Parkinsonism and Related Disorders 21 (2015) 1330e1335

Contents lists available at ScienceDirect

Parkinsonism and Related Disorders

journal homepage: www.elsevier.com/locate/parkreldis

Intertemporal choice in Parkinson's disease and restless legs

syndrome
Mohamed Al-Khaled a, 1, Marcus Heldmann a, b, 1, Inga Bolstorff a, Johann Hagenah c,
Thomas F. Mnte a, b, *
a
Department of Neurology, University of Lbeck, Lbeck, Germany
b
Institute of Psychology II, University of Lbeck, Lbeck, Germany
c
Department of Neurology, Westkstenklinikum Heide, Heide, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Background: Impulse control disorders in Parkinson's disease are a potential consequence of dopami-
Received 30 December 2014 nergic therapy. Impulse control problems might be revealed by intertemporal choice tasks which entail
Received in revised form to forgo an immediately available reward in favor of a larger but later reward. The steepness of the
7 September 2015
discounting curve can be quantied by the parameter k.
Accepted 9 September 2015
Methods: Participants (37 Parkinson patients [13 de novo, 24 medicated], 24 patients with restless legs
syndrome, and 22 controls) were offered 54 choices between immediate smaller rewards and delayed
Keywords:
larger and the k value was estimated from the participants' responses. Participants had the chance of
Pathological gambling
Delay discounting
winning one of their decisions. None of the participants had impulse control disorders.
Parkinson disease Results: Unmedicated Parkinson patients had a higher discounting rate than controls and medicated
Restless legs syndrome patients with restless legs syndrome. The k values of medicated Parkinson patients and patients with
Impulse control disorder restless legs syndrome did not differ from those of controls. No correlation was found between the k
Dopamine agonist value and the dopamine agonist dose.
L-Dopa Conclusion: Impulsive decision making in patients with Parkinson's disease may occur as part of the
disease rather than as a consequence of dopamine agonist therapy.
2015 Elsevier Ltd. All rights reserved.

1. Introduction the prefrontal cortex) which allows us to forego an immediate

Several studies have demonstrated an association between im-
pulse control disorders (ICD) in patients with Parkinson's disease
(PD) and therapy with dopamine agonists (DAs) and, albeit less
pronounced, L-DOPA [1e4]. Research has also revealed an associa-
tion between the development of impulse control disorders in
patients with restless legs syndrome (RLS) and treatment with DAs
[5]. The DOMINION study [6] involving 3090 patients with PD re-
ported that 13% of patients who were treated with DAs developed
impulse control disorders including pathological gambling. Patho-
logical gambling can be characterized as a disturbance of the bal-
ance between an immediate reward process (associated with parts
of the limbic system) favoring the selection of immediately avail-
able rewards and a delayed gratication process (associated with
* Corresponding author. Department of Neurology University of Lbeck, Ratze-
burger Allee 160, 23538 Lbeck, Germany.
E-mail address: Thomas.muente@neuro.uni-luebeck.de (T.F. Mnte).
1
Both authors contributed equally.
reward in order to wait for later, greater reinforcements [7]. The
interplay of these opposing mechanisms can be investigated by
delay discounting or intertemporal choice paradigms, in which
participants choose between a smaller immediate reward and a
larger delayed reward.
Importantly, it has been demonstrated that future rewards are
discounted roughly following a hyperbolic function. A simple
equation which captures real-life discounting quite well is the
following [8]:
A
V
1 kD
where V is the present discounted value of a delayed reward, A is
the amount of the delayed reward, k is the delay discount rate, and
D (days) is the duration of the delay. The delay discount rate k
shows great interindividual variation and indicates the steepness of
the discounting curve. For example, assuming that k 0.016, the
present value of an V80 reward that is available after a delay of 30

http://dx.doi.org/10.1016/j.parkreldis.2015.09.026
1353-8020/ 2015 Elsevier Ltd. All rights reserved.
 M. Al-Khaled et al. / Parkinsonism and Related Disorders 21 (2015) 1330e1335 1331

