Research Articles, Behavioral/Cognitive

Chronic Deep Brain Stimulation of the Human Nucleus Accumbens Region Disrupts the Stability of Intertemporal Preferences

Ben J. Wagner, Canan B. Schüller, Thomas Schüller, Juan C. Baldermann, Sina Kohl, Veerle Visser-Vandewalle, Daniel Huys, Milena Marx, Jens Kuhn and Jan Peters
Journal of Neuroscience 25 October 2023, 43 (43) 7175-7185; https://doi.org/10.1523/JNEUROSCI.0138-23.2023

Abstract

When choosing between rewards that differ in temporal proximity (intertemporal choice), human preferences are typically stable, constituting a clinically relevant transdiagnostic trait. Here we show, in female and male human patients undergoing deep brain stimulation (DBS) of the anterior limb of the internal capsule/NAcc region for treatment-resistant obsessive-compulsive disorder, that long-term chronic (but not phasic) DBS disrupts intertemporal preferences. Hierarchical Bayesian modeling accounting for temporal discounting behavior across multiple time points allowed us to assess both short-term and long-term reliability of intertemporal choice. In controls, temporal discounting was highly reliable, both long-term (6 months) and short-term (1 week). In contrast, in patients undergoing DBS, short-term reliability was high, but long-term reliability (6 months) was severely disrupted. Control analyses confirmed that this effect was not because of range restriction, the presence of obsessive-compulsive disorder symptoms or group differences in choice stochasticity. Model-agnostic between- and within-subject analyses confirmed this effect. These findings provide initial evidence for long-term modulation of cognitive function via DBS and highlight a potential contribution of the human NAcc region to intertemporal preference stability over time.

SIGNIFICANCE STATEMENT Choosing between rewards that differ in temporal proximity is in part a stable trait with relevance for many mental disorders, and depends on prefrontal regions and regions of the dopamine system. Here we show that chronic deep brain stimulation of the human anterior limb of the internal capsule/NAcc region for treatment-resistant obsessive-compulsive disorder disrupts the stability of intertemporal preferences. These findings show that chronic stimulation of one of the brain's central motivational hubs can disrupt preferences thought to depend on this circuit.

Introduction

Humans continuously maneuver their world weighting the future against the present. The degree of temporal discounting of future rewards (as assessed via intertemporal choice tasks) is at least in part a stable trait (Kirby, 2009) with relevance for many mental disorders (Amlung et al., 2019). For example, steep discounting of value over time is a hallmark of substance use disorders (Bickel et al., 2014; Amlung et al., 2017), gambling disorder (Dixon et al., 2003; Wiehler and Peters, 2015), and a range of other disorders (Amlung et al., 2019; Lempert et al., 2019).

Multiple neural systems contribute to human self-control, including PFC regions involved in cognitive control, and regions of the mesolimbic and mesocortical dopamine system (e.g., ventral striatum and ventromedial PFC) involved in reward valuation (Peters and Büchel, 2011; Bartra et al., 2013; McClure and Bickel, 2014). Earlier studies focused on characterizing potentially dissociable striatal and prefrontal value signals during intertemporal choice (McClure et al., 2004; Kable and Glimcher, 2007). This debate has ultimately led to a revised view of self-control, according to which lateral PFC exerts top-down control over ventromedial PFC in support of self-controlled choices (Hare et al., 2009; Figner et al., 2010; Peters and D'Esposito, 2016), where both striatal reward responses (Hariri et al., 2006) and cortico-striatal connectivity (van den Bos et al., 2014) are associated with temporal discounting in cross-sectional analyses. Despite these findings and direct evidence for a contribution of the ventral striatum to intertemporal choice in rodents (Pisansky et al., 2019), direct evidence in humans is lacking.

At the same time, deep brain stimulation (DBS) to different subcortical target sites has gained increasing attention in psychiatry in the context of different mental disorders (Lee et al., 2019; Krauss et al., 2021). This includes, for example, depression (Dandekar et al., 2018), obsessive-compulsive disorder (OCD) (Kohl et al., 2014; Kuhn and Baldermann, 2020; Wu et al., 2021), substance use disorders (Peisker et al., 2018; Mahoney et al., 2020), and schizophrenia (Nucifora et al., 2019). While clinical outcomes are in many cases promising, one oftentimes neglected issue concerns potentially more subtle cognitive effects that might arise as a result of chronic stimulation.

Here we address this issue by capitalizing on the rare opportunity to longitudinally follow patients undergoing therapeutic DBS of the anterior limb of the internal capsule (ALIC)/NAcc region for treatment-resistant OCD. OCD is assumed to be associated with a dysregulation in fronto-striatal circuits (Robbins et al., 2019), which might be normalized via ALIC/NAcc DBS (Figee et al., 2013; Smolders et al., 2013; Smith et al., 2020; Wu et al., 2021). In the context of a DBS treatment-efficacy study (Huys et al., 2019), we comprehensively examined short-term (acute) and long-term (chronic) effects of DBS on intertemporal preference. Analyses focused on three aspects. First, we analyzed group differences in temporal discounting between patients with OCD and controls, which previously yielded heterogeneous results. Second, we directly tested for effects of acute DBS on intertemporal choice by comparing ON versus OFF stimulation periods in patients. Finally, effects of long-term (i.e., chronic) DBS on intertemporal choice were examined by comparing long-term reliability estimates of temporal discounting between controls and patients undergoing DBS. These analyses leveraged hierarchical Bayesian parameter estimation using generative models to directly estimate posterior distributions of test-retest reliabilities, separately for each group, and were corroborated by model-agnostic control analyses.

Materials and Methods

Participants

All participants provided written informed consent before participation, and the study procedure was approved by the Ethics Committee of the Medical Faculty of the University of Cologne. Table 1 provides an overview of demographics data for patients and controls.

View this table:
Table 1.

Demographic data per groupa

Overall procedure

Female and male human patients with OCD and matched controls (for inclusion criteria and demographics, see below) completed three separate testing sessions, where they completed a temporal discounting task (see below). One initial testing session (T1) corresponded to the pre-DBS session in patients, and the first testing session in controls. This was followed by two follow-up sessions at T2, T2.1, and T2.2. The interval between T1 and T2 sessions in days was matched between groups (mean [range] patients: 203 [155-260], controls: 206 [164-247]). The T2.1 and T2.2 sessions were spaced within a week. Here, patients performed DBS on versus off sessions in counterbalanced order with at least 24 h wash-out period, whereas controls performed simply two seperate testing sessions. A subset of patients (n = 7) completed all three testing sessions. Two additional patients completed T1 but only one of the two T2 sessions, yielding a total n = 9 for pre- versus post-DBS analyses in patients. Eight additional patients were only tested at T2, yielding a total of n = 15 for on versus off DBS analyses.

In total, n = 28 controls participated (see below). One control participant missed the final testing session, yielding n = 27 for the corresponding analyses. The Covid-19 pandemic broke out while this study was being conducted. Therefore, 10 of the controls completed testing at T2 online, as opposed to in the laboratory (for control analyses, see below).

