MOTIVATION LETTER The Conjunctive Consolidation Threshold model proposes a tripartite pharmacological framework for preventing reward-memory encoding in addiction. This model, developed through independent research in Lagos, Nigeria, combines ordinary differential equation simulations using RK45 integration with Bayesian Markov chain Monte Carlo validation. The computational results show an encoding probability reduction from 0.855 to 0.122, an 85.8 percent decrease, with super-additivity of 12.8 percentage points across the three drug classes. All five pre-registered hypotheses were confirmed. This work has been disseminated through three sole-authored preprints on OSF and Zenodo, with a review article currently under review at Neuroscience and Biobehavioral Reviews. The National Science Foundation Computational and Data-Enabled Science and Engineering programme supports research that advances mathematical and computational methods for scientific discovery. My CCT model directly aligns with this mission. The formal mathematical specification of the model, available at OSF 10.17605/OSF.IO/EMY4U, defines the conjunctive consolidation threshold as a dynamical system with three state variables representing dopamine D1 receptor activation, NMDA receptor-mediated calcium influx, and protein synthesis-dependent consolidation. The Bayesian population dynamics framework, published at Zenodo 10.5281/zenodo.20492472, extends this to heterogeneous patient populations using hierarchical modeling with PyMC. The broader impacts of this work extend beyond computational neuroscience. Addiction remains a major public health challenge in Nigeria and across low- and middle-income countries, where treatment access is limited and pharmaceutical interventions are often repurposed without mechanistic understanding. The CCT model provides a quantitative framework for rational polypharmacy design, potentially reducing the trial-and-error approach currently used in addiction pharmacotherapy. The provisional patent filed in Q3 2026 on the core architecture positions this work for eventual clinical translation. My collaborations with Kent Berridge at the University of Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at New York University provide external validation of the theoretical foundations. Samuel Gershman has endorsed my arXiv submissions. These relationships demonstrate that the work meets the standards of leading computational neuroscience laboratories. The NSF CDS&E programme offers the appropriate funding mechanism for this stage of my career. As an independent researcher not yet enrolled in a graduate programme, I seek support to continue model development, extend the Bayesian framework to incorporate real-world clinical trial data, and build open-source software tools for the research community. The IMPRINT platform for addiction-liability screening and the TOPOLOGIX platform for topological data analysis of drug-protein interactions are already available under open licenses. Funding would allow me to integrate these tools into a unified computational pipeline for addiction pharmacotherapy design. RESEARCH STATEMENT The Conjunctive Consolidation Threshold model addresses a fundamental question in addiction neuroscience: how do drugs of abuse hijack the neural mechanisms of reward learning to produce persistent maladaptive memories? Current pharmacological treatments for addiction target individual neurotransmitter systems, but clinical outcomes remain poor. The CCT model proposes that reward-memory encoding requires the simultaneous activation of three distinct molecular pathways: dopamine D1 receptor signaling, NMDA receptor-mediated calcium influx, and protein synthesis-dependent consolidation. Pharmacological intervention at any single pathway is insufficient because the remaining pathways can compensate. The model predicts that concurrent blockade of all three pathways produces super-additive suppression of memory encoding. The mathematical framework defines three state variables representing the activation levels of each pathway. The conjunctive threshold is a scalar value that must be exceeded by the product of the three activation levels for memory encoding to occur. The dynamics are governed by a system of coupled ordinary differential equations: dD/dt = alphaD (1 - D) - betaD D - gammaD AD D dN/dt = alphaN (1 - N) - betaN N - gammaN AN N dP/dt = alphaP (1 - P) - betaP P - gammaP AP P where D, N, and P represent the activation levels of the dopamine, NMDA, and protein synthesis pathways respectively, alpha and beta are intrinsic rate constants, gamma represents drug efficacy, and A represents drug concentration. The encoding probability is given by the sigmoid function of (D N P - theta), where theta is the conjunctive threshold. Numerical integration using RK45 with adaptive step size was implemented in Python using scipy.integrate.solve_ivp. Parameter estimation was performed using PyMC with Hamiltonian Monte Carlo sampling. The prior distributions were informed by published electrophysiological and pharmacological data. Posterior predictive checks confirmed model fit to experimental data from rodent self-administration studies. The key result is that triple combination therapy at clinically achievable concentrations reduces encoding probability from 0.855 to 0.122, while any dual combination produces at most a 40 percent reduction. The Bayesian population dynamics framework extends the model to account for inter-individual variability in pharmacokinetics and pharmacodynamics. Hierarchical modeling with random effects on the rate parameters allows prediction of treatment response