days is V54. A higher k value indicates a steeper discount rate and a
preference for a devaluation of future rewards. Research has found
that pathological gamblers and patients with substance addiction
(e.g, tobacco, alcohol, heroin, and cocaine) have higher k values
than HCs [9e13]. Note, that there is no best choice in intertemporal
choice paradigms. Rather, these paradigms are about individual
preferences of the participants.
Several studies have looked at delay discounting in PD [14e16].
While these previous studies [14e16] clearly suggest abnormalities
in intertemporal choice behavior in PD, the pattern of results raises
some questions. First, the fact that two studies [14,15] did not nd
signicantly elevated discount rates in PD without ICD, whereas
Milenkova et al. [16] did, calls for a replication of Milenkova et al.
[16] in a new sample of PD without ICD. Moreover, Voon et al. [14]
described an effect of dopaminergic medication on discount rate
(albeit only in PD with ICD), whereas Milenkova et al. [16] did not
nd such an effect. To tackle these questions, we performed the
current study. To assess, whether an elevated discount rate might
be considered a trait feature of PD, we investigated delay dis-
counting in de novo, i.e. previously unmedicated PD patients. To
test the question whether DA medication in and of itself leads to
elevated discounting rates, we also assessed patients with restless
legs syndrome on dopamine agonist medication. In addition to
these groups, we also investigated PD patients on dopaminergic
medication and a control group.
2. Participants and methods
All participants had normal or corrected-to-normal vision, no
history of pathological gambling, hypersexuality, compulsive
buying, or binge eating as assessed by clinical interview and no
depression (score < 18 on Beck Depression Inventory II [17]).
Screening for cognitive decits in PD was carried out with the
Parkinson Neuropsychometric Dementia Assessment (PANDA) [18].
PD patients were recruited from the local outpatient clinic and
diagnosed in accordance with the UK brain bank criteria [19]. Of the
37 PD patients, 13 were unmedicated de novo patients. In addition,
24 patients with RLS diagnosed according to Allen et al. [20] and
medicated with L-DOPA and/or DAs, participated in the study.
Finally, 22 unmedicated HCs with a similar age range were included
(Table 1).
2.1. Neuropsychological testing
Neuropsychological tests (Table 2) covered the following
cognitive domains: Attention was tested with a simple speeded
reaction time test that required a button press to a visual stimulus
(Alertness test with and without warning tone) as well as a go/nogo
test. Executive function was tested with a version of the color word
interference test (FarbeWort-Interferenz-Test, FWI), a verbal
uency test (Regensburger Wortssigkeitstest, RWT) and two
tests assessing deductive reasoning (LPS 3 and 4). Memory was
assessed with the California Verbal Learning test (CVLT). Finally, the
PANDA test, an instrument for the assessment of cognitive abilities
in PD [18], was administered.
2.2. Intertemporal choice
We used a variant of the task described by Kirby et al. [9] with 54
instead of 27 choices to allow a better differentiation of k-values.
The additional choices covered intermediate k-values. The order of
the trials was xed (Table 3) and did not correlate with the size of
the rewards or k-values. A computerized presentation was used
with the display comprising the smaller immediate reward (on the
left of a xation point), the larger delayed reward (to the right) and
the delay for the delayed reward (below the xation point). There
was no time limit for making the decision. The next choice was
presented 2 s after a reaction. The participants were informed to
make each decision as if it were real. At the end of the participation,
they were allowed to throw a die; if they scored a 6, they were
allowed to draw a number between 1 and 54 (each number rep-
resented 1 decision). Participants then received a reward based on
their choice for that particular decision. If they had chosen the
smaller immediate reward, they were given cash. If they had cho-
sen the delayed reward, the respective sum was paid by bank
transfer after the specied delay.
2.3. Statistical analysis
Four different k values were calculated per subject using the
method described by Kirby et al. [9] a global k value based on all
decisions and one k value each for small, medium, and large
rewards.
To investigate the potential impact of group and reward size on
impulsive behavior, an ANOVA with the between-subjects factor
group (4 levels: de novo PD, medicated PD, RLS, HC) and the within-
subjects factor reward size (3 levels: k values for small, medium,
and large reward sizes) was calculated. Group differences between
categorical data were tested with the chi-square test. Following the
hypothesis that DA treatment is linked to increased k values,

Table 1
Demographic and clinical characteristics.

