Deep brain stimulation, or DBS for short, is used to treat movement disorders like Parkinson’s disease in patients who are responding inadequately to medications. It requires implanting an electrode into the brain and using electrical stimulation aimed at a specific cluster of brain cells to reduce unwanted symptoms. DBS helps to normalize abnormal brain activity similar to a pacemaker resetting an abnormal heart rhythm. Scientists are currently studying whether DBS might also help people with obsessive-compulsive disorder, depression, Alzheimer’s disease or other disorders that affect thinking.

To alter human behavior and treat disorders that affect thinking, DBS will have to be delivered at precise time points as the brain processes information. One potential target is for DBS in both movement and thinking disorders is the subthalamic nucleus. This is a small almond-shaped cluster of brain cells that helps people stop movements. Recent studies suggest it also may play a role in processing emotions, controlling inappropriate responses, and psychiatric illnesses.

Now, Patel et al. show that the subthalamic nucleus helps people decide what to do in the face of uncertainty and that targeting this brain structure with DBS can shift a person’s decision-making. In the experiments, patients with Parkinson’s disease who were awake and undergoing surgery to implant the DBS electrodes also played a computerized gambling game. Patel et al. recorded the electrical activity in the brain cells of the patient’s subthalamic nucleus during the game. The experiments showed that when patients were faced with a decision with 50/50 odds, the pattern of electrical activity in the cells of their subthalamic nucleus reveals their choice about 500 milliseconds before they act on it.

After their surgeries, patients engaged in the same gambling game. This time, Patel et al. specifically targeted the decision-related activity in their subthalamic nucleus with DBS. This caused the patients to make fewer risky decisions in the game. The experiments show DBS can change decision-making behavior in humans. Newer DBS technology may be even more effective at treating brain disorders and cause fewer side effects. Further study into how the brain processes information will also help scientists to better target DBS and possibly treat a broader range of diseases.

Deep brain stimulation (DBS) is a remarkable therapy that has revolutionized the potential for treating neurological and neuropsychiatric illness by directly modifying neural function, though the underlying mechanism of action remains unknown. In its simplest form, DBS can be thought of as a pacemaker for the brain. Current DBS devices deliver continuous electrical stimulation to a targeted brain region to modify or reset abberant neural activity or synchrony (Herrington et al., 2016a). A major area of current research is in identifying more refined methods of stimulation delivery by exploring the timing of stimulation delivery, multi-site stimulation, and real-time sensing and stimulation.

The subthalamic nucleus (STN) is a small almond-shaped nucleus in the basal ganglia classically known for its role in inhibiting motor responses as part of the indirect pathway (Schmidt and Berke, 2017; Schmidt et al., 2013). More recently, a growing body of literature has begun to uncover a more nuanced role of the STN in higher-order cognitive processes such as, emotional processing (Le Jeune et al., 2009; Le Jeune et al., 2008; Drapier et al., 2006; Eitan et al., 2013), response inhibition (Frank et al., 2007; Cavanagh et al., 2011; Isoda and Hikosaka, 2008), and even psychiatric illness (Mallet et al., 2008).

The STN is also an important deep brain stimulation (DBS) target for the treatment of movements disorders such as Parkinson’s Disease (PD). DBS surgery provides one of only a few opportunities to record neuronal responses in humans subjects engaged in cognitive tasks (Zénon et al., 2016; Herz et al., 2016; Cavanagh et al., 2011; Zaghloul et al., 2012). In addition, researchers can leverage implanted DBS electrodes as a neuromodulation tool to study the role of the STN in an extra-operative setting. As such, numerous researchers have utilized this approach to interrogate the function of the STN in conflict (Frank et al., 2007; Schroeder et al., 2002), decision-making (Seymour et al., 2016; Seinstra et al., 2016; Wylie et al., 2010; Zaehle et al., 2017; Cavanagh et al., 2011), and emotional processing (Le Jeune et al., 2008; Le Jeune et al., 2009).

Studies of STN stimulation using current generation DBS systems have been limited by the systems’ design to deliver continuous stimulation. This is a critical limitation to using DBS to explore dynamic aspects of cognition and might be an important source of variability on DBS effects reported in the literature. We hypothesized that targeting stimulation to specific temporal windows during the evolution of a cognitive process (e.g., decision-making), rather than delivering long periods of continuous stimulation, will be critical to understanding the cognitive function and developing new neuromodulation therapies going forward.

In this study, we explore the role of the STN in making decisions under conditions of uncertainty. We employed single-neuronal recordings and intermittent electrical stimulation in human subjects while they engaged in a financial decision-making task (Patel et al., 2012). We find that the STN is selectively activated during a brief window for high-uncertainty trials from single-neuronal data. To assess the role of the STN in decision-making during this brief temporal window, we built a custom device which allowed us to precisely deliver intermittent stimulation within short temporal windows. We found that a brief, high-frequency stimulation pulse delivered prior to the choice period promoted a reduction in risk-seeking behavior. To the best of our knowledge, this is the first study to apply intermittent DBS in humans actively engaged in a cognitive task and the first to demonstrate a reduction in risk-seeking behavior following STN DBS.

