Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 1 of 35

1

Dopaminergic modulation of the trade-off between

probability and time in economic decision-making

Gonzalo Arrondo, PhDa,b,c*

, Maite Aznárez-Sanado, PhDa*

, Maria A. Fernández-

Seara, PhDa, Joaquín Goñi, PhD

a, Francis R. Loayza, PhD

a, Ewa Salamon-Klobut,

MDb, Franz H. Heukamp, PhD

b** and Maria A. Pastor, MD, PhD

a**

aFunctional Neuroimaging Laboratory, Division of Neurosciences, Center for Applied

Medical Research (CIMA), University of Navarra, 31008 Pamplona, Spain.

b Managerial Decision Sciences, IESE Business School, University of Navarra, 08034

Barcelona, Spain

c Psychiatry Department, University of Cambridge, United Kingdom.

*Dr G. Arrondo and Dr M. Aznárez-Sanado contributed equally to this article.

**Dr M. A. Pastor and Dr F. H. Heukamp contributed equally to this article.

Corresponding Author:

Dr Maria A. Pastor, MD, PhD

Functional Neuroimaging Laboratory, Division of Neurosciences, Center for Applied

Medical Research (CIMA), University of Navarra

Pio XII 55, 31008-Pamplona (Navarra) Spain.

Email: mapastor@unav.es

Tel.: +34 948 25 54 00; Fax: +34 948 29 65 00

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 2 of 35

2

ABSTRACT

Studies on animals and humans have demonstrated the importance of dopamine in

modulating decision-making processes. In this work, we have tested dopaminergic

modulation of economic decision-making and its neural correlates by administering

either placebo or metoclopramide, a dopamine D2-receptor antagonist, to healthy

subjects, during a functional MRI study. The decision-making task combined

probability and time delay with a fixed monetary reward. For individual behavioral

characterization, we used the Probability Time Trade-off (PTT) economic model,

which integrates the traditional trade-offs of reward magnitude-time and reward

magnitude-probability into a single measurement, thereby quantifying the subjective

value of a delayed and probabilistic outcome. A regression analysis between BOLD

signal and the PTT model index permitted to identify the neural substrate encoding

the subjective reward-value. Behaviorally, medication reduced the rate of temporal

discounting over probability, reflected in medicated subjects being more prone to

postpone the reward in order to increase the outcome probability. In addition,

medicated subjects showed less activity during the task in the postcentral gyrus as

well as frontomedian areas, whereas there were no differences in the ventromedial

orbitofrontal cortex (VMOFC) between groups when coding the subjective value. The

present study demonstrates by means of behavior and imaging that dopamine

modulation alters the probability-time trade-off in human economic decision-making.

Keywords: probability time trade-off; dopamine antagonist; D2 receptor; impulsivity;

temporal discounting; decision making

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 3 of 35

3

INTRODUCTION

Reports on animals and humans have established the importance of the dopaminergic

system in reward-encoding and decision-making processes. Studies on macaques have

shown that neurons capable of encoding value-related elements are located in the

striatum (Lau and Glimcher, 2008; Samejima et al., 2005). In humans, the striatum is

not only active during the evaluation of the primary reward characteristics (Knutson et

al., 2007; McClure et al., 2007) but it also encodes subjective value (De Martino et

al., 2009; Hsu et al., 2009; Kable and Glimcher, 2007). Dopaminergic treatment used

in Parkinson’s disease (PD) induces impulsivity among other adverse effects

(Ceravolo et al., 2010; Merims and Giladi, 2008). The modification of dopamine

levels in PD patients with impulse control disorders (ICD) has been associated with a

devaluation of a delayed reward (Voon et al., 2010). The acute administration of

levodopa to healthy controls had an effect on reward-related behavior, predisposing

subjects to select shorter delay options although, the administration of a dopamine

antagonist, haloperidol, did not produce significant behavioral changes (Pine et al.,

2010). Previous work has also shown that acute administration of dopamine

antagonists effectively modified reward related behaviors such as gambling

(Tremblay et al., 2010) and instrumental learning (Eisenegger et al., 2014; Pessiglione

et al., 2006). In rats decision-making under uncertainty was influenced by a striatal

D2/3 receptor antagonist, increasing risk aversion (Cocker et al., 2012; St Onge and

Floresco, 2009). A proposed mechanism for these effects has been to consider

dopamine a neurotransmitter driving motivation towards rewards (Berridge et al.,

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 4 of 35

4

2009). Use of different dopamine antagonists in decision-making paradigms may help

to further unravel the role of dopamine in reward related human behaviors.

A choice often involves both the passage of time before an outcome is reached and

uncertainty regarding the outcome. Examples are investment decisions that lead to

uncertain payoffs in the future, or consumption decisions that involve comparing

alternative paths with different delays and uncertain gratification. However, most

studies in the field of economics have focused on the study of the trade-off between

reward magnitude and probability, and between reward magnitude and delay

separately. Reflections on the primacy of time or probability discounting have been

mainly based on parallels in the trade-off between reward magnitude and probability,

and reward magnitude and delay; for a review see Berns et al. (Berns et al., 2007). In

this work, we aimed to elucidate which brain areas are involved during the decision-

making process when both delay and risk are being concurrently evaluated. Aside

from the intrinsic increase in value when a reward is received with higher probability,

which can thus be traded against an increase in delay, a trade-off between delay and

risk is central in many instances of decision making in modern society. Fields where a

trade-off between time and probability of reward is needed when making decisions

include project management where more time intensive procedures may increase the

probability of success of a project, in environmental economics where more time

consuming applications of policy decisions can modify the chances of obtaining the

desired outcome and in health economics where the length of treatments can influence

the probability of a preferred outcome. This kind of interaction between time and risk

in decision making is also in line with classical foraging theory models in which the

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 5 of 35

5

probability of a reward is weighted against the effort or time required for an animal to

obtain it (Real and Caraco, 1986).

To shed light on the role of dopamine in the trade-off between probability and delay,

we measured the impact of a dopamine D2 receptor (D2R) antagonist,

metoclopramide, on the behavioral and functional outcome of a probability-time

trade-off paradigm, using functional magnetic resonance imaging (fMRI).

Metoclopramide, a derivative of benzamide, acts by antagonizing the dopamine D2

receptor in the peripheral and central nervous systems, as it easily crosses the blood

brain barrier (Liu et al., 2009). Its short-term-use good safety profile (Friedman et al.,

2011) and short lasting effect led us to consider it as an interesting D2R antagonist. In

addition, we have previously demonstrated that a single dose of metoclopramide is

capable of altering cerebral perfusion in humans (Fernandez-Seara et al., 2010).