Patients with OCD

Patients with OCD eligible for DBS had to meet the DSM-IV criteria for OCD, a Yale-Brown Obsessive Compulsive Scale >25, at least one cognitive-behavioral therapy (minimum of 45 sessions), at least two unsuccessful treatments with a serotonin reuptake inhibitor and one unsuccessful augmentation with lithium, neuroleptics, or buspirone. Patients were excluded because of drug, medication or alcohol abuse, suicidal ideation, mental retardation, pregnancy or breastfeeding, and schizophrenia. Disease duration was on average 27.59 ± 13.02 years (range 6-48 years). The mean age at onset for OCD was 16.2 ± 9.25 years. Individual patient clinical data are provided in Table 2.

View this table:
Table 2.

Overview of sex, age, disease duration before surgery, and stimulation parameters (monopolar, all bilateral, except for Patient 13) of DBS patients with OCD

Controls

In total, n = 28 controls participated in the study (for demographic data, see Table 1). Exclusion criteria were drug, medication or alcohol abuse, or the diagnosis of a psychiatric disorder. Controls were screened for OCD-symptoms via the Obsessive-Compulsive Inventory-Revised (OCI-R) questionnaire (Foa et al., 2002). Here 9 of 28 subjects scored above the threshold (>21) for possible OCD.

Temporal discounting task

Before the first testing session, participants completed a short adaptive pretest to estimate the individual discount rate (k). This discount rate was used to construct a set of 140 participant-specific trials using MATLAB (version 8.4.0, The MathWorks). The primary task then consisted of choices between an immediate smaller-sooner (SS) reward of 20€ and participant specific larger-but-later (LL) rewards delivered after some delay (1, 2, 7, 14, 30, 90, or 180 d). In 70 trials, LL amounts were uniformly spaced between 20.5€ and 80€. In the remaining 70 trials, LL amounts were uniformly spaced around each estimated indifference point per delay (based on the pretest discount rate). In case indifference points were >80€, only uniformly spaced LL amounts were used. Trials were presented in a pseudorandomized order. Following each task completion, one trial was randomly selected and paid out immediately in cash (in case of a SS choice) or via a timed bank transfer (LL choice).

DBS

DBS was applied to the ALIC/NAcc region. Details on electrode placement (including reconstruction of electrode positions), surgical procedure, and adjustment of stimulation parameters are available elsewhere (Huys et al., 2019). Final stimulation amplitudes ranged from 2.6 to 4.8 V, and pulse-width was set between 60 and 150 µs (Table 2). The frequency of DBS was 130 Hz, except for two patients who received 150 Hz stimulation and one patient who received 90 Hz stimulation.

Computational modeling

To assess intertemporal choice, we applied a standard single parameter hyperbolic discounting model as follows: SV(LLt)=At1 + exp(k) * Dt(1)

Here, A is the numerical reward amount of the LL option on trial t. The discount-rate (k), here modeled in log-space, reflects the steepness of the hyperbolic discounting curve, with greater values corresponding to steeper discounting. Delay D of the LL option is expressed in days. SV then corresponds to the subjective (discounted) value of the delayed (LL) option. We then used softmax action selection (Sutton and Barto, 1998) (Eq. 2) to model the probability of selecting the LL option on trial t. Here, β is an inverse temperature parameter, modeling choice stochasticity. For β = 0, choices are random; and as β increases, choices become more dependent on option values as follows: P(LLt)=exp(β * SV(LLt))exp(β * SV(SS)) + exp(β * SV(LLt))(2)

In a first step, models were fit to all trials from all participants, separately per group and time point, using a hierarchical Bayesian modeling approach. We included separate group-level Gaussian distributions for log(k) and β for patients and controls, and/or T1 and T2 or within T2 (T2.1, T2.2) time points (for information on prior distributions, see below).

Second, we implemented a fully generative approach by drawing discount-rate parameters from multivariate Gaussian distributions, to obtain posterior distributions for the reliability (covariance). That is, for the case of two time points T1 and T2, log(k) values for a given participant i are drawn from a bivariate Gaussian as follows: [log(ki,t1)log(ki,t2)] ∼ MVNormal([kmeant1kmeant2],Sk)(3) with variance-covariance matrix Sk: Sk=(ksd,t1kcovkcovksd,t2)(4)

The diagonal contains the variances and the off-diagonal elements contain the covariances. The test-retest correlation coefficient ρ is directly estimated using the following: Sk[1,2]=ρ[k] * ksd,t1 * ksd,t2(5) with Sk[2,1]=Sk[1,2]

Separate models were used to estimate long-term reliability (T1/pre-DBS and T2/post-DBS in controls and patients, respectively) and short-term reliability (i.e., T2.1/DBS on and T2.2/DBS off), separately in patients and controls.

Hierarchical Bayesian parameter estimation

Parameter estimation was performed using Markov Chain Monte Carlo as implemented in the JAGS software package (Plummer, 2003) (version 4.3) in combination with R (version 3.4) and the R2Jags package. For group-level means, we used uniform priors defined over numerically plausible parameter ranges ([−20, 3] for log(k); [0, 10] for β and [−1,1] for ρ). We initially ran two chains with a varying burn-in period of at least 20,000 samples and thinning of two. Chain convergence was then assessed via the Gelman-Rubinstein convergence diagnostic R̂, and sampling was continued until 1≤R̂≤1.01 for all group-level and individual-subject parameters; 10,000 additional samples were then retained for further analysis. We report posterior group distributions for all parameters of interest as well as their 85% and 95% highest density intervals (HDIs). For group comparisons (patients vs controls) and time effects (T1 vs T2), we report Bayes factors for directional effects for the hyperparameter difference distributions, estimated via kernel density estimation using R (version 4.01) and RStudio (version 1.3).

Analysis of group differences

We report posterior difference distributions and the corresponding 85% and 95% HDIs to characterize differences between patients and controls, changes from T1 to T2 or within T2 (T2.1, T2.2; on/off DBS). We then estimate Bayes factors for directional effects. These were computed as the ratio of the integral of the posterior difference distribution from 0 to +∞ versus the integral from 0 to –∞. Using common criteria (Beard et al., 2016), we considered Bayes factors between 1 and 3 as anecdotal evidence, Bayes factors >3 as moderate evidence, and Bayes factors >10 as strong evidence. Bayes factors >30 and 100 were considered as very strong evidence.

Model-based choice inconsistency

For both model-based and model-agnostic within-participant changes, we leveraged the fact that participants completed the exact same 140 choices during each testing session. To examine model-based changes in preferences, we extracted individual-participant median discount-rates log(k) and decision noise parameters (softmax β) from our hierarchical Bayesian model estimated on T1 data. We then used these parameters to compute choice probabilities for each T1 choice. To examine model-based preference changes from T1 to T2, we then subtracted the T1 choice probability from the corresponding observed choices at T2 (0 for SS and 1 for LL choices). We then computed a choice inconsistency score as the mean of the absolute differences between T1 choice probabilities and T2 choices.