The task is analogous to the classic card game, War. Each player was dealt a card – the player with the highest card won (Figure 1a). Subjects were first presented with their card. They were then prompted to make either a $5 or $20 wager based on the perceived value of their hand. After a wager was selected, the opponent’s hand was revealed followed by visual feedback on the outcome of the trial. For each trial, subjects either won or lost the wagered amount. To simplify the game, we reduced the deck to even cards from 2 through 10 of one suit. Thus, if a subject was dealt a 10-card, the optimal choice would be to place a $20 wager as the outcome is likely to be positive or at worst a draw. Conversely, if the subject received a 2-card the optimal choice would be to place a $5 wager. Uniquely, there is no optimal strategy for the 6-card – the outcome is probabilistically equal.

Figure 1

Download asset Open asset

Figure 1—source data 1

SQLite database containing two tables: behavior and spikes.

The behavior table contains fields for ID, subject, structure (brain region), session (session numer), startIndex, endIndex, startTimestamp (time in seconds for beginning of trial), endTimestamp (time in seconds for end of trial), fixationTimestamp (time in seconds for the fixation cue), pcardTimestamp (time in seconds for card presentation), buttonTimestamp (time in seconds for the gocue), choiceTimestamp (time in seconds when the button press was registered), fulldeckTimestamp (time in seconds for when the opponent’s card is revealed), feedbackTimestamp (time in seconds for feedback), reactionTime (time in seconds), trialNumber (trial number), conditionNumber (condition number), pcard (player’s card), ccard (opponent’s card), choice (wager), winLoss (outcome), value (value of wager). The spikes table contains spiking data for each trial and can be referenced to the behavioral table ID variable by trialID.

https://doi.org/10.7554/eLife.36460.004

Download elife-36460-fig1-data1-v1.db

Signals of decision-making in the STN

We collected behavioral and neurophysiological data from six subjects (five men, one woman; $63.2\pm 6.8$ years old; mean $\pm$ Sc.D.; Table 1) that underwent DBS surgery for PD. On average subjects performed 1.83 sessions of the gambling task with an average of 105.2 trials per session.

Table 1

		Mean	Standard Deviation
Neuroimaging (n = 24)	Age (years)	36	7.5
Intraoperative (n = 6)	Age (years)	63.2	6.8
	Disease Duration (years)	8.2	3.3
	Levodopa dose (mg, daily)	530	300

The gambling task was designed such that on any given trial a positive outcome was probabilistically weighted by the subject’s card. As such, we expected an engaged participant to display longer reaction times for trials in which the outcome was unpredictable; whereas on predictable trials we expected behavior to converge to an optimal strategy resulting in shorter reaction times. We found such a trend ($F_{4,10}=10.2$, $p=4.0\times {10}^{-4}$; ANOVA; Figure 1b). Specifically, 6-card trials had the highest average reaction time ($1.16\pm .19$s and $1.33\pm .61$s; mean $\pm$ Sc.D., respectively) consistent with the unpredictable nature of the outcome (i.e. an equal chance of winning and losing). Similarly, reaction times for the most predictabletrials were amongst the lowest.

First, we examined behavior on trials in which subjects were dealt a 6-card. A behavioral deviation from a 50/50 betting strategy on these trials would indicate a risk-seeking or risk-averse bias. Overall, we found that subjects had a risk-averse bias placing a high wager only 24% of the time (${\chi }_{1,11}^2=42.24$, $p=1.44\times {10}^{-5}$; Figure 1c). In this study, all intraoperative subjects were off dopaminergic medications at least 12 hr prior to surgery. The low-dopamine state may have contributed to subject’s risk-avoidant behavior (St Onge et al., 2011; Claassen et al., 2011).

Current models suggest that STN activity inhibits responses during cognitively demanding situations (Frank, 2006; Frank et al., 2007). This inhibition may serve to allow for additional time to process internal and environmental information before ultimately arriving at and executing a decision. To explore this hypothesis in our study we leveraged the intrinsic symmetry of the behavioral paradigm, and divided trials into low and high cognitive demand. The 10- and 2-cards are extreme situations in which the player is probabilistically likely or unlikely to win, respectively — we call these low-uncertainty trials. Conversely, on the 6-card trials the player has an equal probability of winning and losing and there is no optimal strategy — we call these high-uncertainty trials.

