Lydia Qualls Senior Honors Thesis FINAL

advertisement
DOPAMINE AND LEARNING IN ICD PATIENTS
1
Running Head: DOPAMINE AND LEARNING IN ICD PATIENTS
Effects of Dopamine on Reward Learning in Parkinson’s Patients with Impulse Control
Disorders
Lydia Qualls
Vanderbilt University
Author Note:
Lydia R. Qualls, Department of Psychology, Vanderbilt University.
Honors Thesis completed under the advisement of Dr. David Zald. The author would also like to
acknowledge Dr. Greg Samanez Larkin for his invaluable help and Dr. Daniel Claassen and Dr.
Scott Wylie for allowing me to collect data in the Movement Disorders Lab.
DOPAMINE AND LEARNING IN ICD PATIENTS
Abstract
This study investigated how dopamine agonist medication differentially affects reward learning
in Parkinson’s patients with and without Impulse Control Disorders (ICDs). We tested 16
patients (8 female, 5 with ICDs, mean age = 62.1) on and off of their dopamine agonist
medication using a dynamic foraging task with probability reversals. We hypothesized that
patients with ICDs, but not patients without ICDs, would have worse task performance on
medication than off medication. Paired samples t-tests confirmed our hypothesis – that task
performance was significantly worse on medication than off for ICD patients [t(4) = 2.86, p =
.046], but not for non-ICD patients [t(10) = 1.67, p = .126]. This suggests that ICD patients are
more vulnerable to medication effects that cause aberrant reward learning, which could be the
basis of their impulse control behaviors.
2
DOPAMINE AND LEARNING IN ICD PATIENTS
3
Effects of Dopamine on Reward Learning in Parkinson’s Patients with Impulse Control
Disorders
Many people can be impulsive from time to time; however, for some people impulsive
behavior is pathological. The DSM-IV-TR defines an impulse control disorder as “a failure to
resist an impulse, drive, or temptation to perform an act that is harmful to the person and others”
(p. 663). These impulse control disorders (ICDs) can manifest themselves in ways that have
serious consequences, such as pathological gambling and infidelity to one’s partner. It is
important to determine the cause of a person’s impulse control problem so that it can be
effectively treated. One neurotransmitter, or chemical that influences activity in the brain,
believed to be associated with impulsivity is dopamine. Dopamine originates from a region in the
brain known as the midbrain, which contains the dopamine-producing regions of the substantia
nigra and ventral tegmental area.
In their 2011 review, Antonelli, Ray, & Strafella describe three contributors to
impulsivity in Parkinson’s disease: an underlying diathesis to impulsivity, the disease itself, and
the various treatments for the disease. Pharmacological studies of Parkinson’s patients have
implicated dopamine in eliciting impulsivity. This relationship was first observed in patients
being treated with L-Dopa, a precursor to dopamine. While on this treatment, some patients
displayed abnormal decision making on a betting task, as well as other examples of impulsive
behavior (Cools, Barker, Sahakian, & Robbins, 2003). When investigating impulsive behaviors
in Parkinson’s disease, Ferrara and Stacey (2008) identified three main types of impulse control
problems: reward-seeking behaviors, punding behaviors, and hedonistic homeostatic
dysregulation syndrome. Reward-seeking behaviors can be defined as taking unnecessary risks
for a larger gain and are generally the focus of literature on dopamine-related impulsivity.
DOPAMINE AND LEARNING IN ICD PATIENTS
4
Punding behaviors are those in which a person is inexplicably fixated on something and performs
useless, repetitive behaviors for no discernible reason. Hedonistic homeostatic dysregulation is a
behavioral syndrome which is characterized by self-medication and addiction to dopaminergic
medications in the context of Parkinson’s disease (Ferrara & Stacy, 2010). Bódi et al. (2009)
found that giving L-Dopa to never-medicated Parkinson’s patients caused them to show increased
processing of rewarded stimuli and novelty-seeking, as well as an increased correlation between
the two processes. However, it decreased the correlation between punishment processing and
harm avoidance. The net effect of these changes was to motivate the participant to seek out
reward and novelty without being affected by adverse events, which matches the behavior
pattern of patients with ICDs (Bódi et al., 2009).
The Parkinson’s patients who develop ICDs differ markedly in their behavioral reactions
from Parkinson’s patients who do not develop these disorders. These impulse control disorders
are extreme reward-seeking behaviors that involve biases towards instant gratification with little
thought for the long-term consequences. Impulsive actions such as these are often called
“behavioral addictions,” which draws a link to the influence of dopamine on addictive behavior.