Characteristics PD de novo PD medicated RLS HC

N 13 N 24 N 24 N 22

Age (yrs), mean (SD) 69.9 (11) 67.2 (11.8) 68.4 (6.5) 69.3 (8.1)
Sex w/m 4/9 6/18 16/8 13/9
Education (yrs), mean (SD) 12.6 (4.2) 13.7 (3) 12 (2.4) 14 (3.8)
Disease duration (yrs), mean (SD) 2.2 (1.2) 6.4 (4.4) 13.8 (12.4) e
Symptom onset to diagnosis (yrs), mean (SD) 0.2 (0.5) 5.5 (4.2) 7.2 (3.6) e
Familiar history for movement disorder, n (%) 0 2 (7.7) 7 (27) 1 (4)
Reported Smoking, n (%) 1 (7.7) 3 (12.5) 1 (8.3) 3 (9.1)
Alcohol consumption (occasionally), n (%) 2 (15.3) 6 (25) 7 (19) 10 (36.6)
Reported Sleep disturbance, n (%) 3 (23) 16 (67) 21 (88) 6 (27)
DA-LEDD (mg), mean (SD) e 158.5 (118) 66 (69) e
Total-LEDD (mg), mean (SD) e 440 (247) 123 (99) e
UPDRS III, mean (SD) 23.5 (10.5) 21.1 (6.7) e e
Hoehn and Yahr, mean (SD) 1.5 (0.5) 2.0 (0.6) e e
BDI II, mean (SD) 6.9 (8.7) 8.4 (7.2) 8.5 (7.5) 6.6 (6.1)

DA-LEDD, dopamine agonist-L-DOPA equivalent daily dose; LEDD, L-DOPA equivalent daily dose; UPDRS, Unied Parkinson's Disease Rating Scale; BDI II, Beck Depression
Inventory II.
1332 M. Al-Khaled et al. / Parkinsonism and Related Disorders 21 (2015) 1330e1335

Table 2
Neuropsychological tests.

Tests PDm (n 15) RLS (n 22) HC (n 18) PDm vs HC df 31, t RLS vs HC df 38, t

PANDA cognitive, mean (SD 22.6 (6) 21.9 (5.5) 25.4 (4) 1.54 2.32*
FWI (Colorname reading) 37 (8.9) 36 (6.3) 34 (5.4) 0.86 1.08
FWI (Color-specify) 54 (13.3 47 (7.8) 51 (10.9) 1.30 0.75
FWI 96 (23.5) 83 (14.1) 91 (27.6) 0.56 1.16
TAP alertness w/o warning tone 315 (75.5) 291 (56.5) 307 (55.3) 0.34 0.90
TAP alertnes with warning tone 304 (70.9) 288 (61.5) 306 (50.3) 0.09 1.08
TAP go/no-go 602 (79) 636 (89.9) 615 (67.5) 0.50 0.84
CVLT list A 1 5 (1.8) 6 (2.1) 6 (2.5) 1.33 0
CVLT list A 1-5 46 (11.4) 49 (13.1) 51 (14.9) 1.09 0.44
CVLT list B 5 (1.8) 5 (2.3) 6 (2.2) 1.43 1.40
CVLT short-delay free recall 8 (3.9) 10 (3.7) 9 (4.3) 0.70 0.77
CVLT short-delay cued recall 9 (3.1) 11 (3.3) 11 (3.7) 1.69 0
CVLT long-delay free recall 9 (3.5) 10 (3.7) 10 (4.4) 0.72 0
CVLT long-delay cued recall 10 (3.2) 11 (3.4) 11 (3.8) 0.82 0
CVLT, yes/no recognition 15 (1.6) 14 (2.6) 14 (1.7) 1.73 0
RWT M 16 (5.3) 16 (4.7) 17 (5.4) 0.53 0.61
RWT G-R 18 (5.9) 18 (4.4) 18 (5.4 0 0
RWT food 29 (6.1) 31 (6) 30 (9.1) 0.37 0.40
RWT sports-fruits 21 (5.4) 20 (5.3) 20 (4.4) 0.57 0
LPS 3 20 (6.1) 19 (5.3) 20 (4.6) 0 0.63
LPS 4 23 (7) 24 (3.8) 25 (4.6) 0.94 0.73

