Data-driven control of primate prefrontal neural activity using patterned microstimulation

PROJECT SUMMARY Artificially controlling neural activity in real-time is a central challenge in neuroscience and neuroengineering, with significant implications for understanding brain function and developing treatments for neurological and psychiatric disorders. Despite advances in high-precision neural perturbation technologies, a critical gap exists in predicting and shaping the effects of these perturbations on neural population activity and behavior. Our proposed research addresses this gap by developing a data-driven framework for real-time neural control, integrating theoretical models, empirical analyses, and advanced stimulation techniques. Among circuit perturbation methods, the effects of electrical stimulation remain particularly difficult to predict. However, this technique is widely used in clinical treatments of neurological and psychiatric disorders, including Parkinson’s disease, essential tremor, mood disorders, and obsessive-compulsive disorder. To maximize translational impact and showcase the strengths of our theoretical framework, we focus on multi-electrode patterned microstimulation sequences for circuit perturbations. In Aim 1, we investigate the spatial and temporal summation rules governing multi-electrode microstimulation in monkey prefrontal cortex. By delivering spatiotemporally patterned stimulation and measuring population responses, we aim to uncover the principles of nonlinear summation and interaction effects, such as whether responses are stronger for spatially or temporally proximal stimulations and for stimulation of high-impact “hub” electrodes. We also examine trial-to- trial variability in stimulation responses as a function of network state, offering insights into how pre-existing activity modulates stimulation effects. In Aim 2, we introduce a novel data-driven algorithm called Input/Output Control (I/O-Ctrl) that predicts optimal stimulation patterns to evoke specific neural activity states. This approach requires no prior knowledge of circuit dynamics or connectomes, instead relying on short training sequences of stimulation-response data. By applying I/O-Ctrl to prefrontal circuits, we aim to develop a generalizable framework for targeted neural modulation. In Aim 3, we extend these insights to behavioral control, exploring how neural perturbations in prefrontal cortical circuits influence decision-making in primates. Through closed-loop adaptive stimulation, we aim to identify neural mechanisms that bias choices and examine the causal relationships between population activity patterns and behavior. The proposed work is transformative, addressing key challenges in understanding and controlling neural circuits. By combining computational modeling, multi-electrode stimulation methods, and real-time neural control frameworks, this research has the potential to advance brain-computer interfaces, inform treatments for motor and cognitive disorders, and uncover fundamental principles of brain function. These efforts will significantly enhance our ability to manipulate and interpret neural circuits, with broad implications for neuroscience, medicine, and artificial intelligence.