Page 1
Accepted Manuscript
Dopamine enhances willingness to exert effort for reward in Parkinson’s disease
Dr. Trevor T-J. Chong, Valerie Bonnelle, Sanjay Manohar, Kai-Riin Veromann, KinanMuhammed, George K. Tofaris, Michele Hu, Masud Husain
PII: S0010-9452(15)00127-6
DOI: 10.1016/j.cortex.2015.04.003
Reference: CORTEX 1444
To appear in: Cortex
Received Date: 8 December 2014
Revised Date: 6 March 2015
Accepted Date: 9 April 2015
Please cite this article as: Chong TT-J, Bonnelle V, Manohar S, Veromann K-R, Muhammed K, TofarisGK, Hu M, Husain M, Dopamine enhances willingness to exert effort for reward in Parkinson’s disease,CORTEX (2015), doi: 10.1016/j.cortex.2015.04.003.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.
Page 2
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
1
Dopamine enhances willingness to exert effort for reward
in Parkinson’s disease
Trevor T-J Chong a,b
, Valerie Bonnelle a, Sanjay Manohar
a, Kai-Riin Veromann
a, Kinan
Muhammed a,b
, George K Tofaris b, Michele Hu
b, Masud Husain
a,b
Submitted as a NOTE (3000 words) to Cortex
a Department of Experimental Psychology, University of Oxford, Oxford OX1 3UD
b Nuffield Department of Clinical Neurosciences, John Radcliffe Hospital, Oxford OX3 9DU
Corresponding Author:
Dr Trevor T-J Chong
Nuffield Department of Clinical Neurosciences
John Radcliffe Hospital
Oxford OX3 9DU
United Kingdom
+44 (0) 1865 618634
[email protected]
Running Title: Dopamine in Effort-Based Decision-Making
Page 3
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
2
Abstract
Parkinson’s disease (PD) is traditionally conceptualised as a disorder of movement, but
recent data suggest that motivational deficits may be more pervasive than previously thought.
Here, we ask whether subclinical deficits in incentivised decision-making are present in PD
and, if so, whether dopaminergic therapy ameliorates such deficits. We devised a novel
paradigm in which participants decided whether they were willing to squeeze a hand-held
dynamometer at varying levels of force for different magnitudes of reward. For each
participant, we estimated the effort level at which the probability of accepting a reward was
50% – the effort ‘indifference point’. Patients with PD (N = 26) were tested ON and OFF
their usual dopaminergic medication, and their performance compared to those of age-
matched controls (N = 26). No participant was clinically apathetic as defined by the Lille
Apathy Rating Scale. Our data show that, regardless of medication status, patients with PD
chose to engage less effort than controls for the lowest reward. Overall, however, dopamine
had a motivating effect on participants’ choice behaviour – patients with PD chose to invest
more effort for a given reward when they were in the ON relative to OFF dopamine state.
Importantly, this effect could not be attributed to motor facilitation. We conclude that deficits
in incentivised decision-making are present in PD even in the absence of a clinical syndrome
of apathy when rewards are low, but that dopamine acts to eliminate motivational deficits by
promoting the allocation of effort.
Keywords
Dopamine, effort, reward, decision-making, Parkinson’s disease
Abbreviations
PD = Parkinson’s disease; MoCA = Montreal Cognitive Assessment; LARS = Lille Apathy
Rating Scale; DASS = Depression Anxiety Stress Scale; UPDRS = Unified Parkinson’s
Disease Rating Scale; LE = Levodopa equivalence; MVC = Maximal Voluntary Contraction;
ANOVA = Analysis of Variance
Page 4
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
3
1. Introduction
Parkinson’s disease (PD) is a prototypical model of striatal dysfunction. The accompanying
dopaminergic depletion is traditionally considered one of the underlying mechanisms that
contributes to the cardinal motor symptoms of bradykinesia, rigidity and tremor (Jankovic,
2008). Recently, however, some authors have proposed that at least some Parkinsonian motor
symptoms may represent a deficit in ‘implicit’ motor motivation. For example, one study
reported that patients with PD had similar kinematic parameters to controls, but were more
likely to move slowly when the energetic demands of a movement increased (Mazzoni,
Hristova, & Krakauer, 2007). They therefore conceptualised Parkinsonian bradykinesia as a
shift in the balance between the perceived reward of reaching the target endpoint and the
amount of effort required to achieve a movement of normal speed. Findings such as this
suggest that motivational deficits may be more pervasive in PD than previously thought.
To determine if an action is worth initiating, one must evaluate the cost of that action – for
example, the effort associated with it – against its potential rewards. Effort is generally
considered aversive and, when given a choice, most animals will usually prefer actions that
are less effortful (Salamone, Correa, Farrar, & Mingote, 2007; Walton, Kennerly,
Bannerman, Phillips, & Rushworth, 2006). Thus, rewards which require less effort are
generally preferred over rewards of identical value which are associated with greater effort
(Hull, 1943). A number of animal studies have implicated dopamine in effort and reward
valuation (Pasquereau & Turner, 2013). In rats, dopamine depletion decreases tolerance for
effort, while drugs enhancing dopamine have the reverse effect (Salamone & Correa, 2002;
Salamone et al., 2007). Human data regarding the involvement of dopamine on effort and
reward integration remain relatively scarce, although there is a growing interest towards
understanding the role of dopamine in cost-benefit integration (Frank, 2005; Wardle,
Treadway, Mayo, Zald, & de Wit, 2011).