We hypothesized that the administration of metoclopramide would increase the

willingness of subjects to wait for a less risky reward (i.e. a reward with higher

probability), when compared to the administration of placebo. It was also expected

that medication would modify activity in striatal regions, which present the highest

density of D2Rs, or in frontal regions involved in decision-making. The VMOFC has

been proposed to be a key region in the valuation of rewards and hence was

considered a priori a key area to search for activations related to the assignation of

subjective value and the difference in this assessment between groups. (Bartra et al.,

2013; Levy and Glimcher, 2012) .

MATERIALS AND METHODS

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 6 of 35

6

Subjects

Two groups of subjects were evaluated, in a single-blind, placebo controlled study.

The medicated group (MG) consisted of fifteen subjects, six females (age=24.0±3.4

years: mean±standard deviation (SD)), who received an oral dose (10mg) of

metoclopramide (commercial name Primperan) one hour prior to entering the

scanner. The pharmacokinetic properties of the drug are included in Supplementary

Material (SM). The second group followed the same experimental procedure, but

taking a placebo (an identical capsule with 10 mg of starch). The placebo group (PG)

consisted of fourteen subjects, ten females (age=23.0 1.9 years). Participants were

recruited among the student body of the Medical School and were randomly allocated

to MG and PG. None of them presented psychiatric disorders according to the Mini

International Neuropsychiatric Interview (Cummings et al., 1994). There were no

significant differences between groups in any of the seven personality variables of the

Temperament and Character Inventory-Revised (TCI-R) (Cloninger et al., 1993) (see

Table S1). When the proportion of men and women was compared across groups with

Fisher’s exact test the result was not significant (exact 2-sided p =0.139). The

protocol was approved by the Ethical Committee of University of Navarra and all

subjects provided written informed consent.

Experimental Setup

Participants were tested inside the scanner. Responses were collected via a response

box (See SM for further details).

Task

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 7 of 35

7

The paradigm consisted of an alternation between a decision-making (DM) and two

control tasks.

Subjects were asked to choose between two prospects (options), A and B, in the DM

task. Options A and B were displayed at the same time (Fig.1). Each option consisted

of an amount x, to be received after a time delay t with probability p. The amount of

money was the same (30€) in both options along the experiment. One of the prospects

(fixed prospect) didn’t change throughout the experiment (30€ available in one month

with a reward probability of 20%) while the alternative prospect showed a longer time

delay (t) and a higher probability (p) (Fig.1). In the alternative option, time varied

from 2-7 months in increments of 1 month and probability from 30%-80% in

increments of 10%. Hence, overall there were 36 different alternative options. The

words “configuration” and “proposal” will be used hereafter to define a specific

combination of the fixed and an alternative option, being Fig. 1 one example. Subjects

were instructed that there were no right or wrong answers, and that one of their DM

choices would be selected randomly to be paid with the actual delay and probability

of success. An analogous task has been previously used in a group of healthy controls

to identify neural correlates of uncertainty in decision making (Goñi et al., 2011).

The summation control consisted of selecting the option which presented the highest

sum of numbers. The visual control task, which showed only “X” letters, consisted of

choosing alternatively one of the two options (Fig.1).

The time interval to answer was 7 seconds in every case. Stimuli were shown for 6

seconds and then replaced by a white cross that was presented for 1 second. This

timing was not influenced by the subject’s response time.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 8 of 35

8

The different tasks were grouped in blocks of three. DM blocks were interleaved with

one of the two control blocks in an alternative fashion.

There were three scanning sessions of 15 minutes duration. Each of the 36 DM

proposals was presented at least once in every session. Overall, each proposal was

presented five times in a pseudo-random order during the experiment. In order to

avoid response biases due to the location of the options, the physical position of the

fixed option appeared an equal number of times on each side of the presentation

screen. Further information can be found in SM, including a schema of the

presentation (Fig. S1).

Behavioral measures

Consistency, i.e. the tendency to provide the same answer to multiple presentations of

the same option, was assessed individually for every proposal. Hence 36 consistency

values were obtained per subject, one for each proposal. Consistency values were

calculated as (100*r2/(r1+r2)), where r2 was the highest number of equal answers

from the total of responses (r1+r2). Response time (RT) was calculated as the time

difference between the beginning of the stimuli presentation and the subject’s

response. The mean RT by proposal was obtained and then the mean RT per subject

was calculated as the average of the 36 proposals RTs. The percentage of correct

answers (accuracy) was obtained from the summation control as an indicator of

maintained attention. Mean group values of consistency, accuracy and RT were

calculated as the average of the mean individual results. The number of omissions

(events with no response) was also counted. A Kolmogorov-Smirnov test was used to

evaluate normality. Comparison between groups was performed using a parametric

two-sample t-test for RT and non-parametric tests (Mann-Whitney U tests) for the

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 9 of 35

9

other three variables (consistency, accuracy and omissions), which were non-normally

distributed.

Predictors of choice preference and RT

A differential valuation of time and risk induced by metoclopramide would lead to a

change in the proportion of choices between the fixed (shorter delay but riskier) and

alternative (longer delay but safer) options. To assess the effect of medication, the

relative frequency of choosing the fixed prospect on each configuration was obtained

for each subject.

A generalized (logistic) mixed effects approximation enabled to assess behavioral

differences between groups. The dependent variable was whether the participant

chose the fixed option on each trial. Independent variables were probability (20-80%)

and time delay (2-7 months) of the alternative option, and the subject’s group

(placebo/medicated). Gender was incorporated as a nuisance variable and subject was

the random factor.

A similar linear mixed model was created to investigate whether RTs differed

between groups. The dependent variable was the RT of the choice for each

presentation of a prospect. Independent variables were the same as previously

described. A more in-depth description on the mixed models is included in SM.

