BRS1005_proof ■ 14 February 2017 ■ 1/7

Brain Stimulation xxx (2017) 1e7

55
Contents lists available at ScienceDirect 56
57
58
Brain Stimulation 59
60
journal homepage: http://www.journals.elsevier.com/brain-stimulation 61
62
63
64
65
1 The human globus pallidus internus is sensitive to rewards e Evidence 66
2 67
3 Q3 from intracerebral recordings 68
4 69
5 Q2 Thomas F. Münte a, b, *, 1, Josep Marco-Pallares c, d, 1, Seza Bolat a, Marcus Heldmann a, 70
6 € tz Lütjens e, Wido Nager f, Kirsten Müller-Vahl g, Joachim K. Krauss e
Go 71
7 72
a
8 Department of Neurology, University of Lübeck, Lübeck, Germany
b 73
Institute of Psychology II, University of Lübeck, Lübeck, Germany
9 c 74
Department of Basic Psychology, University of Barcelona, Barcelona, Spain
10 d
Cognition and Brain Plasticity Group, Bellvitge Biomedical Research Institute - IDIBELL, L'Hospitalet de Llobregat, Barcelona, Spain 75
11 e
Department of Neurosurgery, Medical School Hannover, Hannover, Germany 76
12 f €n Klinik Eilbek, Hamburg, Germany
Scho
g
77
13 Department of Psychiatry, Medical School Hannover, Hannover, Germany
78
14 79
15 80
16 a r t i c l e i n f o a b s t r a c t 81
17 82
18 Article history: Background: The globus pallidus internus (GPi) is the final output relay of the basal ganglia for the
Received 3 September 2016 83
19 control of movements but has also been shown to belong to a second pathway projecting to the lateral
Received in revised form habenula. This latter pathway is related to reward processing. 84
20 18 December 2016 85
Method: This prompted us to record, in eight patients receiving deep brain stimulation of the GPi for the
21 Accepted 7 January 2017 86
Available online xxx
alleviation of various movement disorders, local field potentials (LFP) while these patients performed a
22 lottery task. The task entailed choosing between a higher and a lower number, which changed their color 87
23 after the patient's choice with red (green) signaling a loss (win, in Euro cents) corresponding to the 88
Keywords:
24 chosen number. 89
Globus pallidus internus
25 Deep brain stimulation Results: Surface recordings showed a feedback related negativity from a frontal midline site, while time 90
26 Reward domain averages in the GPi showed differential modulation depending on the valence of the stimulus 91
27 Oscillations with polarity inversion indicating that this reward-modulated activity was indeed generated locally. 92
28 Local field potentials Furthermore, wavelet decomposition of the LFP showed a reward-related response in the high beta/low
93
29 gamma range.
94
Conclusion: We conclude that human GPi is involved in reward processing, possibly in relation to the
30 95
lateral habenula.
31 96
© 2017 Elsevier Inc. All rights reserved.
32 97
33 98
34 99
35 100
36 101
37 1. Introduction projects to the motor part of the thalamus which connects to 102
38 cortical output areas as well as to subcortical motor structures [5]. 103
39 The globus pallidus internus (GPi) is the final output relay of the The motor functions of the GPi are undisputed and in fact this 104
40 basal ganglia (BG) for the control of movements [1,2]. The “standard structure has become a target for deep brain stimulation (DBS) in a 105
41 model” of the BG posits that two main projections from the stria- variety of motor disorders such as Parkinson's disease [6], dystonia 106
42 tum team up to facilitate the appropriate cortical motor command, [7], and more recently Huntington's disease [8]. In PD and dystonia, 107
43 while suppressing competing response tendencies [3,4]. Both, the DBS effectively ameliorates motor dysfunctions likely by modifying 108
44 direct and the indirect pathway converge on the GPi which in turn altered neuronal activity. Recently, a number of investigations in 109
45 macaque monkeys have targeted another pathway passing through 110
46 the GPi [9] which is related to reward processing. Besides to the 111
47 thalamus, the GPi also projects to the lateral habenula (LHb) located 112
Abbreviations: LFP, local field potential; GPi, globus pallidus internus; DBS, deep
48 medially over the posterior thalamus [10]. The seminal studies of 113
brain stimulation.
49 * Corresponding author. Department of Neurology, University of Lübeck, Ratze- Hikosaka and co-workers [11,12] have demonstrated that this 114
50 burger Allee 160, 23538 Lübeck, Germany. pathway is used to encode nonmotor signals linked to expected 115
51 E-mail address: Thomas.muente@neuro.uni-luebeck.de (T.F. Münte). rewards. Using antidromic stimulation, Hong and Hikosaka (2008)
1 116
Equal contribution.
52 117
53 118
http://dx.doi.org/10.1016/j.brs.2017.01.004
54 1935-861X/© 2017 Elsevier Inc. All rights reserved. 119