All values are recorded in mean and standard deviation (SD). PANDA (Parkinson neuropsychometric dementia assessment), FWI (FarbeWort Interferenztest, Color-
eWordeInterference Test), TAP (Testbatterie zur Aufmerksamkeits-Prfung, Attention Test Battery), CVLT (California Verbal Learning Test), RWT (Regensburger Wort-
ssigkeitstest, Word Fluency Test), LPS (Leistungsprfsystem, System for Cognitive Abilities).
Data for de novo PD patients are not given as the neuropsychological examination was carried out only in 4 members of that group.
*Signicant p < 0.05.

associations between k values and L-DOPA equivalent daily dose
(LEDD) values were tested with Pearson's product-moment coef-
cient. Spearman's rank correlation was used to investigate the
association between global k values and Unied Parkinson's Dis-
ease Rating Scale (UPDRS) scores.
3. Results
3.1. Demographic data
No differences between groups were found for the variables age
(F 0.2, P > 0.8) and BDI (F 0.73, P > 0.53, see Table 1). There were
also no differences between groups for reported smoking habits
(c2 0.32, P > 0.95), drug use (c2 2.51, P > 0.47), and alcohol
consumption (c2 2.03, P > 0.57). However, a group difference was
found for reported sleep disturbances (c2 23.19, P < 0.001), which
is obviously caused by the high number of patients in the RLS (20 of
24 patients) and the medicated PD (14 of 24 patients) groups
reporting such problems.
3.2. Neuropsychological results
Results of the neuropsychological tests are shown in Table 2
including the t-statistics (unpaired t-test). With the exception of a
signicantly lower PANDA score in the RLS patients compared to
the HC group, none of the comparisons reached statistical
signicance.
3.3. Intertemporal choice task
Global k differed between groups (F 3.24, df 3,79; P 0.026).
Post hoc tests revealed that the de novo PD group had increased
global k values compared with the HC group (P 0.031) and the RLS
group (P 0.033), but not with the medicated PD group (P 0.24;
see also Fig. 1a). The remaining group comparisons did not reach
signicance (all P > 0.6). The two-way factorial ANOVA, taking the
impact of reward size on k values into account, resulted in a sig-
nicant main effect of group (F 3.65; P 0.016; df 3,79), thus
conrming the results of the previous global k ANOVA. The main
effect of reward magnitude (F 2.99; P 0.053; df 2.158) just
failed to become signicant, whereas the interaction group x
reward magnitude (F 0.503; P 0.8; df 6.158) was clearly not
signicant. As shown in Fig. 1b, for large and small rewards, the
pattern of results was similar to the global k analysis, revealing
signicantly larger k values for the de novo PD group compared to
the HC and RLS groups (large reward: de novo PD vs. HC, P 0.014;
de novo PD vs. RLS, P 0.018; small reward: de novo PD vs. HC,
P 0.026; de novo PD vs. RLS, P 0.043), but no signicant dif-
ferences to the medicated PD group. For medium rewards, no dif-
ferences between groups were signicant (all P > 0.05).
3.4. Correlation analysis
The correlation analysis showed no signicant relation between
DA-LEDD and global k values in the pooled medicated PD group and
RLS group (r 0.1), nor in the medicated PD group (r 0.05)
and the RLS group (r 0.16) alone (all P > 0.4). The correlation
analysis for an association between global k and UPDRS revealed a
negative relationship for the medicated PD group (rho 0.63,
P 0.0015) but a positive, although non-signicant, relationship
for the de novo PD group (rho 0.2, P > 0.05).
We also explored the correlation between global k and disease
duration and did not nd any relationship in the medicated PD
(r 0.09) and RLS (r 0.12) groups. Finally, in an exploratory
fashion we also correlated results from the neuropsychological
tests and k in the pooled medicated and RLS group as well as in
both groups separately. We did not obtain any signicant
correlation.
4. Discussion
To disentangle the effects of disease status (PD, RLS, control) and
dopaminergic treatment on the development of impulsive behavior
as reected in intertemporal choice, we studied medicated and
unmedicated (de novo) PD patients, RLS patients and a healthy
control group.
 M. Al-Khaled et al. / Parkinsonism and Related Disorders 21 (2015) 1330e1335 1333

Table 3
Choice paradigms.

