
Fig. 4. Actually judgments of the 21 subjects about the credibility of the 30
different presented newspaper headlines for the three credibility groups “+”,
“0” and “−”. Missing responses lead to totals of less than 100% for “true”
and “false” judgments.
and were also delayed compared to expected, unequivocal
judgments. The individuals’ biases BMcalculated by Eq. (1)
varied between −0.40 and 0.80 with a mean of 0.07. The
standard deviation σB(Eq. (2)), that represents an individ-
ual’s suggestibility in terms of integrating magazine brand
information into the judgments, varied between 0.027 and
0.24 (mean 0.09, median of 0.07). Individual differences of
the subjects’ susceptibility for the biases introduced by the
differentmagazine environmentsM= A, B, C and D are illus-
trated in Fig. 7. Subject “1” illustrates a typical suggestible
subject that showed a distinct variability (σB=0.16) in the
news media specific biases BM. Subject “2” is representative
for subjects that were practically not susceptible to the con-
founding information, i.e. the magazine logos (σB=0.07).
We found no significant influence of the credibility rat-
ings CMto BMby a linear multivariate analysis (MANOVA)
and no correlations between CMand BMby linear regres-
Fig. 5. Distribution of the mean judgment bias over all 30 judgments of
credibility categories “+”, “−”, “0” for all 21 subjects concerning the 4
different magazines (median A: 0.06; B: 0.03; C: 0.03; D: 0.00).
418 M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421
Fig. 6. Frequency distribution of the reaction times of the subjects for press-
ing the “true” (T) or “false” (F) buttons on the response box. The most
frequent responses were T for statements of the credible group “+” and F
for statements of the incredible group “−”, here denoted as “+, T” and “−,
F” (blue and grey curve), respectively. The answers did not differ in reac-
tion times between both categories. The ambivalent statements “0” were
answered nearly equally distributed as T or F (dashed lines). These answers
had been given significantly delayed relative to the unequivocal responses
(“+, T” and “−, F”). (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of the article.).
sion (RA=0.18, PA=0.44; RB=0.18, PB=0.42; RC=0.15,
PC=0.50; RD=0.03, PD=0.89; RM: magazine specific
regressioncoefficientbetweenCMand BM,PMerrorprobabil-
ity, respectively). Even if only strongly susceptible subjects
(σB> median 0.07, N=10) were included into the regression
analysis,we found no significantrelationbetween a magazine
specific bias and the rated credibility of the correspond-
Fig.7. Individualresults ofthe judgmentbias BMfortwobehaviorally differ-
ent,but typicalsample subjects. The stronger suggestiblesubject “1” showed
a higher variability in the magazine brand specific bias BM(BA=0.58,
BB=0.24, BC= 0.4, BD= 0.26, B= 0.37, σB=0.16), than the weaker sug-
gestible subject “2”, who responded for all four magazines with nearly
equal biases close to zero (BA=−0.17, BB=−0.06, BC=−0.06, BD=0.0,
B=−0.07, σB=0.07).
ing magazine (RA=0.08, PA=0.82; RB=0.16, PB=0.65;
RC=0.04, PC=0.90; RD=0.22, PD=0.55).