The pathognomonic striatal dysfunction in PD makes it an excellent model with which to
study the effect of dopamine on incentivised decision-making in humans. It remains poorly
understood how PD affects the valuation of an action’s costs and benefits, and how that may
subsequently affect choice behaviour. Although several studies in PD have examined
impairments in decision-making and reward (e.g., Bódi et al., 2009; Cools, Barker, Sahakian,
& Robbins, 2003; Czernecki et al., 2002; Frank, Seeberger, & O'Reilly, 2004; Mimura, Oeda,
& Kawamura, 2006; Porat, Hassin-Baer, Cohen, Markus, & Tomer, 2014), relatively few
Page 5
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
4
have explicitly examined effort-based motivational deficits (e.g., Porat et al., 2014; Schmidt
et al., 2008). Given the large animal literature postulating the role of striatal dopamine in
incentivisation, we hypothesise that motivational deficits are likely present at least
subclinically in PD, and independent of a clinical syndrome of apathy in which amotivation is
a defining characteristic (Pluck & Brown, 2002). Moreover, we predict that dopamine should
ameliorate these motivational deficits by promoting the allocation of effort.
Here, we report the results of a novel paradigm in which participants decided whether to
accept or reject a potential reward based on the effort that would be required to obtain it. An
important feature of our design was that it allowed us to focus on the effects of dopamine on
participants’ choices. This contrasts with many previous studies, especially those in animals,
which have inferred the motivational effects of dopamine on behaviour by examining the
effort manifest in the actions themselves (see Salamone et al., 2007 for review). By analysing
participants’ choices, we were able to calculate for each stake the effort level at which
participants considered an action not worth pursuing – their ‘effort indifference points.’ We
could then quantify the effect of PD and dopaminergic medication on shifting the position of
these indifference points relative to healthy controls.
Page 6
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
5
2. Material and Methods
2.1. Participants
This study was approved by the local institutional review board, and patients did not receive
financial compensation for their participation in the study. Patients with PD were recruited
through a tertiary hospital and community support groups. All patients were reviewed by at
least two consultant neurologists (TC and one other), and had a confirmed diagnosis of
idiopathic PD. They were excluded if they had a history of stroke, depression, impulse
control disorder, cognitive impairment (Montreal Cognitive Assessment (MoCA) score
<26/30) or musculoskeletal disease that would have interfered with their ability to perform
our task. Patients were on levodopa-containing compounds (n = 10), dopamine agonists (n =
5, including pramipexole, ropinirole, rotigotine), or combinations of both (n = 11). Clinical
severity was assessed with the Unified Parkinson’s Disease Rating Scale (UPDRS) (Fahn,
Elton, & Committee, 1987). We screened for apathy and depression with the Lille Apathy
Rating Scale (LARS) (Sockeel et al., 2006) and Depression Anxiety Stress Scale (DASS)
(Brown, Korotitsch, Chorpita, & Barlow, 1997; Lovibond & Lovibond, 1995), respectively.
Table 1 summarises the characteristics of our final sample of 26 patients. An equal number of
age- and education-matched controls was recruited through the local participant pool. Control
participants were excluded if they had a history of neurological illness, but exclusion criteria
were otherwise identical to those for patients.
2.2. Method
Participants were seated in front of a computer running Psychtoolbox
(http://psychtoolbox.org) implemented in Matlab (MathWorks, USA). They registered their
responses using two hand-held dynamometers (SS25LA, BIOPAC Systems, USA).
At the beginning of each session, the dynamometers were calibrated to each participant’s
maximal voluntary contraction (MVC). Participants alternately squeezed the left and right
dynamometers as strongly as possible, and the maximum contraction reached over three trials
was taken as each participant’s MVC for that hand. This procedure normalised subsequent
responses to each participant’s maximum force.
During the experiment, participants were presented with cartoons of apple trees, and were
instructed to accumulate as many apples as possible based on the combinations of stake and
Page 7
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
6
effort that were presented (Figure 1). Potential rewards were indicated by the number of
apples on the tree (1, 3, 6, 9, 12, 15), while the associated effort was indicated by the height
of a yellow bar positioned on the tree trunk, and ranged over six levels as a function of
participants’ MVCs (60%, 70%, 80%, 90%, 100%, 110%). By referencing the effort levels in
each session to each individual’s maximum force, we were able to normalise the difficulty of
each level across sessions and across individuals. Participants were familiarised with the
effort required for each level prior to commencing the experiment.
----------------------------------
Insert Figure 1 about here
----------------------------------
On each trial, participants had to decide whether they were willing to exert the specified level
of effort for the specified stake. If they judged the particular combination of stake and effort
to be ‘not worth it,’ they selected the ‘No’ response, and the next trial would commence. If,
however, they decided to engage in that trial, they selected the ‘Yes’ option. The tree would
subsequently reappear on the left or right of the screen (selected at random), corresponding to
the hand to be used for response execution. Participants then had five seconds to squeeze the
dynamometer to reach the target effort level. Apples could only be acquired if the target
effort level was reached; if participants failed to do so, no apples were received. If they
rejected a particular combination of effort and reward, they were instructed that a different
tree would subsequently appear and they were to proceed with the same process. At the
conclusion of the trial, they received feedback on their performance. Combinations of stake
and effort were presented according to an adaptive staircase algorithm (see Supplementary
Material).
After an initial practice block of 36 trials, participants completed five experimental blocks of
36 trials, separated by rest breaks. They were tested in two sessions approximately one week
apart. In one (‘ON’) session, patients were tested while taking their usual dopaminergic
medication; and, in the other (‘OFF’), patients were tested after overnight withdrawal of
medication. The order of ON and OFF sessions was counterbalanced across patients. Control
Page 8
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
7
participants performed two identical sessions to exclude the possibility of strategic changes
across sessions.