Voon et al. (2010) compared Parkinson’s patients with ICDs, Parkinson’s patients without ICDs,
and healthy controls on behavioral impulsivity by using intertemporal choice tasks, in which
participants chose between immediate gratification in the form of a small reward or delayed
gratification of a much larger reward. The Parkinson’s patients with ICDs and the Parkinson’s
patients without ICDs were tested on and off of dopamine agonist treatment. The Parkinson’s
patients with ICDs had developed them after starting on dopamine treatment, and were on the
same treatment during the experiment as they were on when they developed the ICDs. These
patients were matched to non-ICD Parkinson’s patients of the same age and gender and on the
DOPAMINE AND LEARNING IN ICD PATIENTS
5
same dopamine treatment. Parkinson’s patients with ICDs had faster reaction times on dopamine
agonist treatment as compared to those patients without ICDs; they also had a greater bias for
instant gratification and greater choice impulsivity in the form of steeper temporal discounting,
compared to non-ICD patients. Dopamine medication status (on or off medication) also had a
greater effect on Parkinson’s patients with ICDs when it came to making a quick decision
between two high-conflict choices (Voon et al., 2010). This study suggests that unknown
individual variations such as baseline dopamine level, dopamine treatment sensitivity, or greater
dopamine receptor affinity might exist between Parkinson’s patients who develop ICDs while on
dopaminergic medication and those who do not.
Dopamine agonists are the dopaminergic medication that has been found to most disrupt
reward learning. Dopamine encodes reward by phasic firing in response to novel or unexpected
reward targets and shows transient dips when expected rewards do not occur (Schultz, Dayan, &
Montague, 1997). Voon et al. found in 2010 that dopamine agonists were more likely to disrupt
learning because they cause a stronger striatal prediction error, causing excessive weighting of
gain cues. Others have hypothesized that dopamine agonists affect dopamine receptors in a way
that levadopa medication does not, continually stimulating dopamine receptors and thereby
blocking phasic dips in dopamine that encodes an important component of the learning signal.
Since these phasic dips act as an error cue, preventing these phasic dips in dopamine inhibits
punishment processing (Frank, Seeberger, & O'Reilly, 2004; Frank, Samanta, Moustafa, &
Sherman, 2007 as cited in Antonelli et al., 2011). In individuals with Parkinson’s disease that are
more sensitive to changes in phasic dopamine firing, dopamine agonist medication would be
very likely to cause an impulse control disorder.
DOPAMINE AND LEARNING IN ICD PATIENTS
6
In the review by Antonelli et al. (2011), the authors distinguish between two types of
impulsivity: motor and cognitive. Motor impulsivity is the preference for performing previously
learned actions over a required action (or the inhibition of an action) despite signals to the
contrary. Cognitive impulsivity is characterized by more than one factor, including altered
decision making and risk taking. According to Antonelli, altered decision making results from a
change in the interactions between the amygdala-controlled impulsive, immediate behavior
system and the slower, more future-oriented prefrontal-controlled system. If the amygdala
system is too active and the prefrontal system is not active enough, this leads to more impulsive
decision making. A related component is risk taking. There are two different sets of conditions
under which different types of risk taking occurs. One type is explicit risk-taking, which happens
in circumstances where the probabilities of risk are known. The other type is ambiguous risk
taking, when the probabilities of the outcomes are unknown. This type of risk taking is mediated
by the ventral frontostriatal loop, in which the orbitofrontal cortex and amygdala serve to
respond to uncertainty and the dorsal striatum responds to reward anticipation (Antonelli et al.,
2011). It is this type of cognitive impulsivity in ambiguous risk taking with which the present
study is concerned.
Rutledge et al. (2009) looked at how dopaminergic drugs affect reinforcement learning
rates and perseveration in Parkinson’s patients. Since loss of dopaminergic neurons in the
substantia nigra is a hallmark trait of Parkinson’s disease, and these neurons are thought to play a
central role in reinforcement learning, it stands to reason that reinforcement learning in
Parkinson’s patients might be deficient. In this study, Parkinson’s patients performed a dynamic
foraging task both on and off medication. Investigators found that the patients on the drugs had
an increased learning rate and that dopaminergic drugs selectively increased learning rates for
DOPAMINE AND LEARNING IN ICD PATIENTS
7
positive outcomes only. Choice perseveration also increased with Parkinson’s disease (compared
to elderly controls) and decreased when the patients were on medication (Rutledge et al., 2009).