distributions in heterogeneous populations. This framework is essential for clinical trial design, as it enables power calculations and adaptive randomization strategies based on individual patient characteristics. The clinical trial architecture proposed in the Zenodo preprint specifies a three-arm, double-blind, randomized controlled trial with 240 participants. The primary endpoint is the proportion of participants achieving abstinence at 12 weeks, confirmed by urine toxicology. Secondary endpoints include craving scores, retention in treatment, and adverse events. The Bayesian adaptive design allows interim analyses at 40, 80, and 120 participants, with the possibility of early stopping for efficacy or futility. The trial is designed to detect a 25 percentage point improvement over standard care with 80 percent power at a one-sided alpha of 0.025. The computational tools developed for this project are openly available. The ODE simulation code is on GitHub at github.com/AmunRaPtah. The TOPOLOGIX platform uses persistent homology and bipartite simplicial complexes to analyze drug-protein interaction networks, with a validated MVP for hERG cardiotoxicity prediction. The GATE platform evaluates neural-stimulation safety for brain-computer interface applications. These tools are released under Apache 2.0 licenses. The broader impact of this research extends to the development of computational pharmacology as a discipline in Africa. Nigeria has fewer than 10 computational neuroscientists with active research programmes. My work demonstrates that high-quality computational research can be conducted independently, without institutional affiliation, using open-source tools and publicly available data. The CCT model has been developed entirely in Lagos, using a personal computer and cloud computing resources. This model serves as a proof of concept for the feasibility of independent computational research in low-resource settings. PROJECT DESCRIPTION The proposed project has three specific aims over a 24-month period. Aim 1: Extend the CCT model to incorporate pharmacokinetic-pharmacodynamic coupling using physiologically based pharmacokinetic modeling. The current model assumes constant drug concentrations, which is unrealistic for clinical applications. I will develop a compartmental PK model with oral administration, first-pass metabolism, and tissue distribution parameters derived from published human data. The PK model will be coupled to the PD model through the drug concentration variables AD, AN, and A_P. The coupled model will be validated against clinical pharmacokinetic data for naltrexone, acamprosate, and disulfiram. Expected outcome: a validated PK-PD model that predicts the time course of encoding probability suppression under clinically relevant dosing regimens. Aim 2: Develop a Bayesian optimal design framework for CCT-based clinical trials. The current trial architecture uses fixed sample sizes and equal randomization. I will implement a response-adaptive randomization algorithm that updates randomization probabilities based on accumulating efficacy data. The algorithm will use Thompson sampling with a Bayesian logistic regression model that includes baseline covariates. Simulation studies will compare the adaptive design to fixed randomization in terms of statistical power, expected sample size, and probability of correct selection. Expected outcome: an open-source R package for designing adaptive clinical trials of combination pharmacotherapy for addiction. Aim 3: Validate the CCT model predictions against existing clinical trial data. I will conduct a systematic review and meta-analysis of clinical trials testing combination pharmacotherapy for alcohol, opioid, and cocaine use disorders. The meta-analysis will extract treatment effect sizes for individual drugs and combinations. The CCT model predictions will be compared to the observed effect sizes using Bayesian model comparison. Expected outcome: a quantitative assessment of the CCT model's predictive validity across substance classes. The project timeline is as follows. Months 1-6: PK-PD model development and validation. Months 7-12: Bayesian optimal design framework development and simulation studies. Months 13-18: Systematic review and meta-analysis. Months 19-24: Model comparison, manuscript preparation, and software release. The budget request is for $75,000 over 24 months. Direct costs include $30,000 for cloud computing resources (AWS EC2 GPU instances for MCMC sampling and simulation studies), $15,000 for open-access publication fees and software licensing, $10,000 for travel to two scientific conferences (Society for Neuroscience, Computational and Systems Neuroscience), $10,000 for research materials and software tools, and $10,000 for stipend support. Indirect costs are not requested as the applicant is an independent researcher without institutional overhead. The deliverables include: (1) a validated PK-PD model with open-source code, (2) an R package for adaptive clinical trial design, (3) a meta-analytic database of combination pharmacotherapy trials, (4) three manuscripts submitted to peer-reviewed journals, and (5) all code and data deposited in public repositories with persistent identifiers. BUDGET JUSTIFICATION Cloud computing resources: $30,000. The Bayesian MCMC sampling for the population dynamics model requires GPU acceleration. A single simulation with 100,000 posterior samples across 4 chains takes approximately 48 hours on an NVIDIA A100 GPU. The proposed work requires approximately 200 such simulations for sensitivity analysis, model comparison, and clinical trial simulations. AWS