----------------------------------
Insert Figure 2 about here
----------------------------------
3. Results
For each stake, we estimated the effort level at which the probability of accepting an offer
was 50% (i.e., the effort ‘indifference point’). For each participant, we fitted a logistic
function to the choice probability data at each effort level (Figure 2). The effort indifference
points thus derived for each participant were then plotted against their corresponding stake
magnitudes. We then compared the effort indifference points for PD ON, PD OFF and
controls with repeated-measures ANOVAs.
3.1. Control Data
First, we ensured that control performance did not differ across testing sessions (Figure 3). A
repeated-measures ANOVA on effort indifference points with the factors of Session (First,
Second) and Stake (Levels 1-6) showed a significant main effect of Stake (F(5, 125) = 47.90,
p < .001), with Bonferroni-corrected contrasts revealing significant differences at each
successive Stake Level (all p < 0.05). Importantly, neither the main effect of Session (F(1,
25) = 0.59) nor its interaction with Stake (F(5, 125) = 1.54) was significant, indicating no
differences in control performance across Sessions 1 and 2. We therefore collapsed the
control data across the two sessions for subsequent analyses.
----------------------------------
Insert Figure 3 about here
----------------------------------
Page 9
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
8
3.2. Patient Data – ON vs OFF
To compare the effect of drug on effort indifference points, we performed a similar two-way
repeated-measures ANOVA, with the factors of Drug (ON, OFF) and Stake (1-6) (Figure
4A). This revealed a significant main effect of Drug, F(1, 25) = 25.9, p < .001, such that
patients ON medication were willing to invest more effort than those OFF, as reflected by a
higher mean effort indifference point (M 4.34 ± SE 0.10 vs 3.89 ± 0.13). The main effect of
Stake was also significant, F(5, 125) = 111.2, p < .001, with Bonferroni-corrected contrasts
demonstrating significant differences between all pairings of Stake (p < .001). The interaction
between Drug and Stake was not significant (F(5, 125) = 1.26).
----------------------------------
Insert Figure 4 about here
----------------------------------
To determine if maximal force output was modulated by dopamine, we compared MVCs ON
and OFF medication. Importantly, they were not significantly different (OFF 355 ± 24N vs
ON 361 ± 23N, t(25) = -1.34). There was also no significant effect of time-on-task, which we
used to examine the effect of fatigue on motor performance (see Supplementary Material).
Furthermore, there was no correlation between shifts in effort indifference points and
improvements in motor severity on the motor subscale (Part III) of the UPDRS (r = 0.22, p =
.28; see Supplementary Material). Thus, the shift of effort indifference points ON medication
was not simply attributable to a capacity to exert greater force or reductions in motor severity.
Given the association between dopamine and impulse control disorders (Weintraub et al.,
2010), could the incentivising effect of dopamine be mediated by lower risk aversion? We
analysed the proportion of trials in which patients engaged in effort levels beyond their
capacity to perform (i.e., Effort Level 6, or 110% MVC). Importantly, there was no
significant difference in this parameter ON versus OFF medication (t(25) = -1.59).
Furthermore, there was no effect of drug on the proportion of accepted trials in which patients
failed to reach the target effort level (t(25) = 0.17), and no effect of drug on failure rates or
trial history (see Supplementary Material).
Page 10
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
9
3.3. Patient vs Control Data
Next, we compared patient performance with that of controls (Figure 4B). An ANOVA
showed a significant effect of Stake (F(5, 250) = 106.96, p < .001) but not of Group, which
was qualified by a significant interaction (F(5, 250) = 9.62, p < .001). Patients ON dopamine
invested less effort than controls for the lowest Stake (2.42 ± 0.24 vs 3.19 ± 0.19, p < .05).
However, quite the opposite was found for higher Stakes (levels 4-6), at which controls were
actually willing to exert less effort than patients ON medication (Stake Level 4, ON 4.89 ±
0.11 vs Control 4.40 ± 0.11, p < .005; Level 5, ON 5.05 ± 0.12 vs Control 4.62 ± 0.12, p <
.05; Level 6, ON 5.26 ± 0.13 vs Control 4.75 ± 0.13, p < .01). Notably, there was no
significant difference in MVCs between patients ON medication and controls (Patients 360 ±
23N vs Controls 350 ± 24N, t(50) = 0.31).
For patient performance OFF medication vs controls (Figure 4C), the analogous ANOVA
demonstrated a significant effect of Stake (F(5, 250) = 111.90, p < .001), with a non-
significant main effect of Group (F(1, 50) = 2.70). Again, the two-way interaction was
significant (F(5, 250) = 6.12, p < .001), with Bonferroni-corrected comparisons revealing that
patients OFF medication were willing to expend less effort than controls, but only for the
lowest two stakes (Stake Level 1, OFF 3.19 ± 0.20 vs Control 2.31 ± 0.21, p < .005; Level 2,
3.80 ± 0.16 vs 3.33 ± 0.16, p < .05). MVCs between patients OFF medication and controls
were not significantly different (Patients 354 ± 23N vs Controls 350 ± 24N, t(50) = 0.129).
Page 11
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
10
4. Discussion
Few studies to date have examined impairments in effort-based decision-making in PD (e.g.,
Porat et al., 2014; Schmidt et al., 2008). Our data reveal two key findings. First, patients with
PD, regardless of medication status, were willing to invest less effort than their healthy
counterparts for the lowest reward. Second, dopamine exerted a motivating influence on
choice behaviour. Specifically, patients with PD chose to invest more effort for a given stake
when they were ON medication relative to OFF. Importantly, the incentivising effect of
dopamine cannot simply be due to motor facilitation, as there were no significant differences
in MVC across drug session, or between patients and controls. Furthermore, the shift in effort
indifference points from OFF to ON was not correlated with improvements in clinical motor
severity as measured by the motor section of the UPDRS.