Because controls showed lower decision noise (β) compared with patients, which was also reflected in an overall somewhat lower model fit in patients (Table 3), for additional control analyses, we also matched groups on β.

View this table:
Table 3.

Mean (SD) proportion of correctly predicted choices of the full hierarchical Bayesian model (hyperbolic discounting + softmax + correlation in ρ log(k) and ρ softmax β) that jointly estimates discount rate, softmax β, and the intraindividual correlation of both parametersa

Model agnostic analysis of indifference points

Model-based analyses rely on specific mathematical assumptions regarding the shape of the discounting function. Furthermore, they can be affected by potential group differences in model fit. Therefore, we additionally examined a model-agnostic measure of within-participant changes in preferences. To this end, we fit sigmoid functions (see Eq. 3) to the choice data of each delay D per participant and time point T as follows: P(LLD,T)=11 + exp(−(AD,T−c)*b)(6)

We modeled the probability to choose the delayed reward for delay D at time point T for each participant as a sigmoid function of the absolute LL reward amount A. Here, c is the inflection point of the sigmoid (corresponding to the indifference point, i.e., the point of subjective equivalence between the delayed reward and the immediate reward at the respective delay D), and b is the slope.

For delays with only larger-later choices, the indifference point was set to the midpoint between the immediate reward (20€) and the smallest available LL reward. For delays with only SS choices, the indifference point was conservatively set to max(LL). These rules were also applied in cases where there was only a single noisy LL or SS choice for a given delay. Using this procedure, we computed 196 indifference points in controls and 63 indifference points in patients. Six indifference points in controls and two in patients with OCD could not be estimated. We then computed the mean absolute deviation in indifference points between T1 and T2 as a model-agnostic measure of preference consistency.

Permutation-based group comparisons

Model-based and model-agnostic consistency measures (see previous sections) were compared between groups using permutation tests. To this end, we compared the observed group difference in preference consistency to a null distribution of preference consistency that was obtained by randomly shuffling group labels 10,000 times, and computing the group difference for these shuffled data. Significance was assessed using a two-tailed p value of 0.05.

Parameter recovery analyses

To ensure that the parameters underlying the data-generating process could be recovered using our modeling procedures, we generated 10,000 datasets simulated from the posterior distribution of our generative modeling approach. Thirty of these simulated datasets were randomly selected and refit with the same model (Haines et al., 2020). Parameter recovery was then assessed in two ways. For relevant group-level parameters, log(k) and ρ log(k), we examined whether the estimated 95% highest posterior density intervals contained the true generating parameters. For subject-level parameters, we examined correlation coefficients of generating versus estimated single-subject parameters, pooled across all 30 simulations.

Code and data availability

Model code is available on the Open Science Framework (https://osf.io/53mkd/). Raw choice data are available from Zenodo.org (https://doi.org/10.5281/zenodo.7559218) for researchers meeting the criteria for access to confidential data.

Results

Single-subject choice data, separately for T1 and T2 sessions, are shown in Figure 1 for individual patients with OCD, and in Figure 2 for individual control participants. In a first step, data were then modeled for each time point separately using hierarchical Bayesian parameter estimation using a hyperbolic discounting model with softmax action selection. This model accounted for ≥80% of choices in both patients and controls across all testing sessions (Table 3). In line with previous work (Sohn et al., 2014), patients with OCD exhibited increased discounting (a higher discount rate log(k)), both pre-DBS at T1 (Fig. 3A; the directional Bayes factor [dBF] for a group difference <0 was 35.57; 95% HDI of the difference distribution: −2.28 to 0.04) and post-DBS at T2 (Fig. 3B; dBF [< 0] = 15.85; 95% HDI = −2.95 to 0.34).

Figure 1.
Figure 1.

Single-subject choice data for all n = 9 patients with pre- (T1) and post-DBS data (T2). Green and red points represent LL and SS choices, respectively, across LL amounts (y axis) and delays (x axis). Black circles represent estimated indifference points.

Figure 2.
Figure 2.

Single-subject choice data for all n = 28 controls. Green and red points represent LL and SS choices, respectively, across LL amounts (y axis) and delays (x axis). Black circles represent estimated indifference points.

Figure 3.
Figure 3.

Group-level effects in intertemporal choice. A, At T1 (pre-DBS), patients (n = 9) discounted delayed rewards steeper compared with controls (n = 28) (dBF: controls < patients = 35.75; HDI = −2.28 to 0.04). B, Comparison of pooled T2 sessions in controls (n = 28, T2.1 and T2.2) and pooled DBS on and DBS off sessions at T2 in patients (n = 9). Higher log(k) values in patients were ∼15.85 times more likely than lower values (95% HDI of the difference distribution: −2.95 to 0.34). C, There was no evidence for a systematic change in log(k) across T2 sessions in controls (dBF = 0.73). D, Likewise, there was no systematic effect of acute DBS at T2 in patients (on vs off, n = 15, dBF = 0.90). E, F, If anything, both controls (E, n = 28) and patients (F, n = 9) tended to discount rewards steeper after 6 months (controls: T1 < T2; dBF = 4.15; 95% HDI = −2.22 to 0.90; patients: pre-DBS< post-DBS; dBF = 10.87; 95% HDI = −2.10 to 0.41). All panels represent posterior distributions smoothed via kernel density estimation in R. Thin (thick) horizontal lines indicate the 95% (85%) HDIs.

Controls and patients showed no systematic short-term change in discounting (log(k)) between the two testing sessions at T2 (Fig. 3C,D). Since this comparison corresponds to comparing on versus off DBS sessions in patients, we found no systematic effect of phasic DBS on temporal discounting. With respect to the long-term changes in temporal discounting, if anything, rewards were discounted somewhat steeper after 6 months in controls (Fig. 3E; dBF [<0] = 4.15; 95% HDI = −2.22 to 0.90) or following 6 months of continuous DBS in patients (Fig. 3F; dBF [< 0] = 10.87; 95% HDI = −2.10 to 0.41), a pattern observed previously in healthy control participants (Kirby, 2009).

We applied both inter- and intra-individual analytical approaches to study DBS effects on the stability of intertemporal preferences. First, we used a generative modeling approach, modeling log(k) values across multiple time points using multivariate Gaussian distributions (Haines et al., 2020) (see Materials and Methods). This allowed us to directly estimate a posterior distribution for the test-retest correlation ρ of log(k), separately for patients and controls, and for short-term (T2.1 vs T2.2) and long-term (T1 vs T2) reliability. In controls, the test-retest reliability of log(k) was high, both across T2 sessions (Fig. 4A: 1-week short-term reliability) and between T1 and pooled T2 data (Fig. 4B, 6 month long-term reliability). In both cases, the mean posterior distribution of the reliability coefficient ρ was ≥ 0.8, indicating high reliability. In patients, short-term reliability was likewise high (Fig. 4D). In stark contrast, long-term reliability in patients was completely disrupted (Fig. 4E). For short-term reliability in patients and controls, the mean posterior distribution of the reliability coefficient ρ was ≥ 0.95 and ≥ 0.94 respectively, indicating excellent reliability. In contrast, for long-term reliability in patients, ρ was ≈–0.30, and the 95% HDI overlapped with 0 (Fig. 4F).