3.3. Cortical activity
The fMRI analysis performed on the single subject level
(one-factorial ANOVA) revealed significant BOLD signal
modulations by the factor Mfor about half of the exam-
ined subjects. In cases with significant effects, these changes
occurred consistently within the medial prefrontal cortex,
quite in correspondence to our hypothesis. For illustration
of the spatial distribution of significant cortical activity mod-
ulations on the single subject level, the individual results of
subject 1 are given in Fig. 8, upper row, right panel. The
left panel (upper row) represents the magazine-specific time
course of the BOLD signal within the medial prefrontal cor-
tex (MNI coordinates: x=6,y=55, z=2). The height of the
signal peaks of the hemodynamic response was in good lin-
ear correspondence to the respective response biases BM
(compare Figs. 7 and 8). In this subject, magazine brand
logo A produced the largest cortical “framing effect” cor-
responding to the strongest bias shift BA, and logo C the
second largest effect in good correspondence to the second
largest shift BC, etc. Matching these findings, the second sub-
ject, whose judgments were not influenced by the magazine
brands, showed no relevant significant modulations of the
cortical activity (compare Figs. 7 and 8, lower row, right
panel). The brand-specific hemodynamic response curves
showed no event-related significant modulation within the
same area. The regression group analysis provided evidence
thattheresultsof the two illustrativesubjectspresentedabove
could be generalized. Regression analysis demonstrated that
the degree of the subjects’ susceptibility to a judgment bias,
expressed by σB, and the corresponding individuals’ brain
activitymodulations induced by the varying magazine brands
highly correlated for all subjects. At a significance thresh-
old of (p< 0.001, FWE-corrected, corresponding to Z>5.70)
we obtained the following five correlating activation epicen-
ters (voxel clusters): superior temporal gyrus, BA 38 (x=46,
y=6,z=−12) (Z= 5.82); cingulate gyrus, BA 24 (−2, −2,
38)(Z=5.78); superior temporal gyrus, BA 22 (−58, 10, −2)
(Z=5.75); parietal lobe, precuneus, BA 7 (−10, −60, 60)
(Z=5.73); medial frontal gyrus, BA 10 (2, 48, 6) (Z=5.73).
According to our hypothesis the significant correlations pre-
dominantly occurred within the medial prefrontal cortex.
Fig. 9A shows the results of the linear regression for all
21 subjects of the medial frontal gyrus (x=2,y= 48, z= 6).
Fig.9B represents the spatial extent of the voxels that showed
significant correlations (p<0.0001, uncorrected, minimum
cluster size 200 voxels).
4. Discussion
By presenting completely fictitious statements, our binary
credibility judgments required a special type of decision: the
M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421 419
Fig. 8. Illustrative single subject results. Left panel: news magazine specific activation changes within the medial prefrontal cortex during credibility judgment
(MNI coordinates: x=6,y= 55, z=2) for two subjects “1” and “2”. The letters A, B, C and D denote the four different news magazine brands (confounding
implicitstimuli). The explicit stimulus material (judged statements)was the same for allfour conditions A, B, Cand D. Right panel: mapof brain regionsshowing
significantlinear correlations betweenthe magazine specific cortical activationchanges andthe corresponding responsebias BM(Fig. 7) during credibility rating.
likelihoodof correctness of the presented statement is derived
by evaluating its plausibility on the background of explicit
current knowledge. Because it is not possible to “know for
sure”aboutthetruth of a fictitious statement, the setting guar-
anteed that the subject’s decision could not be based on a
direct memory recall. This is why uncertainty arose and an
evaluation process was triggered. Because of the preceding
normalization of the ratings, we would assume an average
bias Baround 0 for representative subjects, while “gullible”
subjects would obtain positive and “distrustful” persons neg-
ative values for B(B>0 or <0, respectively). If the magazine
type had no influence on the perception and processing of
the news statements in the sense of a framing effect, there
should have been no significant differences between the four
BMvalues, i.e. BA=BB=BC=BD, because the formal state-
ments and their explicit information to be evaluated were the
same in all four magazines. Quite contrary, we found highly
significant intra-individual variations in BM(expressed by
σB), reflecting a strong influence of the confounding framing
information, i.e. the magazine brands. One potential mech-
anism for magazine brand specific judgment shifts could be
that memory contents reflecting a subject’s previously per-
ceived credibility of a particular magazine interfere with the
plausibility evaluation process. In this case, one should find a
correlation between the magazine specific credibility ratings
CMand the magazine specific bias shift BM. However, our
results of the regression analysis provide no evidence for this
hypothesis.
Another potential explanation for the magazine brand-
specific bias shifts can be deduced by looking at the specific
functionsassociated with the activity oftheventromedialpre-
frontal cortex. For subjects showing higher susceptibility to
a judgment bias (B>median 0.07, N=10), we found cere-
bral activity consistently increased during decision-making
in those regions associated with self-reflection, rewards
and the integration of emotions into decision-making, i.e.