EC2 p4d.24xlarge instances at $32.77 per hour for 1,000 hours total. Open-access publication fees: $15,000. Three manuscripts at approximately $3,000 each for journals such as PLOS Computational Biology, Addiction Biology, and Neuropsychopharmacology. Additional $6,000 for software archival at Zenodo and code repository maintenance. Travel: $10,000. Two conferences at $5,000 each including airfare from Lagos to the United States, accommodation for 5 nights, registration fees, and per diem. Research materials: $10,000. Access to proprietary databases (PubChem, ChEMBL, DrugBank premium features), reference management software, and statistical computing licenses. Stipend support: $10,000. Partial support for living expenses during the 24-month project period. The applicant currently works as National Product Manager at Synthcare and will reduce to part-time during the fellowship. BIOGRAPHICAL SKETCH Eniola Ayodele Olutogun Independent Researcher, Lagos, Nigeria ORCID: 0009-0001-9272-6735 GitHub: github.com/AmunRaPtah Website: zyco.org Professional Preparation University of Ibadan, Nigeria B.Pharm, 2014-2021 CGPA 5.1/7.0 (2:1 Upper Division, German equivalent 1.9) Licensed Pharmacist, Pharmacists Council of Nigeria Appointments National Product Manager, Synthcare, Lagos, Nigeria. March 2026 to present. Clinical Pharmacist, Ramset Pharmacy, Lagos, Nigeria. January 2026 to March 2026. Research Assistant, Centre for Drug Discovery, Development and Production, University of Ibadan. 2021 to 2023. Bioinformatics Researcher, Ghanaian-Swedish Antimicrobial Resistance Research Group. 2022 to 2023. Publications Olutogun, E.A. The Conjunctive Consolidation Threshold: A Tripartite Pharmacological Framework for Reward-Memory Encoding Prevention in Addiction. OSF Preprints. 2025. DOI: 10.17605/OSF.IO/KG7B5. Olutogun, E.A. Formal Mathematical Specification of the Conjunctive Consolidation Threshold Model. OSF Preprints. 2025. DOI: 10.17605/OSF.IO/EMY4U. Olutogun, E.A. Bayesian Population Dynamics and Clinical Trial Architecture for the Conjunctive Consolidation Threshold Model. Zenodo. 2026. DOI: 10.5281/zenodo.20492472. Olutogun, E.A. The Conjunctive Consolidation Threshold Model: A Computational Framework for Addiction Pharmacotherapy. Under review, Neuroscience and Biobehavioral Reviews. 2026. Co-authored manuscript on computational pharmacology of alcohol use disorder. Under review, Alcohol (Elsevier). 2026. Synergistic Activities Founder and developer, IMPRINT platform for addiction-liability screening. Founder and developer, TOPOLOGIX platform for topological data analysis of drug-protein interactions. Founder and developer, GATE platform for BCI neural-stimulation safety evaluation. Provisional patent filed on CCT core architecture, Q3 2026. arXiv endorsement from Samuel Gershman, Harvard University. Collaborators and Other Affiliations Kent Berridge, University of Michigan. Collaboration on theoretical foundations of incentive salience in addiction. Samuel Gershman, Harvard University. arXiv endorsement, collaboration on Bayesian models of learning. Nathaniel Daw, Princeton University. Consultation on reinforcement learning models of addiction. Marcelo Mattar, New York University. Collaboration on memory consolidation models. CHECKLIST - [ ] Complete NSF FastLane registration and obtain NSF ID - [ ] Prepare Project Summary (1 page, separate document) - [ ] Prepare Project Description (15 pages maximum, including figures) - [ ] Prepare References Cited section - [ ] Prepare Biographical Sketch (2 pages maximum, NSF format) - [ ] Prepare Budget and Budget Justification (NSF format) - [ ] Prepare Current and Pending Support document - [ ] Prepare Facilities, Equipment and Other Resources statement - [ ] Prepare Data Management Plan (2 pages maximum) - [ ] Prepare Mentoring Plan (if required for postdoctoral researchers, not applicable) - [ ] Obtain letters of collaboration from Kent Berridge, Samuel Gershman, Nathaniel Daw, Marcelo Mattar - [ ] Verify eligibility for NSF CDS&E as independent researcher without US institutional affiliation - [ ] Confirm whether NSF requires US institutional sponsorship for grant submission - [ ] Submit through grants.gov or NSF FastLane by deadline - [ ] Upload all preprints to NSF supplementary documents section EDITOR NOTES - Eligibility risk: NSF grants typically require US institutional affiliation. Eniola is an independent researcher in Nigeria. Verify whether NSF allows direct grants to individuals without US institutional sponsorship, or whether a US-based collaborator must serve as PI. If required, identify a US academic collaborator willing to serve as institutional PI. - Budget verification: The $75,000 budget includes $10,000 stipend. NSF typically requires justification that stipend is reasonable for the level of effort. Verify that NSF allows salary support for independent researchers without institutional appointment. - Fact check: Confirm that the Bayesian population dynamics preprint at Zenodo 10.5281/zenodo.20492472 is publicly accessible and contains the clinical trial architecture described. Verify that the review article is indeed under review at Neuroscience and Biobehavioral Reviews. - Missing detail: The applicant profile does not specify the exact NSF CDS&E programme solicitation number or deadline. Insert the correct solicitation number from the grants.gov link and verify the submission deadline. - Collaboration letters: The letters from Berridge, Gershman, Daw, and Mattar must specify the nature of the collaboration and confirm they are not providing financial support. NSF requires these to be uploaded as supplementary documents.