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 2/7

2 T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7

1 first identified GPi neurons that projected to the LHb. In a second were implanted in all patients bilaterally in the posteroventral 66
2 step, the behavior of these neurons was studied in a saccade task lateral “motor” portion of GPi guided by CT-stereotactic surgery, 67
3 involving rewards. Indeed, LHb-projecting GPi neurons showed magnetic resonance imaging and microelectrode recording. All 68
4 firing patterns that varied as a function of expected rewards. Tourette patients and two patients with dystonia were implanted 69
5 Anatomical studies in monkeys have suggested that LHb-projecting under general anesthesia because of the severity of their movement 70
6 neurons are distinct from other motor-related GPi neurons [13]. disorder. Preliminary target coordinates were x ¼ 20, y ¼ 3, and 71
7 According to these studies, LHb projecting neurons amount to z ¼ 0, related to the midcommissural point. In all patients the 72
8 about 10% of the total neuronal population of the GPi. Using a Medtronic model 3387 (Medtronic Neurological Division, MN, USA) 73
9 different approach, Karadi and co-workers [14,15] recorded single DBS electrode was used which harbors 4 contacts of 1.27 mm 74
10 neuron activity of the GP in rats and rhesus monkeys to the diameter and 1.5 mm length spaced 1 mm apart. Postoperative 75
11 administration of glucose, as well as gustatory, and olfactory stereotactic CT confirmed appropriate DBS electrode placement 76
12 stimuli. In monkeys, visual presentation of food and nonfood ob- with the lowest active contact located at 17.4e22 mm lateral to the 77
13 jects was also used. In rats as well as monkeys about 15% of the GPi intercommissural line, 2e4 mm anterior to, and 2e4 mm below the 78
14 neurons were glucose-sensitive, while 19% and 16% of the cells midcommissural point in standard Montreal Neurological Institute 79
15 responded to gustatory and olfactory stimulation, respectively. (MNI) stereotactic space coordinates [21]. There were no intra- 80
16 Obviously, these responses can be viewed as being reward-related operative or postoperative adverse events. All patients benefitted 81
17 and the location of glucose-sensitive neurons was similar to the from chronic stimulation with improvement in the corresponding 82
18 LHb-projecting GPi neurons identified by Hong and Hikosaka scores. 83
19 [11,12]. 84
20 In the present investigation, we recorded from DBS electrodes 2.2. Stimuli 85
21 which were temporarily externalized prior to implantation of the 86
22 pulse generator while the patient is performing a cognitive task. To We used an established gambling task [16,17] in which the 87
23 further investigate the role of the GPi in reward processing, we used numbers 5 and 25 were presented in white on a black background 88
24 a simple lottery task that has been used for electrophysiological and in one of the possible orders, [5 25] or [25 5]. Participants selected 89
25 neuroimaging studies [16e20]. This task requires subjects to select one of the numbers by pressing a spatially corresponding button 90
26 the smaller or the greater of two numbers displayed on a computer with the left or right index finger. One second after the choice, one 91
27 screen. A change of color of the numbers indicates whether the of the numbers turned green, while the other changed to red. If the 92
28 chosen number led to a monetary win (change to green) or a loss number selected by the participant changed to red (green), this 93
29 (change to red) for the subject. The change of color, i.e. the feedback signaled a loss (gain) of the corresponding amount of money (in 94
30 signal indicating wins or losses, is the critical reward indicating Euro cents). Two seconds later, a new pair of numbers appeared on 95
31 stimulus in this type of study. In averaged scalp surface event- the screen for the next trial. Patients were provided with an initial 96
32 related potentials (ERPs) losses are associated with a phasic nega- sum of 10 V and were encouraged to gain as much as possible. They 97
33 tive component peaking at about 280 ms post onset of the feedback were familiarized with the task during a brief practice block. Pa- 98
34 signal, variously termed medial frontal negativity or feedback tients took part in 5e15 blocks of 40 trials each. The probability to 99
35 related negativity (FRN). This FRN is superimposed on a positive win/lose was 50%. The experiment also contained trials in which 100
36 going component present for both, wins and losses. Furthermore, the numbers not only changed color (to indicate wins and losses) 101
37 time-frequency analyses has revealed an increase of theta power but also doubled in size. These “boost” trials were included to 102
38 for losses and an increase of beta-power for wins [16,17]To the provide unexpected wins and losses (for a similar approach: [17]) 103
39 extent to which the GPi is sensitive to rewards we expected a which are known from animal experiments to lead to prominent 104
40 change both in the oscillatory and averaged local field potential modulations of dopaminergic activity [22]. The boost trials 105
41 (LFP). occurred randomly interspersed with standard trials in 12% of the 106
42 trials. The number of trials was insufficient to compute reliable 107
43 2. Method average LFPs and time-frequency analyses, however. Therefore, we 108
44 will not report data from the boost trials. Recording sessions 109
45 All procedures were approved by the ethical review board of differed somewhat in duration between patients, depending on 110
46 Hannover Medical School. The study was performed in compliance their clinical condition. 111
47 with the Declaration of Helsinki. 112
48 113
2.3. Data acquisition and analysis
49 2.1. Patients, surgery, and electrode localization 114
50 115
We obtained bipolar recordings between different contacts of
51 Eight patients were included in the study (mean age 35.5, range 116
the DBS-electrodes.2 Data were sampled at 1000 points/s and data
52 17e54). All patients were right-handed. Of these, 4 patients suf- 117
analysis was performed using EEG-Lab [24] and ERP-Lab (http://
53 fered from Tourette syndrome (mean age 32.2, range 19e47, dis- 118
www.erpinfo.org/erplab/erplab-toolbox) software. To increase
54 ease duration 4.5e37 years), one had generalized dystonia (age 17, 119
signal-to-noise ratio, we averaged over small (5 cents) and large (25
55 disease duration 17 years), 2 had cervical dystonia (age 17 and 57, 120
cents) wins and small and large losses, from 100 to 1000 ms after
56 disease duration 15 years in one patient, unclear in the other) and 121
57 one patient had severe dystonic tremor (age 54, disease duration 12 122
58 years). In the Tourette patients, the Yale Global Tic Severity Scale 2
123
An important issue with regard to bipolar recordings of LFPs, discussed for
59 (YGTSS) scores were between 12 and 23 for the motor part example by Herrojo-Ruiz et al. [23], is that polarity cannot be defined due to the
124
60 (maximum possible score: 25) and between 32 and 44 for the total measurement of the “differential” electrical potential from 2 contacts located in or 125
61 tic score (maximum possible: 50). Medication at the time of the near the target structure. Moreover, depending on the precise position of the 126
62 recording included antipsychotics (3 patients), serotonin reuptake recording sites, a polarity reversal between subjects might occur. In the present set 127
of data we carefully examined the single patients' data and found that in spite of
63 inhibitors (2), benzodiazepines (2), and anticholinergics(1).Patients 128
this potential problem very similar findings could be obtained in the different
64 gave informed consent prior to participation in the study which patients owing to the very stable localization of the contacts. Therefore we found it 129
65 had been approved by the local ethics committee. DBS electrodes justified to obtain group averages for the different bipolar derivations. 130