A notable feature of our paradigm, and one of its significant strengths, is that it allowed us to
dissect out choice behaviour from motor preparation and execution. Many studies, in
particular those in animals, infer the effect of dopamine on effort by observing the effort
manifest in the behaviour itself (see Salamone et al., 2007 for review). A recent study in
healthy adults, for example, reported that dopamine augments response vigour in proportion
to average reward rate (Beierholm et al., 2013). In contrast to these previous studies,
however, our paradigm demonstrates that the incentivising effect of dopamine is evident even
during choice behaviour – i.e., prior to an action being initiated.
The question of how dopamine modulates aberrant cost-benefit integration in PD has not
been extensively explored. The finding that patients ON medication were willing to exert
greater force relative to OFF supports animal data showing that increasing dopaminergic tone
enables high-effort behaviours and increases tolerance of effort expenditure (Cagniard et al.,
2006; Niv, Daw, Joel, & Dayan, 2007; Robbins & Everitt, 1992; Wardle et al., 2011).
Critically, this incentivising effect of dopamine is independent of any motor changes which
might have occurred between the OFF and ON sessions. This is an important consideration,
given that a recent study in PD found that the greater number of key-presses that patients
exerted for reward when medicated was related to an improvement in their motor symptoms
(Porat et al., 2014). Our study builds on these previous findings by showing that the
motivational effect of dopamine on effort-based choices can occur independent of motor
facilitation, as measured by either motor strength (MVC) or the clinical severity of motor
signs (UPDRS).
Page 12
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
11
Studies of disordered motivation in PD often focus on clinically apathetic patients (e.g.,
Dujardin et al., 2007). Here, we show that patients with PD, who were neither clinically
apathetic nor depressed, and regardless of medication status, were less motivated than
controls to invest effort when the rewards were low. This confirms that Parkinsonian striatal
dysfunction is sufficient to cause an imbalance in the estimation of an action’s expected
value, and is consistent with animal studies showing that dopamine antagonism or depletion
reduces willingness to work for reward (Salamone et al., 2007). Although we only found a
reduction in motivation for the lowest levels of reward, any potential differences at higher
stakes in the comparison of PD OFF vs controls could very well have been obscured by a
saturation effect at the highest levels of effort. It should also be noted that our finding of
lower effort indifference points in patients vs controls for low stakes occurred despite the
LARS scores between the two groups being statistically similar and within the normal range.
This result therefore emphasises that motivational deficits may be present subclinically in PD
for low rewards, but that they are detectable with a sufficiently sensitive measure.
Finally, it is worth considering why participants in our task may have been willing to trade
effort for fictive rewards. There is of course a considerable literature that supports the view
that effort carries a value cost, and discounts the subjective value of potential rewards (e.g.,
Botvinick, Huffstetler, & McGuire, 2009). Complementing this literature is a considerable
volume of evidence showing that real and fictive rewards are discounted similarly in
behavioural paradigms (Hinvest & Anderson, 2010; Madden, Begotka, Raiff, & Kastern,
2003; Matusiewicz, Carter, Landes, & Yi, 2013). Furthermore, fMRI studies have shown that
real and fictive rewards recruit overlapping neural regions (Bickel, Pitcock, Yi, & Angtuaco,
2009). In light of these findings, we therefore expected our participants to discount effort
even in the presence of fictive rewards, as they in fact ultimately did.
Together, our findings show that deficits in incentivised decision-making are present in PD
when rewards are low even in the absence of a clinical syndrome of apathy, but that
dopamine acts to ameliorate motivational deficits by promoting the allocation of effort. This
echoes recent reports that Parkinsonian movement shares many attributes with healthy
behaviour (Desmurget et al., 2004), with a reduced motor drive being central to certain
Parkinsonian motor symptoms (Kojovic et al., 2014; Mazzoni et al., 2007). The
pervasiveness of motivational impairments in PD invites reconsideration of the degree to
which Parkinsonian hypokinesia is due simply to motor dysfunction versus a primary
Page 13
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
12
motivational deficit. These contributions are not mutually exclusive, and both might be
important in determining the surface manifestations of dopaminergic deficits in PD.
Page 14
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
13
Figure Legends
Figure 1. Summary of a typical trial.
Stakes were indicated by the number of apples on the tree (1, 3, 6, 9, 12, 15), while the
associated effort was indicated by the height of a yellow bar positioned at one of six levels on
the tree trunk (corresponding to MVCs of 60%, 70%, 80%, 90%, 100%, 110%). On each
trial, participants decided whether they were willing to exert the specified level of effort for
the specified stake. If they judged the particular combination of stake and effort to be ‘not
worth it,’ they selected the ‘No’ response. If, however, they decided to engage in that trial,
they selected the ‘Yes’ response, and then had to squeeze a hand-held dynamometer with a
force sufficient to reach the target effort level. Participants received visual feedback of their
performance, as indicated by the height of a red force feedback bar. To reduce the effect of
fatigue, participants were only required to squeeze the dynamometers on 50% of accepted
trials. At the conclusion of each trial, participants were provided with feedback on the
number of apples gathered.
Figure 2. An example of the fitted probability functions for a representative participant.
Logistic functions were used to plot the probability of engaging in a trial as a function of the
effort level for each of the six stakes. Each participant’s effort indifference points – the effort
level at which the probability of engaging in a trial for a given stake is 50% (indicated by the
dashed line) – were then computed.
Figure 3. Effort indifference points plotted as a function of stake for healthy controls in
Sessions 1 and 2.
