Towards Precision Neuromodulation: A Control-Theoretic Framework for Dissociating Therapeutic Seizure Dynamics from Cognitive Disruption

Author: Rekha Boodoo‑Lumbus

Affiliation: RAKHEE LB LIMITED, United Kingdom

© 2026 Rekha Boodoo‑Lumbus / RAKHEE LB LIMITED.

All Rights Reserved (including images and graphics)

Abstract

Electroconvulsive therapy (ECT) remains the most effective intervention for treatment-resistant depression (TRD), yet its clinical utility is fundamentally constrained by cognitive side effects. This article reframes ECT not as a procedure, but as a high amplitude system-level perturbation acting on a nonlinear neural network. Drawing on principles from neuroscience, physics, and control theory, we argue that cognitive adverse effects are not incidental but intrinsic to the global synchrony required for seizure mediated remission. We formalise the neuromodulation challenge as a control problem: how to induce therapeutic plasticity while preserving functional network segregation. Finally, we propose a cross-disciplinary roadmap for next-generation interventions that move beyond binary seizure based models toward adaptive, patterned neuromodulation operating closer to the brain’s native dynamics.

Introduction: The “Gold Standard” Paradox

In the treatment of severe, treatment-resistant psychiatric disorders, electroconvulsive therapy (ECT) continues to demonstrate unmatched efficacy, with response rates frequently exceeding those of pharmacological and non-convulsive neuromodulation approaches. Despite this clinical success, ECT remains fundamentally constrained by its cognitive side effect profile, most notably retrograde amnesia and executive dysfunction. Contemporary clinical refinements, such as unilateral electrode placement and ultra-brief pulse durations, treat these deficits as extrinsic procedural artefacts to be minimised through technical optimisation. However, despite decades of adjusting delivery parameters, these refinements have produced diminishing returns. This plateau suggests that the dominant source of cognitive disruption is not a "technical" flaw in electrode geometry or pulse duration, but a structural limitation inherent to the therapy’s primary mechanism of action.

This article advances the thesis that cognitive side effects are not accidental by-products, but emergent consequences of the global synchrony required for seizure mediated remission. We argue that the information processing capacity of the medial temporal lobe is physically and computationally incompatible with the high amplitude, low specificity perturbations induced by current ECT protocols. By reframing ECT through the lens of nonlinear dynamics and control theory, we can reconceptualise neuromodulation as a problem of constrained optimisation. This shift enables the rational design of next-generation interventions that preserve therapeutic state-transitions while respecting the functional segregation required for cognitive integrity.

Section 1: Mechanisms of ECT at the Systems Level

1.1 Neurobiological Cascades and Forced Synchronisation

ECT induces a cascade of neurobiological changes, including robust upregulation of neurotrophic factors such as brain-derived neurotrophic factor (BDNF), modulation of the hypothalamic pituitary adrenal axis, and widespread alterations in functional connectivity (Leaver et al., 2022). At the systems level, these effects are hypothesised to arise from a forced global synchronisation of neural activity, effectively disrupting maladaptive attractor states associated with depressive pathology (van der Lande et al., 2024).

From a dynamical systems perspective, the generalised seizure functions as a high energy perturbation that transiently destabilises pathological network regimes, allowing the system to reorganise into a more adaptive configuration (Deco et al., 2014). This “reset” mechanism explains both the rapid antidepressant efficacy of ECT and its resistance to conventional dose-response optimisation (Krystal et al., 2003). Crucially, this reorganisation is not a targeted correction of specific neural circuits, but a statistical disruption of the entire system. By inducing a seizure, ECT increases the probability that the brain will 'exit' a pathological state by temporarily erasing the network specificity that holds it there. The clinical 'response' is therefore a statistical byproduct of global instability, which explains why cognitive specialised circuits, such as those for memory, are inevitably caught in the same destabilisation

1.2 The Cost of Global Synchrony

The therapeutic benefits of global synchrony come at a measurable cognitive cost. Neuroimaging and neuropsychological studies consistently implicate the hippocampus and medial temporal lobe, regions critical for episodic memory and contextual integration, as particularly vulnerable to ECT induced disruption (Abbott et al., 2014). The generalised seizure collapses functional segregation within these circuits, impairing the balance between local specialisation and global integration that characterises healthy brain networks (Fide et al., 2025).