Figure 4.
Figure 4.

Reliability of intertemporal choice (discount rate log(k)). A, B, Contour plots of the fitted multivariate Gaussian and individual median discount-rate parameters log(k) for T1 and T2 sessions in controls (A,B) and patients (D,E). x and y axes with regard to the short-term data refer to the comparison within T2. A, Control session T2.1 (x axis) versus T2.2 (y axis). With regard to the long-term data (B,E), axes refer to T1 (x axis) versus pooled T2.1 and T2.2 data (y axis). D, T2.1 (DBS on) versus T2.2 (DBS off) log(k) values in patients at T2. E, In the long-term condition data, x axis refers to discount rates in the T1 (pre-DBS) session; y axis refers to the pooled T2 sessions (post-DBS). C, F, ρ log(k) hyperparameter capturing the correlation of short- and long-term reliability of intertemporal preference in controls (C) and patients (F). All parameters were estimated from our hierarchical Bayesian model. Note the difference in x axis scaling in C, F.

We then conducted a range of control analyses. First, to ensure that results were not affected by the different ranges of log(k) values in the two groups, we drew a subset of n = 9 participants from the control group, matched on the exact range of log(k) values in patients. This subset of controls still exhibited significant long-term reliability (Fig. 5A, mean ρ = 0.55), and the 95% HDI of the posterior distribution of ρ was > 0. Second, to ensure that results are not affected by potential group differences in decision noise (β parameter), we likewise drew a subset of n = 14 controls matched to the patients on β. This subset of controls still exhibited significant long-term reliability (Fig. 5A, mean ρ = 0.77), and the 95% HDI of the posterior distribution of ρ was again > 0 (Fig. 5A). Third, the loss of long-term reliability was also not generally because of the presence of OCD symptoms, as the n = 14 controls with the highest OCI-R scores (in a range similar to some patients with OCD) (e.g., Pinto et al., 2014) still exhibited significant long-term reliability (mean ρ = 0.94) with the 95% HDI again not overlapping with 0.

Figure 5.
Figure 5.

Control analyses. A, Posterior distributions of ρ log(k) (i.e., the test-retest reliability of the discount rate log(k)) for both short-term and long-term reliabilities and various subgroups of the control group. First row, Controls matched on the range of log(k) values of the patient group, to account for effects of log(k) variance on reliability. Second row, Controls matched on decision noise of the patient group to account for effects of decision-noise on reliability. Third row, Controls with high OCI-R scores (mean [range] = 23.57 [14–40) to account for effects of OCD symptoms on reliability. Fourth row, All control participants. B, Difference between long-term and short-term reliability for all control subgroups. C, Group difference (controls vs patients) in short-term versus long-term reliability of all control subsamples. Here, values > 0 reflect lower long-term reliability in patients versus controls.

Figure 6.
Figure 6.

A, Difference in long- versus short-term reliability (i.e., test-retest correlation coefficient ρ log(k)) in controls (green) and patients (blue). In controls, the 95% HDI overlapped with 0 (95% HDI –0.93 –0.02), there was no overlap in patients. B, The group difference of the distributions shown in A. The 95% HDI did not overlap with zero (0.08-1.65; dBF = 36.78), indicating a reliable group difference.

To directly quantify group differences in reliabilities between controls and patients, we next computed differences in long- versus short-term reliability posterior distributions for each group, and then compared these difference distributions across groups. If group differences in long-term versus short-term reliability are similar across groups, this difference distribution should not be different from zero. In contrast, if the disrupted long-term stability in patients is greater than what would be expected, given the control data, this difference distribution should be reliably >0. In all control groups, long-term reliability was numerically lower than short-term reliability (Fig. 5B), but the group difference in this difference was reliably >0 in all cases (Fig. 5C), reflecting a disrupted long-term reliability in the patient group. Figure 6 summarizes these results for the full control sample (n = 28). This shows a much more substantial difference in long- versus short-term reliability in patients (Fig. 6A), resulting in a reliable group difference in reliabilities, with the 95% HDI of the difference distribution not overlapping with 0 (Fig. 6B). There was no difference in short-term reliability between patients and controls. This argues against the idea that differences in reliability are driven or amplified by differences in short-term reliability.

To evaluate parameter recovery of our generative modeling approach, we randomly selected 30 of 10,000 simulated datasets across the entire posterior distribution for both short- and long-term reliability from our patient population. These synthetic datasets were then refitted with the same model. Furthermore, we randomly sampled nine participants from our short-term data to demonstrate that high correlation coefficients can still be accurately recovered, even with a smaller sample size. Our results demonstrate that our model accurately recovers the true parameters across the range of parameters of interest in our original data. Specifically, the generating hyperparameters mostly aligned with refitted 95% HDIs (in 93% of all cases), and we observed high correlations > 0.9 between generating and fitted single-subject parameters. Our findings are summarized in Table 4.

View this table:
Table 4.

Parameter recovery resultsa

In a last step, we tested whether a disruption of preference stability (as reflected in the disruption of long-term reliability) would also manifest at the level of individual decisions. Using both model-based and model-agnostic techniques (via indifference points and standard statistical analyses, see Materials and Methods), preference stability at the individual-participant level was quantified. As a model-based measure of stability, the mean deviation of T2 choice probabilities from predictions of a model fitted to T1 data was computed (Fig. 7A). This inconsistency measure was greater in patients versus controls (permutation test, Fig. 7B). As a model-agnostic measure of stability, the mean absolute deviation of T1 and T2 indifference points was computed (Fig. 7C). This inconsistency measure was again greater in patients versus controls (permutation test, Fig. 7D). In the light of the slightly poorer model fit in patients versus controls (Table 3), which was reflected in overall lower softmax β parameters in patients (Fig. 8A), three different control subgroups were selected based on different cutoff values of β (Fig. 8A). Model-based choice inconsistency for a cutoff of β = 0.5 is shown in Figure 8B. Across three cutoff values, permutation tests again confirmed lower choice consistency in patients versus controls (Fig. 8C–E), illustrating that these effects are not driven by group differences in β.

Figure 7.
Figure 7.

Within-participant changes in intertemporal preferences. A, C, Model-based T1-T2 choice inconsistency (deviation between T2 choices and model-predictions based on a model fitted to T1 data) and model-agnostic (MA) T1-T2 choice inconsistency (mean absolute deviation of indifference points between T1 and T2). B, D, Permutation test for model-based and model-agnostic inconsistency scores, respectively. Histograms represent null distributions of mean group differences across 10,000 randomly shuffled group labels. Red vertical lines indicate observed group differences. Red horizontal line indicates 95% HDI.

Figure 8.
Figure 8.