Effort indifference points divide the stake-effort space into a sector in which participants are
willing to engage in an effortful response (below the curve) from a sector that is judged ‘not
worth the effort’ (above the curve). Control performance was identical between sessions 1
and 2. Error bars indicate ± 1 SEM.
Figure 4. Effort indifference points plotted as a function of stake for patients and
controls.
(A) Regardless of medication status, patients had significantly lower effort indifference
points than controls for the lowest reward. However, for high rewards, effort indifference
points were significantly higher for patients when they were ON medication, relative not only
to when they were OFF medication, but even compared to healthy controls. Inset: For clarity,
PD data are replotted against control performance for patients (B) ON medication and (C)
OFF medication. Shading denotes effort indifference points being greater for patients than
controls (orange), or less for patients than controls (yellow). Error bars indicate ± 1 SEM.
Page 15
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
14
Table 1.
Summary of participant demographics (means ± SD).
Patients with PD Healthy Controls Group Difference
N 26 26 -
Age (years) 66.6 (± 6.8) 66.2 (± 6.4) t(50) = 0.23, p = .82
Gender (M:F) 17:9 15:11 χ2 = 0.08, p = .78
LARS a
-28 (± 4.2) -29 (± 5.7) U = 277, p = .23
Depression Score on DASS b
2.00 (± 2.23) 1.5 (± 1.84) U = 295, p = .41
MoCA Scores c 28.2 (± 1.3) 28.2 (± 1.7) t(50) = 0.09, p = .93
UPDRS III (ON, OFF) d
ON: 21.6 (± 11.7)
OFF: 31.9 (± 13.6)
N/A -
Hoehn & Yahr Stage d 1.85 (± 0.54) N/A -
Disease Duration (years) 5.1 (± 3.1) N/A -
Levodopa Equivalence (mg) e 538 (± 275) N/A -
Interval between sessions
(days)
7.8 (± 1.7) 7.2 (± 0.8) t(50) = 1.51, p = .14
Average time since last dose
(hours)
ON: 2.28 (± 0.97)
OFF: 13.4 (± 3.4)
N/A -
a Normal range < -16 (Sockeel et al., 2006)
b Normal range = 0-9 (Lovibond & Lovibond, 1995)
c MoCA normal range 26-30.
d Clinical severity was assessed with the motor section (Part III, items 18-31) of the Unified
Parkinson’s Disease Rating Scale (UPDRS) (Fahn et al., 1987) and the modified Hoehn and
Yahr scale. See Supplementary Table 1 for a full summary of patients’ UPDRS data.
e Levodopa equivalence (LE) scores were calculated based on standard formulae (Tomlinson
et al., 2010). Patients were on levodopa-containing compounds (n = 10), dopamine agonists
(n = 5), or combinations of both (n = 11).
Page 16
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
15
References
This study was funded by a grant from The Wellcome Trust to MH and by a Neil Hamilton
Fairley Fellowship, National Health and Medical Research Council, Australia to TC.
Page 17
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
16
References
Beierholm, U., Guitart-Masip, M., Economides, M., Chowdhury, R., Düzel, E., Dolan, R., &
Dayan, P. (2013). Dopamine modulates reward-related vigor.
Neuropsychopharmacology, 38(8), 1495-1503.
Bickel, W., Pitcock, J., Yi, R., & Angtuaco, E. (2009). Congruence of BOLD response across
intertemporal choice conditions: Fictive and real money gains and losses. Journal of
Neuroscience, 29(27), 8839-8846.
Bódi, N., Kéri, S., Nagy, H., Moustafa, A., Myers, C. E., Daw, N., . . . Gluck, M. A. (2009).
Reward-learning and the novelty-seeking personality: a between- and within-subjects
study of the effects of dopamine agonists on young Parkinson's patients. Brain,
132(9), 2385-2395.
Botvinick, M., Huffstetler, S., & McGuire, J. (2009). Effort discounting in human nucleus
accumbens. Cognitive, Affective, & Behavioral Neuroscience, 9(1), 16-27.
Brown, T., Korotitsch, W., Chorpita, B., & Barlow, D. (1997). Psychometric properties of the
Depression Anxiety Stress Scales (DASS) in clinical samples. Behaviour Research
and Therapy, 35, 79-89.
Cagniard, B., Beeler, J., Britt, J., McGehee, D., Marinelli, M., & Zhuang, X. (2006).
Dopamine scales performance in the absence of new learning. Neuron, 51(5), 541-
547.
Cools, R., Barker, R., Sahakian, B., & Robbins, T. (2003). L-Dopa medication remediates
cognitive inflexibility, but increases impulsivity in patients with Parkinson's Disease.
Neuropsychologia, 41, 1431-1441.
Czernecki, V., Pillon, B., Houeto, J., Pochon, J., Levy, R., & Dubois, B. (2002). Motivation,
reward, and Parkinson's disease: influence of dopatherapy. Neuropsychologia, 40(13),
2257-2267.
Page 18
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
17
Desmurget, M., Gaveau, V., Vindras, P., Turner, R., Broussolle, E., & Thobois, S. (2004).
On-line motor control in patients with Parkinson's disease. Brain, 127, 1755-1773.
Dujardin, K., Sockeel, P., Devos, D., Delliaux, M., Krystkowiak, P., Destee, A., & Defebvre,
L. (2007). Characteristics of apathy in Parkinson's disease. Movement Disorders,
22(6), 778-784.
Fahn, S., Elton, R., & Committee, M. o. t. U. D. (1987). Unified Parkinson's disease rating
scale. In S. Fahn, C. Marsden, M. Goldstein & D. Calne (Eds.), Recent developments
in Parkinson's disease II (pp. 153-163). New York: MacMillan.