Probability Time Trade-off model (PTT)

In order to quantify the attractiveness of each option, we used a derivation of the

model denominated Probability Time Trade-off (PTT) (Baucells and Heukamp, 2012;

Baucells et al., 2009) which permits to integrate the magnitudes of probability p ,

time t and reward amount x . This model links the trade-offs of reward magnitude-

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 10 of 35

10

time and reward magnitude-probability. It consists of a function on the option

presented, ),,( tpxV that provides a score of attractiveness for an economic option

characterized by an amount of money x happening with probability p after a number

of months t . The PTT model is conformed as the product of two functions, ),,( tpxw

and )(xv , which quantify the probability time-tradeoff given an amount of money and

a power-law modulation of the amount respectively. It can be expressed as:

PTT: xexvtpxwtpxV ptxr ))ln()(()(),,(),,( ,

where )(xr is a probability discount function that regulates the trade-off between

probability and time. It is a direct measure of the discounting effect of time delay on

probability, given an amount of money x . Smaller outcomes of r(x) indicate that a

subject attaches less value to time and thus is more likely to wait for a probability

increase. Exponent is a curvature parameter that expresses the subjects’ sensitivity

to the choice parameters, independently of the probability-time trade-off. Finally,

parameter is the scaling factor that modulates the power-law relationship between

x and the overall option attractiveness.

In the decision-making experiment carried out in this paper, the amount of money x

was intentionally kept constant to 30€. By doing so, the power-law term x becomes

a constant that is not necessary to include, since modulation of the reward amount

remains out of the spectrum of the experiment. Hence, in the context of our

experiment, the PTT model can be simplified to:

PTT simplified: ))ln((),(),( prtV

eetpwtpV O ,

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 11 of 35

11

where, as stated above, the term x is not present, and the function )(xr is now a

coefficient parameter r , not dependent on any reward magnitude x . We denote by

OV the exponent of the function which defines the attractiveness of the chosen option.

In summary, all the influence of varying x is removed in this simplified version of

the model.

During the decision making experiment, the outcome of participants behavior consists

of a sequence choices },...,,{ 21 tOOO where each choice iO is either the fixed

option ( FOi ) or the alternative option ( AOi ) and has certain exponent of

attractiveness ( OV will be either FV or AV )

The outputs of the simplified PTT model are estimations of the attractiveness of one

option. Every time the two options are presented (fixed and alternative), the simplified

PTT will provide two values related to the attractiveness of each option ( FV and AV

respectively), based on prospect dependent parameters ( t and p ) and on subject

dependent parameters ( r and ). The estimated probabilities for the fixed and

alternative outcomes to happen ( FP̂ and AP̂ ) can be estimated by applying a softmax

decision rule, namely:

)/()/(

)/(

)/()/(

)/(

ˆ1ˆ

,ˆ

AF

A

AF

F

VV

V

FA

VV

V

F

ee

ePP

ee

eP

where β permits to quantify the degree of additional stochasticity of the subject’s

behavior with respect to the PTT evaluation. In order to estimate the three parameters

of the model (r, β and ) for each subject’s sequence of choices },...,,{ 21 tOOO , a

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 12 of 35

12

maximum likelihood criterion was used. The likelihood function to be minimized

was:

Likelihood function: t

i

OiPL

1

ˆlnln

Lln gets closer to zero as the probabilities inferred from the PTT model explain better

the subject’s behavior.

After the estimation procedure, normality was assessed with a Kolmogorov-Smirnov

test. Since none of r, , β and Lln were normally distributed in at least one of the

groups, non-parametric tests (Mann-Whitney U test) were employed for the purpose

of group comparisons. (See SM for further details on the estimation of the PTT and

also for a comparison between the mixed effects and PTT approximations for the

modeling of the behavior).

Scanning procedure

fMRI scanning was performed with a 3.0 Tesla MR scanner (Siemens TRIO,

Germany). 320 volumes were acquired in every session using an EPI sequence

(resolution=3×3×3mm3, TR/TE=3000/30ms). An anatomical image was also acquired

(1x1x1mm3) (see SI for more information). More detailed MRI parameters can be

found in SM.

fMRI data analysis

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 13 of 35

13

Data were analyzed using the Statistical Parametric Mapping program, version SPM8

(see SM for detailed pre-processing steps).

Individual task-related activation for each condition (DM, summation and visual

controls) was evaluated as the convolution of the hemodynamic response function

(Canonical HRF) with a boxcar function of duration equivalent to the subject’s RT for

each decision event. The onset time corresponded to the beginning of each proposal

presentation. Hence, the analysis was an event-related design which included 3 events

of duration equal to each RT within every pseudo-block.

A parametric regressor coding the subjective attractiveness of the selected prospect

was introduced in the DM condition. This regressor was obtained using the simplified

PTT model (Eq.1), employing the parameters r and estimated for each subject (see

Methods). The PTT regressor value was calculated individually for every subject and

proposal, taking into account the corresponding probability and time-delay of the

chosen prospect. Regions where neural activity showed a positive correlation with

subjective value were obtained for each subject.

To make inferences at the population level, individual contrast images were

incorporated into random effects models. Group analyses were carried out for the

placebo group as a portrait of the effects in the normal population and additionally we

searched for differences between the placebo and medicated groups to study the effect

of metoclopramide administration. Contrasts of interest were the difference between

the task and the summation control events, and the neural correlate of subjective value

using the PTT model. Additionally, we include in SM the comparison between the

task and the visual control condition, between the task and the mean of both control

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 14 of 35

14

conditions, and between the task and the summation control including sex as a

covariate.

Statistical significance was set at FWE corrected p-value of 0.05 at the voxel level,

although uncorrected maps (p<0.001, k>25) will also be shown for further exploration

of group effects. In the case of the PPT regressor, a predefined area of interest was

created, consisting in an 8 mm radius sphere with its center located at x=0, y=36, z=-

10.5 (Levy and Glimcher, 2012).Two approaches were used to analyze the pattern of

results within this region. Firstly, a small volume correction (SVC) of FWE at the

peak level of 0.05 was carried out as implemented in SPM. Secondly, we extracted

the mean Beta values of all voxels within this a-priori defined region of interest

(describing the strength of the correlation between the BOLD signal and the PTT

regressor), using Marsbar (Brett et al., 2002). This mean value per subject was

compared between groups with a two-sample t-test (See SM for further details on the

ROI creation and voxel selection). Coordinates of local maxima and anatomical

nomenclature were assessed using the SPM Anatomy Toolbox V18 (Eickhoff et al.,

2007).