Order SIR LDR Delay (days) K-value LDR size

1 39 50 64 0.004407 M
2 40 85 16 0.070313 L
3 21 30 39 0.010989 S
4 52 75 40 0.011058 L
5 22 25 136 0.001003 S
6 49 60 89 0.002522 M
7 9 25 10 0.177778 S
8 25 60 14 0.1 M
9 73 80 137 0.0007 L
10 47 50 160 0.000399 M
11 28 30 179 0.000399 S
12 33 80 14 0.101732 L
13 24 25 184 0.000226 S
14 20 55 10 0.175 M
15 80 85 157 0.000398 L
16 34 50 30 0.015686 M
17 67 75 119 0.001003 L
18 34 35 186 0.000158 S
19 31 85 7 0.248848 L
20 19 25 53 0.005958 S
21 41 60 42 0.011034 M
22 14 25 19 0.041353 S
23 72 75 161 0.000259 L
24 54 60 111 0.001001 M
25 41 75 20 0.041463 L
26 25 30 80 0.002500 S
27 54 55 117 0.000158 M
28 78 80 162 0.000158 L
29 13 30 3 0.435897 S
30 20 55 7 0.250000 M
31 11 30 7 0.246753 S
32 62 80 66 0.004399 L
33 46 55 109 0.001795 M
34 50 85 25 0.028000 L
35 15 35 13 0.102564 S
36 32 60 2 0.437500 M
37 21 35 24 0.027778 S
38 22 50 18 0.070707 M
Fig. 1. A Global k-mean values in healthy controls (HC), de novo PD patients (PDdn),
39 55 75 61 0.005961 L
medicated PD patients (PDm), and patients suffering from restless-legs syndrome
40 40 55 62 0.006048 M
(RLS). B Discounting rates (k-value) for large (amount between 75 V and 85 V), me-
41 29 80 10 0.175862 L
dium (amount between 50 V and 60 V), and small (amount between 25 V and 35 V)
42 27 35 67 0.004422 S
delayed rewards in healthy controls (HC), de novo PD patients (PDdn), medicated PD
43 27 50 21 0.040564 M
patients (PDm), and patients suffering from restless-legs syndrome (RLS).
44 21 25 106 0.001797 S
45 40 75 2 0.437500 L
46 24 35 29 0.015805 S
47 53 55 151 0.000250 M signicantly steeper discounting curves than both, PD without ICD
48 69 85 91 0.002548 L and controls, whereas the discount rate was only slightly, and non-
49 55 60 130 0.000699 M signicantly elevated in PD without ICD relative to controls.
50 54 80 30 0.016049 L
Using the same discounting task as Housden et al. [15] our group
51 17 35 15 0.070588 S
52 71 85 110 0.001793 L [16] obtained data from PD patients without overt ICD on and off
53 27 30 158 0.000703 S dopaminergic medication as well as from control participants.
54 29 50 26 0.027851 M Here, PD patients showed signicantly and markedly elevated
SIR indicates small immediate reward; LDR, large delayed reward; K-value, delay discount rates relative to controls regardless of whether they were
discounting rate at which the two decisions are of equal subjective value; LDR size, on their best medication or off (12 h since last intake of medi-
the size of the delayed reward was classied as small (S, between 25 and 35 V), cation). The missing medication effect in Milenkova et al. [16] could
medium (M, between 50 and 60 V), and large (L, between 75 and 85 V).
have been due to several factors: Firstly, the delay between the last
intake of medication and the test could have been too short given
the long half live of dopamine agonists. Secondly, the results could
Previous studies of intertemporal choice in PD had given mixed point towards an elevated discount rate as a genuine feature
results. Voon et al. [14] assessed two groups of PD patients (with (endophenotype) of PD. Thirdly, the missing medication effect
and without ICD) as well as normal control subjects using a might also be due to the fact that the population studied by Mile-
feedback-based intertemporal choice task. PD patients were tested nkova et al. [16] did not have overt impulse control disorders.
twice, i.e. once with their usual dopaminergic medication and once On the basis of these previous results, we predicted that treat-
off medication. While a medication effect was found in PD patients ment with DAs and/or L-DOPA should lead to a higher k value, i.e. a
with ICD, no effect of dopaminergic medication was found in PD steeper discounting of future rewards. Alternatively, we considered
patients without impulse problems. The intertemporal choice of the possibility that steeper discounting of future reward might, at
Voon et al. [14] used delays of only 7e28 s which mimics common least in part, be a trait feature of PD. Unlike our earlier study [16]