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 3/7

T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7 3

1 66
2 67
3 68
4 69
5 70
6 71
7 72
8 73
9 74
10 75
11 76
12 77
13 78
14 79
15 80
16 81
17 82
18 83
19 84
20 85
21 86
22 87
23 88
24 89
25 90
26 91
27 92
28 93
29 94
30 95
31 96
32 97
33 Fig. 1. Time domain averages of surface activity (Fz referenced to the mean activity at the two mastoid processes) and local field potentials from the GPI for Tourette (left) and 98
34 Dystonia patients (right). To account for inter-individual amplitude differences, data were normalized as described in Mas-Herrero et al. [26]. Time 0 marks the time of the color 99
35 change, i.e. the time at which participants knew whether or not they had won or lost the sum indicated by the chosen number. 100
36 101
37 102
38 the feedback. Trials with amplitudes greater than þ-100 mV were permutations between the two conditions (2^8 ¼ 256) and only 103
39 automatically rejected. those clusters with p < 0.05 corrected were reported. 104
40 In addition to standard analysis in the time-domain, we also Surface EEG was recorded in all patients from Fz (all patients) 105
41 examined time-frequency behavior of the electrical activity (see and Cz (4 patients) electrodes referenced to the average activity at 106
42 Ref. [16] for a more extensive description). To this end 4 s epochs the two mastoid processes. Because of the dressing of the wound, 107
43 were generated (2000 ms before and after the stimulus-onset). we did not record from more surface electrodes in the current 108
44 Single-trial data was convoluted using a complex Morlet wavelet: study. Surface EEG activity was analysed as described for the LFP. 109
45 110
46 3. Results 111
47  1=2 t 2 112
48 wðt; f0 Þ ¼ 2ps2t e2s2
t e2ipf0 t 3.1. Behavioral data 113
49 114
50 with the relation sf0f (where sf ¼ 4ps
1 ) set to 6.7 [25]. Changes in
t
Participants selected the higher number more often than the 115
51 time varying energy (square of the convolution between wavelet lower number (67.4% SD 17.1%, comparison with chance level, 116
52 and signal) in the studied frequencies (from 1 Hz to 40 Hz; linear t(7) ¼ 2.87, p < 0.05). The mean reaction time of participants was 117
53 increase) with respect to baseline (300 to 0 ms with regard to 1717 ms SD 1123 ms with no difference between selection for the 118
54 stimulus onset) were computed for each trial and averaged sepa- high and low numbers (t(7) ¼ 0.97, n.s.). 119
55 rately for each stimulus class. 120
56 Given the considerable inter-individual amplitude we normal- 3.2. Averaged LFP 121
57 ized the activity of the two conditions for each participant. Each 122
58 time point was transformed into z-scores by subtracting the mean Fig. 1 and Fig. 2 show the LFPs for the bipolar derivations be- 123
59 and dividing by the standard deviation of the two conditions [26]. tween contacts 2 and 0 (top) and the difference between contacts 3 124
60 We then determined the point-by-point statistical t-test among and 2 (bottom) for the right and left GPi in addition to the surface 125
61 conditions. ERP from the Fz electrode for the two groups (Fig. 1) and group 126
62 In the time-frequency data, in order to address the multiple- averaged including statistics (Fig. 2). The specific derivations (2e0, 127
63 comparison problem, we used the nonparametric permutation 3e2) were chosen to include a common contact (2) and to provide 128
64 test proposed by Maris and Oostenveld [27], clustering sensor- evidence for a phase reversal. A mean of 98 trials (range: 21e296) 129
65 time-frequency points (two-sided). We performed all possible per patient were averaged for the gain condition, whereas 94 trials 130