the ventral parts of the medial prefrontal cortex (VMPFC)
(ID 11, BA 10) [2,17,31,46]. This region is of particular
importance for our social behavior, observing social conven-
tions and processing of emotions and feelings [7]. Dama-
sio and co-workers concluded from lesion studies that the
VMPFC is part of a system that stores information about past
rewards and punishments [2–4]. Based on a special frame-
work for the integration of emotions for decision-making
(“Somatic Marker Hypothesis”, [5,6]), these authors inter-
preted decision-making as a process influenced by marker
signals arising in (somatic) bioregulatory processes. From
this point of view, the magazine-specific differences in
response behavior can be ascribed to a framing effect that
is based on particular emotions individually associated to
the four magazine brands. The latter hypothesis would be
compatible with recent studies providing evidence that con-
sumerproduct brands employed as emotionalized stimuli can
specifically modulate cortical activation in the VMPFC and,
thus, buying behavior. Accordingly, and in analogy to the
“Coca-Cola” test [8], McClure et al. describe a consistent
neural response in the VMPFC correlating with the subjects’
behavioural preferences for different beverages (Coca-Cola®
and Pepsi®)[29]. For products of equal quality, distinguish-
able only by brand information, we had recently revealed a
non-linear winner-take-all effect during decision-making in
favor of a subject’s favorite brand. This behavior was charac-
terized by increased activation in the VMPFC and other brain
areas [9]. Further, a number of previous studies demonstrated
thatemotionalized selfandother subjectivejudgments specif-
ically activated the medial prefrontal cortex [30,34].
420 M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421
Fig. 9. (A) Results of the regression group analysis for all 21 subjects. Con-
cerning our hypothesis, we found a strong linear correlation between the
cortical activity changes in the VMPFG (assessed by F-values, single sub-
ject ANOVA) and the individual’s suggestibility, expressed by the standard
deviation σBof the magazine specific judgment bias BM(r=0.88). Each
point reflects data from a single participant (N=21). The activity modula-
tions were temporally associated with the subjects’ judgment response, i.e.
pressing the “true” or “false” button. (B) Spatial distribution of significant
correlations (p< 0.00001, uncorrected, minimum cluster size 200 voxels) of
the regression group analysis for all 21 subjects (x=2,y= 48, z=6).
5. Conclusion
We conclude that parts of the medial prefrontal cortex,
structures particularly involved in processing emotions, also
play a key role in the integration of implicit decision relevant
framing information during decision-making. This framing
effect can strongly confound the individual’s decisions on a
neural level and a behavioral level as well. In the context of
our specific experimental design, the individual differences
in processing the judgment task within the prefrontal cortex
canbe interpreted as a neural correlate of the individuals’ dif-
ferent framing effects induced by the magazine brand during
the perception and judgment of news paper statements. Here,
we describe a neural correlate of framing effects as had been
postulated by behavioral decision theorists.
Acknowledgements
This work was supported by grants of the Stiftung
Neuromedizin—Medical Foundation, M¨
unster, the Gesell-
schaftzur F¨
orderungder Westf¨
alischenWilhelms-Universit¨
at
zu M¨
unster e. V and the German Ministry for Education and
Research(BMBF).Thestudy wasnot supported by industrial
grants. We are greatly indebted to Prof. Dr. Dieter Ahlert and
Prof. Dr. Walter Heindel for their continuous support. We
acknowledge the invaluable help of Benjamin Gess and Hos-
sein Ghodrati in supporting the time consuming fMRI data
processing,Dipl. Phys. Jens Sommer in supporting the devel-
opmentof the stimulation design and SveaPolster-Broughton
for revising the manuscript.
References
[1] J. Ashburner, P. Neelin, D.L. Collins, A. Evans, K. Friston, Incorpo-
rating prior knowledge into image registration, Neuroimage 6 (1997)
344–352.
[2] A. Bechara, H. Damasio, A.R. Damasio, Emotion, decision making
and the orbitofrontal cortex, Cereb. Cortex 10 (2000) 295–307.
[3] A. Bechara, H. Damasio, A.R. Damasio, G.P. Lee, Different contri-
butions of the human amygdala and ventromedial prefrontal cortex
to decision-making, J. Neurosci. 19 (1999) 5473–5481.
[4] A. Bechara, H. Damasio, D. Tranel, A.R. Damasio, Deciding advan-
tageously before knowing the advantageous strategy, Science 275
(1997) 1293–1295.