The finding that Parkinson’s patients on-drug had a higher learning rate than elderly and off-drug
Parkinson’s patients, who had the same rate, suggests that L-Dopa medication may be
“overdosing” the reward learning mechanism, which is relatively undamaged by the extensive
loss of dopamine neurons in the substantia nigra that occurs early in the course of Parkinson’s
disease. This may highlight the role of the ventral tegmental area in the midbrain, another
dopaminergic area correlated with learning, which is relatively undamaged in early Parkinson’s
disease. The selectivity of the drug increasing positive outcomes again illustrates why some
patients develop ICDs on medication. If patients are more sensitive to reward than loss, they may
be biased to take riskier options on gambling tasks.
Figure 1. Model of the “overdose” effect dopaminergic medication may have on reward
learning.
The Rutledge et al. study used patients who were on several different kinds of treatments,
making it difficult to tell which neurotransmitter system or which part of the system was being
altered to produce the differences in learning rates. They also only examined the behavior of
DOPAMINE AND LEARNING IN ICD PATIENTS
8
Parkinson’s patients as a whole, without using any behavioral subcategories. In our study, we
wanted to see if there are any differences in reward learning between Parkinson’s patients with
ICDs and those without, and how this differed on and off dopamine agonist medication. We
predicted that patients with ICDs would perform worse on the dynamic foraging task when on
dopamine agonists than patients without ICDs due to the impulsive behavior the drug activates.
Methods
Participants
The participants consisted of 36 adults with Parkinson’s disease, both with and without
impulse control disorders, who were recruited from a population of Parkinson’s patients visiting
the clinic at the Vanderbilt Medical Center. All the participants were on medications for their
Parkinson’s disease. Each participant gave written informed consent before starting the
experiment and was screened for any signs of dementia (Montreal Cognitive Assessment Task,
Nasreddine & Phillips, 2005), depression (CES-D, Yesavage et al., 1983) , and intellectual
disabilities (American National Adult Reading Test, Uttl, 2010). For our analyses, we used the
subset of the participants who were on dopamine agonist medication (n= 16, 8 female, 5 with
ICDs, mean age = 62.1). Most patients had either ropinirole or pramipexole as their dopamine
agonist medication. Other demographic information is included in Table 1.
Design
We performed a repeated measures design in which each participant performed the task
off medication, on agonist medication, on carbadopa-levodopa medication, and on both agonist
and carbadopa-levodopa medications. However, not all participants were taking both agonist and
carbadopa-levodopa medication, so 19 participants only made two visits, one off medication and
one on. Of our subsample, 13 patients were on both carbadopa-levodopa and dopamine agonist
DOPAMINE AND LEARNING IN ICD PATIENTS
9
medication and 3 patients were on dopamine agonist only. Order of medication was randomized
and participants came back on separate days to do each separate trial. The withdrawal period
from agonist medication was an average of 36 hours and the withdrawal period from carbadopalevodopa medication was an average of 12 hours.
Task
The dynamic foraging task, called the CrabgameV (shown in Figure 2 below), was
programmed in Eprime 2 and consists of a total of six hundred choice trials. In the task,
participants have to select a red buoy or a green buoy that represents a crab trap. A running tally
of crabs caught is displayed at the top of the screen. Each trap has a different probability of
success (also shown in Figure 2), and the probabilities switch after a variable number of trials.
Figure 2. Dynamic foraging task – Crab Task. Adapted from “Dopaminergic Drugs
Modulate Learning Rates and Perseveration in Parkinson’s Patients in a Dynamic Foraging
DOPAMINE AND LEARNING IN ICD PATIENTS 10
Task,” by R. B. Rutledge, S. C. Lazzaro, B. Lau, C. E. Myers, M. A. Gluck, and P. W. Glimcher,
2009, Journal of Neuroscience, 29, p. 15105. Copyright 2009 by Society for Neuroscience.
Procedure
Before performing the task the first time, each participant was screened for depression,
dementia, and other comorbidities. Each time a participant came to the lab, they performed the
same task procedure. They were seated at a computer and told: “You will be playing a game
where you will fish for crabs. Your goal is to catch as many crabs as you can. Just like in real
life, once a crab is caught in a trap it will remain there until taken out. You will make your
choices by pressing one of two keys on the hand grips.” The experimenter demonstrated the best
way to hold the hand grips and the participant practiced pressing each of the buttons with their
thumb. The task typically took each participant 20 minutes to perform.