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 4/7

4 T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7

1 which yielded significant differences between gain and loss con- 66

2 ditions (right panel, p < 0.05 corrected for multiple comparisons). 67
3 68
4 4. Discussion 69
5 70
6 The current set of data provides clear evidence for a reward 71
7 sensitivity of the human GPi echoing previous studies in monkeys 72
8 [9,11,12]. Averaged LFPs showed differences between gain and loss 73
9 trials bilaterally, i.e. irrespective of the response hand (which was 74
10 always the right hand in the current study). This and the fact that 75
11 the effects clearly appear about 400 ms after the choice response 76
12 shows that these differences cannot be attributed to motor pro- 77
13 cesses. The polarity reversal of the gain/loss difference in the 78
14 averaged LFP between derivation 2e0 and 3-2 demonstrates that 79
15 this valence effect must be generated locally in the GPi or in close 80
16 vicinity. Thus, valence of a decision outcome is coded in the GPi. 81
17 A similar conclusion can be reached based on the wavelet- 82
18 transformed time-frequency analyses. Here, a burst of high beta/ 83
19 low gamma activity could be observed following gains but not 84
20 losses in the 400e600 ms time-window. Beta activity as a response 85
21 to positive outcomes has also been observed in recordings from the 86
22 Ncl. accumbens [28]. 87
23 In a recent review [18], we have discussed the functional role of 88
24 high beta and have proposed that this activity might serve the 89
25 coupling of fronto-striatal areas. Unexpected and/or highly relevant 90
26 positive outcomes lead to a power increase for high beta which 91
27 might serve this coupling. Indeed, a number of studies have pro- 92
28 vided evidence that this activity has a role in the functional 93
29 coupling of distant areas [29]. These findings have been corrobo- 94
30 rated by computational models [30] that demonstrated that oscil- 95
31 lations in the beta range (<30 Hz) can serve to synchronize distant 96
32 Fig. 2. Time domain averages of surface activity (Fz referenced to the mean activity at brain areas. Gamma oscillations (30e80 Hz) on the other hand 97
33 the two mastoid processes) and local field potentials from the GPI for all the partici- appear to synchronize activity at a local level. Based on a model by 98
34 pants. To account for inter-individual amplitude differences, data were normalized as Axmacher et al. [31] we hypothesized that high beta/low gamma 99
35 described in Mas-Herrero et al. [26]. Time 0 marks the time of the color change, i.e. the activity transmits a motivational value signal from medial frontal to 100
time at which participants knew whether or not they had won or lost the sum indi-
36 reward related brain structures. 101
cated by the chosen number. The surface ERP showed a typical feedback-related
37 negativity. The LFP from the GPI showed significant modulations as a function of It is important to distinguish between different stages of reward 102
38 reward valence. Waveforms were baseline-corrected with regard to the mean ampli- processing: The delivery of a reward (outcome of an action) is an 103
39 tude in the time-window 100 to 0 ms and statistically compared as described in the important event. Often, however, a reward does not come unex- 104
40 methods. The bars underneath each diagram mark the time-windows in which LFPs to 105
pected but is presignalled by cues. Thus, an expectation phase
win and loss trials were significantly different. Please observe the polarity reversal for
41 the GPI electrodes. precedes the outcome. Animal research has revealed that midbrain 106
42 dopaminergic neurons show an increase in firing following a 107
43 reward (see e.g. Refs. [32,33]). Importantly, this increase in firing 108
44 (range: 18e286) were obtained in the loss condition. The surface can be anticipated to follow a cue after Pavlovian conditioning [34]. 109
45 ERP showed a typical finding, i.e. a more prominent negativity for In humans, fMRI has been used to uncover the extensive neural 110
46 the losses relative to the wins that corresponds to the FRN network underlying reward expectation in humans: Robust acti- 111
47 component. A distinct bilateral pattern of LFP modulations as a vations were obtained in orbitofrontal cortex (OFC), anterior 112
48 function of gain and loss was apparent in the GPi. Significant dif- cingulate cortex (ACC), supplementary motor area (SMA), insula, 113
49 ferences emerged between the two conditions bilaterally for the postcentral gyrus, medial temporal gyrus (MTG), precuneus, infe- 114
50 derivation 2e0 between 500 and 600 ms (Fig. 2). In addition, there rior parietal lobule (IPL), superior parietal lobule (SPL), thalamus, 115
51 was a clear polarity inversion of the gain/loss effect between the putamen, caudate, globus pallidus and nucleus accumbens (NAcc) 116
52 two derivations which was significant on the right side. Such a [35e45]. Interestingly, the structures processing reward outcome 117
53 phase reversal indicates that the electrophysiological activity are very similar to those found for the processing reward antici- 118
54 originates in the target structure or in close vicinity and cannot be pation: prefrontal cortex (PFC), ACC, inferior frontal gyrus (IFG), 119
55 due to far distance volume conduction. insula, posterior cingulate cortex (PCC), fusiform gyrus, amygdala, 120
56 thalamus, globus pallidus and striatum [36,42,46e49]. In the pre- 121
57 sent study, we focused on reward outcome, as the probabilistic 122
58 nature of the task did not allow for the development of grounded 123
59 3.3. Time frequency analysis of LFP expectations. Hikosaka's group [9,11,12] on the other hand showed 124
60 that GPi neurons coded reward expectations. Thus, future experi- 125
61 The results from the time frequency analysis are depicted in ments need to address reward expectation as well using patients 126
62 Fig. 3 for the bipolar derivation between contacts 2 and 0 where receiving DBS of the GPi. 127
63 differences between gain and loss trials emerged most clearly. The Obviously, an important question to answer in future studies is 128
64 most prominent feature of the time-frequency plot is a burst of high the relation of the current findings to the series of studies on 129
65 beta/low gamma (20e30 Hz) around 400e600 ms for the gains, reward-sensitive neurons in monkeys from the Hikosaka group 130