RESULTS

Behavioral measures

Individual behavioral measures in the placebo and the medicated groups (PG and MG

respectively) were used to evaluate possible differences at the group level in terms of

consistency, response time and level of attention.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 15 of 35

15

The PG and MG group median consistency of the responses was 94% (range 83-

100%, SD=5). Statistical differences (p=0.029) were found between PG

(median(range):92(83-100)) and MG (median(range):98(83-99)). MG showed higher

consistency in reproducing the same answer in each of the five repetitions per

proposal. There were no statistical differences between the mean RT of the two

groups (p=0.454; RT_placebo= 2.7±0.9s; RT_medicated 2.5±1.0s) or in the number

of omissions in the DM condition (p=0.630; median (range) omissions_placebo: 0 (0-

2); median (range) omissions_medicated: 0 (0-3)).

All subjects performed the summation control task with high accuracy. Indeed, no

statistical differences (p=0.140) were found between PG (median (range):98.4 (93.7-

100)) and MG (median (range):98.9 (95.8-100)).

Summarizing MG and PG showed no statistical differences in RT, the number of

omissions or the summation control’s accuracy, indicating a similar level of attention

during all the scanning procedure.

Predictors of choice preference and RT

The choice-dependent outcomes of the simplified PTT were used as a neural correlate

in order to characterize grey matter clusters whose BOLD activity codifies the level of

attractiveness of the option selected at a time.

Plots of the mean frequency of responses are depicted in Fig. 2 (A1 for PG and A2 for

MG). As shown in these plots, the frequency of choosing the alternative option

increases with higher reward probabilities and decreases with increasing reward

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 16 of 35

16

delays. However, subjects in the MG showed less modulation in their choices,

preferring the safest and most delayed option more frequently than subjects in the PG.

This is also supported by the behavioral comparisons performed with the generalized

linear mixed model (Table 1A). Higher probabilities (p<0.001) and shorter delays

(p=0.04) led to higher preference for the alternative option. However, being

medicated reduced significantly the rate of choosing the short-term option (p=0.05).

Influence of delay in choice varied between PG and MG (p=0.03), while this was not

observed with probability (p=0.4). These results suggest that medication led to a

reduction of the subject's sensitivity to delay, while it did not have such an impact on

sensitivity to probability, which remained high.

On the other hand, medication did not influence RTs (p=0.2) (Table1B). While time

delay (p<0.001) and probability (p<0.001) had a significant effect on RTs, their effect

did not differ between PG and MG (p=0.53 for delay; p=0.89 for probability).

Probability Time Trade-off Model (PTT)

Table 2 shows the values of r and which best describe each individual’s behavior.

As explained in Methods, r, the probability discount rate, represents the relative

weight of time versus probability in the subjective value of the prospect. For r equal

0, time does not change the subjective value, however for large values of r, the

subjective value decreases steeply with time. Thus r is a measure of the subject’s

willingness to wait for a higher probability reward. On the other hand, expresses the

subject’s general sensitivity to the choice parameters, but it does not affect the relative

weights of probability and time dimensions in the decision. Therefore, for the purpose

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 17 of 35

17

of this study, we will focus on the values obtained for the parameter r. Regarding the

goodness of fit Lln , the closer it is to zero, the better the simplified PTT model is

able to predict the subject’s empirical behavior.

MG showed a higher preference for choosing the alternative option when compared to

PG and thus a significantly lower value of r (p=0.034; r_PG median (range):0.178

(0.041-0.524); r_MG median (range):0.067 (0.000-2.000)). Medicated subjects were

more willing to wait in order to increase the probability of the reward, which was

reflected in a reduced weight of increasing delay on the subjective valuation of a

given prospect, measured by the parameter r.

The goodness of the fit was better in MG than in PG (p=0.007). This difference can

be understood as the result of the more predictable and thus easier to model behavior

shown by medicated subjects, who were more consistent in their choices.

fMRI data

When we examined changes in activation during the decision making task, the

placebo group showed an activation cluster including Pre-SMA and anterior cingulate

cortex corrected for multiple comparisons (Fig. 2 B1 in main article, table S2).

Additionally, the exploratory analysis with uncorrected p<0.001, revealed two large

clusters of activation in the dorsolateral prefrontal cortex (precentral, inferior and

middle frontal gyri), and in the insula (Fig. 2 B2 and table S3). In the comparison

between placebo and medicated groups the postcentral gyrus showed statistically

higher activity in the placebo group compared to the medicated group (Fig. 2 C1 and

table S4). Additionally, when the uncorrected map was explored, the SMA-ACC

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 18 of 35

18

cluster appeared more active in the placebo group (Fig 2 C2 and table S5). There were

no areas showing higher activation in the medicated group. The comparison of the

task activation to either the visual control condition or both control conditions did not

change the differential overall pattern of results between groups (See SM). When sex

was included as a covariate relevant clusters were located in the same areas than in

the analysis without the covariate (most notably the precentral gyrus and SMA),

however peak values were slightly reduced. In the case of the highest peak the t

values decreased from t=5.74 to t=5.52. This was enough to hinder multiple testing

correction (t threshold = 5.67).

The inclusion of the PTT parametric regressor in the SPM analysis allowed us to

evaluate the areas coding the subjective value of the chosen option. No areas

correlated with subjective value after correcting for multiple tests at the whole brain

level in PG. An exploratory evaluation of the uncorrected results showed activations

in areas including the VMOFC, striatum, parietal cortex (angular gyrus) and

cerebellum (Fig. S2 and table S7). More importantly, the regional analysis showed a

significant activation within the VMOFC after small volume correction (FWE at the

peak level of 0.05, corresponding to a t value of 3.41 and an uncorrected p of 0.002)

in PG (Fig. 2 D, table S6). The voxel-wise comparison between placebo and

medicated groups showed no significant differences at the whole brain level or within

the VMOFC area when using SVC. However the comparison of the beta values of the

correlation averaged within the sphere (and compared with a 2-sample t-test) showed

a trend towards significance (t=1.7, p=0.101), with higher correlation values in the

placebo than the medicated groups.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 19 of 35

19

DISCUSSION

The administration of a single dose of metoclopramide, a D2R antagonist, to healthy

volunteers led to behavioral and brain activity changes during a decision-making task.

Behaviorally, medicated subjects presented less modulation in their choices. They

preferred the most delayed option in the majority of the configurations, showing a

reduced influence of increasing delay on their final choice when compared to PG. Our

study is the first to describe a differential effect of dopamine in probability and delay

discounting.