Graph-theoretical analyses of brain organisation suggest that optimal cognitive function emerges near a critical regime balancing segregation and integration (Bullmore & Sporns, 2009). ECT’s high amplitude perturbation risks pushing the system beyond this regime, resulting in over-integration and loss of network specificity (Zhou et al., 2021). Cognitive side effects, therefore, reflect a fundamental trade-off between therapeutic reorganisation and informational precision (Sackeim et al., 2007). From this perspective, cognitive side effects are not collateral damage but the predictable consequence of using global synchronisation as a therapeutic lever in a system whose normal function depends on finely tuned segregation. This trade off is not simply biological; it is ultimately rooted in how external energy interacts with neural tissue.

Section 2: Energy, Fields, and Coupling to Neural Tissue

2.1 The Physics of Electrochemical Coupling

To understand why different neuromodulation modalities converge on similar cognitive risks despite distinct delivery mechanisms, it is necessary to examine how external energy interacts with neural tissue at a physical level. Neuromodulation fundamentally involves the association of externally applied energy fields, electrical, magnetic, or mechanical, to the brain’s intrinsic electrochemical gradients (Alipour & Hajipour-Verdom, 2021). Rather than framing interventions as “electrical” or “chemical,” neural tissue can be understood as a nonlinear, excitable medium in which external fields influence ion channel dynamics, membrane potentials, and synaptic efficacy through multiple physical mechanisms, including capacitive currents, induced electric fields, and acoustic radiation forces (Blackmore et al., 2019).

This perspective dissolves the false dichotomy between electricity and chemistry, emphasising instead the mode, scale, and pattern of energy interaction (Parpura et al., 2012). Mainly due to the brain’s extracellular milieu is a conductive, anisotropic volume conductor, any sufficiently intense external field will propagate and summate across large cortical territories. Consequently, regardless of modality, perturbations strong enough to enforce large scale synchrony will converge on similar system-level consequences: widespread disruption of functional segregation and the cognitive operations that depend on it.

2.2 Thresholds of Plasticity

The central engineering challenge in neuromodulation is identifying parameter regimes, defined by amplitude, spatial extent, and temporal patterning, that cross plasticity thresholds without collapsing functional segregation (Deng et al., 2011). ECT delivers electric fields well above neural activation thresholds across large brain volumes, ensuring seizure induction but sacrificing specificity. The result is a supra threshold perturbation that reliably triggers the neurotrophic and connectivity changes required for remission, yet simultaneously erases the finely tuned local processing that supports memory and executive function.

Emerging alternatives such as magnetic seizure therapy (MST) demonstrate that increased focality can reduce the volume of tissue subjected to global synchrony while preserving seizure mediated plasticity (Borrione et al., 2020). These findings suggest that side effect profiles scale directly with the spatial footprint of enforced synchrony. Precision oriented design principles therefore become essential: the goal is to locate the narrow window in parameter space where the minimum energy required for therapeutic state transition is delivered without overshooting into pathological over-integration.

Section 3: Precision Neuromodulation as Control Theory

3.1 Pathological Attractors and State Transitions

Once neuromodulation is understood as a problem of driving state transitions in a nonlinear, high-dimensional system under biological constraints, control theory becomes not a metaphor but a formal necessity. Psychiatric disorders can be conceptualised as stable, maladaptive attractor states within high-dimensional neural state spaces (Friston, 2010). From this perspective, neuromodulation aims not to excite neurons per se, but to drive controlled state transitions between network regimes. ECT achieves this through a large, indiscriminate perturbation that kicks the system out of its depressive attractor; precision neuromodulation seeks to achieve the same transition with minimal collateral disruption to neighbouring basins of attraction (Honey et al., 2007). In high dimensional state space, the challenge is acute: small perturbations rarely suffice to escape a deep attractor, while large ones risk destabilising unrelated circuits. Control theory supplies the mathematical language to quantify this trade off and to design inputs that steer trajectories efficiently.

3.2 The Control Objective

Effective control of neural systems requires moving beyond open loop stimulation toward adaptive strategies informed by real time system state (Zanos et al., 2018). Closed loop sensing, using biomarkers such as oscillatory phase, individual alpha frequency, or network coherence, allows interventions to be timed and patterned to exploit endogenous dynamics (Arns et al., 2023). Adaptive patterning, rather than fixed pulses, enables resonance with specific circuit motifs, such as theta-gamma cross scale interaction implicated in memory and mood regulation (Buzsáki & Watson, 2012; Harris & Gordon, 2015). The formal control objective can thus be framed as a constrained optimisation problem: minimise cognitive disruption (loss of network segregation) subject to successfully driving the system from a pathological attractor to a healthy one. This formulation makes explicit why ECT’s efficacy and side-effect profile have remained inseparable, it is an open-loop, high-gain controller operating far from the brain’s native dynamics, and clarifies the path forward for truly precision neuromodulation.