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 5/7

T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7 5

1 66
2 67
3 68
4 69
5 70
6 71
7 72
8 73
9 74
10 75
11 76
12 77
13 78
14 79
15 80
16 81
17 82
18 83
19 84
20 85
21 86
22 87
23 88
24 89
25 90
26 91
27 92
28 93
29 94
30 95
31 96
32 97
33 98
34 99
35 100
36 101
37 102
38 103
39 Fig. 3. Time frequency analysis of the GPI data (bipolar derivation between contacts 2 and 0). The third column shows the difference in power for gain minus loss trials. Time 104
0 marks the time of the color change of the numbers, i.e. the time at which participants knew whether or not they had won or lost the sum indicated by the chosen number. Clearly,
40 105
gains were associated with increased high beta/low gamma activity between 400 and 600 ms. The rightmost column shows significant differences as revealed by permutation
41 testing (see methods).
106
42 107
43 108
44 [9,11,12] which, at this point, is suggestive but not proven. The GPI Ref. [54] for first evidence). One challenge of such studies is the 109
45 neurons in the monkey studies were identified by antidromic limited time allowed for such recordings in a clinical setting (see, 110
46 stimulation in the lateral habenula, were only a minority of the GPI [28]), but clearly an extension of our research to Parkinson patients, 111
47 neurons and located near the border of the GPI. Wickens [50] asked in whom intraoperative recordings could be obtained, is warranted. 112
48 whether these GPI cells transmit a reward-negative signal to the Another way to pinpoint the role of the GPi-lateral habenula 113
49 habenula. He further postulated reward-positive cells in the GPI as pathway in humans would be connectivity analysis in fMRI [55,56]. 114
50 well. A demonstration of such cells in humans would require micro- A further task for future studies would be to systematically map 115
51 recordings which are only possible during the operation before reward-related responses with regard to electrode locations. Sub- 116
52 replacement of the initial electrode by the stimulation electrode. divisions of the GPI have been discussed on the basis of differential 117
53 Because of their severe movement disorders, the majority of the response of motor symptoms to stimulation of different parts of the 118
54 patients included in the current study were operated while under GPI. In dystonia, it has been shown that stimulation in ventral GPI is 119
55 general anesthesia. Therefore it was not possible to obtain micro- more effective for dystonic symptoms than more dorsal stimulation 120
56 recordings intraoperatively from this group of patients in order to [57,58], whereas clinical studies in Parkinson's disease found that 121
57 establish the relationship between the LFP and single cell re- stimulation of ventral regions of the GPI were more beneficial for 122
58 sponses. Comparative recordings with micro- and macro- rigidity and the alleviation of dyskinesias, while dorsal stimulation 123
59 electrodes have been done, for example, in Parkinson's patients improved akinesia but worsened dyskinesias [59]. Such functional 124
60 from the subthalamic nucleus [51] and the GPI [52] and in a patient subdivisions have recently also been suggested based on diffusion- 125
61 with depression for a reward task from the ventral striatum [53]. tensor-imaging based tractography [60] with regard to motor 126
62 One task for future studies would therefore be to perform similar projections. The limited number of patients in the current study 127
63 reward-processing experiments intraoperatively in Parkinson pa- does not allow to determine the relationship between reward- 128
64 tients and to compare micro- and macroelectrode results (see related responses in the GPI and electrode location. Based on 129
65 130

