Measuring Autonomy in Patients with Deep Brain Stimulation

DBS patient reports may point to a pre-reflective level of autonomy experience that current frameworks do not capture.

Mingqian Li & Caitlin MaceHistory and Philosophy of Science, University of Pittsburgh

PSA 2026, San Diego

Poster forum, Friday, November 20

6:30 to 9:00 pm, The Westin San Diego Gaslamp Quarter

Tell us what you think

Thank you for scanning the code. The form takes about a minute and has no required questions. Objections are especially welcome.

Open the feedback form Email Mingqian

Full text

Working draft, November 2026.

DBS provides electrical stimulation with electrodes implanted in target brain regions. It is a standard treatment for Parkinson’s disease and is under investigation for psychiatric conditions (Lozano et al., 2019). Patients frequently report confusion about their own mental states: “I don’t recognize myself anymore” (Schüpbach et al., 2006, as quoted in Goddard, 2017). Recent clinical evidence supports these concerns: Schüpbach et al. (2006) document patients’ feelings of strangeness and unfamiliarity with themselves despite motor improvement, and Gilbert et al. (2017) report patients describing the experience as “it’s not the me that went into the surgery”.

Current neuroethical accounts frame these concerns through the lens of agency or autonomy. Frankfurt (1971) argues that free will requires second-order volitions, a framework later adopted as a model of autonomy. Christman (2009) locates autonomy in a condition of non-alienation upon historically sensitive self-reflection regarding the processes through which one’s values were formed. Applied to DBS, Goddard (2017) argues that the salient ethical concern is DBS’s impact on relational agency, understood as the capacity for embodied, relationally constituted autonomous action. Lipsman & Glannon (2013) examine the implications of DBS for perceptions of personal identity, agency, and free will. Sharp and Wasserman (2016) apply historicist conditions on moral responsibility to DBS, arguing that certain neurological modifications can undermine responsibility. Their analysis concerns responsibility rather than autonomy directly, but shares the requirement that the agent’s reflective relation to the causal history of change be intact. Bergamin et al. (2022) analyze autonomy across psychiatric contexts and identify a shared dependence on reflective competence as a criterion. Despite their differences, these accounts converge on a shared structure: autonomy is assessed through the agent’s capacity for reflective endorsement or non-alienation.

If autonomy is assessed exclusively through reflective endorsement, disruption that does not impair reflective capacity will go undetected. It is then necessary to look at the empirical evidence about whether DBS impairs reflective capacity. Witt et al. (2008), in a multicentre randomized controlled trial comparing STN-DBS with best medical therapy, find that overall cognition is not reduced by stimulation, although selective declines in frontal cognitive functions are observed. Trenado et al. (2023) directly test metacognition in PD patients under STN-DBS (n = 12). Their overall conclusion is that STN-DBS has no significant effect on metacognition. However, the correlation between confidence and actual accuracy holds only for DBS-OFF, not for DBS-ON, suggesting a subtle decoupling of self-monitoring from performance under stimulation. Gilbert et al. (2017) find that patients report self-estrangement after DBS, while family members often notice personality changes that patients themselves do not fully recognize. Pham et al. (2015) similarly report that caregivers identify increases in impulsivity that patients’ self-assessments do not capture. Zuk et al. (2023) report that researchers working with next-generation DBS observed personality, mood, and behavioral changes in their patients, suggesting that such changes may be more common than the existing literature reflects. Taken together, this evidence suggests that the reflective capacities on which current autonomy accounts depend are largely preserved under STN-DBS, yet patients continue to report disruptions in their sense of self. This discrepancy indicates that something is being missed.

If the disruption is not at the reflective level, we need a framework that distinguishes levels of agency experience. Synofzik et al. (2008) propose that agency itself has two components. One is reflective, the action is conceptually attributed to the agent through interpretive judgment. The other is pre-reflective, in that the action is felt as self-caused. Synofzik, Vosgerau, and Newen (2008) call the latter the feeling of agency. By this demarcation, the feeling of agency arises from pre-reflective sensorimotor processing and multisensory integration, while JoA is responsible for reflective and inferential attribution. In practice, agency is measured through both subjective and objective routes. Subjective assessment relies on patient self-report, caregiver observation, and neuropsychological scales. Gilbert, Viaña, and Ineichen (2021) review more than 1,500 articles and find that current instruments rely heavily on self-report. Objective assessment involves behavioral measurement of sub-capacities such as attentional control, decision-making, and goal-directed action. But both routes capture what the agent can report or what behavior makes visible, and both ultimately measure JoA. There is no established method for directly measuring SoA, which operates at a pre-reflective level through sensorimotor feedback and is not accessible to the agent’s own reflective judgment. In other neurological conditions, SoA and JoA can be disrupted in different ways: in schizophrenia with passivity experiences, patients lack the feeling that their actions are self-generated, and on the two-step account a further disturbance at the level of judgment leads them to attribute those actions to external forces (Frith et al., 2000; Synofzik et al., 2008). This raises the possibility that DBS may also produce a dissociation between SoA and JoA, but this framework has not yet been applied to DBS.