[5] A.R. Damasio, Descartes’ Error: Emotion, Reason, and the Human
Brain, Putnam, New York, 1994.
[6] A.R. Damasio, The somatic marker hypothesis and the possible func-
tions of the prefrontal cortex, Philos. Trans. R. Soc. Lond. B Biol.
Sci. 351 (1996) 1413–1420.
[7] A.R. Damasio, D. Tranel, H. Damasio, Individuals with sociopathic
behavior caused by frontal damage fail to respond autonomically to
social stimuli, Behav. Brain Res. 41 (1990) 81–94.
[8] L. De Chernatony, M.H.B. Mc Donald, Creating Powerful Brands,
third ed., Butterworth-Heinemann, 2003.
[9] M. Deppe, W. Schwindt, H. Kugel, H. Plassmann, P. Kenning, Non-
linear responses within the medial prefrontal cortex reveal when
specific implicit information influences economic decision making,
J. Neuroimaging 15 (2005) 171–182.
[10] M. Deppe, J. Sommer, R. Merz, M. Sand, ShowPics, A Computer
Program for Realtime Neuropsychological Stimulation, Developed
at the Department of Neurology, University of Muenster, Germany,
2003.
[11] M. Deppe, O. Steinstr¨
ater, S. Knecht, The T2T-Database Java Applet,
Neuroimage 19 (2003) 48.
[12] R.J. Dolan, R. Lane, P. Chua, P. Fletcher, Dissociable temporal lobe
activations during emotional episodic memory retrieval, Neuroimage
11 (2000) 203–209.
[13] S. French, Decision Theory: An Introduction to the Mathematics of
Rationality, Ellis Horwood, 1988.
M. Deppe et al. / Brain Research Bulletin 67 (2005) 413–421 421
[14] K.J. Friston, Statistical parametric mapping and other analyses of
functional imaging data, in: A.W. Toga, J.C. Mazziotta (Eds.), Brain
Mapping: The Methods, Academic Press, New York, 1996, pp.
363–386.
[15] K.J. Friston, J. Ashburner, C.D. Frith, J.B. Poline, J.D. Heather, R.S.
Frackowiak, Spatial registration and normalization of images, Hum.
Brain Mapp. 2 (1995) 165–189.
[16] K.J. Friston, A.P. Holmes, K.J. Worsley, J.P. Poline, C.D. Frith, R.S.J.
Frackowiak, Statistical parametric maps in functional imaging: a gen-
eral linear approach, Hum. Brain Mapp. 2 (1995) 189–210.
[17] W.J. Gehring, A.R. Willoughby, The medial frontal cortex and the
rapid processing of monetary gains and losses, Science 295 (2002)
2279–2282.
[18] P.W. Glimcher, A. Rustichini, Neuroeconomics: the consilience of
brain and decision, Science 306 (2004) 447–452.
[19] V. Goel, R.J. Dolan, Explaining modulation of reasoning by belief,
Cognition 87 (2003) B11–B22.
[20] K. Gross, L. D’Ambrosio, Framing emotional response, Polit. Psy-
chol. 25 (2004) 1–29.
[21] C.A. Holloway, Decision Making Under Uncertainty: Models and
Choices, Prentice-Hall, 1979.
[22] D. Kahneman, P. Slovic, A. Tversky, Judgement Under Uncertainty:
Heuristics and Biases, Cambridge University Press, Cambridge,
1982.
[23] D. Kahneman, A. Tversky, Variants of uncertainty, Cognition 11
(1982) 143–157.
[24] D. Kahneman, A. Tversky, Choices, values, and frames, Am. Psy-
chol. 39 (1984) 341–350.
[25] B. Knutson, G.W. Fong, S.M. Bennett, C.M. Adams, D. Hommer, A
region of mesial prefrontal cortex tracks monetarily rewarding out-
comes: characterization with rapid event-related fMRI, Neuroimage
18 (2003) 263–272.
[26] D.C. Krawczyk, Contributions of the prefrontal cortex to the neural
basis of human decision making, Neurosci. Biobehav. Rev. 26 (2002)
631–664.