Modeling
We used a computational reinforcement learning model to estimate three performance
related measures. We fit the model from Rutledge et al. to the choice data using MATLAB
scripts. Three free parameters in the model were estimated from the data as measures of
performance (learning rate, noise parameter, and choice perseveration parameter). For each
individual and each session, the model fit a single value to each of the three free parameters that
best fit the choices that individuals made in the session. The best fitting parameter estimates were
those that minimized the difference between the observed choices that individuals made and the
predicted choices that would be made by the model assuming those parameters (i.e., maximum
likelihood). The reinforcement model is composed of two main equations, a value function and a
decision function. The value function is an exponential function that describes the weight given
to recent outcomes (learning rate) and is used to update value estimates over time. The values
DOPAMINE AND LEARNING IN ICD PATIENTS 11
generated by the value function for each of the two options are then used in a decision function
to determine choice. The decision function is a softmax/logistic function that determines the
strength of the response bias for the option with a high subjective value on that trial. It includes
the noise parameter and the perseveration parameter. In general the larger the difference between
the values of the two options, the higher the probability of selection the option with high
subjective value. Individuals with higher slopes (noise parameter) are much more likely to
choose the higher value option even for small differences in value (i.e., this can be considered
exploitative). Individuals with higher perseveration values are more likely to choose the same
option repeatedly independent of the current or recent value of that choice.
Analysis
We analyzed MATLAB output using the statistics software SPSS, version 21. From the
output, we took the number for percent rich, which is the percentage at which a participant chose
the more rewarded option, regardless of whether the individual trial was rewarded. This was the
primary measure of a participant’s ability to learn about which reinforcement condition was
currently more rewarding. The output also provided numbers for learning rate and perseveration.
As stated above, we concentrated our analysis on participants on dopamine agonist, which is
where we expected to see an interaction with ICD status.
Results
We first performed ANCOVAs to determine which measures of task performance were
influenced by medication state and ICD status. We included age, years since diagnosis, daily
medication dosage equivalent, IQ, and behavioral impulsivity as measured by the Barrett
Impulsivity Scale (Barratt, Stanford, Dowdy, Liebman, & Kent, 1999) in our analyses as
covariates. In these analyses, we were unable to find any main effects or interactions among the
DOPAMINE AND LEARNING IN ICD PATIENTS 12
variables. We then performed ANOVAs since the covariates did not explain a significant amount
of the variance. Average values for main test parameters for ICD and non-ICD participants are
shown in Table 2 (off medication) and Table 3 (on medication). For percent rich, the main effect
of medication state was significant [F(1, 14) = 4.89, p = .044, ηp2 = .259], but the main effect of
ICD status on task performance was non-significant [F(1, 14) = 1.86, p = .194, ηp2 = .117], and
the interaction effect of medication state and ICD status was non-significant [F(1, 14) = .006, p =
.938, ηp2 = .000]. For learning rate, the main effect of medication state on task performance was
non-significant [F(1, 14) = 1.75, p = .207, ηp2 = .111], the main effect of ICD group was nonsignificant [F(1, 14) = .806, p = .385, ηp2 = .054], and the interaction of medication state and
ICD status was non-significant [F(1, 14) = 2.43, p = .141, ηp2 = .148]. For perseveration, the
main effect of medication state on task performance was non-significant [F(1, 14) = .889, p =
.362, ηp2 = .060], the main effect of ICD group was non-significant [F(1, 14) = .817, p = .381,
ηp2 = .055], and the interaction of medication state and ICD status was non-significant [F(1, 14)
= .000, p = .987, ηp2 = .000].
Because we had hypothesized that ICD patients would perform worse on dopamine
agonist medication than off dopamine agonist medication, and that this effect would not be seen
for non-ICD patients, we decided to investigate what was driving the significant main effect of
medication state on percent rich by examining this relationship directly with paired samples ttests. Using the percent rich measure of task performance, we found the hypothesized
relationship to be confirmed. ICD patients did significantly worse while on their dopamine
agonist medication [t(4) = 2.86, p = .046], (M = 57.5, SD = 6.15, M = 54.6, SD = 5.04), whereas
task performance for non-ICD patients was not significantly different on versus off medication
[t(10) = 1.67, p = .126], (M = 60.5, SD = 4.19, M = 57.8, SD = 4.68). However, an independent
DOPAMINE AND LEARNING IN ICD PATIENTS 13
samples t-test shows that the difference in performance of ICD patients compared to non-ICD
patients was not significant [t(14) = .079, p = .938].