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 6/7

6 T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7

1 observations reporting the development of impulse control disor- [2] Alexander GE, Crutcher MD. Functional architecture of basal ganglia circuits: 66
neural substrates of parallel processing. Trends Neurosci 1990;13:266e71.
2 ders after GPi stimulation in patients suffering from Parkinson's 67
[3] Albin RL, Young AB, Penney JB. The functional anatomy of basal ganglia dis-
3 Disease there is some evidence for a causal role of the GPI in reward orders. Trends Neurosci 1989;12:366e75. 68
4 processing [61]. In order to investigate this causal role systemati- [4] Mink J. The basal ganglia: focused selection and inhibition of competing motor 69
5 cally a further approach could be to employ bipolar stimulation of programs. Prog Neurobiol 1996;50:381e425. 70
[5] Parent M, Le vesque M, Parent A. The pallidofugal projection system in pri-
6 the different possible combination of contacts in order to interfere mates: evidence for neurons branching ipsilaterally and contralaterally to the 71
7 with reward processing. As animal studies particularly stress a thalamus and brainstem. J Chem Neuroanat 1999;16:153e65. 72
8 negative reward signal that might be used to learn action e [6] Odekerken VJ, van Laar T, Staal MJ, Mosch A, Hoffmann CF, Nijssen PC, et al. 73
Subthalamic nucleus versus globus pallidus bilateral deep brain stimulation
9 outcome relations, paradigms such as reversal learning [62] and for advanced Parkinson's disease (NSTAPS study): a randomised controlled
74
10 stimulus-response learning [63] might significantly add to our trial. Lancet Neurol 2013;12:37e44. 75
11 understanding of the GPI's function in reward processing. [7] Volkmann J, Mueller J, Deuschl G, Kühn AA, Krauss JK, Poewe W, et al. Pallidal 76
neurostimulation in patients with medication-refractory cervical dystonia: a
12 randomised, sham-controlled trial. Lancet Neurol 2014;13:875e84.
77
13 5. Limitations [8] Gonzalez V, Cif L, Biolsi B, Garcia-Ptacek S, Seychelles A, Sanrey E, et al. Deep 78
14 brain stimulation for Huntington's disease: long-term results of a prospective 79
open-label study. J Neurosurg 2014;121:114e22.
15 As with every technique, the recording of LFPs from DBS pa- 80
[9] Hikosaka O, Nakamura K, Nakahara H. Basal ganglia orient eyes to reward.
16 tients has certain limitations. These days, many patients are oper- J Neurophysiol 2006;95:567e84. 81
17 ated on in a single session, i.e. the pulse generator is implanted
[10] Lecourtier L, Kelly PH. A conductor hidden in the orchestra? Role of the 82
habenular complex in monoamine transmission and cognition. Neurosci
18 right away. This precludes larger sample sizes, which in principle 83
Biobehav Rev 2007;31:658e72.
19 would be desirable. Also, it is obvious that the target structure is [11] Hong S, Hikosaka O. The globus pallidus sends reward-related signals to the 84
20 used as a site for stimulation because of a known, as is the case, e.g., lateral habenula. Neuron 2008;60:720e9. 85
21 [12] Hong S, Hikosaka O. Diverse sources of reward value signals in the basal 86
with the subthalamic nucleus in Parkinson's disease, or suspected, ganglia nuclei transmitted to the lateral habenula in the monkey. Front Hum
22 as is the case, e.g., with the GPi in Tourette's syndrome, dysfunction Neurosci 2013;7:778. 87
23 of the structure. At the same time, experiments are performed to [13] Parent M, Levesque M, Parent A. Two types of projection neurons in the in- 88
24 ternal pallidum of primates: single-axon tracing and threedimensional 89
gain insight into the normal cognitive function of the target reconstruction. J Comp Neurol 2001;439:162e75.
25 structure. In this situation, it might be advantageous to include a [14] Karadi Z, Faludi B, Lenard L, Czurko  A, Niedetzky C, Vida I, et al. Glucose- 90
26 rather heterogeneous group of patients sharing the target structure sensitive neurons of the globus pallidus: II. Complex functional attributes. 91
27 Brain Res Bull 1995;37:157e62. 92
but only partially sharing the pathophysiology as was done In the [15] Lena
rd L, Karadi Z, Faludi B, Czurko  A, Niedetzky C, Vida I, et al. Glucose-
28 present study. For obvious regions it will be impossible to obtain sensitive neurons of the globus pallidus: I. Neurochemical characteristics.
93
29 data from normal healthy participants which is why the current Brain Res Bull 1995;37:149e55. 94