A, Individual-subject softmax β parameters (mean of T1 and T2 posterior medians) modeling decision noise. To account for effects of group differences in β, in additional analyses, controls matched on β to the patients were selected across various β cut-offs (β = 0.5: red dotted line, n = 17; β = 0.41: green dotted line, n = 14; β = 0.7: blue dotted line, n = 19). B, Individual-subject model-based T1-T2 choice inconsistency in β-matched controls (β cutoff = 0.5) and patients. C–E, Permutation test results for group differences in model-based choice inconsistency (histograms of group differences with 10,000 randomly shuffled group labels) across all cut-offs (C: β = 0.5, D: β = 0.41, E: β = 0.7). In line with the analysis across the whole sample, patients showed significantly increased model-based inconsistency for all cut-offs (C: p = 0.018, D: p = 0.023, E: p = 0.01). C–E, Vertical lines indicate the observed group difference. Horizontal red lines indicate the 95% HDI of the null distribution.

Discussion

We capitalized on the rare opportunity to longitudinally follow a group of patients with OCD receiving DBS to the ALIC/NAcc region, to examine chronic stimulation effects on temporal discounting. Crucially, according to current models of intertemporal choice (Peters and Büchel, 2011) and self-control (Hare et al., 2009; Figner et al., 2010), the ventral striatum/NAcc regions play a central role in this process in both rodents and humans (Da Costa Araújo et al., 2009; Bartra et al., 2013; McClure and Bickel, 2014; Pisansky et al., 2019), but direct evidence for a mechanistic role in human intertemporal choice is lacking. Our data show that chronic DBS to the ALIC/NAcc region disrupts the stability of intertemporal preferences, both reflected in group-level changes in reliability and individual-level changes in preferences.

Control analyses fit in with previously reported increases in temporal discounting and elevated impulsivity in patients with OCD (Sohn et al., 2014; Ong et al., 2019), while contrasting in part with other studies that do not find any group differences between patients and controls (Carlisi et al., 2017; Norman and Carlisi, 2017; Steinglass et al., 2017). Thus, overall studies to date suggest that associations of OCD and intertemporal choice are heterogeneous and possibly related to contributions of specific OCD subtypes or comorbidities (Pinto et al., 2014).

Generative hierarchical Bayesian models then allowed us to estimate posterior distributions of test-retest correlations (Haines et al., 2020), separately for short-term and long-term stability, and for patients and controls. This revealed a reduction in long-term reliability of log(k) in patients undergoing DBS. Notably, this effect was robust to a number of control analyses, controlling for (e.g., effects of the observed range of log(k) values, choice stochasticity effects, and the presence of OCD symptoms). Analyses of within-subject preference stability confirmed these results. Both model-based and model-agnostic measures confirmed a reduction in long-term preference stability in patients undergoing DBS compared with controls, and these analyses were again confirmed when groups were matched on choice stochasticity.

These findings suggests that, in addition to short-term plasticity processes (Kuhn and Baldermann, 2020), long-term ALIC/NAcc DBS (Denys et al., 2010) can interfere with the expression of intertemporal preferences that are thought to rely on this same circuitry (Hariri et al., 2006; Kable and Glimcher, 2007; Peters and Büchel, 2011; McClure and Bickel, 2014). While earlier reports noted effects of acute stimulation on risk-taking and impulsivity (Luigjes et al., 2011; Nachev et al., 2015) (albeit with acute block-wise stimulation protocols that differ substantially from our approach, where ON and OFF sessions were separated by 24 h wash-out periods), our longitudinal analysis revealed changes only following prolonged stimulation and does not suggest a specific direction of change. Rather, the data might suggest a role of the ALIC/NAcc region in maintaining preference stability over time. However, our data do not rule out the possibility that these effects are unspecific, in the sense that any chronic DBS protocol might have elicited the same effects. For example, long-term DBS of the subthalamic nucleus is associated with increased plasticity (i.e., widespread structural changes from sensory motor to PFC and limbic areas) (van Hartevelt et al., 2014).

The exact cellular mechanisms underlying DBS effects remain speculative and poorly understood (Figee et al., 2013; Lozano and Lipsman, 2013; Robbins et al., 2019). Potential effects range from DBS acting as an informational lesion, to changes in inter-regional functional connectivity (Figee et al., 2013) and a general modulation of oscillatory activity and in consequence pathologic circuitry (Lozano and Lipsman, 2013).

How might the NAcc region support preference stability in intertemporal choice? First, it is well established that NAcc activity tracks subjective reward valuation across a range of different reinforcement learning and decision-making tasks and different classes of reinforcers (Bartra et al., 2013; Clithero and Rangel, 2014), including intertemporal choice tasks (Kable and Glimcher, 2007; Basar et al., 2010; Peters and Büchel, 2011). Long-term DBS to this region might induce local plasticity, and thus disrupt or alter value representations in this circuit. Additionally (or alternatively), DBS might disrupt interactions of this circuit with other regions supporting intertemporal choice, such as lateral PFC and/or ventromedial PFC. In line with this view, DBS of the NAcc has been shown to affect activation dynamics in frontostriatal circuits (Figee et al., 2013, 2014) and similar network interactions have been suggested to support successful self-control (Hare et al., 2009; Peters and D'Esposito, 2016).

A number of limitations of the present study deserve discussion. First, as noted above, the lack of a control group that received DBS to another target site limits the interpretability of our findings. Effects might reflect general DBS effects, rather than effects specific to the ALIC/NAcc region. However, given that DBS stimulation sites differ for different disorders, such studies would then require additional controls for differences in symptoms. Additional longitudinal studies are clearly required to clarify this. Second, we focused on intertemporal choice, a stable decision trait with relevance for many mental disorders (Amlung et al., 2019; Lempert et al., 2019). Future work is required to establish whether the observed effects extend to other tasks or preferences. Third, we did not obtain neural read-outs, such as fMRI data that might have allowed us to examine systems-level effects of DBS in individual patients (Figee et al., 2013; Smolders et al., 2013). The degree to which the observed effects are because of changes in neural systems beyond the stimulated target site therefore remain elusive.

In conclusion, the present findings yield an example for a subtle alteration in higher cognitive function following long-term (chronic) DBS to ALIC/NAcc. These data further reveal a potential contribution of this area to the maintenance of preference stability over time in the context of intertemporal choice that further studies might elaborate on.

Footnotes

  • C.B.S. was supported by Walter and Marga Boll Foundation 210-06-16 and Institutional Strategy of the University of Cologne within German Excellence Initiative ZUK 81/1. J.C.B. was supported by Deutsche Forschungsgemeinschaft Project-ID 431549029–SFB 1451. J.P. was supported by Deutsche Forschungsgemeinschaft PE 1627/5-1.

  • J.K. has occasionally received honoraria from AstraZeneca, Lilly, Lundbeck, and Otsuka Pharma for lecturing at conferences and financial support to travel; and financial support for investigator-initiated trials from Medtronic Europe SARL (Meerbusch, Germany). The remaining authors declare no competing financial interests.

