1
17
18
19
14
15
16
20
21
22
9
10
11
12
13
4
5
6
2
3
7
8
Hidehiko Takahashi, Saori Fujie, Colin Camerer, Ryosuke Arakawa,
Harumasa Takano, Fumitoshi Kodaka, Hiroshi Matsui, Takashi Ideno,
Shigetaka Okubo, Kazuhisa Takemura, Makiko Yamada, Yoko Eguchi, Toshiya Murai,
Yoshiro Okubo, Motoichiro Kato, Hiroshi Ito, Tetsuya Suhara
Supplementary Methods
Materials and Methods
Participants
Nineteen healthy male volunteers [mean age 30.9 ± (s.d.) 13.5 years] were studied. Based on unstructured psychiatric screening interviews, none of them met the criteria for any psychiatric disorder. None of the subjects were under the influence of alcohol at the time, nor did they have a history of psychiatric disorder, significant physical illness, head injury, neurological disorder, or alcohol or drug dependence. All subjects were right-handed according to the Edinburgh Handedness Inventory. All subjects underwent MRI to rule out cerebral anatomic abnormalities. After complete explanation of the study, written informed consent was obtained from all subjects. The study was approved by the Ethics and
Radiation Safety Committee of the National Institute of Radiological Sciences, Chiba,
Japan.
Experimental procedure
1

18
19
20
21
22
15
16
17
12
13
14
10
11
7
8
9
1
2
3
4
5
6
The behavioral loss aversion parameter for each subject was determined outside the PET scanner, based on a staircase procedure suggested by Tversky and Kahneman.
1
Participants were presented with options between a mixed gamble (gain–loss) and a “stay” option on a computer monitor (Supplementary Figure 1). Each mixed gamble had a 50% chance of losing a fixed amount of X and a 50% chance of gaining Y. A “stay” option was described as a mixed gamble that had a 50% chance of losing 0 yen and a 50% chance of gaining 0 yen (i.e, getting 0 yen for sure). We used 4 different possible losses (-X): -2500 yen, -5000 yen, -10000 yen, and -15000 yen. In each trial, the participants chose between a mixed gamble and a “stay” option. The relative position (left and right) of the two options was randomized to counterbalance for order effects. The participants were instructed as follows:
“Two options of a mixed gamble will be presented to you. Make a choice between the two options according to your preference by pressing the right or left button. There is no correct answer and no time limit. Once you make a choice, the next pair of options will be presented.”
Each time a choice was made between a mixed gamble and a “stay” option in a trial, the amount of possible gain Y in the next trial was adjusted and ten trials of mixed gambles with possible loss (-X) were iterated to successively narrow the range including the amount of possible gain to make up for a 50% chance of losing X. The adjustments in the amount of Y were done in the following manner. The initial range of Y was set between 0.5*X and
10*X. The range was divided into thirds. The one-third and two-thirds intersecting points of the initial range were used as possible gains Y in trials 1 and 2. If the participant accepted the mixed gamble of the two-thirds and rejected that of the one-third in trials 1 and 2, the
2

18
19
20
21
22
15
16
17
12
13
14
10
11
7
8
9
1
2
3
4
5
6 middle third portion of the initial range was used as a range for trials 3 and 4. If the participant accepted both mixed gambles of the thirds, the lower third part was then used as a range. If the participant rejected both mixed gambles of the thirds, the upper third part was then used. The new range was again divided into thirds and the same procedure was iterated until the participant completed trial 10. The mean of the final range was used for the amount of gain Y final
to make up for a 50% chance of losing X. Once Y final
was estimated for a given loss (-X), the gambles with the next loss (-X) were chosen for estimation, and so on. The order of X was randomized across the participants.
Behavioral data estimation
The amount of gain Y final
to make up for a 50% chance of losing X is expressed as Y final
=
λ*X, where λ is the loss aversion parameter. This parameter λ is similar to the parameter in prospect theory but makes the common simplifying assumptions of a linear rather than curvilinear value function, and identical decision weights for a 0.5 probability of a gain or loss.
2
(Note that the decision weight need not be equal to 0.5, as long as it is assumed to be the same for the gain probability and the loss probability.) Based on previous literature, we confined the range of λ from 0.5 to 10 during estimation procedure. A smaller value of λ
(closer to 0.5) means less loss aversion (actually, loss-seeking) and a higher value (closer to
10) means more loss aversion. The parameter was estimated by a least squares method.
PET scanning
PET studies were performed on ECAT EXACT HR+ (CTI-Siemens, Knoxville, TN). The
3