real life situations. Housden et al. [15] used the delay discounting but in line with data by Housden et al. [15] medicated PD pa-
task introduced by Kirby [13]. PD patients with ICD had tients without ICD showed a moderate, non-signicant elevation of
1334 M. Al-Khaled et al. / Parkinsonism and Related Disorders 21 (2015) 1330e1335
k compared to the healthy controls in the present study. Plugging
the resulting k-values for medicated PD patients and control par-
ticipants into the hyperbolic function given in the introduction and
assuming a reward of 100 Euros and a delay of 50 days, medicated
PD patients value this future reward as being equivalent to 64 Euros
at present, whereas it is worth 74 Euros for normal controls. Note,
that the k-values obtained for medicated PD patients and healthy
controls are in the same range as those obtained by both previous
studies [15,16].
Interestingly, the RLS patients did not show any tendency for an
elevated delay discounting rate in spite of a recent study by Voon
et al. [5] that suggests ICD in RLS patients with dopaminergic
treatment in about 7%.
The most important nding of the present study concerns the
signicantly elevated discount rate in de novo PD patients who had
never received dopaminergic medication prior to our testing. They
showed a k-value of about 0.02 which translates into a devaluation
of 100 Euros to 50 Euros in a period of 50 days. This nding suggests
that an elevated discounting rate might be a trait marker of PD
rather than the mere result of DA treatment. The elevated dis-
counting rates in the Milenkova-study [16] for PD patients on and
off their medication as well as the (non-signicant) increase of
discounting rates of medicated PD patients in the current study as
well as in Housden et al. [15] would be compatible with such a view.
The ndings of a greatly elevated k-value for PD patients with
ICD, as demonstrated by Voon et al. [14] for very short delays and by
Housden et al. [15] for delays equivalent to the present study
further suggests that impulsivity as measured by delay discounting
tasks might be particularly altered in this subgroup of patients. So
far, a sensitivity of k-value to medication status has only been
demonstrated in this subgroup.
A functional magnetic resonance imaging study [21] showed
that participants being administered the D2/D3 agonist prami-
pexole and those receiving a placebo had more activity in the nu-
cleus accumbens to cues signaling potential rewards than to cues
signaling no reward. Whereas this imaging study suggests a
changed neural response to reward cues due to dopamine agonists,
the present study did not suggest a major effect of DAs and/or L-
DOPA on intertemporal choice. In addition, the k value did not
correlate with the DA-LEDD in the PD group and in the RLS group.
An Italian study found that the frequency of impulsivity in drug-
nave patients with PD was similar to that of HCs [22]. Similarly, the
present study did not nd differences in reported smoking and
alcohol consumption, which may be considered markers for
impulsive behavior, between the de novo PD patients and the other
groups.
Pathological gambling and other ICDs occur in a minority of
patients receiving DA therapy which suggests that besides DAs
other factors, for example specic genetic predispositions [23],
might be important for the development of full-blown ICDs. Pre-
vious research has shown that dopaminergic depletion affects not
only the nigrostriatal motor system but also the mesocorticolimbic
circuit [24], which modulates control behavior and impulsivity.
Interestingly, Joutsa et al. [25] recently showed that k shows a
positive correlation with left caudate dopaminergic terminal
function as assessed by [18F]uorodopa PET. In PD patients
showing impulse control disorders k was further correlated with
greater dopaminergic terminal function in the anterior putamen. As
this structure is functionally connected to the pre-supplementary
motor area, greater presynaptic dopaminergic availability in the
anterior putamen may contribute to the emergence of impulse
control disorders in PD. These ndings as well as the results of the