  • Correspondence should be addressed to Ben J. Wagner at ben_jonathan.wagner{at}tu-dresden.de

References

  1. ↵
    2. Amlung M,
    3. Marsden E,
    4. Holshausen K,
    5. Morris V,
    6. Patel H,
    7. Vedelago L,
    8. Naish KR,
    9. Reed DD,
    10. McCabe RE
    (2019) Delay discounting as a transdiagnostic process in psychiatric disorders: a meta-analysis. JAMA Psychiatry 76:1176–1186. https://doi.org/10.1001/jamapsychiatry.2019.2102 pmid:31461131
    OpenUrlPubMed
  2. ↵
    2. Amlung M,
    3. Vedelago L,
    4. Acker J,
    5. Balodis I,
    6. MacKillop J
    (2017) Steep delay discounting and addictive behavior: a meta-analysis of continuous associations. Addiction 112:51–62. https://doi.org/10.1111/add.13535 pmid:27450931
    OpenUrlCrossRefPubMed
  3. ↵
    2. Bartra O,
    3. McGuire JT,
    4. Kable JW
    (2013) The valuation system: a coordinate-based meta-analysis of BOLD fMRI experiments examining neural correlates of subjective value. Neuroimage 76:412–427. https://doi.org/10.1016/j.neuroimage.2013.02.063 pmid:23507394
    OpenUrlCrossRefPubMed
  4. ↵
    2. Basar K,
    3. Sesia T,
    4. Groenewegen H,
    5. Steinbusch HW,
    6. Visser-Vandewalle V,
    7. Temel Y
    (2010) Nucleus accumbens and impulsivity. Prog Neurobiol 92:533–557. https://doi.org/10.1016/j.pneurobio.2010.08.007 pmid:20831892
    OpenUrlCrossRefPubMed
  5. ↵
    2. Beard E,
    3. Dienes Z,
    4. Muirhead C,
    5. West R
    (2016) Using Bayes factors for testing hypotheses about intervention effectiveness in addictions research. Addiction 111:2230–2247. https://doi.org/10.1111/add.13501 pmid:27347846
    OpenUrlCrossRefPubMed
  6. ↵
    2. Bickel WK,
    3. Koffarnus MN,
    4. Moody L,
    5. Wilson AG
    (2014) The behavioral- and neuro-economic process of temporal discounting: a candidate behavioral marker of addiction. Neuropharmacology 76:518–527. https://doi.org/10.1016/j.neuropharm.2013.06.013 pmid:23806805
    OpenUrlCrossRefPubMed
  7. ↵
    2. Carlisi CO
    , MRC AIMS Consortium (2017) Comparison of neural substrates of temporal discounting between youth with autism spectrum disorder and with obsessive-compulsive disorder. Psychol Med 47:2513–2527. https://doi.org/10.1017/S0033291717001088 pmid:28436342
    OpenUrlPubMed
  8. ↵
    2. Clithero JA,
    3. Rangel A
    (2014) Informatic parcellation of the network involved in the computation of subjective value. Soc Cogn Affect Neurosci 9:1289–1302. https://doi.org/10.1093/scan/nst106 pmid:23887811
    OpenUrlCrossRefPubMed
  9. ↵
    2. Da Costa Araújo S,
    3. Body S,
    4. Hampson CL,
    5. Langley RW,
    6. Deakin JF,
    7. Anderson IM,
    8. Bradshaw CM,
    9. Szabadi E
    (2009) Effects of lesions of the nucleus accumbens core on intertemporal choice: further observations with an adjusting-delay procedure. Behav Brain Res 202:272–277. https://doi.org/10.1016/j.bbr.2009.04.003 pmid:19463712
    OpenUrlCrossRefPubMed
  10. ↵
    2. Dandekar MP,
    3. Fenoy AJ,
    4. Carvalho AF,
    5. Soares JC,
    6. Quevedo J
    (2018) Deep brain stimulation for treatment-resistant depression: an integrative review of preclinical and clinical findings and translational implications. Mol Psychiatry 23:1094–1112. https://doi.org/10.1038/mp.2018.2 pmid:29483673
    OpenUrlCrossRefPubMed
  11. ↵
    2. Denys D,
    3. Mantione M,
    4. Figee M,
    5. van den Munckhof P,
    6. Koerselman F,
    7. Westenberg H,
    8. Bosch A,
    9. Schuurman R
    (2010) Deep brain stimulation of the nucleus accumbens for treatment-refractory obsessive-compulsive disorder. Arch Gen Psychiatry 67:1061–1068. https://doi.org/10.1001/archgenpsychiatry.2010.122 pmid:20921122
    OpenUrlCrossRefPubMed
  12. ↵
    2. Dixon MR,
    3. Marley J,
    4. Jacobs EA
    (2003) Delay discounting by pathological gamblers. J Appl Behav Anal 36:449–458. https://doi.org/10.1901/jaba.2003.36-449 pmid:14768665
    OpenUrlCrossRefPubMed
  13. ↵
    2. Figee M,
    3. Luigjes J,
    4. Smolders R,
    5. Valencia-Alfonso CE,
    6. van Wingen G,
    7. Kwaasteniet B,
    8. Mantione M,
    9. Ooms P,
    10. Koning P,
    11. Vulink N,
    12. Levar N,
    13. Droge L,
    14. van den Munckhof P,
    15. Schuurman PR,
    16. Nederveen A,
    17. van den Brink W,
    18. Mazaheri A,
    19. Vink M,
    20. Denys D
    (2013) Deep brain stimulation restores frontostriatal network activity in obsessive-compulsive disorder. Nat Neurosci 16:386–387. https://doi.org/10.1038/nn.3344 pmid:23434914
    OpenUrlCrossRefPubMed
  14. ↵
    2. Figee M,
    3. Koning P,
    4. Klaassen S,
    5. Vulink N,
    6. Mantione M,
    7. van den Munckhof P,
    8. Schuurman R
    (2014) Deep brain stimulation induces striatal dopamine release in obsessive-compulsive disorder. Biol Psychiatry 75:647–652. https://doi.org/10.1016/j.biopsych.2013.06.021 pmid:23938318
    OpenUrlCrossRefPubMed
  15. ↵
    2. Figner B,
    3. Knoch D,
    4. Johnson EJ,
    5. Krosch AR,
    6. Lisanby SH,
    7. Fehr E,
    8. Weber EU
    (2010) Lateral prefrontal cortex and self-control in intertemporal choice. Nat Neurosci 13:538–539. https://doi.org/10.1038/nn.2516 pmid:20348919
    OpenUrlCrossRefPubMed
  16. ↵
    2. Foa EB,
    3. Huppert JD,
    4. Leiberg S,
    5. Langner R,
    6. Kichic R,
    7. Hajcak G,
    8. Salkovskis PM
    (2002) The Obsessive-Compulsive Inventory: development and validation of a short version. Psychol Assess 14:485–496. https://doi.org/10.1037/1040-3590.14.4.485
    OpenUrlCrossRefPubMed
  17. ↵
    2. Haines N,
    3. Kvam PD,
    4. Irving LH,
    5. Smith C,
    6. Beauchaine TP,
    7. Pitt MA,
    8. Ahn W,
    9. Turner B