We examined single-neuronal data from the STN using standard stereotactic and intraoperative microelectrode mapping procedures. We collected 27 well-isolated neurons with an average of $3.1\pm 1.1$ (mean $\pm$ Sc.D.) neurons per subject (Figure 2—figure supplement 1). All analyses were performed on normalized and pooled spiking data. We apriori selected a 500 ms window during the choice period based on previous findings (Patel et al., 2012) and explored the relationship between STN activity and the level of uncertainty on a given trial. To do this, we applied a regression model predicting z-scored spike counts as a function of the card value and wager. Interestingly, we found an interaction effect between card value and wager ($F_{9,1450}=2.55$, $p=0.02$; ANOVA) but no main effects for card value ($F_{9,1450}=1.69$, $p=0.14$) or wager ($F_{9,1450}=2.68$, $p=0.10$). Further exploration revealed a significant effect on high-uncertainty trials ($t_{1450}=-2.38$, $p=0.01$; xtbfFig. 2a; Figure 2—figure supplements 2, 3 and 4) which was not present on low-uncertainty trials ($t_{1450}=-1.02$, $p=0.30$; $t_{1450}=-0.16$, $p=0.86$; 2- and 10-cards respectively; Figure 2b). Interestingly, we found trending activity for the 4- and 8-trials ($t_{1450}=-1.730$, $p=0.08$; $t_{1450}=-1.78$, $p=0.07$) which contain an intermediate degree of uncertainty. No other stimulus epoch correlated with subject behavior (Table 2).

Figure 2 with 4 supplements see all

Download asset Open asset

Table 2

Epoch	Window	$\Delta$	T	P
Fixation	0–500 ms	−1.52	1.25	0.23
Card	0–500 ms	−1.34	0.70	0.49
Choice	0–500 ms	−1.79	1.31	0.21
Choice	500–1000 ms	−3.75	2.96	0.01
Feedback	0–500 ms	0.17	−0.17	0.86

This signal is unlikely to represent an overt finger movement because our task design balances the presentation of the $five and $20 wagers equally to the left- and right-hand side of the screen. Also, we found no difference in activity between wagers centered on the button press ($F_{9,1450}=0.24$, $p=0.98$; Figure 2a,b) suggesting this signal was not movement-related. In addition, there was no relationship between the wager ($t_{23}=0.09$, $p=0.92$) or the outcome ($t_{23}=0.71$, $p=0.48$) on the previous trial.

Lastly, we found that z-scored reaction times on high-uncertainty trials were longer when subjects placed a high vs. low wager ($t_6=-3.28$, $p=0.01$; Figure 2c). There was no difference in reaction times on low-uncertainty trials for high vs. low wager ($t_9=1.17$, $p=0.27$; Figure 2d).

Effects of intermittent STN stimulation on behavior

We have shown that STN activity within a brief temporal window during the choice period predicts the upcoming wager selectively for high but not low-uncertainty trials. Interestingly, previous human neurophysiology studies have described similar conflict signals arising earlier during the stimulus presentation epoch (Zaghloul et al., 2012; Sheth et al., 2012). To explore this discrepancy, we used intermittent DBS to test whether altering STN activity during this finite time window would alter the subject’s ultimate decision using intermittent DBS. We recruited 13 subjects (12 men, one woman; $62.6\pm 7.4$ years old; mean $\pm$ Sc.D.; Table 3) who had previously undergone STN DBS surgery for PD. All subjects had completed surgery at least 6 months prior tithe study.

Table 3

	Mean (n = 13)	Standard Deviation
Age (years)	62.6	7.4
Disease Duration (years)	15.5	5.6
Time since surgery (years)	3.9	2.5
Levodopa dose (mg, daily)	575	310
Therapeutic left voltage (volts)	2.9	0.8
Therapeutic right voltage (volts)	2.9	0.7
Therapeutic frequency (Hz)	180	14.3
Study voltage (volts)	1.0	0.9

Through patients’ implanted DBS electrodes we applied intermittent electrical stimulation to the STN while subjects were engaged in the same gambling task. Specifically, we applied one of three different stimulation categories randomly on 6-card trials, either giving: no stimulation, 1 s of stimulation during the fixation epoch, or 1 s of stimulation prior to the choice period. To control for observational effects of turning on/off the stimulator (e.g. feeling a sensation when the stimulator turns on), we systematically lowered the voltage setting—blinded to the subject—to a sub-threshold level prior to each experimental session. In addition, we characterized the latency from the trigger to current delivery and found it to be 174 $\pm$0.002 ms ($n$=26; mean $\pm$ Sc.D.; Figure 3—figure supplement 1). All other settings (e.g. electrode contacts, frequency, and pulse-width) were unaltered from therapeutic levels and were returned to normal following the study.

On average subjects performed 2 sessions of the gambling task with an average of 108 trials per session. Similar to the intraoperative experiment, we found that subjects demonstrated understanding of the underlying structure of the task ($F_{4,26}=5.83$, $p=0.0002$; ANOVA; Figure 3a). The fastest reaction times were observed for low-uncertainty trials ($1.19\pm .76$ seconds, $1.11\pm .71$ seconds; mean $\pm$ Sc.D.; 2- and 10-cards respectively); and on average, the high-uncertainty trials wreathe slowest ($1.46\pm 1.16$ seconds; mean $\pm$ Sc.D.). Unlike during the intraoperative sessions, subjects were on their clinical regimen of dopamine replacement therapy during this experiment. We did not observe the same risk-averse behavior on 6-card trials (${\chi }_{1,27}^2=34.13$, $p=0.13$; Figure 3b).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

SQLite database containing a single table: behavior.