present study suggest that PD itself and maybe differentially
weighted neurodegeneration in the different dopaminergic path-
ways may lead to impulsive discounting and may constitute a risk
factor for pathological gambling.
Impulse control disorders in PD have been the focus of intense
research recently owing to their clinical importance. Wiehler and
Peters [26] recently reviewed the evidence for a relationship be-
tween delay and probability discounting in pathological gambling
(as one example of impulse control disorders) and found that none
of the three available studies on this matter revealed a correlation
between the two, while pathological gambling affected both, delay
and probability discounting but mostly the former. Thus, it appears
that different paradigms assess different aspects of reward based
behavior. This may explain the differences between the current
study and a previous study by Torta et al. [27] who employed the
Cambridge Gamble Task, a task requiring risky decision-making in
the face of explicit outcome probabilities, in PD. These authors
found more impulsive behavior in PD patients. Moreover, patients
on higher doses of dopaminergic medication showed more pro-
nounced impulsivity than patients on lower doses. Thus, further
studies providing a face to face comparison of probabilistic
gambling tasks and delay discounting tasks in PD are needed to
address these discrepancies.
5. Conclusion
The present study shows that impulsive decision making in
patients with PD may not be a side effect of dopaminergic treat-
ment but may rather be a trait marker of PD, as the de novo PD
group showed a signicantly steeper discount function than HC
participants. Moreover, there was no difference between medicated
PD patients and de novo PD patients. Further investigations taking
the state of the dopaminergic system into account, e.g. by using
PET-based methods (cf. [25]), are needed to further explore
impulsive decision making in PD.
Financial disclosures
Thomas F. Mnte has received support from DFG and BMBF.
Johann Hagenah has received support from BachmanneStrauss
Dystonia & Parkinson Foundation (BSDPF).
Author roles
Mohamed Al-Khaled contributed the organization of the
research project, recruited participants, reviewed the statistical
analysis, researched literature, and wrote the rst draft. Marcus
Heldmann contributed the study design, did the statistical analysis
and co-wrote the rst draft. Inga Bolstorff recruited participants,
acquired the data and participated in the analysis. Thomas F. Mnte
contributed the conception and the organization of the study,
reviewed the statistical analysis, and wrote the nal draft. Johann
Hagenah contributed the study design, recruited participants,
reviewed the statistical analysis, and critiqued the manuscript.
Acknowledgment
We thank all participants in this study and Anke Wilhoeft for
conducting part of the neuropsychological assessment.
References
[1] D. Weintraub, A.D. Siderowf, M.N. Potenza, J. Goveas, K.H. Morales, J.E. Duda,
et al., Association of dopamine agonist use with impulse control disorders in
parkinson disease, Arch. Neurol. 63 (2006) 969e973.
[2] A.H. Evans, A.P. Strafella, D. Weintraub, M. Stacy, Impulsive and compulsive
behaviors in Parkinson's disease, Mov. Disord. 24 (2009) 1561e1570.
[3] S.S. O'Sullivan, A.H. Evans, A.J. Lees, Dopamine dysregulation syndrome: an
overview of its epidemiology, mechanisms and management, CNS Drugs 23
 M. Al-Khaled et al. / Parkinsonism and Related Disorders 21 (2015) 1330e1335
(2009) 157e170.
[4] V. Voon, S.H. Fox, Medication-related impulse control and repetitive behaviors
in parkinson disease, Arch. Neurol. 64 (2007) 1089e1096.
[5] V. Voon, A. Schoerling, S. Wenzel, V. Ekanayake, J. Reiff, C. Trenkwalder, et al.,
Frequency of impulse control behaviours associated with dopaminergic
therapy in restless legs syndrome, BMC Neurol. 11 (2011) 117.
[6] D. Weintraub, J. Koester, M.N. Potenza, A.D. Siderowf, M. Stacy, V. Voon, et al.,
Impulse control disorders in parkinson disease: a cross-sectional study of
3090 patients, Arch. Neurol. 67 (2010) 589e595.