30 [16] Marco-Pallares J, Cucurell D, Cunillera T, García R, Andre s-Pueyo A, Münte TF, 95
technique should be supplemented by surface electrophysiology et al. Human oscillatory activity associated to reward processing in a gambling
31 and functional imaging. 96
task. Neuropsychologia 2008;46:241e8.
32 [17] Marco-Pallare s J, Cucurell D, Cunillera T, Kra
€mer UM, C amara E, Nager W, et al. 97
33 Genetic variability in the dopamine system (dopamine receptor D4, catechol- 98
34 Contribution of authors O-methyltransferase) modulates neurophysiological responses to gains and
99
losses. Biol Psychiatry 2009;66:154e61.
35 [18] Marco-Pallare s J, Münte TF, Rodríguez-Fornells A. The role of high-frequency 100
36 Thomas F. Münte e Conceived the study, wrote the manuscript. oscillatory activity in reward processing and learning. Neurosci Biobehav Rev 101
37 Josep Marco-Pallares e Performed parts of the analysis, critically 2015;49:1e7. 102
reviewed the manuscript. [19] Gehring WJ, Willoughby AR. The medial frontal cortex and the rapid pro-
38 cessing of monetary gains and losses. Science 2002;295:2279e82. 103
39 Seza Bolat e Recorded parts of the data, performed clinical [20] Camara E, Kr€ amer UM, Cunillera T, Marco-Pallare s J, Cucurell D, Nager W, et al. 104
40 testing, critically reviewed the manuscript. The effects of COMT (Val108/158Met) and DRD4 (SNP -521) dopamine ge- 105
41 Marcus Heldmann e Recorded parts of the data, performed notypes on brain activations related to valence and magnitude of rewards.
106
Cereb Cortex 2010;20:1985e96.
42 parts of the analysis, critically reviewed the manuscript. €necker T, Kupsch A, Kühn AA, Schneider GH, Hoffmann KT. Automated 107
[21] Scho
Go€ tz Lütjens e Recruited the patients, performed clinical as- optimization of subcortical cerebral MR imaging-atlas coregistration for
43 108
44 sessments, critically reviewed the manuscript. improved postoperative electrode localization in deep brain stimulation. Am J
109
Wido Nager - Recorded parts of the data, performed clinical Neuroradiol 2009;30:1914e21.
45 [22] Schultz W, Tremblay L, Hollerman JR. Reward processing in primate orbito- 110
46 testing, critically reviewed the manuscript. frontal cortex and basal ganglia. Cereb Cortex 2000;10:272e84. 111
Joachim K. Krauss - Conceived the study, recruited the patients, [23] Herrojo Ruiz M, Huebl J, Scho € necker T, Kupsch A, Yarrow K, Krauss JK, et al.
47 112
critically reviewed the manuscript. Involvement of human internal globus pallidus in the early modulation of
48 cortical error-related activity. Cereb Cortex 2014;24:1502e17. 113
49 All authors have approved the final version of the manuscript. [24] Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single- 114
trial EEG dynamics including independent component analysis. J Neurosci
50 115
Meth 2004;134:9e21.
51 Declaration of interest [25] Tallon-Baudry C, Bertrand O, Delpuech C, Pernier J. Oscillatory gamma-band 116
52 (30e70 Hz) activity induced by a visual search task in humans. J Neurosci 117
53 The authors declare that they have no financial interests 1997;17:722e34. 118
[26] Mas-Herrero E, Zatorre RJ, Rodriguez-Fornells A, Marco-Pallare s J. Dissociation
54 involved in the preparation of this article. between musical and monetary reward responses in specific musical anhe- 119
55 donia. Curr Biol 2014;24:699e704. 120
56 [27] Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and MEG- 121
Acknowledgments data. J Neurosci Meth 2007;164:177e90.
57 [28] Münte TF, Heldmann M, Hinrichs H, Marco-Pallares J, Kr€ amer UM, Sturm V,
122
58 TFM and MH are supported by the DFG and the BMBF. JMP is et al. Nucleus accumbens is involved in human action monitoring: evidence 123
59 Q1 from invasive electrophysiological recordings. Front Hum Neurosci 2008;1:11. 124
supported by a grant from the Spanish government (PSI2012- [29] Steriade M. Grouping of brain rhythms in corticothalamic systems. Neuro-
60 37472). science 2006;137:1087e106.
125
61 [30] Kopell N, Ermentrout GB, Whittington MA, Traub RD. Gamma rhythms and 126
62 beta rhythms have different synchronization properties. Proc Natl Acad Sci U. 127
References S. A 2000;97:1867e72.
63 128
[31] Axmacher N, Cohen MX, Fell J, Haupt S, Dümpelmann M, Elger CE, et al.
64 [1] DeLong MR. Activity of pallidal neurons during movement. J Neurophysiol Intracranial EEG correlates of expectancy and memory formation in the hu- 129
65 1971;34:414e27. man hippocampus and nucleus accumbens. Neuron 2010;65:541e9. 130