The behavior table contains fields for ID, subject, session (session number), reactionTime (time in seconds), trialNumber (trial number), conditionNumber (condition number), pcard (player’s card), ccard (opponent’s card), leftChoice (value of wager mapped to left button), stimulation (stimulation condition), choice (wager), winLoss (outcome), value (value of wager).

https://doi.org/10.7554/eLife.36460.015

Download elife-36460-fig3-data1-v1.db

Guided by our neurophysiological findings, we expected that modulation of intrinsic decision signaling prior to the choice period would selectively bias subject behavior. As such, we expected no difference when stimulation was delivered during the fixation period compared to when it was omitted. The data confirmed this hypothesis ($F_{2,28}=2.93$, $p=0.05$; ANOVA). In contrast, when stimulation was delivered prior to the choice period, we found that on average subjects had a strong risk-averse bias and placed a high wager only 33.0 $\pm$4.83% (mean $\pm$ s.e.m.) of the time, on average an absolute 15% less than the no stimulation group ($t_{28}=2.77$, $p=0.009$; Figure 3c). Importantly, there was no difference between the omitted and fixation stimulation conditions ($t_{28}=0.14$, $p=0.88$), on which subjects placed a high wager on average of 48.3 $\pm$5.92% and 49.2 $\pm$ 5.6% of the time (mean $\pm$ s.e.m.), respectively.

To further explore the effects of intermittent stimulation on decision-making we more closely examined the effects within individual subjects. To do so, we plotted each subject’s average high wager percentage when stimulation was omitted and delivered at the choice period (Figure 3d). Overall, we found subjects spanned a large range in baseline tendency for placing high wagers, ranging from 11% to 100%. We found that 7 out of 11 subjects displayed a reduction in risk-seeking behavior (Table 4). Of the seven subjects the average magnitude of change was 12.8% (range: 1–18%). Interestingly, we observed that the magnitude of the reduction in risk-seeking behavior correlated with their initial starting point ($t_7=2.46$, $p=0.05$; Figure 3e). For the three subjects that showed an increase in risk-seeking behavior, the average magnitude of change was 12.1% (range: 2–22%). The same correlation did not appear to exist in the this group ($t_2=1.13$, $p=0.46$), though the sample size is limited. One subject experienced no change in either direction from the stimulation.

Table 4

No stim	pre-Choice	Change
1.00	0.83	blue-0.16
0.75	0.60	blue-0.15
0.71	0.57	blue-0.14
0.57	0.38	blue-0.18
0.52	0.40	blue-0.12
0.44	0.56	red0.11
0.35	0.38	red0.02
0.28	0.17	blue-0.11
0.23	0.23	0.00
0.21	0.20	blue-0.01
0.11	0.33	red0.22

Lastly, we explored whether stimulation had an effect on subject’s reaction time performance. We found that there was no overall main effect of stimulation epoch ($F_{2,577}=1.37$, $p=0.25$; ANOVA) or wager ($F_{1,577}=0.31$, $p=0.57$; ANOVA) on reaction time. However, we did find an interaction effect between stimulation epoch and wager ($F_{2,577}=4.15$, $p=0.01$; Figure 3f). Specifically, during the choice period stimulation condition, reaction times were faster when subjects placed a high wager compared with a low wager ($t_{22}=3.72$, $p=0.001$), supporting previous findings (Frank et al., 2007). No similar differences were observed for the omitted and fixation stimulation conditions.

We used a multi-modal approach consisting of single-neuronal recordings and intermittent stimulation to characterize the neurophysiological role of the STN in decision-making under uncertainty. To do so, we used a financial decision-making task designed to interrogate risk-taking behavior. Using this task, we categorized trials into high and low-uncertainty. We defined high-uncertainty as trials in which the probability of a positive and negative outcome are equal. As a result there was no optimal behavioral strategy. Conversely, low-uncertainty trials were cases in which the outcome was heavily biased towards or against a positive outcome. On these trials, subject behavior was reliably stereotyped towards the most appropriate wager to maximize gains or minimize losses. We found that on high-uncertainty trials STN neural activity encoded the upcoming decision in a discrete 500 ms temporal window immediately before the choice period. In a recent functional imaging study, Fleming et al. found a bilateral increase in BOLD response localized to the STN selectively for high-uncertainty trials where subjects responded against a status-quo bias. Although their study uses perceptual decisions, we demonstrate that the same underlying mechanism may extend to value-based decisions made under conditions of uncertainty. Other studies have observed similar neural responses in the STN following conflict related encoding (Cavanagh et al., 2011; Zaghloul et al., 2012) and control signal encoding (Isoda and Hikosaka, 2008; Wiecki and Frank, 2013).Unfortunately, the task design in the present study does not let us dissect the influence of conflict, control, and uncertainty on the observed neural responses reported here. This remains an open question within the STN literature body. It is worth noting the possibility that the observed STN neural response in this experiment is a combination of conflict and control. More specifically, a departure from a prepotent response (i.e. placing a high wager) induces STN activity and allows for the recruitment of control centers to mediate a new decision. This would be supported by computational models (Wiecki and Frank, 2013) and experimental data (Coulthard et al., 2012) but require further investigation to tease apart.