[7] S.M. McClure, D.I. Laibson, G. Loewenstein, J.D. Cohen, Separate neural systems
value immediate and delayed monetary rewards, Science 306 (2004)
503e507.
[8] J.E. Mazur, Fixed and variable ratios and delays: further tests of an equivalence
rule, J. Exp. Psychol. Anim. Behav. Process 12 (1986) 116e124.
[9] K.N. Kirby, N.M. Petry, W.K. Bickel, Heroin addicts have higher discount rates
for delayed rewards than non-drug-using controls, J. Exp. Psychol. Gen. 128
(1999) 78e87.
[10] B. Reynolds, K. Karraker, K. Horn, J.B. Richards, Delay and probability dis-
counting as related to different stages of adolescent smoking and non-
smoking, Behav. Process. 64 (2003) 333e344.
[11] B. Reynolds, Do high rates of cigarette consumption increase delay dis-
counting? A cross-sectional comparison of adolescent smokers and young-
adult smokers and nonsmokers, Behav. Process. 67 (2004) 545e549.
[12] B. Reynolds, J.B. Richards, K. Horn, K. Karraker, Delay discounting and prob-
ability discounting as related to cigarette smoking status in adults, Behav.
Process. 65 (2004) 35e42.
[13] K.N. Kirby, N.M. Petry, Heroin and cocaine abusers have higher discount rates
for delayed rewards than alcoholics or non-drug-using controls, Addiction 99
(2004) 461e471.
[14] V. Voon, B. Reynolds, C. Brezing, C. Gallea, M. Skaljic, V. Ekanayake, et al.,
Impulsive choice and response in dopamine agonist-related impulse control
behaviors, Psychopharmacol. Berl. 207 (2010) 645e659.
[15] C.R. Housden, S.S. O'Sullivan, E.M. Joyce, A.J. Lees, J.P. Roiser, Intact reward
learning but elevated delay discounting in parkinson's disease patients with
impulsive-compulsive spectrum behaviors, Neuropsychopharmacology 35
(2010) 2155e2164.
[16] M. Milenkova, B. Mohammadi, K. Kollewe, C. Schrader, A. Fellbrich,
1335
M. Wittfoth, et al., Intertemporal choice in parkinson's disease, Mov. Disord.
26 (2011) 2004e2010.
[17] A.T. Beck, R.A. Steer, R. Ball, W. Ranieri, Comparison of Beck depression in-
ventories -Ia and -II in psychiatric outpatients, J. Pers. Assess. 67 (1996)
588e597.
[18] E. Kalbe, P. Calabrese, N. Kohn, R. Hilker, O. Riedel, H.U. Wittchen, et al.,
Screening for cognitive decits in Parkinson's disease with the Parkinson
neuropsychometric dementia assessment (Panda) instrument, Park. Relat.
Disord. 14 (2008) 93e101.
[19] A.J. Hughes, S.E. Daniel, L. Kilford, A.J. Lees, Accuracy of clinical diagnosis of
idiopathic parkinson's disease: A clinico-pathological study of 100 cases,
J. Neurol. Neurosurg. Psychiatry 55 (1992) 181e184.
[20] R.P. Allen, D. Picchietti, W.A. Hening, C. Trenkwalder, A.S. Walters,
J. Montplaisi, Restless legs syndrome: diagnostic criteria, special consider-
ations, and epidemiology. A report from the restless legs syndrome diagnosis
and epidemiology workshop at the national institutes of health, Sleep. Med. 4
(2003) 101e119.
[21] Z. Ye, A. Hammer, E. Camara, T.F. Munte, Pramipexole modulates the neural
network of reward anticipation, Hum. Brain Mapp. 32 (2012) 800e811.
[22] A. Antonini, C. Siri, G. Santangelo, R. Cilia, M. Poletti, M. Canesi, et al.,
Impulsivity and compulsivity in drug-naive patients with Parkinson's disease,
Mov. Disord. 26 (2011) 464e468.
[23] J.Y. Lee, E.K. Lee, S.S. Park, J.Y. Lim, H.J. Kim, J.S. Kim, et al., Association of DRD3
and GRIN2B with impulse control and related behaviors in Parkinson's dis-
ease, Mov. Disord. 24 (2009) 1803e1810.
[24] H. Braak, E. Ghebremedhin, U. Rub, H. Bratzke, K. Del Tredici, Stages in the
development of parkinson's disease-related pathology, Cell Tissue Res. 318
(2004) 121e134.
[25] J. Joutsa, V. Voon, J. Johansson, S. Niemela , J. Bergman, V. Kaasinen, Dopami-
nergic function and intertemporal choice, Transl. Psychiatry 5 (2015) e491.
[26] A. Wiehler, J. Peters, Reward-based decision making in pathological gambling:
the roles of risk and delay, Neurosci. Res. 90 (2015) 3e14.
[27] D.M. Torta, L. Castelli, M. Zibetti, L. Lopiano, G. Geminiani, On the role of
dopamine replacement therapy in decision-making, working memory, and
reward in Parkinson's disease: does the therapy-dose matter? Brain Cogn. 71
(2009) 84e91.