Regarding brain activity, the present work strengthens the notion that the medial

prefrontal cortex (MPFC) and orbitofrontal cortex (OFC) are critical components of a

cortico-subcortical dopaminergic system that encodes subjective value of delayed and

probabilistic rewards (Levy and Glimcher, 2012). The activation clusters identified

for PG in the comparison between task and control matched areas reported previously

to be involved in reward value coding. Specifically the activation cluster that survived

correction for multiple testing was located in medial frontal regions including Pre-

SMA and ACC, areas which are involved in the process of reward valuation

independently of the type of reward (Sescousse et al., 2010). When a regressor coding

the subjective value of the chosen option was used, we found an activation in the

ventromedial orbitofrontal cortex. This is in line with our initial hypothesis based on

the existing abundant data supporting that this region plays a key role in the

assignation of a common subjective value to rewards of any type (Bartra et al., 2013;

Levy and Glimcher, 2012). Moreover, when the uncorrected map was explored two

clusters fell within the striatum, (specifically within the caudate nucleus and

pallidum), another key region in reward valuation (Bernacer et al., 2013). Overall, the

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 20 of 35

20

regions found activated in the contrast between task and control as well as correlated

to the subjective value regressor are part of corticostriatal circuits (Postuma and

Dagher, 2006). Both MPFC and OFC are innervated by dopaminergic neurons of the

ventral tegmental area, presenting D2R (Woodward et al., 2009) and are connected to

the limbic system (Craig, 2009; Kim et al., 2011); whereas dopaminergic impaired

modulation of these regions has also been associated with compulsive behaviors and

impulsivity ((Volkow et al., 1996), see the model proposed by Volkow et al., 2011

(Volkow et al., 2011)).

Behavioral effects of metoclopramide administration were revealed using two

different approximations. In the mixed effects model there was an interaction between

the factor “time” and the factor “group”, indicating that if a log-linear influence of

time and probability on behavior was assumed, time had a reduced influence in the

choices of those subjects taking metoclopramide. As an alternative to the previous

analysis we also modeled behavior using the PTT model, which characterizes time as

a non-linear modulator on risk perception. According to the parameter estimates

obtained with the PTT model medicated subjects differed from the placebo group in

their r values, indicating that time had a reduced effect on how they perceived the

probability of the reward and hence they were more prone to choose a delayed

reward.

Behavioral differences due to metoclopramide are in line with previous studies, where

dopamine agonists led to opposite behavioral changes, that is, an increase in

impulsiveness and in the rate of temporal discounting. When PD patients that suffered

from impulse control deficits participated in an intertemporal choice task, they were

less prone to wait for the reward after the intake of dopamine agonists (Voon et al.,

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 21 of 35

21

2010). Impulsive PD patients on dopamine agonists also preferred risky over safer

options more frequently than healthy controls or other PD patients (Voon et al., 2011).

Similar effects have been described in healthy controls who received levodopa (Pine

et al., 2010). Participants showed a higher temporal discount rate, choosing more

often options with short delays under the levodopa condition. In the same study, the

administration of haloperidol, a dopamine antagonist, resulted in no behavioral

changes, probably due to a reduction in alertness. The behavioral changes described in

our work cannot be attributed to medication-related changes in alertness or attention

since no differences in RT, omissions or accuracy were found between groups. The

differential effect of metoclopramide compared to haloperidol could be related to the

higher metoclopramide substantia nigra tissue binding (Chen et al., 2011), in spite of

its lower relative potency as D2R antagonist.

PG and MG showed differences in brain activity when carrying out the decision

making task involving reward, probability and time. Medication led to a decrease in

activation in the postcentral gyrus, while further exploration of the uncorrected results

also showed lower activity in frontomedian areas within the SMA-ACC complex.

In the PG we found significant correlation between the BOLD signal and the

subjective value regressor (PTT regressor) in the VMOFC. The behavioral differences

between the groups suggest a differential valuation of the reward, induced by

metoclopramide. However we did not find differences in correlation in this area

between groups. Hence, our behavioral results indicate an abnormal evaluation of

time due to metoclopramide which is reflected by a reduced effect of time on decision

making and the imaging results show reduced activation during the task condition,

without evidence for activation differences between groups when an individual

repressor coding subjective value is used. Our behavioral results could be explained

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 22 of 35

22

as a consequence of reducing dopamine levels in the striatum through the

administration of metoclopramide, which in turn led to a change in how participants

perceived the passing of time and the motivational salience of rewards. The “wanting”

of a reward is a dissociable aspect of pleasure and reward processing with its neural

substrate in the striatum (Berridge et al., 2009). Higher dopaminergic levels in the

striatum increase the incentive salience of a reward and the urging that individuals

have for it (Kirsch et al., 2007). The reduced salience would make subjects less

willing to make an effort to obtain it (Wardle et al., 2011), and, as shown in our

results, ease the wait as a consequence of the attenuation of the urge to receive it, in

line with proposals differentiating between effort and delay (Prevost et al., 2010).

Also, Parkinson’s disease patients, who suffer from dopamine depletion, have been

shown to underestimate the passing of time (Pastor et al., 1992) and our results could

be similarly explained, as medicated participants seemed to underestimate time and

therefore were more willing to wait for the reward.

It must be noted however that the evidence for associating behavioral changes to

changes in striatal dopamine signaling is only indirect as we did not found differences

in subcortical activity between groups. Nevertheless, the same dose of

metoclopramide changed cerebral blood flow in the striatum during a resting-state

study, indicating that this compound has indeed effects in this region (Fernandez-

Seara et al., 2010). This altered baseline state could be affecting the amplitude of

fMRI signals observed during the task and masking differences in task evoked neural

activity, between groups (Cohen et al., 2002).

While similar results would be expected if metoclopramide led to an increase in risk

aversion, as participants would wait longer to avoid risky choices (St Onge and

Floresco, 2009), the results of the Generalized Linear Mixed Model are more in line

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 23 of 35

23

with the former explanation. Moreover, in a recent work levodopa administration to

healthy subjects did not change their perception of risk (Symmonds et al., 2013).

In our study there was a lack of significant differences within the VMOFC between

groups in the correlation between BOLD signal and the subjective value regressor.