Section 4: Comparative Survey of Convergent Technologies and a Unified Design Framework

Comparing neuromodulation modalities through the lens of scale of synchrony reveals a continuum rather than a hierarchy. ECT occupies the extreme of global, high amplitude synchronisation, reliably driving therapeutic state transitions at the cost of widespread network desegregation. In contrast, repetitive transcranial magnetic stimulation (rTMS) and focused ultrasound operate at lower amplitudes with markedly greater spatial specificity, while deep brain stimulation targets discrete nuclei yet still propagates through distributed networks. Magnetic seizure therapy (MST) occupies an instructive intermediate position, achieving seizure induction with a reduced spatial footprint of synchrony and correspondingly milder cognitive side effects (Lozano et al., 2019; Borrione et al., 2020).

This reframing highlights that therapeutic efficacy and cognitive risk are jointly determined by how broadly and coherently a modality perturbs network dynamics (Pascual-Leone et al., 2000; Loo et al., 2012). Viewed through this lens, differences between modalities are best understood not in terms of technological sophistication, but in terms of how precisely they can perturb network dynamics without inducing pathological levels of synchrony.

The preceding analysis implies that progress in neuromodulation will depend less on discovering new stimulation modalities than on integrating biological, physical, and computational constraints into a unified control framework. The future of neuromodulation therefore lies in precision non-implantable approaches that integrate insights across disciplines. Neuroscience must identify biomarkers signalling sufficient plasticity without over-integration (Vogelstein et al., 2014). Physics and engineering must refine field shaping and phased-array technologies to achieve deep, focal targeting (Caruso et al., 2016). Biology and chemistry offer complementary levers, such as pharmacologically assisted plasticity, that lower the energetic cost of state transitions and thereby reduce the need for disruptive stimulation (Fröhlich, 2016; Sunderam et al., 2010).

Conclusion

A truly side effect free ECT is likely physically implausible, as its therapeutic mechanism, large scale synchrony, is fundamentally incompatible with the functional segregation required for cognition. However, by reframing neuromodulation as a control problem in a nonlinear system, we can design next generation interventions that approximate ECT’s efficacy while operating closer to the brain’s native dynamics. The path forward lies not in eliminating perturbation, but in making it precise, adaptive, and proportionate. This reframing has implications beyond electroconvulsive therapy itself. It suggests that the historical search for a “gentler” ECT has been misdirected, insofar as it treats cognitive side effects as removable artefacts rather than as emergent properties of global network control. By contrast, a control theoretic perspective clarifies why seizure based efficacy and cognitive disruption have remained tightly linked despite decades of technical refinement.

More broadly, this framework challenges prevailing assumptions in psychiatric neuromodulation by shifting the focus from stimulation modality to system dynamics. Therapeutic success becomes a question of how precisely an intervention can bias network trajectories without collapsing functional segregation, rather than how forcefully it can disrupt neural tissue. This shift reframes future innovation as a problem of constrained optimisation under biological limits, rather than one of incremental technological escalation. In this sense, the contribution of the present framework is not the proposal of a specific alternative intervention, but the articulation of a principled boundary: any approach capable of matching ECT’s efficacy must either reproduce its large scale synchrony or find new ways to lower the energetic cost of state transitions. Progress will therefore depend on integrating advances in neuroscience, physics, and control theory to design interventions that operate within this boundary, rather than attempting to bypass it.

References

1. Abbott, C. C., et al. (2014). Electroconvulsive therapy and hippocampal volume. Biological Psychiatry, 75(1), 1-2.

   
   

2. Alipour, M., & Hajipour-Verdom, A. (2021). Neurobiophysics: An Introduction to the Physics of the Brain. Springer.

   
   

3. Arns, M., et al. (2023). Alpha peak frequency-based Brainmarker-I as a method to stratify to pharmacotherapy and brain stimulation treatments in depression. Nature Mental Health, 1(12), 1023-1032.