Frank, M. (2005). Dynamic dopamine modulation in the basal ganglia: a neurocomputational
account of cognitive deficits in medicated and non-medicated Parkinsonism. Journal
of Cognitive Neuroscience, 17, 51-72.
Frank, M., Seeberger, L., & O'Reilly, R. (2004). By carrot or by stick: cognitive
reinforcement learning in parkinsonism. Science, 306, 1904-1943.
Hinvest, N., & Anderson, I. (2010). The effects of real versus hypothetical reward on delay
and probability discounting. Quarterly Journal of Experimental Psychology A, 63,
1072-1084.
Hull, C. (1943). Principles of behavior. New York: Appleton-Century.
Jankovic, J. (2008). Parkinson's disease: clinical features and diagnosis. Journal of
Neurology, Neurosurgery and Psychiatry, 79, 368-376.
Kojovic, M., Mir, P., Trender-Gerhard, I., Schneider, S. A., Pareés, I., Edwards, M. J., . . .
Jahanshahi, M. (2014). Motivational modulation of bradykinesia in Parkinson’s
disease off and on dopaminergic medication. Journal of Neurology, 1-10. doi:
10.1007/s00415-014-7315-x
Lovibond, S., & Lovibond, P. (1995). Manual for the Depression Anxiety Stress Scales (2nd
ed.). Sydney: Psychology Foundation.
Page 19
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
18
Madden, G., Begotka, A., Raiff, B., & Kastern, L. (2003). Delay discounting of real and
hypothetical rewards. Experimental and Clinical Psychopharmacology, 11, 139-145.
Matusiewicz, A., Carter, A., Landes, R., & Yi, R. (2013). Statistical equivalence and test-
retest reliability of delay and probability discounting using real and hypothetical
rewards. Behavioural Processes, 100, 116-122.
Mazzoni, P., Hristova, A., & Krakauer, J. (2007). Why don't we move faster? Parkinson's
disease, movement vigor, and implicit motivation. Journal of Neuroscience, 27(27),
7105-7116.
Mimura, M., Oeda, R., & Kawamura, M. (2006). Impaired decision-making in Parkinson's
disease. Parkinsonism and Related Disorders, 12, 169-175.
Niv, Y., Daw, N., Joel, D., & Dayan, P. (2007). Tonic dopamine: Opportunity costs and the
control of response vigor. Psychopharmacology, 191(3), 507-520.
Pasquereau, B., & Turner, R. (2013). Limited encoding of effort by dopamine neurons in a
cost-benefit trade-off task. Journal of Neuroscience, 33(19), 8288-8300.
Pluck, G., & Brown, R. (2002). Apathy in Parkinson's disease. Journal of Neurology,
Neurosurgery and Psychiatry, 73, 636-642.
Porat, O., Hassin-Baer, S., Cohen, O. S., Markus, A., & Tomer, R. (2014). Asymmetric
dopamine loss differentially affects effort to maximize gain or minimize loss. Cortex,
51, 82-91.
Robbins, T., & Everitt, B. (1992). Functions of dopamine in the dorsal and ventral striatum.
In T. Robbins (Ed.), Seminars in the Neurosciences (Vol. 4, pp. 119-127). London:
Saunders.
Salamone, J., & Correa, M. (2002). Motivational views of reinforcement: implications for
understanding the behavioral functions of nucleus accumbens dopamine. Behavioural
Brain Research, 137(1-2), 3-25.
Page 20
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
19
Salamone, J., Correa, M., Farrar, A., & Mingote, S. (2007). Effort-related functions of
nucleus accumbens dopamine and associated forebrain circuits. Psychopharmacology
(Berl), 191, 461-482.
Schmidt, L., d'Arc, B., Lafargue, G., Galanaud, D., Czernecki, V., Grabli, D., . . . Pessiglione,
M. (2008). Disconnecting force from money: effects of basal ganglia damage on
incentive motivation. Brain, 131, 1303-1310.
Sockeel, P., Dujardin, K., Devos, D., Denève, C., Destée, A., & Defebvre, L. (2006). The
Lille apathy rating scale (LARS), a new instrument for detecting and quantifying
apathy: validation in Parkinson's disease. Journal of Neurology, Neurosurgery and
Psychiatry, 77(5), 579-584.
Tomlinson, C., Stowe, R., Patel, S., Rick, C., Gray, R., & Clarke, C. (2010). Systematic
review of levodopa dose equivalency reporting in Parkinson's disease. Movement
Disorders, 25(15), 2649-2685.
Walton, M., Kennerly, S., Bannerman, D., Phillips, P., & Rushworth, M. (2006). Weighing
up the benefits of work: Behavioral and neural analyses of effort-related decision
making. Neural Networks, 19, 1302-1314.
Wardle, M., Treadway, M., Mayo, L., Zald, D., & de Wit, H. (2011). Amping up effort:
Effects of d-Amphetamine on human effort-based decision-making. Journal of
Neuroscience, 31(46), 16597-16602.
Weintraub, D., Koester, J., Potenza, M., Siderowf, A., Stacy, M., Voon, V., . . . Lang, A.
(2010). Impulse control disorders in Parkinson disease: a cross-sectional study of
3090 patients. Archives of Neurology, 67, 589-595.
Page 21
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
Page 22
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
Page 23
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
Page 24
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
Page 25
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
1
Supplementary Material
S1. Supplementary Method
For the main experiment, participants were instructed that the aim of this task was to gather
as many apples as they could based on the combinations of effort and reward presented on
each trial. Prior to commencing the task, participants were first familiarised with the amount
of force required to achieve each effort level. During these preliminary trials, images of trees
without apples were presented. They were told that the height of the horizontal bar on the tree
trunk was proportional to the amount of force they would need to exert in the main
experiment. However, they were not explicitly informed about the percentage MVC
corresponding to each level. In this preliminary phase, participants had the opportunity to
familiarise themselves with the amount of force required for each effort level by squeezing
the dynamometers on separate trials to attempt to achieve each target effort level (two
familiarisation trials per effort level).