We could further discover the dissociation if we look back at the clinical evidence through the distinction between SoA and JoA. Frank et al. (2007) show that STN-DBS selectively impairs the ability to slow down in the face of decision conflict, effectively altering the causal parameters of action selection. Such disruption in action preference reflects a change in the sensorimotor process of action selection, without involving reflective endorsement or conscious monitoring. Maith et al. (2025) use a neuro-computational basal ganglia model, fitted to dystonia patient data, to show that GPi-DBS shifts decision-making from goal-directed to habitual control. This finding may generalize to other DBS targets, though the model does not directly address conscious processing. These results suggest that DBS can alter the causal structure of action selection at a level prior to reflective attribution, which is the level at which the feeling of agency operates.

The current definition of SoA tracks only felt causal ownership, but patient reports indicate that someone can reflectively endorse an action as theirs while the action does not feel like theirs. This is what DBS patients appear to report when they describe feeling like themselves yet not quite themselves. Kraemer (2013) raises the valuable distinction between felt authenticity and felt alienation, capturing an instantaneous feeling of whether the experience is familiar to the agent or not. But she does not conceptualize it as an independent layer within the agency or autonomy framework; rather, she frames it under the umbrella of personal authenticity and alienation. It is still not clear what patients are actually reporting since neither SoA nor JoA captures the immediate sense of whether an action fits one’s intentional structure.

If agency has a two-level structure in SoA and JoA, there is no conceptual reason why autonomy should have only the single level of reflective endorsement. Schlimme (2013), in a phenomenological analysis of suicidal experience, describes a pre-reflective sense of self-determination as distinct from reflective judgment. Although his context differs from DBS, the structural distinction between pre-reflective and reflective self-determination is directly relevant. Existing literature touches on pre-reflective autonomy but has not treated it as an independent concept. One might argue that reflective endorsement is sufficient for autonomy assessment, and that what patients report is simply a byproduct of adjusting to neurological change. But DBS clinical reports are evidence for precisely this missing level since what patients experience is a change in their sense of autonomy that persists even when reflective endorsement is intact. One might further argue that autonomy is by definition a reflective concept, and that its assessment should operate only at the reflective level. But this would imply that autonomy is only present at the moments when the agent actively reflects on and endorses their actions. Between those moments, the agent continues to act, make decisions, and navigate the world. If autonomy requires reflective endorsement for each action, then the vast majority of human behavior falls outside its scope. In practice, agents rely on continuous pre-reflective feedback to maintain a sense that their actions fit their intentions. Reflective endorsement is intermittent; the experience of acting autonomously is not. This suggests that autonomy cannot be fully accounted for at the reflective level alone. This paper establishes the gap that makes such a concept necessary without developing it further. Recognizing this gap has clinical implications for how we assess and support patient autonomy under DBS. Current approaches focus primarily on the loss of autonomy and tend to intervene after stimulation has ended, helping patients adapt to changes retrospectively. But if there is a pre-reflective level of autonomy experience, working at the level of feeling during the modulation process itself, helping patients recognize and navigate their immediate sense of autonomy, may be equally valuable.

References

Poster image credits

Background brain
Andreas Vesalius, De humani corporis fabrica (1543). Courtesy of the U.S. National Library of Medicine. Public domain.
Electrode
Hand and conductor after Luigi Galvani, De viribus electricitatis in motu musculari commentarius (1791), Fig. 23. Public domain.
DBS system illustration
Shamir, R. R., Noecker, A. M., & McIntyre, C. C. (2014), via Wikimedia Commons, licensed under CC BY 3.0. Background removed.
Data figures
Redrawn from values reported in Trenado et al. (2023), published under CC BY 4.0.

About me

I am Mingqian Li. I work in clinical epilepsy research and ask how agency and autonomy change under neuromodulation.

Interests. Neuroethics, neuromodulation, and clinical neuroscience.

Looking for. A clinical research coordinator (CRC) position in neuromodulation. I plan to apply to MD-PhD programs.