    (2020) Theoretically informed generative models can advance the psychological and brain sciences: lessons from the reliability paradox. PsyArXiv.
  18. ↵
    2. Hare TA,
    3. Camerer CF,
    4. Rangel A
    (2009) Self-control in decision-making involves modulation of the vmPFC valuation system. Science 324:646–648. https://doi.org/10.1126/science.1168450 pmid:19407204
    OpenUrlAbstract/FREE Full Text
  19. ↵
    2. Hariri AR,
    3. Brown SM,
    4. Williamson DE,
    5. Flory JD,
    6. Wit H,
    7. Manuck SB
    (2006) Preference for immediate over delayed rewards is associated with magnitude of ventral striatal activity. J Neurosci 26:13213–13217. https://doi.org/10.1523/JNEUROSCI.3446-06.2006 pmid:17182771
    OpenUrlAbstract/FREE Full Text
  20. ↵
    2. Huys D,
    3. Kohl S,
    4. Baldermann JC,
    5. Timmermann L,
    6. Sturm V,
    7. Visser-Vandewalle V,
    8. Kuhn J
    (2019) Open-label trial of anterior limb of internal capsule-nucleus accumbens deep brain stimulation for obsessive-compulsive disorder: insights gained. J Neurol Neurosurg Psychiatry 90:805–812. https://doi.org/10.1136/jnnp-2018-318996 pmid:30770458
    OpenUrlAbstract/FREE Full Text
  21. ↵
    2. Kable JW,
    3. Glimcher PW
    (2007) The neural correlates of subjective value during intertemporal choice. Nat Neurosci 10:1625–1633. https://doi.org/10.1038/nn2007 pmid:17982449
    OpenUrlCrossRefPubMed
  22. ↵
    2. Kirby KN
    (2009) One-year temporal stability of delay-discount rates. Psychon Bull Rev 16:457–462. https://doi.org/10.3758/PBR.16.3.457 pmid:19451368
    OpenUrlCrossRefPubMed
  23. ↵
    2. Kohl S,
    3. Schönherr DM,
    4. Luigjes J,
    5. Denys D,
    6. Mueller UJ,
    7. Lenartz D,
    8. Visser-Vandewalle V,
    9. Kuhn J
    (2014) Deep brain stimulation for treatment-refractory obsessive-compulsive disorder: a systematic review. BMC Psychiatry 14:214. https://doi.org/10.1186/s12888-014-0214-y pmid:25085317
    OpenUrlCrossRefPubMed
  24. ↵
    2. Krauss JK,
    3. Lipsman N,
    4. Aziz T,
    5. Boutet A,
    6. Brown P,
    7. Chang JW,
    8. Davidson B,
    9. Grill WM,
    10. Hariz MI,
    11. Horn A,
    12. Schulder M,
    13. Mammis A,
    14. Tass PA,
    15. Volkmann J,
    16. Lozano AM
    (2021) Technology of deep brain stimulation: current status and future directions. Nat Rev Neurol 17:75–87. https://doi.org/10.1038/s41582-020-00426-z pmid:33244188
    OpenUrlPubMed
  25. ↵
    2. Kuhn J,
    3. Baldermann JC
    (2020) Elucidating neural network changes induced by deep brain stimulation for OCD. Brain 143:1293–1296. https://doi.org/10.1093/brain/awaa108 pmid:32438409
    OpenUrlPubMed
  26. ↵
    2. Lee DJ,
    3. Lozano CS,
    4. Dallapiazza RF,
    5. Lozano AM
    (2019) Current and future directions of deep brain stimulation for neurological and psychiatric disorders. J Neurosurg 131:333–342. https://doi.org/10.3171/2019.4.JNS181761 pmid:31370011
    OpenUrlCrossRefPubMed
  27. ↵
    2. Lempert KM,
    3. Steinglass JE,
    4. Pinto A,
    5. Kable JW,
    6. Simpson HB
    (2019) Can delay discounting deliver on the promise of RDoC? Psychol Med 49:190–199. https://doi.org/10.1017/S0033291718001770 pmid:30070191
    OpenUrlCrossRefPubMed
  28. ↵
    2. Lozano AM,
    3. Lipsman N
    (2013) Probing and regulating dysfunctional circuits using deep brain stimulation. Neuron 77:406–424. https://doi.org/10.1016/j.neuron.2013.01.020 pmid:23395370
    OpenUrlCrossRefPubMed
  29. ↵
    2. Luigjes J,
    3. Mantione M,
    4. van den Brink W,
    5. Schuurman PR,
    6. van den Munckhof P,
    7. Denys D
    (2011) Deep brain stimulation increases impulsivity in two patients with obsessive-compulsive disorder. Int Clin Psychopharmacol 26:338–340. https://doi.org/10.1097/YIC.0b013e32834af505 pmid:21857527
    OpenUrlPubMed
  30. ↵
    2. Mahoney JJ,
    3. Hanlon CA,
    4. Marshalek PJ,
    5. Rezai AR,
    6. Krinke L
    (2020) Transcranial magnetic stimulation, deep brain stimulation, and other forms of neuromodulation for substance use disorders: review of modalities and implications for treatment. J Neurol Sci 418:117149. https://doi.org/10.1016/j.jns.2020.117149 pmid:33002757
    OpenUrlPubMed
  31. ↵
    2. McClure SM,
    3. Bickel WK
    (2014) A dual-systems perspective on addiction: contributions from neuroimaging and cognitive training. Ann NY Acad Sci 1327:62–78. https://doi.org/10.1111/nyas.12561 pmid:25336389
    OpenUrlCrossRefPubMed
  32. ↵
    2. McClure SM,
    3. Laibson DI,
    4. Loewenstein G,
    5. Cohen JD
    (2004) Separate neural systems value immediate and delayed monetary rewards. Science 306:503–507. https://doi.org/10.1126/science.1100907 pmid:15486304
    OpenUrlAbstract/FREE Full Text
  33. ↵
    2. Nachev P,
    3. Lopez-Sosa F,
    4. Gonzalez-Rosa JJ,
    5. Galarza A,
    6. Avecillas J,
    7. Pineda-Pardo JA,
    8. Lopez-Ibor JJ,
    9. Reneses B,
    10. Barcia JA
    , (2015) Dynamic risk control by human nucleus accumbens. Brain 138:3496–3502. https://doi.org/10.1093/brain/awv285 pmid:26428667
    OpenUrlCrossRefPubMed
  34. ↵
    2. Norman LJ,
    3. Carlisi CO
    (2017) Neural dysfunction during temporal discounting in paediatric attention-deficit/hyperactivity disorder and obsessive-compulsive disorder. Psychiatry Res Neuroimaging 269:97–105. https://doi.org/10.1016/j.pscychresns.2017.09.008 pmid:28988149
    OpenUrlCrossRefPubMed
  35. ↵
    2. Nucifora FC,
    3. Woznica E,
    4. Lee BJ,
    5. Cascella N,
    6. Sawa A
    (2019) Treatment resistant schizophrenia: clinical, biological, and therapeutic perspectives. Neurobiol Dis 131:104257. https://doi.org/10.1016/j.nbd.2018.08.016 pmid:30170114