Furthermore, Cavanaugh et al. have previously shown that increases in local-field potential oscillations in the medial prefrontal cortex and STN correlate with trial-by-trial decision conflict and that continuous electrical stimulation through implanted DBS electrodes can prevent adjustments in decision thresholds ultimately resulting in rapid or impulsive decision-making (Cavanagh et al., 2011). In contrast, in our data the application of intermittent DBS prior to the choice period resulted in an increasein risk-averse decisions and in reaction times for those decision. One potential explanation for these seemingly conflicting findings is that the effects of stimulation may vary depending on the duration of stimulation. It has previously been suggested that short bursts of high-frequency STN stimulation serve to increase local firing rates which are subsequently silenced with prolonged stimulation (Lee et al., 2009). Continuous, high-frequency stimulation has also been proposed to act as an informational lesion, essentially overwriting the normal time-varying activity of the target (Herrington et al., 2016b). Our finding also appears to correspond to the observed neurophysiological data from this study, where a slight increase in overall STN activity during the choice period correlates with placing a low wager. An interesting limitation of the stimulation study is that stimulation was only delivered on the high-uncertainty trials, limiting our ability to understand the constraints of its effect on modifying behavior on medium- or low-uncertainty trials. We would hypothesize the effect would be limited or not present on low-uncertainty trials given that no differential encoding was observed, however this remains to be studied.

Interestingly, our findings differ from other human neurophysiology studies in which conflict activity was observed during the stimulus presentation, as opposed to the choice period, in the dorsal anterior cingulate (Sheth et al., 2012) and the STN (Zaghloul et al., 2012). To further explore the temporal dynamics of the observed signal, we performed a second experiment in which we applied intermittent STN stimulation through implanted DBS electrodes selectively during high-uncertainty trials. Stimulation was delivered either during the fixation period, choice period, or it was omitted. This technique is uniquely different than previous studies using DBS as a method to interrogate neural circuits because we implemented a system for rapidly turning on and off the implanted device, permitting us to time-lock delivery to specific task-epochs. This approach may further reduce confounding effects of long-term stimulation, such as carry-over effects. As a result, we found that intermittent stimulation prior to the choice period—the same interval during which we observed the neurophysiological decision signal from the first experiment—selectively altered subject behavior. No differences were observed in subject behavior when stimulation was omitted or delivered during the fixation period. We found that stimulation prior to the choice period interrupted subjects’ ability to appropriately slow responses when betting against their bias (i.e. when they placed a high bet), resulting in a shortened reaction time, consistent with previous work (Frank et al., 2007; Cavanagh et al., 2011).

Although we attempted to reduce confounding effects by demonstrating both physiological and stimulation evidence to support our claim, our experimental design has several fundamental limitations. In the first experiment, we perform intraoperative recordings in patients undergoing a neurosurgical procedure. Naturally, there are several limitations for performing studies in the operating room, such as the length of each experimental session. For this reason, the total number of trials and neurons we are able to record can often be limited. In this study, we focus our neurophysiological findings to population responses. Despite this limitation, however, we find the reported effects to be consistent across the population. In addition, we compensate for this limitation by developing a novel stimulation method to carefully test the relationship between our neurophysiological findings and subject behavior. Furthermore, the subjects in this study all suffer from advanced PD, a disease known to affect natural reward processing. For obvious reasons, these experiments are constrained to populations requiring neurosurgical treatment, and direct comparisons to a healthy population are limited to behavioral measures.

In conclusion, we provide functional imaging and neurophysiological evidence in human subjects demonstrating the critical role of the STN in encoding decisions under conditions of uncertainty. Moreover, we demonstrate that electrical stimulation of the STN within a finite temporal window can selectively bias subject behavior towards more risk-averse decisions. Together, this provides evidence for the role of precision neuromodulation approaches and closed-loop deep brain stimulation for the advancement of neurological and neuropsychiatric therapies.

Data curation and sharing is currently in process for the entire DARPA SUBNETS project. The underlying data for Figures 1, 2, and 3 in this study have been provided.