18
19
20
21
22
15
16
17
12
13
14
10
11
7
8
9
1
2
3
4
5
6 system provides 63 planes and a 15.5-cm field of view. To minimize head movement, head fixation device (Fixster, Stockholm, Sweden) was used.
attenuation correction was performed using a specific radioactivities a
A transmission scan for
68
Ge-
68
Ga source. Dynamic PET scans were performed after a bolus i.v. injection of 189.1 ± 3.9 MBq of (S,S)-[ 18
F]FMeNER-D
2
with
(328.3 ± 239.9 GBq/µmol). Brain radioactivities were measured from 0 to 90 min (1 min×10, 2 min×15, and 5 min×10), followed by scanning from 120 to
180 min (10 min×6). MRI was performed on Gyroscan NT (Philips Medical Systems,
Best,
The Netherlands) (1.5 T). T1-weighted images of the brain
Quantification of NET were obtained for all subjects.
Scan parameters were 1-mm-thick, 3-dimensional T1 images with a transverse plane
(repetition time/echo time, 19/10 milliseconds; flip angle, 30°; scan matrix, 256 × 256 pixels; field of view, 256 × 256 mm; number of excitations, 1).
All emission scans were reconstructed with a Hanning filter (cutoff frequency: 0.4 cycle/pixel). All MR images were co-registered to the summated PET images (120-180 min) of (S,S)-[ 18 F]FMeNER-D
2
using a statistical parametric mapping (SPM5) system. For the quantification of NET binding, the PMOD software package (PMOD Technologies,
Zurich, Switzerland) was used. Given our hypothesis from the previous literature
5 high NET expression in the brain,
6
and the
the region-of-interest (ROI) was set on the thalamus.
Although previous studies suggest that the amygdala might also play a role in loss aversion, its small structure and relatively low expression of NET prevented reliable measurement of its NET binding for the current study. The thalamic ROIs were defined on co-registered
4

10
11
7
8
9
1
2
3
4
5
6
MR images using the fusion tool of PMOD by an investigator (S.F.) who was blinded to the behavioral data. ROIs of the thalamus consisted of three axial slices. The middle slice was set at the level of the middle portion of the third ventricle in the superior-inferior direction with reference to coronal and sagittal slices. Regional radioactivity was calculated for each frame, corrected for decay, and plotted versus time. The average values of right and left
ROIs were used to increase the signal-to-noise ratio for the calculations.
(S,S)-[
18
F]FMeNER-D
2
bindings were expressed as binding potentials relative to non-displaceable binding (BP
ND
).
3 BP
ND of (S,S)-[ 18 F]FMeNER-D
2
in the thalamus was calculated using the area under the curve (AUC) ratio method. We used the caudate as reference brain region because of its negligible NET density.
4
In the AUC ratio method,
BP
ND
can be expressed as:
12
BP
ND
＝ AUC
(thalamus)
/ AUC
(caudate)
- 1
19
20
21
22
13
14
15
16
17
18 where AUC
(thalamus)
is the area under the time-activity curve of the thalamus, and
AUC
(caudate)
is that of the caudate. An integration interval of 120–180 min was used in this method because the specific binding (target region minus caudate) has a peak during this period of PET measurement.
7
More details of PET scanning and the quantification of NET binding were described elsewhere.
6,7
Correlation coefficients between BP
ND
of
(S,S)-[
18
F]FMeNER-D
2
in the thalamus and the behavioral parameter (λ) were calculated using SPSS (SPSS, Chicago, IL). For display purposes, each summed PET image (120–180 min) of (S,S)-[ 18 F]FMeNER-D
2
was spatially normalized to the Montreal Neurological
Institute stereotaxic brain, and an image of the average of the normalized PET summed
PET images of the 19 subjects was created (Supplementary Figure 2).
5

1
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
5
6
7
8
2
3
4
9
10
11
12
13
14
Supplementary References
1.
2.
3.
4.
5.
6.
7.
Tversky A, Kahneman D. Advances in prospect theory: Cumulative representation of uncertainty.
Sokol-Hessner P, Hsu M, Curley NG, Delgado MR, Camerer CF, Phelps EA.
Thinking like a trader selectively reduces individuals' loss aversion.
Sci USA 2009; 106 : 5035-5040.
Arakawa R, Okumura M, Ito H, Seki C, Takahashi H, Takano H
J Nucl Med 2008; 49 : 1270-1276.
et al.
Quantitative analysis of norepinephrine transporter in the human brain using PET with
(S,S)-18F-FMeNER-D2.
Proc Natl Acad
Norepinephrine transporter occupancy by antidepressant in human brain using positron emission tomography with (S,S)-[18F]FMeNER-D2. Psychopharmacology
(Berl) 2010;
J Risk Uncertain
210 : 331-336.
1992; 5 : 297-323.
Tom S, Fox C, Trepel C, Poldrack R. The neural basis of loss aversion in decision-making under risk. Science 2007; 315 : 515-518.
Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN
Consensus nomenclature for in vivo imaging of reversibly binding radioligands.
Cereb Blood Flow Metab 2007; 27 : 1533-1539.
Schou M, Halldin C, Pike VW, Mozley PD, Dobson D, Innis RB et al.
(S,S)-[18F]FMeNER-D2. Eur Neuropsychopharmacol 2005; 15 : 517-520.
Sekine M, Arakawa R, Ito H, Okumura M, Sasaki T, Takahashi H
et al.
J
Post-mortem human brain autoradiography of the norepinephrine transporter using
et al.
6

4
5
6
7
8
1
2
3
Supplementary Figure 1 Options between a mixed gamble (gain–loss) and a “stay” option on a computer monitor. Participants were required to make a choice between a mixed gamble that had a 50% chance of losing 5000 yen, a 50% chance of gaining 15000 yen, and a “stay” option that was described as a mixed gamble that had a 50% chance of losing 0 yen and a 50% chance of gaining 0 yen, according to their preferences.
7

1
2
3
Supplementary Figure 2 Average of spatially normalized summed PET image (120–180 min) of (S,S)-[ 18 F]FMeNER-D
2
8