62
Percent Rich
60
58
56
Percent Rich - On
54
Percent Rich - Off
52
50
48
ICD
non-ICD
ICD Status
Figure 3. Differences between medication state effect (on or off medication) on task
performance (percent rich) in ICD versus non-ICD patients. Error bars represent standard errors.
Discussion
We found a main effect of medication state on percent rich, though we did not observe a
main effect of ICD status or an ICD status by medication state interaction. We did not find
significant differences between groups on learning rate or perseveration, either. However, given
the medium to large effect sizes for learning rate and for the ICD status by medication state
interaction on percent rich, these may reach statistical significance with more participants, given
that the study was underpowered at the current sample size. The lack of findings in perseveration
suggests that this may not be an area in which impulse control behaviors cause reward learning
impairment. In follow-up t-tests within groups, we found a suggestive difference in drug effects
on percent rich between ICD and non-ICD patients. Although there was a lack of a statistically
significant difference in the independent samples t-test, it may again due to the small sample
DOPAMINE AND LEARNING IN ICD PATIENTS 14
sizes used in the present study. We hope to continue this study in the future and greatly increase
the sample size.
The suggestive evidence that ICD patients may be more susceptible to dopaminergic drug
effects is partially consistent with the “overdose” hypothesis which would predict worse
performance on agonist medication. Since task performance relies on being able to consider
which choice is more likely to be rewarding, and participants who act impulsively on medication
are not able to consider which choice is more likely to carry reward, their percentage of rich
choices suffers. This implies that the impulse control behaviors seen in some patients on
dopamine agonist treatment are a result of aberrant reward learning brought on by the drug. Only
some patients on dopaminergic medications develop ICDs, and even then it is most often patients
taking dopamine agonist medication that develop these disorders. The fact that task performance
(as measured by the percentage of rich choices a patient makes) while on dopamine agonists
decreases most significantly for patients with these disorders illustrates that it is the medication,
in addition to individual differences in dopaminergic brain biology, that causes the impulse
control problems.
Another component of the larger study which at the time of this writing was not far along
enough to include was a behavioral and imaging study examining how dopamine agonists impact
cognitive functioning, specifically, learning, effort, and risk-taking. Both ICD and non-ICD
patients have been included. Resting state MRIs will be performed on and off dopamine agonist
medication to examine how dopamine affects cerebral blood flow. This should allow us to see
how both baseline and medicated dopamine signaling differs between ICD and non-ICD
participants, and may provide a clue as to what underlying differences may predispose a
Parkinson’s patient to develop an ICD after treatment with dopamine agonist medication.
DOPAMINE AND LEARNING IN ICD PATIENTS 15
Participants will also undergo an [18F]fallypride PET scan looking at the density of
D2/D3 dopamine receptors in the striatum. Dalley et al. (2007) found that a decreased amount of
D2/D3 receptors in the striatum was related to trait impulsivity in rats, suggesting that fewer of
these receptors in the striatum might be a biological predisposition to impulsive personality. This
PET imaging portion of the study could reveal a difference of type and number of dopamine
receptors present in the striatum between patients with ICDs and those without. In addition to
trait impulsivity, differing receptor densities could also cause differences in reactions to
dopamine agonist medications. Both ropinirole and pramipexole dopamine agonists are selective
to D2 (and D3 in the case of pramipexole) and do not directly affect D1 receptors (Ahlskog,
2003). ICD patients could have a pattern of D2/D3 receptor density that causes them to respond
to dopamine agonist medication with increased problematic impulse control behavior. These next
steps using imaging are crucial in determining how differences in impulse control behavior arise
from the dopamine system in Parkinson’s disease and possibly in other impulse control
populations so that the condition can be addressed with the best treatment possible.
DOPAMINE AND LEARNING IN ICD PATIENTS 16
References
Ahlskog, J. E. (2003). Slowing Parkinson’s disease progression: Recent dopamine agonist trials.
Neurology, 60(3), 381–389. doi:10.1212/01.WNL.0000044047.58984.2F
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders
(4th ed., text rev.). Washington, DC: Author.