Trials were presented according to an adaptive staircase algorithm, in which combinations
of stake and effort were presented depending on participants’ previous choices
(Christopoulos, Tobler, Bossaerts, Dolan, & Schultz, 2009). If a particular combination of
stake and effort was declined on one trial, a higher stake or lower effort level was presented
on a subsequent trial (stake and effort levels were adjusted alternately). The opposite would
occur if a combination was accepted. Three randomly interleaved staircases were used so that
participants were unaware of the algorithm.
The advantage of such a design, in contrast to the approach of randomly sampling the entire
stake-effort space, is that it substantially reduced any effect of learning on task performance –
an important consideration given that dopamine is thought to be involved in reinforcement
and associative learning (Wise, 2004). By using a staircase algorithm, we were able to
converge efficiently on participants’ indifference points, which were therefore derived
independently of any associative learning between stimulus and reward, or force and effort.
In estimating participants’ effort indifference points, choice data were fitted using the
Palemedes toolbox (www.palamedestoolbox.org), a set of routines implemented in Matlab for
analysing psychophysical data (Kingdom & Prins, 2010; Prins & Kingdom, 2009). These
routines were used to fit a logistic function to choice data, characterised by 4 parameters (α, β,
Page 26
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
2
γ, λ).
We used free parameters for threshold (α) and slope (β). Fixed parameters were used for the
guess and lapse rates, based on the veridical distribution of choice responses for each
participant, with γ corresponding to the minimum, and λ being 1 minus the maximum. Thus,
the logistic functions fitted for each participant correspond to their actual responses.
S2. Supplementary Analyses on Patient Data
S2.1. Comparison of Indifference Point Slopes
In the principal analyses described in the main text, the interaction between Drug (ON, OFF)
and Stake (1-6) was not statistically significant (F(5,125) = 1.26). Graphically, however,
there appears to be a potential interaction between the two variables. As suggested by an
anonymous reviewer, a more sensitive test for an interaction might be to determine the slope
of indifference point curves for individual subjects by regressing them against a non-linear
(sigmoidal) function, and to then compare these slopes in the OFF and ON sessions with a
paired t-test. Doing so revealed that the slopes between the OFF and ON sessions were not
significantly different (t(25) = 1.39). This is in keeping with the absent statistical interaction
between Drug and Stake, and suggests that the difference between drug sessions is best
described as an upward shift of indifference point curves from OFF to ON.
S2.2. Failure Rates
As described in the Method section of the manuscript, participants were rewarded only on
trials in which they successfully achieved the target effort level. To determine if failure rates
differed as a function of Group or Effort, we compared failure rates in patients with a two-
way ANOVA on the factors of Drug (ON, OFF) and Effort level (1-6). This analysis revealed
a main effect of Effort, such that failure rates increased as a function of effort level (Effort
level 1, 1.4 ± 1.1%; Effort level 2, 0.6 ± 0.4%; Effort level 3, 7.2 ± 4.3%; Effort level 4, 32.6
± 6.1 %; Effort level 5, 60.2 ± 8.6%; Effort level 6, 98.0 ± 2.0%, p < .001). Critically,
however, neither the main effect of Drug nor its interaction with Effort was significant (both
p > .90), demonstrating that failure rates did not differ across the ON and OFF sessions.
Page 27
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
3
We also asked whether failure rates may have been correlated with motor severity as
measured on the motor section (Part III) of the UPDRS. We therefore performed a correlation
analysis between failure rates and scores on the motor section of the UPDRS in each of the
ON and OFF sessions. However, there was no correlation between failure rates and motor
scores in either of the two sessions (ON, r = 0.07, p = .74; OFF, r = -0.07, p = .75), nor was
there a correlation between changes in failure rates (ON > OFF) and improvements in motor
scores (ON > OFF) across the two drug sessions (r = 0.12, p = .56).
S2.3. Trial History
To determine if performance on one trial (i.e., failure or success to reach the required target
force) influenced decisions on subsequent trials, we first determined the probability of a
participant accepting a given offer in the trial immediately following failed versus successful
attempts at gathering a reward. We then conducted a 2 x 2 ANOVA on the factors of Drug
(ON, OFF) and Trial History (preceding trial failed vs successful). This analysis
demonstrated no significant main effects or interactions – specifically, the probability of a
patient accepting an offer was the same, regardless of whether they failed or succeeded on the
preceding trial (Drug, F(1,25) = 0.84, p = .37; Trial History, (F(1,25) = 0.28, p = .60; Drug x
Trial History, F(1,25) = 0.50, p = .49). This shows that the effect of trial history did not
significantly differ between patients ON and OFF medication.
S2.4. Effect of Motor Improvement on Effort Indifference Points
It is conceivable that the amount of effort patients were willing to exert was correlated with
the severity of their motor impairment. To address this question, we examined whether effort
indifference points were correlated with the motor section (Part III) of the UPDRS, for each
of the ON and OFF sessions separately. Importantly, there was no correlation between effort
indifference points and the severity of motor symptoms in either session (ON, r = -0.14, p =
.50; OFF, r = 0.03, p = .89).
On a related point, could the increase in effort indifference points from the OFF to ON
sessions be accounted for by a corresponding reduction in the severity of motor symptoms?