However, the comparison of the betas of the correlation averaged within the ROI

between groups was close to significance, indicating that the lack of evidence could

be due to low statistical power. Other possible limitations of this study include the

fact that gender distribution across groups was unequal or that a personality trait that

could be influencing results, as impulsivity, was not measured directly. These and

other aspects are further discussed in SM. Regarding gender distribution, we included

sex as a covariate in the behavioural analysis, and hence results cannot be attributed to

the different proportion of men and women in both groups. We also included sex as a

covariate in an additional neuroimaging analysis comparing task vs. summation

control. After including sex as a covariate the main difference between groups did not

survive the threshold for multiple comparisons correction. However,; the pattern of

activation in the uncorrected map remained mostly unchanged. Several previous

studies have used single challenges of dopaminergic medications in order to study

their effect in similar paradigms. In their study, Pine et al. described a more marked

decrease in activity under the levodopa condition when rewards became more delayed

in putamen, insula, subgenual cingulate and lateral OFC (Pine et al., 2009) whereas

other authors did not found an effect of levodopa in risk aversion (Symmonds et al.,

2013). On the other hand, tolcapone administration diminished functional

connectivity between putamen and pregenual rACC (Kayser et al., 2012). Although

similar brain areas have been related to coding value of delayed and probabilistic

rewards, the direction of BOLD changes when such valuation is altered by

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 24 of 35

24

dopaminergic drugs is still not clear. The expression of impulsivity in humans has

been related to striatal dopamine release mediated in part by inhibitory midbrain

autoreceptors (Buckholtz et al., 2010). The diversity of responses could be explained

by differences in the drugs affinity for the autoreceptors, and could also be related to

the higher autoreceptors sensitivity to increases in dopamine in healthy subjects with

already optimized levels of dopamine (Cools and D'Esposito, 2011). Further studies

about specific effects of different dopamine modulators on brain activity related to

subjective value coding are needed. The use of drugs that target specific dopaminergic

receptors, as done in this work, may help to elucidate human decision-making

processes.

The novelty of this study lies in characterizing by means of behavior and imaging

how the trade-off between time and probability is modulated by a single dose of the

dopamine D2 receptor antagonist, metoclopramide. Altogether, our results support the

involvement of limbic associated prefrontal areas in coding the subjective value of

options that involve delayed and risky rewards. D2R blockade induced subjects to

choose more delayed but less risky options in line with the conceptual role of

dopamine as a motivational agent

CONFLICT OF INTEREST

The authors do not have any conflict of interest to declare.

ACKNOWLEDGEMENTS

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 25 of 35

25

This work was funded by a grant from the Foundation for Applied Medical Research

(FIMA, University of Navarra) and EUROCORES (the European Science

Foundation). M. A-S was a Government of Navarra Predoctoral Fellow (2007-2010).

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 26 of 35

26

REFERENCES

Bartra, O., McGuire, J.T., Kable, J.W., 2013. The valuation system: a coordinate-based meta-analysis of BOLD fMRI experiments examining neural correlates of subjective value. Neuroimage 76, 412-427.

Baucells, M., Heukamp, F., 2012. Probability and Time Trade-off. Management Science 58, 831-842.

Baucells, M., Heukamp, F., Villasis, A., 2009. Trading-off probability and time: experimental evidence. IESE Business School Working Paper.

Bernacer, J., Corlett, P.R., Ramachandra, P., McFarlane, B., Turner, D.C., Clark, L., Robbins, T.W., Fletcher, P.C., Murray, G.K., 2013. Methamphetamine-induced disruption of frontostriatal reward learning signals: relation to psychotic symptoms. Am J Psychiatry 170, 1326-1334.

Berns, G.S., Laibson, D., Loewenstein, G., 2007. Intertemporal choice--toward an integrative framework. Trends Cogn Sci 11, 482-488.

Berridge, K.C., Robinson, T.E., Aldridge, J.W., 2009. Dissecting components of reward: 'liking', 'wanting', and learning. Current opinion in pharmacology 9, 65-73.

Brett, M., Anton, J.L., Valabregue, R., Poline, J.B., 2002. Region of interest analysis using an SPM toolbox, 8th International Conference on Functional Mapping of the Human Brain. NeuroImage, Sendai, Japan.

Buckholtz, J.W., Treadway, M.T., Cowan, R.L., Woodward, N.D., Li, R., Ansari, M.S., Baldwin, R.M., Schwartzman, A.N., Shelby, E.S., Smith, C.E., Kessler, R.M., Zald, D.H., 2010. Dopaminergic network differences in human impulsivity. Science 329, 532.

Ceravolo, R., Frosini, D., Rossi, C., Bonuccelli, U., 2010. Spectrum of addictions in Parkinson's disease: from dopamine dysregulation syndrome to impulse control disorders. J Neurol 257, S276-283.

Chen, S., Seeman, P., Liu, F., 2011. Antipsychotic drug binding in the substantia nigra: an examination of high metoclopramide binding in the brains of normal, Alzheimer's disease, Huntington's disease, and Multiple Sclerosis patients, and its relation to tardive dyskinesia. Synapse 65, 119-124.

Cloninger, C.R., Svrakic, D.M., Przybeck, T.R., 1993. A psychobiological model of temperament and character. Arch Gen Psychiatry 50, 975-990.

Cocker, P.J., Dinelle, K., Kornelson, R., Sossi, V., Winstanley, C.A., 2012. Irrational choice under uncertainty correlates with lower striatal D(2/3) receptor binding in rats. J Neurosci 32, 15450-15457.

Cohen, E.R., Ugurbil, K., Kim, S.G., 2002. Effect of basal conditions on the magnitude and dynamics of the blood oxygenation level-dependent fMRI response. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 22, 1042-1053.

Cools, R., D'Esposito, M., 2011. Inverted-U-shaped dopamine actions on human working memory and cognitive control. Biol Psychiatry 69, e113-125.

Craig, A.D., 2009. How do you feel--now? The anterior insula and human awareness. Nat Rev Neurosci 10, 59-70.

Cummings, J.L., Mega, M., Gray, K., Rosenberg-Thompson, S., Carusi, D.A., Gornbein, J., 1994. The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurology 44, 2308-2314.

De Martino, B., Kumaran, D., Holt, B., Dolan, R.J., 2009. The neurobiology of reference-dependent value computation. J Neurosci 29, 3833-3842.

Eickhoff, S.B., Paus, T., Caspers, S., Grosbras, M.H., Evans, A.C., Zilles, K., Amunts, K., 2007. Assignment of functional activations to probabilistic cytoarchitectonic areas revisited. Neuroimage 36, 511-521.

Eisenegger, C., Naef, M., Linssen, A., Clark, L., Gandamaneni, P.K., Muller, U., Robbins, T.W., 2014. Role of Dopamine D2 Receptors in Human Reinforcement Learning. Neuropsychopharmacology.