[27] G.F. Loewenstein, E.U. Weber, C.K. Hsee, N. Welch, Risk as feel-
ings, Psychol. Bull. 127 (2001) 267–287.
[28] F.W. Mast, G. Zaltman, A behavioral window on the mind of the
market: an application of the response time paradigm, Brain Res.
Bull. 67 (2005) 422–427.
[29] S.M. McClure, J. Li, D. Tomlin, K.S. Cypert, L.M. Montague, P.R.
Montague, Neural correlates of behavioral preference for culturally
familiar drinks, Neuron 44 (2004) 379–387.
[30] K.N. Ochsner, K. Knierim, D.H. Ludlow, J. Hanelin, T. Ramachan-
dran, G. Glover, S.C. Mackey, Reflecting upon feelings: an fMRI
study of neural systems supporting the attribution of emotion to self
and other, J. Cogn. Neurosci. 16 (2004) 1746–1772.
[31] J. O’Doherty, M.L. Kringelbach, E.T. Rolls, J. Hornak, C. Andrews,
Abstract reward and punishment representations in the human
orbitofrontal cortex, Nat. Neurosci. 4 (2001) 95–102.
[32] J. O’Doherty, J. Winston, H. Critchley, D. Perrett, D.M. Burt, R.J.
Dolan, Beauty in a smile: the role of medial orbitofrontal cortex in
facial attractiveness, Neuropsychologia 41 (2003) 147–155.
[33] R. Peterson, The neuroscience of investing: FMRI of the reward
system, Brain Res. Bull. 67 (2005) 391–397.
[34] K.L. Phan, T. Wager, S.F. Taylor, I. Liberzon, Functional neu-
roanatomy of emotion: a meta-analysis of emotion activation studies
in PET and fMRI, Neuroimage 16 (2002) 331–348.
[35] J.B. Pochon, R. Levy, P. Fossati, S. Lehericy, J.B. Poline, B. Pillon,
D. Le Bihan, B. Dubois, The neural system that bridges reward and
cognition in humans: an fMRI study, Proc. Natl. Acad. Sci. U.S.A.
99 (2002) 5669–5674.
[36] A.G. Sanfey, J.D. Cohen, Is knowing always feeling? Proc. Natl.
Acad. Sci. U.S.A. 101 (2004) 16709–16710.
[37] P. Slovic, M.L. Finucane, E. Peters, D.G. MacGregor, The affect
heuristic, in: T. Gilovich, D. Griffin, D. Kahneman (Eds.), Heuristics
and Biases: The Psychology of Intuitive Judgements, New York,
2002.
[38] P. Slovic, M.L. Finucane, E. Peters, D.G. MacGregor, Risk as anal-
ysis and risk as feelings: some thoughts about affect, reason, risk,
and rationality, Risk Anal. 24 (2004) 311–322.
[39] D. Tranel, A. Bechara, N.L. Denburg, Asymmetric functional roles
of right and left ventromedial prefrontal cortices in social con-
duct, decision-making, and emotional processing, Cortex 38 (2002)
589–612.
[40] A. Tversky, D. Kahneman, Judgement und uncertainty: heuristics
and biases, Science 185 (1974) 1124–1131.
[41] A. Tversky, D. Kahneman, The framing of decisions and the psy-
chology of choice, Science 211 (1981) 453–458.
[42] K.G. Volz, R.I. Schubotz, D.Y. von Cramon, Predicting events of
varying probability: uncertainty investigated by fMRI, Neuroimage
19 (2003) 271–280.
[43] K.G. Volz, R.I. Schubotz, D.Y. von Cramon, Why am I unsure?
Internal and external attributions of uncertainty dissociated by fMRI,
Neuroimage 21 (2004) 848–857.
[44] K.G. Volz, R.I. Schubotz, D.Y. von Cramon, Variants of uncertainty
in decision making and their neural correlates, Brain Res. Bull. 67
(2005) 403–412.
[45] H. Walter, B. Abler, A. Ciaramidaro, S. Erk, Motivating forces of
human actions: neuroimaging reward and social interaction, Brain
Res. Bull. 67 (2005) 368–381.
[46] D.H. Zald, J.V. Pardo, The neural correlates of aversive auditory
stimulation, Neuroimage 16 (2002) 746–753.