1. 1. Asaad WF
   2. Eskandar EN

   (2008a) A flexible software tool for temporally-precise behavioral control in matlab

   Journal of Neuroscience Methods 174:245–258.

   https://doi.org/10.1016/j.jneumeth.2008.07.014

   • PubMed
   • Google Scholar
2. 1. Asaad WF
   2. Eskandar EN

   (2008b) Achieving behavioral control with millisecond resolution in a high-level programming environment

   Journal of Neuroscience Methods 173:235–240.

   https://doi.org/10.1016/j.jneumeth.2008.06.003

   • PubMed
   • Google Scholar
3. 1. Asaad WF
   2. Santhanam N
   3. McClellan S
   4. Freedman DJ

   (2013) High-performance execution of psychophysical tasks with complex visual stimuli in MATLAB

   Journal of Neurophysiology 109:249–260.

   https://doi.org/10.1152/jn.00527.2012

   • PubMed
   • Google Scholar
4. 1. Cavanagh JF
   2. Wiecki TV
   3. Cohen MX
   4. Figueroa CM
   5. Samanta J
   6. Sherman SJ
   7. Frank MJ

   (2011) Subthalamic nucleus stimulation reverses mediofrontal influence over decision threshold

   Nature Neuroscience 14:1462–1467.

   https://doi.org/10.1038/nn.2925

   • PubMed
   • Google Scholar
5. 1. Claassen DO
   2. van den Wildenberg WP
   3. Ridderinkhof KR
   4. Jessup CK
   5. Harrison MB
   6. Wooten GF
   7. Wylie SA

   (2011) The risky business of dopamine agonists in parkinson disease and impulse control disorders

   Behavioral Neuroscience 125:492–500.

   https://doi.org/10.1037/a0023795

   • PubMed
   • Google Scholar
6. 1. Coulthard EJ
   2. Bogacz R
   3. Javed S
   4. Mooney LK
   5. Murphy G
   6. Keeley S
   7. Whone AL

   (2012) Distinct roles of dopamine and subthalamic nucleus in learning and probabilistic decision making

   Brain 135:3721–3734.

   https://doi.org/10.1093/brain/aws273

   • PubMed
   • Google Scholar
7. 1. Drapier D
   2. Drapier S
   3. Sauleau P
   4. Haegelen C
   5. Raoul S
   6. Biseul I
   7. Peron J
   8. Lallement F
   9. Rivier I
   10. Reymann JM
   11. Edan G
   12. Verin M
   13. Millet B

   (2006) Does subthalamic nucleus stimulation induce apathy in Parkinson's disease?

   Journal of Neurology 253:1083–1091.

   https://doi.org/10.1007/s00415-006-0177-0

   • PubMed
   • Google Scholar
8. 1. Eitan R
   2. Shamir RR
   3. Linetsky E
   4. Rosenbluh O
   5. Moshel S
   6. Ben-Hur T
   7. Bergman H
   8. Israel Z

   (2013) Asymmetric right/left encoding of emotions in the human subthalamic nucleus

   Frontiers in Systems Neuroscience 7:69.

   https://doi.org/10.3389/fnsys.2013.00069

   • PubMed
   • Google Scholar
9. 1. Fleming SM
   2. Thomas CL
   3. Dolan RJ

   (2010) Overcoming status quo Bias in the human brain

   Proceedings of the National Academy of Sciences 107:6005–6009.

   https://doi.org/10.1073/pnas.0910380107

   • PubMed
   • Google Scholar
10. 1. Frank MJ
    2. Samanta J
    3. Moustafa AA
    4. Sherman SJ

    (2007) Hold your horses: impulsivity, deep brain stimulation, and medication in parkinsonism

    Science 318:1309–1312.

    https://doi.org/10.1126/science.1146157

    • PubMed
    • Google Scholar
11. 1. Frank MJ

    (2006) Hold your horses: a dynamic computational role for the subthalamic nucleus in decision making

    Neural Networks 19:1120–1136.

    https://doi.org/10.1016/j.neunet.2006.03.006

    • PubMed
    • Google Scholar
12. 1. Herrington TM
    2. Cheng JJ
    3. Eskandar EN

    (2016a) Mechanisms of deep brain stimulation

    Journal of Neurophysiology 115:19–38.

    https://doi.org/10.1152/jn.00281.2015

    • PubMed
    • Google Scholar
13. 1. Herrington TM
    2. Cheng JJ
    3. Eskandar EN

    (2016b) Mechanisms of deep brain stimulation

    Journal of Neurophysiology 115:19–38.

    https://doi.org/10.1152/jn.00281.2015

    • PubMed
    • Google Scholar
14. 1. Herz DM
    2. Zavala BA
    3. Bogacz R
    4. Brown P

    (2016) Neural Correlates of Decision Thresholds in the Human Subthalamic Nucleus

    Current Biology 26:916–920.

    https://doi.org/10.1016/j.cub.2016.01.051

    • PubMed
    • Google Scholar
15. 1. Isoda M
    2. Hikosaka O

    (2008) Role for subthalamic nucleus neurons in switching from automatic to controlled eye movement

    Journal of Neuroscience 28:7209–7218.

    https://doi.org/10.1523/JNEUROSCI.0487-08.2008

    • PubMed
    • Google Scholar
16. 1. Le Jeune F
    2. Drapier D
    3. Bourguignon A
    4. Péron J
    5. Mesbah H
    6. Drapier S
    7. Sauleau P
    8. Haegelen C
    9. Travers D
    10. Garin E
    11. Malbert CH
    12. Millet B
    13. Vérin M