This is an important consideration, given that a recent study in PD found that patients
performed a higher number of keyboard presses for reward while ON medication relative to
OFF, but that this change was related to improvements in motor symptoms (Porat, Hassin-
Baer, Cohen, Markus, & Tomer, 2014). To address this issue, we performed a correlation
Page 28
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
4
analysis examining changes in effort indifference points from the OFF to ON sessions against
changes in the motor section (Part III) of the UPDRS across those two sessions. Importantly,
however, this correlation was not significant (r = 0.22, p = .28).
In summary, the severity of motor symptoms cannot account for the effort indifference points
in either the ON or OFF sessions, nor could improvement in patient symptoms account for
increases in indifference points from the OFF to ON sessions.
S2.5. Patient Force Output Data
The following analyses were conducted in order to verify that the shifts in indifference points
reported in the main text could not be attributable simply to changes in force output.
S2.5.1. Initial MVCs
To recapitulate, the analyses reported in the main text showed that the MVC at the beginning
of each session did not differ between patients ON and OFF medication, nor did they differ
between the patient and control groups overall. In addition, the time-to-peak contraction did
not differ between the ON and OFF sessions (OFF 2.56 ± 0.1s vs ON 2.53 ± 0.1s, t(25) =
1.04, n.s.). This suggested that differences in indifference points ON and OFF medication,
and between patients and controls, could not simply be due to differences in motor strength at
the beginning of each session.
In addition, we tested for any correlations between changes in patients’ MVCs and shifts in
their indifference points. For each patient, we calculated the difference in MVC between the
ON and OFF sessions. In addition, we calculated differences in their mean effort indifference
point ON vs OFF medication. If increases in MVC accounted for increases in patients’ effort
indifference points, we would expect a correlation between these two variables. However, no
such correlation was found (Pearson’s correlation coefficient -0.004, p = .984).
S2.5.2. Time-on-task Analyses
In order to verify that there were no changes in force output during the experiment, we
compared motor output between the first and second halves of each session (90 trials per
half). Importantly, dopamine did not differentially affect patients’ maximal grip force across
the first and second halves of each session (Drug, F(1, 25) = 2.36; Session Half, F(1, 25) =
0.19; Drug × Session Half, F(1, 25) = 2.03). Similarly, patients’ time-to-peak contraction did
Page 29
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
5
not differ over the course of the experiment as a function of drug (Drug, F(1, 25) = 1.24;
Session Half, F(1, 25) = 0.66; Drug × Session Half, F(1, 25) = 1.55).
In summary, these analyses show that there were no significant differences in patients’ motor
output at the beginning of each experimental session, nor during the sessions themselves.
This therefore excludes the possibility that any shifts of effort indifference points were due to
changes in force output.
S2.6. Effect of Medication Class
Because our study was aimed at examining the effect of dopaminergic medication in general
on decision-making, patients in our study were on different therapeutic regimens (levodopa-
containing compounds alone (n = 10), dopamine agonists alone (n = 5), or combinations of
both (n = 11)). An interesting question is whether performance on our task differed according
to medication subgroup. An ANOVA comparing the between-subjects effect of Medication
Subgroup (levodopa only, dopamine agonists only, both) on the within-subjects factors of
Drug Session (ON, OFF) and Stake Level (1-6) showed that Medication Subgroup was not
involved in any significant main effects or interactions (all F < 1.21). Obviously, however,
this null result should be interpreted with caution given the small and uneven sample sizes in
each subgroup, and it would be useful for future studies to pursue whether drug class has a
differential effect on motivation.
S2.7. Effect of Total Levodopa Equivalence Dose
In addition, we also asked whether dopamine had a dose-dependent effect on effort
indifference points. However, the correlations between mean effort indifference points and
levodopa equivalent dose was not significant, even after performing a partial correlation
controlling for disease duration (r = -0.22, p = .29).
S2.8. Effect of Apathy or Depressive Ratings
One might predict that patients who were towards the more ‘apathetic’ or ‘depressed’ range
on the LARS or DASS respectively may have had lower effort indifference points due to
their subjectively lower motivation. We therefore performed correlation analyses between
participants’ mean effort indifference points and their scores on the LARS and the
Depression subscale of the DASS. However, all Spearman correlation coefficients were not
significant (all p > .05). One potential reason for this is that our participants all scored within
Page 30
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
6
the normal range on these measures, and therefore did not demonstrate the variability that
might be required to reveal such correlations.
Page 31
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
7
Supplementary Table 1.
Summary of UPDRS scores for patients with PD (means ± SD).
UPDRS Patient Scores
Part I 7.0 (± 4.4)
Part II 11.2 (± 4.7)
Part III (ON) 21.6 (± 11.7)
Part III (OFF)
31.9 (± 13.6)
Part IV
2.08 (± 3.5)
Hoehn & Yahr Stage 1.85 (± 0.54)
Page 32
MA
NU
SC
RIP
T
AC
CE
PTE
D
ACCEPTED MANUSCRIPT
8
References
Christopoulos, G., Tobler, P., Bossaerts, P., Dolan, R., & Schultz, W. (2009). Neural
correlates of value, risk, and risk aversion contributing to decision making under risk.
Journal of Neuroscience, 29(40), 12574-12583.
Kingdom, F., & Prins, N. (2010). Psychophysics: A Practical Introduction. London:
Academic Press.
Porat, O., Hassin-Baer, S., Cohen, O. S., Markus, A., & Tomer, R. (2014). Asymmetric
dopamine loss differentially affects effort to maximize gain or minimize loss. Cortex,
51, 82-91.
Prins, N., & Kingdom, F. (2009). Palamedes: Matlab routines for analyzing psychophysical
data. http://www.palamedestoolbox.org
Wise, R. (2004). Dopamine, learning and motivation. Nature Reviews Neuroscience, 5, 483-
494.