Fernandez-Seara, M.A., Aznarez-Sanado, M., Mengual, E., Irigoyen, J., Heukamp, F., Pastor, M.A., 2010. Effects on resting cerebral blood flow and functional connectivity induced by metoclopramide: a perfusion MRI study in healthy volunteers. Br J Pharmacol.

Friedman, B.W., Mulvey, L., Esses, D., Solorzano, C., Paternoster, J., Lipton, R.B., Gallagher, E.J., 2011. Metoclopramide for acute migraine: a dose-finding randomized clinical trial. Ann Emerg Med 57, 475-482 e471.

Goñi, J., Aznarez-Sanado, M., Arrondo, G., Fernandez-Seara, M., Loayza, F.R., Heukamp, F.H., Pastor, M.A., 2011. The neural substrate and functional integration of uncertainty in decision making: an information theory approach. PLoS One 6, e17408.

Hsu, M., Krajbich, I., Zhao, C., Camerer, C.F., 2009. Neural response to reward anticipation under risk is nonlinear in probabilities. J Neurosci 29, 2231-2237.

Kable, J.W., Glimcher, P.W., 2007. The neural correlates of subjective value during intertemporal choice. Nat Neurosci 10, 1625-1633.

Kayser, A.S., Allen, D.C., Navarro-Cebrian, A., Mitchell, J.M., Fields, H.L., 2012. Dopamine, corticostriatal connectivity, and intertemporal choice. J Neurosci 32, 9402-9409.

Kim, M.J., Loucks, R.A., Palmer, A.L., Brown, A.C., Solomon, K.M., Marchante, A.N., Whalen, P.J., 2011. The structural and functional connectivity of the amygdala: from normal emotion to pathological anxiety. Behav Brain Res 223, 403-410.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 27 of 35

27

Kirsch, P., Ronshausen, S., Mier, D., Gallhofer, B., 2007. The influence of antipsychotic treatment on brain reward system reactivity in schizophrenia patients. Pharmacopsychiatry 40, 196-198.

Knutson, B., Rick, S., Wimmer, G.E., Prelec, D., Loewenstein, G., 2007. Neural predictors of purchases. Neuron 53, 147-156.

Lau, B., Glimcher, P.W., 2008. Value representations in the primate striatum during matching behavior. Neuron 58, 451-463.

Levy, D.J., Glimcher, P.W., 2012. The root of all value: a neural common currency for choice. Curr Opin Neurobiol 22, 1027-1038.

Liu, X., Van Natta, K., Yeo, H., Vilenski, O., Weller, P.E., Worboys, P.D., Monshouwer, M., 2009. Unbound drug concentration in brain homogenate and cerebral spinal fluid at steady state as a surrogate for unbound concentration in brain interstitial fluid. Drug Metab Dispos 37, 787-793.

McClure, S.M., Ericson, K.M., Laibson, D.I., Loewenstein, G., Cohen, J.D., 2007. Time discounting for primary rewards. J Neurosci 27, 5796-5804.

Merims, D., Giladi, N., 2008. Dopamine dysregulation syndrome, addiction and behavioral changes in Parkinson's disease. Parkinsonism Relat Disord 14, 273-280.

Pastor, M.A., Artieda, J., Jahanshahi, M., Obeso, J.A., 1992. Time-Estimation and Reproduction Is Abnormal in Parkinsons-Disease. Brain 115, 211-225.

Pessiglione, M., Seymour, B., Flandin, G., Dolan, R.J., Frith, C.D., 2006. Dopamine-dependent prediction errors underpin reward-seeking behaviour in humans. Nature 442, 1042-1045.

Pine, A., Seymour, B., Roiser, J.P., Bossaerts, P., Friston, K.J., Curran, H.V., Dolan, R.J., 2009. Encoding of marginal utility across time in the human brain. J Neurosci 29, 9575-9581.

Pine, A., Shiner, T., Seymour, B., Dolan, R.J., 2010. Dopamine, time, and impulsivity in humans. J Neurosci 30, 8888-8896.

Postuma, R.B., Dagher, A., 2006. Basal ganglia functional connectivity based on a meta-analysis of 126 positron emission tomography and functional magnetic resonance imaging publications. Cereb Cortex 16, 1508-1521.

Prevost, C., Pessiglione, M., Metereau, E., Clery-Melin, M.L., Dreher, J.C., 2010. Separate valuation subsystems for delay and effort decision costs. J Neurosci 30, 14080-14090.

Real, L., Caraco, T., 1986. Risk and Foraging in Stochastic Environments. Annu Rev Ecol Syst 17, 371-390.

Samejima, K., Ueda, Y., Doya, K., Kimura, M., 2005. Representation of action-specific reward values in the striatum. Science 310, 1337-1340.

Sescousse, G., Redoute, J., Dreher, J.C., 2010. The architecture of reward value coding in the human orbitofrontal cortex. J Neurosci 30, 13095-13104.

St Onge, J.R., Floresco, S.B., 2009. Dopaminergic modulation of risk-based decision making. Neuropsychopharmacology 34, 681-697.

Symmonds, M., Wright, N.D., Fagan, E., Dolan, R.J., 2013. Assaying the effect of levodopa on the evaluation of risk in healthy humans. PLoS One 8, e68177.

Tremblay, A.M., Desmond, R.C., Poulos, C.X., Zack, M., 2010. Haloperidol modifies instrumental aspects of slot machine gambling in pathological gamblers and healthy controls. Addict Biol 16, 467-484.

Volkow, N.D., Ding, Y.S., Fowler, J.S., Wang, G.J., 1996. Cocaine addiction: hypothesis derived from imaging studies with PET. J Addict Dis 15, 55-71.

Volkow, N.D., Wang, G.J., Fowler, J.S., Tomasi, D., Telang, F., 2011. Addiction: beyond dopamine reward circuitry. Proc Natl Acad Sci U S A 108, 15037-15042.

Voon, V., Gao, J., Brezing, C., Symmonds, M., Ekanayake, V., Fernandez, H., Dolan, R.J., Hallett, M., 2011. Dopamine agonists and risk: impulse control disorders in Parkinson's disease. Brain 134, 1438-1446.

Voon, V., Reynolds, B., Brezing, C., Gallea, C., Skaljic, M., Ekanayake, V., Fernandez, H., Potenza, M.N., Dolan, R.J., Hallett, M., 2010. Impulsive choice and response in dopamine agonist-related impulse control behaviors. Psychopharmacology (Berl) 207, 645-659.

