Researcher ORCID Identifier
0009-0005-5323-9546
Graduation Year
2025
Document Type
Open Access Senior Thesis
Degree Name
Bachelor of Arts
Department
W.M. Keck Science Department
Second Department
Human Biology
Reader 1
Sarah Marzen
Reader 2
Jenna Monroy
Terms of Use & License Information
Rights Information
© 2025 Richard Ampah
Abstract
This study presents an original interdisciplinary investigation into how reinforcement learning (RL) can model motor and cognitive defects and potentially improve motor and cognitive functions in individuals with cerebral palsy (CP), a non-progressive neurological disorder that impairs movement and adaptability. Integrating computational neuroscience and machine learning, the research applies policy gradient methods and Markov Decision Processes (MDPs) to simulate adaptive learning in agents with and without CP-related constraints.
The central aim is to compare the cumulative rewards of optimal policies, derived from value iteration, and human-like learning policies using the REINFORCE algorithm, both with and without the Bellman baseline. The Bellman baseline serves to reduce gradient variance, simulating efficient cognitive feedback mechanisms. Two probabilistic frameworks are developed: the Innate Policy, representing uniform state knowledge from birth, and the Adaptive Policy, which models real-world learning through stationary state distributions shaped by an agent’s experience.
Simulations across 1,000 randomized environments with 10 states and 11 actions showed that REINFORCE with a Bellman baseline (non-CP model) significantly outperformed its non-baseline counterpart (cpREINFORCE, CP model), with a mean reward difference of 4.13 and a p-value of 1.32 × 10⁻⁴⁹ (p < 0.001). These findings support the hypothesis that baseline-guided reflects the adaptive strategies observed in individuals with CP undergoing effective therapy.
By mathematically modeling motor and cognitive defects, this research provides a novel computational framework to understand and simulate motor and cognitive challenges associated with CP. It advances the intersection of neuroscience and machine learning, with potential applications in clinical treatment design and adaptive rehabilitation strategies targeting neural adaptability.
Recommended Citation
Ampah, Richard, "Optimizing Decision-Making in a Cerebral Palsy Model Using Reinforcement Learning" (2025). Pitzer Senior Theses. 221.
https://scholarship.claremont.edu/pitzer_theses/221
Included in
Applied Mathematics Commons, Artificial Intelligence and Robotics Commons, Bioinformatics Commons, Biomedical Informatics Commons, Cognitive Science Commons, Data Science Commons, Developmental Biology Commons, Disability Studies Commons, Mathematics Commons, Medical Anatomy Commons, Medical Neurobiology Commons, Medical Physiology Commons, Nervous System Commons, Neurology Commons, Neurosciences Commons, Neurosurgery Commons, Other Rehabilitation and Therapy Commons, Pediatrics Commons, Physical Therapy Commons, Physiotherapy Commons, Probability Commons, Theory and Algorithms Commons, Translational Medical Research Commons
Comments
Modeling Neural Adaptation in Cerebral Palsy Through Reinforcement Learning
This research explores how machine learning, particularly reinforcement learning (RL), can model motor and cognitive defects, and simulate and support motor and cognitive development in individuals with cerebral palsy (CP). By combining policy gradient algorithms with biologically inspired modeling, the study examines how learning policies differ in agents with and without CP-related constraints. Through thousands of randomized simulations, it demonstrates the power of adaptive learning and the role of neural feedback in optimizing rehabilitation. This interdisciplinary work bridges computational neuroscience, AI, and clinical theory to offer insights into how artificial agents can mimic, and potentially enhance real-world therapeutic strategies that will improve cerebral function.