    (2009) Subthalamic nucleus stimulation in Parkinson disease induces apathy: a PET study

    Neurology 73:1746–1751.

    https://doi.org/10.1212/WNL.0b013e3181c34b34

    • PubMed
    • Google Scholar
17. 1. Le Jeune F
    2. Péron J
    3. Biseul I
    4. Fournier S
    5. Sauleau P
    6. Drapier S
    7. Haegelen C
    8. Drapier D
    9. Millet B
    10. Garin E
    11. Herry JY
    12. Malbert CH
    13. Vérin M

    (2008) Subthalamic nucleus stimulation affects orbitofrontal cortex in facial emotion recognition: a PET study

    Brain 131:1599–1608.

    https://doi.org/10.1093/brain/awn084

    • PubMed
    • Google Scholar
18. 1. Lee KH
    2. Blaha CD
    3. Garris PA
    4. Mohseni P
    5. Horne AE
    6. Bennet KE
    7. Agnesi F
    8. Bledsoe JM
    9. Lester DB
    10. Kimble C
    11. Min HK
    12. Kim YB
    13. Cho ZH

    (2009) Evolution of Deep Brain Stimulation: Human Electrometer and Smart Devices Supporting the Next Generation of Therapy

    Neuromodulation: Technology at the Neural Interface 12:85–103.

    https://doi.org/10.1111/j.1525-1403.2009.00199.x

    • PubMed
    • Google Scholar
19. 1. Mallet L
    2. Polosan M
    3. Jaafari N
    4. Baup N
    5. Welter ML
    6. Fontaine D
    7. du Montcel ST
    8. Yelnik J
    9. Chéreau I
    10. Arbus C
    11. Raoul S
    12. Aouizerate B
    13. Damier P
    14. Chabardès S
    15. Czernecki V
    16. Ardouin C
    17. Krebs MO
    18. Bardinet E
    19. Chaynes P
    20. Burbaud P
    21. Cornu P
    22. Derost P
    23. Bougerol T
    24. Bataille B
    25. Mattei V
    26. Dormont D
    27. Devaux B
    28. Vérin M
    29. Houeto JL
    30. Pollak P
    31. Benabid AL
    32. Agid Y
    33. Krack P
    34. Millet B
    35. Pelissolo A
    36. STOC Study Group

    (2008) Subthalamic nucleus stimulation in severe obsessive-compulsive disorder

    New England Journal of Medicine 359:2121–2134.

    https://doi.org/10.1056/NEJMoa0708514

    • PubMed
    • Google Scholar
20. 1. Patel SR
    2. Sheth SA
    3. Martinez-Rubio C
    4. Mian MK
    5. Asaad WF
    6. Gerrard JL
    7. Kwon CS
    8. Dougherty DD
    9. Flaherty AW
    10. Greenberg BD
    11. Gale JT
    12. Williams ZM
    13. Eskandar EN

    (2013) Studying task-related activity of individual neurons in the human brain

    Nature Protocols 8:949–957.

    https://doi.org/10.1038/nprot.2013.050

    • PubMed
    • Google Scholar
21. 1. Patel SR
    2. Sheth SA
    3. Mian MK
    4. Gale JT
    5. Greenberg BD
    6. Dougherty DD
    7. Eskandar EN

    (2012) Single-neuron responses in the human nucleus accumbens during a financial decision-making task

    Journal of Neuroscience 32:7311–7315.

    https://doi.org/10.1523/JNEUROSCI.0027-12.2012

    • PubMed
    • Google Scholar
22. 1. Schmidt R
    2. Berke JD

    (2017) A Pause-then-Cancel model of stopping: evidence from basal ganglia neurophysiology

    Philosophical Transactions of the Royal Society B: Biological Sciences 372:20160202.

    https://doi.org/10.1098/rstb.2016.0202

    • PubMed
    • Google Scholar
23. 1. Schmidt R
    2. Leventhal DK
    3. Mallet N
    4. Chen F
    5. Berke JD

    (2013) Canceling actions involves a race between basal ganglia pathways

    Nature Neuroscience 16:1118–1124.

    https://doi.org/10.1038/nn.3456

    • PubMed
    • Google Scholar
24. 1. Schroeder U
    2. Kuehler A
    3. Haslinger B
    4. Erhard P
    5. Fogel W
    6. Tronnier VM
    7. Lange KW
    8. Boecker H
    9. Ceballos-Baumann AO

    (2002) Subthalamic nucleus stimulation affects striato-anterior cingulate cortex circuit in a response conflict task: a PET study

    Brain 125:1995–2004.

    https://doi.org/10.1093/brain/awf199

    • PubMed
    • Google Scholar
25. 1. Seinstra M
    2. Wojtecki L
    3. Storzer L
    4. Schnitzler A
    5. Kalenscher T