    OpenUrlPubMed
  36. ↵
    2. Ong CW,
    3. Graves K,
    4. Berry MS,
    5. Odum AL,
    6. Twohig MP
    (2019) Obsessive-compulsive symptoms are associated with increased delay discounting in a novel hand-washing task. Behav Anal (Wash D C) 19:136–149. https://doi.org/10.1037/bar0000096 pmid:31552288
    OpenUrlPubMed
  37. ↵
    2. Peisker CB,
    3. Schüller T,
    4. Peters J,
    5. Wagner BJ,
    6. Schilbach L,
    7. Müller UJ,
    8. Visser-Vandewalle V,
    9. Kuhn J
    (2018) Nucleus accumbens deep brain stimulation in patients with substance use disorders and delay discounting. Brain Sci 8:104257. https://doi.org/10.3390/brainsci8020021 pmid:29382059
    OpenUrlPubMed
  38. ↵
    2. Peters J,
    3. Büchel C
    (2011) The neural mechanisms of intertemporal decision-making: understanding variability. Trends Cogn Sci 15:227–239. https://doi.org/10.1016/j.tics.2011.03.002 pmid:21497544
    OpenUrlCrossRefPubMed
  39. ↵
    2. Peters J,
    3. D'Esposito M
    (2016) Effects of medial orbitofrontal cortex lesions on self-control in intertemporal choice. Curr Biol 26:2625–2628. https://doi.org/10.1016/j.cub.2016.07.035 pmid:27593380
    OpenUrlCrossRefPubMed
  40. ↵
    2. Pinto A,
    3. Steinglass JE,
    4. Greene AL,
    5. Weber EU,
    6. Simpson HB
    (2014) Capacity to delay reward differentiates obsessive-compulsive disorder and obsessive-compulsive personality disorder. Biol Psychiatry 75:653–659. https://doi.org/10.1016/j.biopsych.2013.09.007 pmid:24199665
    OpenUrlCrossRefPubMed
  41. ↵
    2. Pisansky MT,
    3. Lefevre EM,
    4. Retzlaff CL,
    5. Trieu BH,
    6. Leipold DW,
    7. Rothwell PE
    (2019) Nucleus accumbens fast-spiking interneurons constrain impulsive action. Biol Psychiatry 86:836–847. https://doi.org/10.1016/j.biopsych.2019.07.002 pmid:31471038
    OpenUrlCrossRefPubMed
  42. ↵
    2. Robbins TW,
    3. Vaghi MM,
    4. Banca P
    (2019) Obsessive-compulsive disorder: puzzles and prospects. Neuron 102:27–47. https://doi.org/10.1016/j.neuron.2019.01.046 pmid:30946823
    OpenUrlCrossRefPubMed
  43. ↵
    2. Smith EE,
    3. Schüller T,
    4. Huys D,
    5. Baldermann JC,
    6. Andrade P,
    7. Allen JJ,
    8. Visser-Vandewalle V,
    9. Ullsperger M,
    10. Gruendler TO,
    11. Kuhn J
    (2020) A brief demonstration of frontostriatal connectivity in OCD patients with intracranial electrodes. Neuroimage 220:117138. https://doi.org/10.1016/j.neuroimage.2020.117138 pmid:32634597
    OpenUrlPubMed
  44. ↵
    2. Smolders R,
    3. Mazaheri A,
    4. van Wingen G,
    5. Figee M,
    6. Koning P,
    7. Denys D
    (2013) Deep brain stimulation targeted at the nucleus accumbens decreases the potential for pathologic network communication. Biol Psychiatry 74:e27–e28. https://doi.org/10.1016/j.biopsych.2013.03.012 pmid:23623188
    OpenUrlCrossRefPubMed
  45. ↵
    2. Sohn SY,
    3. Kang J,
    4. Namkoong K,
    5. Kim SJ
    (2014) Multidimensional measures of impulsivity in obsessive-compulsive disorder: cannot wait and stop. PLoS One 9:e111739. https://doi.org/10.1371/journal.pone.0111739 pmid:25372136
    OpenUrlPubMed
  46. ↵
    2. Steinglass JE,
    3. Lempert KM,
    4. Choo TH,
    5. Kimeldorf MB,
    6. Wall M,
    7. Walsh BT,
    8. Fyer AJ,
    9. Schneier FR,
    10. Simpson HB
    (2017) Temporal discounting across three psychiatric disorders: anorexia nervosa, obsessive-compulsive disorder, and social anxiety disorder. Depress Anxiety 34:463–470. https://doi.org/10.1002/da.22586 pmid:28009473
    OpenUrlCrossRefPubMed
  47. ↵
    2. Sutton RS,
    3. Barto AG
    (1998) Reinforcement learning: an introduction to reinforcement learning. Cambridge, MA: Massachusetts Institute of Technology.
  48. ↵
    2. van den Bos W,
    3. Rodriguez CA,
    4. Schweitzer JB,
    5. McClure SM
    (2014) Connectivity strength of dissociable striatal tracts predict individual differences in temporal discounting. J Neurosci 34:10298–10310. https://doi.org/10.1523/JNEUROSCI.4105-13.2014 pmid:25080591
    OpenUrlAbstract/FREE Full Text
  49. ↵
    2. van Hartevelt TJ,
    3. Cabral J,
    4. Deco G,
    5. Møller A,
    6. Green AL,
    7. Aziz TZ,
    8. Kringelbach ML
    (2014) Neural plasticity in human brain connectivity: the effects of long term deep brain stimulation of the subthalamic nucleus in Parkinson's disease. PLoS One 9:e86496. https://doi.org/10.1371/journal.pone.0086496 pmid:24466120
    OpenUrlCrossRefPubMed
  50. ↵
    2. Wiehler A,
    3. Peters J
    (2015) Reward-based decision making in pathological gambling: the roles of risk and delay. Neurosci Res 90:3–14. https://doi.org/10.1016/j.neures.2014.09.008 pmid:25269860
    OpenUrlCrossRefPubMed
  51. ↵
    2. Wu H, et al
    . (2021) Deep brain stimulation for refractory obsessive-compulsive disorder (OCD): emerging or established therapy? Mol Psychiatry 26:60–65. https://doi.org/10.1038/s41380-020-00933-x pmid:33144712
    OpenUrlPubMed
Back to top

In this issue

Email
Print
View Full Page PDF
Citation Tools
Respond to this article
Chronic Deep Brain Stimulation of the Human Nucleus Accumbens Region Disrupts the Stability of Intertemporal Preferences
Ben J. Wagner, Canan B. Schüller, Thomas Schüller, Juan C. Baldermann, Sina Kohl, Veerle Visser-Vandewalle, Daniel Huys, Milena Marx, Jens Kuhn, Jan Peters
Journal of Neuroscience 25 October 2023, 43 (43) 7175-7185; DOI: 10.1523/JNEUROSCI.0138-23.2023
Twitter logo Facebook logo Mendeley logo

Jump to section

Keywords

Responses to this article

Respond to this article

Jump to comment:

No eLetters have been published for this article.