Wardle, M.C., Treadway, M.T., Mayo, L.M., Zald, D.H., de Wit, H., 2011. Amping up effort: effects of d-amphetamine on human effort-based decision-making. J Neurosci 31, 16597-16602.

Woodward, N.D., Zald, D.H., Ding, Z., Riccardi, P., Ansari, M.S., Baldwin, R.M., Cowan, R.L., Li, R., Kessler, R.M., 2009. Cerebral morphology and dopamine D2/D3 receptor distribution in humans: a combined [18F]fallypride and voxel-based morphometry study. Neuroimage 46, 31-38.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 28 of 35

28

TABLES

A. PREDICTION OF CHOICES

Estimate Std. Error z value p-value

Intercept 4.3398 1.0786 4.0230 0.0001

Time 0.2312 0.1147 2.0150 0.0439

Probability -17.4218 1.2722 -13.6940 0.0000

Medication -2.4778 1.2799 -1.9360 0.0529

Male 0.6708 1.2245 0.5480 0.5838

Time*Probability 0.9329 0.2200 4.2410 0.0000

Time*Medication -0.1584 0.0717 -2.2070 0.0273

Probability*Medication 0.8298 0.9454 0.8780 0.3801

B. PREDICTION OF RESPONSE TIMES

Estimate Std. Error z value p-value

Intercept 4292.2881 296.4149 14.481 0.0001

Time -124.2131 26.2135 -4.739 0.0001

Probability -25.4941 2.2307 -11.429 0.0001

Medication -332.4922 366.7485 -0.907 0.2334

Repetition -4.6835 0.231 -20.272 0.0001

Male 391.2917 356.3268 1.098 0.1336

Time*Probability 2.7894 0.4352 6.409 0.0001

Time*Medication -9.2834 14.8725 -0.624 0.534

Probability*Medication 0.2363 1.4873 0.159 0.8856

Table 1. A. Prediction of choices result of a generalized mixed model (Dependent

variable: probability of choosing the fixed option. Independent variables: probability

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 29 of 35

29

and time delay of the alternative option, medication, gender and a random intercept).

Model includes all first-level interactions. Goodness of fit: Akaike Information

Criterion (AIC)=1293. B. Prediction of response times result of a generalized mixed

model (Dependent variable: probability of choosing the fixed option. Independent

variables: probability and time delay of the alternative option, medication, gender,

number of prospect's repetition and a random intercept). Model includes all first-level

interactions. Goodness of fit: AIC=85857.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 30 of 35

30

Group Subject option A r lnL

Placebo p1 151/180 0.113 2.815 1.0 45

Placebo p2 109/179 0.204 1.998 1.0 85

Placebo p3 133/180 0.146 2.539 1.0 60

Placebo p4 53/179 0.524 1.168 1.0 83

Placebo p5 96/180 0.215 1.979 1.0 87

Placebo p6 117/176 0.140 2.500 1.0 63

Placebo p7 83/180 0.271 2.329 1.0 62

Placebo p8 158/179 0.100 2.567 1.0 53

Placebo p9 92/180 0.216 3.734 4.8 89

Placebo p10 180/180 0.041 8.884 2.6 0

Placebo p11 85/180 0.262 3.419 2.2 53

Placebo p12 133/179 0.151 2.671 1.0 53

Placebo p13 65/178 0.333 2.029 1.0 60

Placebo p14 144/180 0.132 2.794 1.0 47

Median 113 0.178 2.55 1 60

Medicated m1 16/180 2.000 0.740 1.0 51

Medicated m2 179/180 0.000 8.336 9.9 6

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 31 of 35

31

Medicated m3 180/180 0.041 8.884 2.6 0

Medicated m4 177/179 0.000 8.032 10.0 11

Medicated m5 179/180 0.049 4.278 1.0 4

Medicated m6 138/180 0.134 2.509 1.0 60

Medicated m7 178/180 0.056 4.137 1.0 7

Medicated m8 165/177 0.072 2.996 1.0 32

Medicated m9 177/178 0.067 8.794 2.8 2

Medicated m10 177/180 0.000 8.336 9.9 6

Medicated m11 135/179 0.156 2.993 1.0 41

Medicated m12 112/180 0.184 2.147 1.0 81

Medicated m13 179/180 0.000 8.335 10.0 6

Medicated m14 109/177 0.208 2.734 1.0 52

Medicated m15 85/180 0.275 1.833 1.0 83

Median 177 0.067 4.137 1 11

Table 2. Individual values which best describe each subject’s behavior. Option A

denotes the fraction between the number of alternative option choices (longer in time

and higher in probability) and the total number of choices provided by the subject.

Parameters r and are the optimal values found that minimize the likelihood function

lnL. According to the PTT model, the parameter r represents the weighting of time

versus probability in the subjective valuation of the decision and quantifies the

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 32 of 35

32

subjects’ sensitivity to the choice variables (see section “Probability Time Trade-off

Model (PTT)” in Results). The parameter represents the degree of stochasticity of

each subject's behavior.

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 33 of 35

33

FIGURE LEGENDS

Figure 1. fMRI paradigm. Visual stimuli: Display showing the DM task (top panel),

summation control (center panel) and visual control (bottom panel) conditions.

Options A and B appeared on the screen at the same time for each task, on the left and

right side respectively. They were presented for six seconds and were replaced by a

white cross of one second duration before the next proposal appeared. Subjects could

make the choice at any time during the seven seconds.

Figure 2: A. Temperature plots of the preference for the alternative option (longer

time delay; higher probability). A1. Placebo Group. A2. Medicated Group. X axis is

the probability of the alternative option while Y axis is the time delay. The color scale

represents the group mean relative frequency of choosing the alternative option. A

bilinear interpolation was carried out for improved visualization. B. Areas showing a

differential activation between task and control in the placebo group. B1. p<0.05,

FWE-corrected at the voxel level. B2. p<0.001 uncorrected; k>25. C. Areas showing

a higher activation during the task in the placebo compared to the medicated group.

C1. FWE-corrected. C2. p<0.001 uncorrected; k>25. D. Areas coding the subjective

value of the chosen option, result of a positive correlation of BOLD signal with the

PTT regressor in the placebo group (Small volume corrected, p-value<0.05, at the

VMOFC region).

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 34 of 35

34

FIGURES

Figure 1

Dopaminergic modulation of the trade-off between probability and time in economic decision-making. Arrondo G, Aznárez-

Sanado, et al. European Neuropsychopharmacology (Accepted on March 2015)

Page 35 of 35

35

Figure 2