   
   

4. Blackmore, J., et al. (2019). Ultrasound Neuromodulation: A Review of Results, Mechanisms and Safety. Ultrasound in Medicine & Biology, 45(7), 1509-1536.

   
   

5. Borrione, L., et al. (2020). Precision non-implantable neuromodulation therapies: a perspective for the depressed brain. Brazilian Journal of Psychiatry, 42(4), 403-419.

   
   

6. Bullmore, E., & Sporns, O. (2009). Complex brain networks: graph theoretical analysis of structural and functional systems. Nature Reviews Neuroscience, 10(3), 186-198.

   
   

7. Buzsáki, G., & Watson, B. O. (2012). Brain rhythms and neural syntax. Neuron, 75(3), 364-375.

   
   

8. Caruso, L., et al. (2016). In vivo magnetic recording of neuronal activity. Cold Spring Harbor Laboratory.

   
   

9. Deco, G., et al. (2014). The dynamics of resting fluctuations in the brain. Nature Reviews Neuroscience, 15(1), 13-26.

   
   

10. Deng, Z. D., et al. (2011). Electric field modeling of ECT and MST. The Journal of ECT, 27(2), 135-143.

    
    

11. Fide, E., Bora, E., & Yener, G. (2025). Network Segregation and Integration Changes in Healthy Aging. European Journal of Neuroscience, 61.

    
    

12. Friston, K. J. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138.

    
    

13. Fröhlich, F. (2016). Network Neuroscience. Academic Press.

    
    

14. Harris, A. Z., & Gordon, J. A. (2015). Long-Range Neural Synchrony in Behavior. Annual Review of Neuroscience, 38(1), 171-194.

    
    

15. Holtzheimer, P. E., & Mayberg, H. S. (2011). Deep brain stimulation for treatment-resistant depression. Annual Review of Medicine, 62, 317-330.

    
    

16. Honey, C. J., et al. (2007). Network structure of cerebral cortex shapes functional connectivity on multiple time scales. PNAS, 104(24), 10240-10245.

    
    

17. Krystal, A. D., et al. (2003). The effect of seizure duration on the antidepressant efficacy of ECT. Archives of General Psychiatry, 60(10), 997-1002.

    
    

18. Leaver, A. M., et al. (2022). Parsing the Network Mechanisms of Electroconvulsive Therapy. Biological Psychiatry, 92(3), 193-203.

    
    

19. Loo, C. K., et al. (2012). Transcranial direct current stimulation (tDCS) in depression. The Journal of ECT, 28(1), 30-37.

    
    

20. Lozano, A. M., et al. (2019). Deep brain stimulation: current challenges and future directions. Nature Reviews Neurology, 15(3), 148-160.

    
    

21. Parpura, V., et al. (2012). Neuromodulation: selected approaches and challenges. Journal of Neurochemistry, 124(4), 436-453.

    
    

22. Pascual-Leone, A., et al. (2000). Transcranial magnetic stimulation in cognitive neuroscience. Current Opinion in Neurobiology, 10(2), 232-237.

    
    

23. Sackeim, H. A., et al. (2007). The cognitive effects of electroconvulsive therapy in community settings. Neuropsychopharmacology, 32(1), 244-254.

    
    

24. Sporns, O. (2011). Networks of the Brain. MIT Press.

    
    

25. Sunderam, S., et al. (2010). Feedback control of seizures. IEEE Reviews in Biomedical Engineering, 3, 116-130.

    
    

26. Uhlhaas, P. J., & Singer, W. (2010). Abnormal neural oscillations and synchrony in schizophrenia. Nature Reviews Neuroscience, 11(2), 100-113.

    
    

27. Van der Lande, G. J. M., et al. (2024). Brain state identification and neuromodulation. Brain Communications, 6(5).

    
    

28. Vogelstein, J. T., et al. (2014). Discovery of brainwide neural-behavioral maps. Science, 344(6182), 386-392.

    
    

29. Zanos, S., et al. (2018). Closed-loop neuromodulation in physiological and disease states. Current Opinion in Biomedical Engineering, 8, 13-22.

    
    

30. Zhou, X., et al. (2021). Optimal Organization of Functional Connectivity Networks. Frontiers in Computational Neuroscience, 15.

Acknowledgements / Transparency Statement

AI‑generated image. Editorial development and conceptual refinement were supported through collaborative use of AI tools, including Copilot, Grok, and Gemini.