    (2016) No effect of subthalamic deep brain stimulation on intertemporal Decision-Making in parkinson patients

    eNeuro 3:ENEURO.0019–16.

    https://doi.org/10.1523/ENEURO.0019-16.2016

    • Google Scholar
26. 1. Seymour B
    2. Barbe M
    3. Dayan P
    4. Shiner T
    5. Dolan R
    6. Fink GR

    (2016) Deep brain stimulation of the subthalamic nucleus modulates sensitivity to decision outcome value in Parkinson's disease

    Scientific Reports 6:32509.

    https://doi.org/10.1038/srep32509

    • PubMed
    • Google Scholar
27. 1. Sheth SA
    2. Mian MK
    3. Patel SR
    4. Asaad WF
    5. Williams ZM
    6. Dougherty DD
    7. Bush G
    8. Eskandar EN

    (2012) Human dorsal anterior cingulate cortex neurons mediate ongoing behavioural adaptation

    Nature 488:218–221.

    https://doi.org/10.1038/nature11239

    • PubMed
    • Google Scholar
28. 1. St Onge JR
    2. Abhari H
    3. Floresco SB

    (2011) Dissociable contributions by prefrontal D1 and D2 receptors to risk-based decision making

    Journal of Neuroscience 31:8625–8633.

    https://doi.org/10.1523/JNEUROSCI.1020-11.2011

    • PubMed
    • Google Scholar
29. 1. Wiecki TV
    2. Frank MJ

    (2013) A computational model of inhibitory control in frontal cortex and basal ganglia

    Psychological Review 120:329–355.

    https://doi.org/10.1037/a0031542

    • PubMed
    • Google Scholar
30. 1. Wylie SA
    2. Ridderinkhof KR
    3. Elias WJ
    4. Frysinger RC
    5. Bashore TR
    6. Downs KE
    7. van Wouwe NC
    8. van den Wildenberg WP

    (2010) Subthalamic nucleus stimulation influences expression and suppression of impulsive behaviour in Parkinson's disease

    Brain 133:3611–3624.

    https://doi.org/10.1093/brain/awq239

    • PubMed
    • Google Scholar
31. 1. Zaehle T
    2. Wagenbreth C
    3. Voges J
    4. Heinze HJ
    5. Galazky I

    (2017) Effects of deep brain stimulation of the subthalamic nucleus on perceptual decision making

    Neuroscience 343:140–146.

    https://doi.org/10.1016/j.neuroscience.2016.11.044

    • PubMed
    • Google Scholar
32. 1. Zaghloul KA
    2. Weidemann CT
    3. Lega BC
    4. Jaggi JL
    5. Baltuch GH
    6. Kahana MJ

    (2012) Neuronal activity in the human subthalamic nucleus encodes decision conflict during action selection

    Journal of Neuroscience 32:2453–2460.

    https://doi.org/10.1523/JNEUROSCI.5815-11.2012

    • PubMed
    • Google Scholar
33. 1. Zénon A
    2. Duclos Y
    3. Carron R
    4. Witjas T
    5. Baunez C
    6. Régis J
    7. Azulay JP
    8. Brown P
    9. Eusebio A

    (2016) The human subthalamic nucleus encodes the subjective value of reward and the cost of effort during decision-making

    Brain 139:1830–1843.

    https://doi.org/10.1093/brain/aww075

    • PubMed
    • Google Scholar

Author details

1. Shaun R Patel

   1. Department Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, United States
   2. Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing

   For correspondence

   shaun.patel@me.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9334-2522

2. Todd M Herrington

   Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Formal analysis, Investigation, Writing—review and editing

   Competing interests

   No competing interests declared

3. Sameer A Sheth

   Department of Neurosurgery, Baylor College of Medicine, Houston, United States

   Contribution

   Data curation, Investigation, Writing—review and editing

   Competing interests

   No competing interests declared

4. Matthew Mian

   Department Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Data curation, Investigation

   Competing interests

   No competing interests declared

5. Sarah K Bick

   Department Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Data curation, Writing—review and editing

   Competing interests

   No competing interests declared

6. Jimmy C Yang

   Department Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Data curation, Writing—review and editing

   Competing interests

   No competing interests declared

7. Alice W Flaherty

   1. Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, United States
   2. Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Data curation, Writing—review and editing

   Competing interests

   No competing interests declared

8. Michael J Frank

   Department of Cognitive, Linguistic and Psychological Sciences, Brown University, Providence, United States

   Contribution

   Formal analysis, Investigation, Writing—review and editing

   Competing interests

   Senior editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0001-8451-0523

9. Alik S Widge

   Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, United States

   Contribution

   Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8510-341X

10. Darin Dougherty

    Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, United States

    Contribution

    Data curation, Funding acquisition, Writing—review and editing

    Competing interests

    No competing interests declared

11. Emad N Eskandar

    Department Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston, United States

    Contribution

    Data curation, Funding acquisition, Investigation, Writing—review and editing

    Competing interests

    No competing interests declared

• 1,474

  views

• 192

  downloads

• 13

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.