Subtherapeutic subcallosal cingulate stimulation modulates mood and learning during decision-making in treatment-resistant depression
DOI: Pending DOI
Synopsis
Deep brain stimulation is a promising intervention for treatment-resistant depression (TRD) [1-2], yet its acute effects on behavior and decision-making remain poorly understood. We investigated how subtherapeutic stimulation of the subcallosal cingulate cortex influences social and reward-based decision-making in humans with TRD. Behaviorally, stimulation improved momentary mood ratings and accelerated response times, with computational modeling showing enhanced sensitivity to social norm violations and reward learning rates. Resting-state local field potential periodic and aperiodic features covaried with behavioral and computational processes, suggesting that neural dynamics contribute to stimulation-induced shifts in decision-making.
Background
Depression is marked by disruptions in affective and cognitive decision-making, including altered responses to social feedback and impaired learning from positive outcomes3. Deep brain stimulation (DBS) targeting the subcallosal cingulate cortex (SCC) has demonstrated therapeutic benefits for treatment-resistant depression (TRD)2. The SCC is positioned within a limbic-cortical network involved in emotion regulation, reward valuation, and behavioral flexibility, with altered patterns of neural activities found in depression [4]. While prior work has focused largely on long-term clinical response to DBS2, few studies have examined how acute, subtherapeutic stimulation modulates decision-making and its associated neural dynamics. To address this gap, we combined behavioral paradigms, computational modeling, and intracranial local field potential (LFP) recordings to characterize the immediate effects of subtherapeutic SCC stimulation on social and reward-based decision processes in humans with TRD.
Methods
We tested twelve participants with TRD undergoing SCC DBS implantation surgery. Ten participants completed two tasks in both pre- and post-stimulation sessions: the Ultimatum Game (UG), in which they accepted or rejected unfair monetary offers from a virtual partner to assess social exchange and fairness sensitivity [5], and the Reversal Learning (RL) task, in which they selected between two “slot machines” with shifting reward contingencies to assess reward learning. Behavioral analyses included mixed-effect regression models and cognitive computational models to evaluate stimulation effects on model-agnostic and model-based metrics. Resting-state local field potentials (LFPs) were recorded from the left SCC, via the implanted DBS device in a subset of five participants before and after subtherapeutic stimulation. Power spectral density was decomposed into aperiodic and periodic components6. Exploratory regression analyses assessed whether resting-state features (spectral exponent, alpha, and beta power) explained additional variance in behavioral and computational outcomes beyond stimulation condition alone.
Results
In the UG (n = 10), acute subtherapeutic stimulation improved momentary mood ratings shown by a mixed-effect linear regression (b = 4.99, p = 0.0002) and increased model-derived sensitivity to social norm violations shown by the Wilcoxon signed-rank test (W = 2.00, p = 0.0059), without altering overall rejection behavior, suggesting greater affective engagement without changes in decision outcome. In the RL (n = 10), stimulation significantly reduced response times in both the mixed block (b = −0.24, p = 0.0012) and reward block (b = −0.18, p = 0.0135), indicating faster responding following stimulation, particularly in reward-rich contexts. No significant effects of stimulation on optimal choice rates were observed. Computational modeling revealed that the best-fitting reinforcement learning model across sessions was a two-learning-rate model (n = 9), with separate reward and punishment learning rates, showing a significant increase in reward learning post-stimulation in the RL task (W = 0.00, p = 0.0039).
Resting-state SCC LFP analysis (n = 5) revealed stimulation-induced decreases in aperiodic exponent (b = −0.13, p = 0.00026) and alpha power (b = −0.083, p = 0.026), and an increase in lower beta power (b = 0.09, p = 0.00066). Exploratory regressions indicated that resting-state spectral features explained additional variance in mood and computational metrics (n = 3). In the UG, both higher average spectral exponent (b = −23.28, p = 7.1 × 10⁻⁶) and alpha power (b = −30.43, p = 0.0001) predicted lower momentary mood ratings, improving model fit compared to the stimulation-only model (adj R² = 0.46–0.51 vs. 0.30). Moreover, higher resting-state spectral exponent was associated with greater utility of offer acceptance (b = 10.33, p = 0.018), and higher alpha power was associated with larger trial-level norm expectation (b = 2.33, p = 0.0001). Together, these findings suggest that stimulation-induced behavioral changes co-occur with meaningful changes in resting-state neural features during social decision-making.
Discussion
Our findings demonstrate that subtherapeutic SCC stimulation modulates decision-making processes and resting-state neural dynamics in individuals with TRD. In summary, subtherapeutic stimulation did not alter what decisions were made but improved mood, enhanced sensitivity to fairness, elicited faster responses, and increased reward learning rates. These effects align with SCC’s known role in affective regulation and cognitive flexibility [7], suggesting engagement of limbic-cortical networks implicated in adaptive behavior [4]. Resting-state LFPs changes post stimulation mirrored neural signatures linked to clinical recovery in prior DBS studies [8], indicating potential electrophysiological markers of behavioral modulation. Together, these findings position SCC as a node that shapes both emotional and computational aspects of decision-making. Future work incorporating concurrent task-based recordings may clarify the neural mechanisms by which SCC stimulation modulates choice and mood.
Disclosures
The authors are grateful to the patients and their families for their collaboration that enabled this study. The authors thank Jessica Alexander and Laureen Pagan for their support with clinical administration and DBS patient coordination. The Summit RC+S and Percept PC devices were provided by Medtronic. The authors also thank the surgical team and the Alpha Omega technical team at Mount Sinai West for their support. This work was supported by the National Institute Health grants R01MH124115 (XG, PRM), UH3NS103550 (HSM), the Hope for Depression Foundation (HSM), and the Red Gates Foundation (PRM). Work on “Computational Roles of Dopamine and Serotonin Transients in Anhedonia in Humans with Treatment Resistant Depression” was supported by Wellcome Leap as part of the Multi-Channel Psych Program (XG, HSM). HSM received consulting and IP licensing fees from Abbott Neuromodulation. XG is a scientific advisor to Soihtu Dx.
References
1. Mayberg, H. S. et al. Deep brain stimulation for treatment-resistant depression. Neuron 45, 651–660 (2005).
2. Alagapan, S. et al. Cingulate dynamics track depression recovery with deep brain stimulation. Nature 622, 130–138 (2023).
3. Lawlor, V. M. et al. Dissecting the impact of depression on decision-making. Psychol. Med. 50, 1613–1622 (2019).
4. Mayberg, H. S. Limbic–cortical dysregulation: a proposed model of depression. J. Neuropsychiatry Clin. Neurosci. 9, 471–481 (1997).
5. Gu, X. et al. Necessary, yet dissociable contributions of the insular and ventromedial prefrontal cortices to norm adaptation: computational and lesion evidence in humans. J. Neurosci. 35, 467–473 (2015).
6. Donoghue, T., Haller, M., Peterson, E. J., Varma, P., Sebastian, P., Gao, R., Noto, T., Lara, A. H., Wallis, J. D., Knight, R. T., Shestyuk, A. & Voytek, B. Parameterizing neural power spectra into periodic and aperiodic components. Nat. Neurosci. 23, 1655–1665 (2020).
7. Lockwood, P. L. & Wittmann, M. K. Ventral anterior cingulate cortex and social decision-making. Neurosci. Biobehav. Rev. 92, 187–191 (2018).
8. Smart, O. L. et al. Initial unilateral exposure to deep brain stimulation in treatment-resistant depression patients alters spectral power in the subcallosal cingulate. Front. Comput. Neurosci. 12, 43 (2018).