Antonelli, F., Ray, N., & Strafella, A. P. (2011). Impulsivity and Parkinson’s disease: more than
just disinhibition. Journal of the neurological sciences, 310(1-2), 202–7.
doi:10.1016/j.jns.2011.06.006
Barratt, E. S., Stanford, M. S., Dowdy, L., Liebman, M. J., & Kent, T. A. (1999). Impulsive and
premeditated aggression: A factor analysis of self-reported acts. Psychiatry Research, 86(2),
163–73. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10397418
Bódi, N., Kéri, S., Nagy, H., Moustafa, A., Myers, C. E., Daw, N., Dibó, G., et al. (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: A Journal of
Neurology, 132(Pt 9), 2385–95. doi:10.1093/brain/awp094
Cools, R., Barker, R. a, Sahakian, B. J., & Robbins, T. W. (2003). L-Dopa medication
remediates cognitive inflexibility, but increases impulsivity in patients with Parkinson’s
disease. Neuropsychologia, 41(11), 1431–1441. doi:10.1016/S0028-3932(03)00117-9
Dalley, J. W., Fryer, T. D., Brichard, L., Robinson, E. S. J., Theobald, D. E. H., Lääne, K., Peña,
Y., et al. (2007). Nucleus accumbens D2/3 receptors predict trait impulsivity and cocaine
reinforcement. Science (New York, N.Y.), 315(5816), 1267–70.
doi:10.1126/science.1137073
DOPAMINE AND LEARNING IN ICD PATIENTS 17
Ferrara, B. J. M., & Stacy, M. (2010). Impulse-Control Disorders in Parkinson’s Disease. CNS
Spectrums, (August 2008), 690–698.
Nasreddine, Z., & Phillips, N. (2005). The Montreal Cognitive Assessment, MoCA: A brief
screening tool for mild cognitive impairment. Journal of the American Geriatrics Society,
53(4). Retrieved from http://stroke.ahajournals.org/content/42/9/2642.short
Rutledge, R. B., Lazzaro, S. C., Lau, B., Myers, C. E., Gluck, M. A, & Glimcher, P. W. (2009).
Dopaminergic drugs modulate learning rates and perseveration in Parkinson’s patients in a
dynamic foraging task. The Journal of Neuroscience: The Official Journal of the Society for
Neuroscience, 29(48), 15104–14. doi:10.1523/JNEUROSCI.3524-09.2009
Schultz, W., Dayan, P., & Montague, P. R. (1997). A Neural Substrate of Prediction and Reward.
Science, 275(5306), 1593–1599. doi:10.1126/science.275.5306.1593
Uttl, B. (2010). North American Adult Reading Test: Age Norms, Reliability, and Validity.
Journal of Clinical and Experimental Neuropsychology, (March 2012), 37–41.
Voon, V., Reynolds, B., Brezing, C., Gallea, C., Skaljic, M., Ekanayake, V., Fernandez, H., et al.
(2010). Impulsive choice and response in dopamine agonist-related impulse control
behaviors. Psychopharmacology, 207(4), 645–59. doi:10.1007/s00213-009-1697-y
Yesavage, J. A, Brink, T. L., Rose, T. L., Lum, O., Huang, V., Adey, M., & Leirer, V. O. (1983).
Development and validation of a geriatric depression screening scale: A preliminary report.
Journal of Psychiatric Research, 17(1), 37–49. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/7183759
DOPAMINE AND LEARNING IN ICD PATIENTS 18
Appendix
Demographics
Age
Years Since Diagnosis
BIS
IQ
Equivalent Daily Dose
ICD
65.33 (8.96)
5.88 (3.81)
72.33 (10.13)
118.01 (7.26)
746.45 (552.75)
Non-ICD
62.23 (8.30)
6.92 (3.88)
58.00 (7.39)
119.13 (6.71)
788.19 (246.44)
Table 1: Participant Demographics
Off Medication
Percent Rich
# of Crabs
Reward
Learning Rate
Noise
Perseveration
ICD
57.18 (5.55)
182.33 (6.83)
28.5 (1.09)
.84 (.25)
2.283 (2.13)
.23 (.80)
Non-ICD
60.23 (4.01)
186.23 (8.57)
29.09 (1.34)
.87 (.17)
4.79 (4.68)
-.06 (.82)
Table 2: Participant performance off medication
On Medication
Percent Rich
# of Crabs
Reward
Noise
Learning Rate
Perseveration
ICD
55.02 (4.60)
177.33 (4.93)
27.72 (.76)
1.65 (1.14)
.73 (.35)
.39 (.32)
Table 3: Participant performance on medication
Non-ICD
57.84 (4.68)
184.00 (10.35)
28.75 (1.62)
3.17 (2.22)
.90 (.18)
-.06 (.90)
Download