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004
 BRS1005_proof ■ 14 February 2017 ■ 7/7

T.F. Münte et al. / Brain Stimulation xxx (2017) 1e7 7

1 [32] Schultz W, Apicella P, Scarnati E, Ljungberg T. Neuronal activity in monkey [50] Wickens J. Toward an anatomy of disappointment: reward-related signals 37
ventral striatum related to the expectation of reward. J Neurosci 1992;12: from the globus pallidus. Neuron 2008;60:530e1.
2 38
4595e610. [51] Trottenberg T, Fogelson N, Kühn AA, Kivi A, Kupsch A, Schneider GH, et al.
3 [33] Schultz W, Apicella P, Ljungberg T, Romo R, Scarnati E. Reward-related activity Subthalamic gamma activity in patients with Parkinson's disease. Exp Neurol 39
4 in the monkey striatum and substantia nigra. Prog Brain Res 1993;99:227e35. 2006;200:56e65. 40
5 [34] Schultz W, Dayan P, Montague PR. A neural substrate of prediction and [52] Weinberger M, Hutchison WD, Alavi M, Hodaie M, Lozano AM, Moro E, et al. 41
reward. Science 1997;275:1593e9. Oscillatory activity in the globus pallidus internus: comparison between
6 [35] Cooper JC, Knutson B. Valence and salience contribute to nucleus accumbens Parkinson's disease and dystonia. Clin Neurophysiol 2012;123:358e68. 42
7 activation. NeuroImage 2008;39:538e47. [53] Lega BC, Kahana MJ, Jaggi J, Baltuch GH, Zaghloul K. Neuronal and oscillatory 43
8 [36] Dillon DG, Holmes AJ, Jahn AL, Bogdan R, Wald LL, Pizzagalli DA. Dissociation activity during reward processing in the human ventral striatum. Neuroreport 44
of neural regions associated with anticipatory versus consummatory phases of 2011;22:795e800.
9 incentive processing. Psychophysiol 2008;45:36e49. [54] Howell NA, Prescott IA, Lozano AM, Hodaie M, Voon V, Hutchison WD. Pre-
45
10 [37] Ernst M, Nelson EE, McClure EB, Monk CS, Munson S, Eshel N, et al. Choice liminary evidence for human globus pallidus pars interna neurons signaling 46
11 selection and reward anticipation: an fMRI study. Neuropsychologia 2004;42: reward and sensory stimuli. Neuroscience 2016;328:30e9. 47
1585e97. [55] Jocham G, Klein TA, Ullsperger M. Differential modulation of reinforcement
12 [38] Galvan A, Hare TA, Davidson M, Spicer J, Glover G, Casey BJ. The role of ventral learning by D2 dopamine and NMDA glutamate receptor antagonism.
48
13 frontostriatal circuitry in reward-based learning in humans. J Neurosci J Neurosci 2014;34:13151e62. 49
14 2005;25:8650e6. [56] Ullsperger M, von Cramon DY. Error monitoring using external feedback: 50
[39] Izuma K, Saito DN, Sadato N. Processing of social and monetary rewards in the specific roles of the habenular complex, the reward system, and the cingulate
15 51
human striatum. Neuron 2008;58:284e94. motor area revealed by functional magnetic resonance imaging. J Neurosci
16 [40] Knutson B, Adams CM, Fong GW, Hommer DW. Anticipation of increasing 2003;23:4308e14. 52
17 monetary reward selectively recruits nucleus accumbens. J Neurosci 2001;21: [57] Houeto JL, Yelnik J, Bardinet E, Vercueil L, Krystkowiak P, Mesnage V, et al. 53
RC159. Acute deep-brain stimulation of the internal and external globus pallidus in
18 54
[41] O'Doherty JP, Deichmann R, Critchley HD, Dolan RJ. Neural responses during primary dystonia: functional mapping of the pallidum. Arch Neurol 2007;64:
19 anticipation of a primary taste reward. Neuron 2002;33:815e26. 1281e6. 55
20 [42] Preuschoff K, Bossaerts P, Quartz SR. Neural differentiation of expected [58] Tisch S, Zrinzo L, Limousin P, Bhatia KP, Quinn N, Ashkan K, et al. Effect of 56
21 reward and risk in human subcortical structures. Neuron 2006;51:381e90. electrode contact location on clinical efficacy of pallidal deep brain stimula- 57
[43] Rademacher L, Krach S, Kohls G, Irmak A, Gründer G, Spreckelmeyer KN. tion in primary generalised dystonia. J Neurol Neurosurg Psychiatry 2007;78:
22 Dissociation of neural networks for anticipation and consumption of mone- 1314e9. 58
23 tary and social rewards. NeuroImage 2010;49:3276e85. [59] Krack P, Pollak P, Limousin P, Hoffmann D, Benazzouz A, Le Bas JF, et al. 59
24 [44] Yacubian J, Gl€
ascher J, Schroeder K, Sommer T, Braus DF, Büchel C. Dissociable Opposite motor effects of pallidal stimulation in Parkinson's disease. Ann 60
systems for gain- and loss-related value predictions and errors of prediction Neurol 1998;43:180e92.
25 in the human brain. J Neurosci 2006;26:9530e7. [60] Rozanski VE, Vollmar C, Cunha JP, Tafula SM, Ahmadi SA, Patzig M, et al.
61
26 [45] Ye Z, Hammer A, C amara E, Münte TF. Pramipexole modulates the neural Connectivity patterns of pallidal DBS electrodes in focal dystonia: a diffusion 62
27 network of reward anticipation. Hum Brain Map 2011;32:800e11. tensor tractography study. Neuroimage 2014;84:435e42. 63
[46] Carter CS, Braver TS, Barch DM, Botvinick MM, Noll DC, Cohen JD. Anterior [61] Moum SJ, Price CC, Limotai N, Oyama G, Ward H, Jacobson C, et al. Effects of
28 cingulate cortex, error detection, and the online monitoring of performance. STN and GPi deep brain stimulation on impulse control disorders and dopa-
64
29 Science 1998;280:747e9. mine dysregulation syndrome. PLoS One 2012;7:e29768. 65
30 [47] Fujiwara J, Tobler PN, Taira M, Iijima T, Tsutsui K-I. Segregated and integrated [62] Hauser TU, Hunt LT, Iannaccone R, Walitza S, Brandeis D, Brem S, et al. 66
coding of reward and punishment in the cingulate cortex. J Neurophysiol Temporally dissociable contributions of human medial prefrontal subregions
31 67
2009;10:3284e93. to reward-guided learning. J Neurosci 2015;35:11209e20.
32 [48] Liu X, Powell DK, Wang H, Gold BT, Corbly CR, Joseph JE. Functional dissoci- [63] Schroll H, Horn A, Gro€schel C, Brück C, Brücke C, Lütjens G, et al. Differential 68
33 ation in frontal and striatal areas for processing of positive and negative contributions of the globus pallidus and ventral thalamus to stimulus- 69
reward information. J Neurosci 2007;27:4587e97. response learning in humans. Neuroimage 2015;122:233e45.
34 70
[49] Nieuwenhuis S, Slagter HA, von Geusau NJA, Heslenfeld DJ, Holroyd CB.
35 Knowing good from bad: differential activation of human cortical areas by 71
36 positive and negative outcomes. Eur J Neurosci 2005;21:3161e8. 72

Please cite this article in press as: Münte TF, et al., The human globus pallidus internus is sensitive to rewards e Evidence from intracerebral
recordings, Brain Stimulation (2017), http://dx.doi.org/10.1016/j.brs.2017.01.004