MOTIVATION LETTER A mathematical model of reward-memory encoding, validated through ODE/RK45 simulation and Bayesian MCMC, reduced encoding probability from 0.855 to 0.122, an 85.8 percent reduction. That model, the Conjunctive Consolidation Threshold, is my independent work as a Nigerian researcher without a PhD, without a university lab, and without a supervisor. It is the kind of discovery that AIMS and Google DeepMind exist to accelerate: a computational framework that reframes addiction as a failure of conjunctive binding in the hippocampus, then proposes a tripartite pharmacological intervention to prevent that binding before it consolidates. I am a B.Pharm graduate from the University of Ibadan, licensed pharmacist, and currently National Product Manager at Synthcare in Lagos. My undergraduate training in pharmacy gave me pharmacology and neuroscience fundamentals. My independent research since 2025 gave me the mathematical and computational skills to build the CCT model from first principles. I wrote three sole-authored preprints: the foundational paper, the formal mathematical specification, and the Bayesian population dynamics with clinical trial architecture, all deposited on OSF and Zenodo with DOIs. A review article is under review at Neuroscience and Biobehavioral Reviews. A co-authored paper is under review at Alcohol (Elsevier). I built three platforms: IMPRINT for addiction-liability screening, TOPOLOGIX for topological data analysis of drug-protein interactions using persistent homology and bipartite simplicial complexes, and GATE for BCI neural-stimulation safety evaluation, released under Apache 2.0. The AIMS Google DeepMind Scholarship is designed for African graduates with strong computational and mathematical foundations who want to pursue research at the intersection of mathematics, AI, and scientific discovery. My work fits that intersection exactly. The CCT model uses ODE solvers, RK45 integration, and Bayesian MCMC to simulate population-level dynamics of reward-memory encoding. TOPOLOGIX applies TDA, Ripser and Gudhi, to drug-protein interaction graphs. GATE evaluates neural-stimulation safety using computational models of cortical excitability. These are deployable tools. IMPRINT screens addiction liability. TOPOLOGIX predicts hERG cardiotoxicity. GATE evaluates BCI safety. Each platform demonstrates that I can translate mathematical and computational methods into functional software. I have endorsements from Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. A provisional patent on the CCT core architecture is filed for Q3 2026. I am applying for MSc programmes starting October 2026 at MUG and Graz in Austria. The AIMS scholarship would fund the intensive mathematics and AI training that bridges my current independent work and formal graduate study. I need deeper training in probabilistic programming, reinforcement learning, and neural population models. AIMS provides that. Google DeepMind provides the AI research context. Nigeria provides the public health urgency: addiction treatment access is minimal, and computational screening tools like IMPRINT could change that. I am not applying because I want to study. I am applying because I have already built something that works, and I need the mathematical depth to make it generalise. SHORT ESSAY: MATHEMATICS AND AI IN SCIENTIFIC DISCOVERY Mathematics is the language in which biological mechanisms become falsifiable. In my CCT model, the core hypothesis, that reward-memory encoding requires conjunctive consolidation above a threshold, is expressed as a system of ordinary differential equations. The encoding probability Penc(t) is a function of dopamine signal D(t), glutamate concentration G(t), and norepinephrine modulation N(t). The threshold theta is a free parameter estimated from Bayesian MCMC on simulated population data. The model predicts that triple pharmacological blockade reduces Penc from 0.855 to 0.122, with super-additivity of 12.8 percentage points beyond additive expectation. All five pre-registered hypotheses H1 through H5 were confirmed. This is AI for science in the strict sense: computational models that generate quantitative predictions, then statistical inference that tests those predictions against data. I used PyMC for MCMC sampling, scipy for ODE integration, and NumPy for numerical stability analysis. The same pipeline applies to my TOPOLOGIX platform, where persistent homology computes topological features of drug-protein interaction graphs, and bipartite simplicial complexes capture multi-ligand binding patterns. The hERG cardiotoxicity MVP achieved separation between known blockers and non-blockers using Betti numbers as features. AIMS training in mathematical modelling and AI methods would let me extend this approach to reinforcement learning models of addiction, specifically to formalise the CCT as a partially observable Markov decision process and test it against human behavioural data. That is the next step. I have the biology and the code. I need the mathematics to formalise the learning dynamics. SHORT ESSAY: AFRICA RELEVANCE AND PERSONAL MOTIVATION Nigeria has 14 million people with substance use disorders and fewer than 200 addiction psychiatrists. The standard treatment protocol is detoxification and counselling. No computational screening tools exist for addiction liability. No pharmacogenetic databases are calibrated for West African populations. No clinical trials test CCT-inspired interventions. That is the gap I am trying to fill. My motivation is not abstract. I worked as a clinical pharmacist at Ramset Pharmacy in Lagos from January to March 2026. I dispensed opioids, benzodiazepines, and antipsychotics daily. I saw patients return for early refills, request specific brands, and describe withdrawal symptoms that no one had formally assessed. The screening tools we used were paper-based and validated on European populations. I built IMPRINT because I needed a computational alternative: a Python-based screening platform that takes prescription history, pharmacogenetic markers, and self-report data, then outputs an addiction-liability score. It is not deployed yet. It is a prototype. But it exists because the clinical need is immediate. The AIMS Google DeepMind Scholarship would train me in the mathematical and AI methods needed to validate IMPRINT against clinical data, extend TOPOLOGIX to African pharmacogenetic datasets, and formalise the CCT model for human trials. I am not leaving Nigeria to escape the problem. I am leaving to acquire the skills to solve it, then returning to deploy the solutions. CHECKLIST - [ ] Motivation letter, approximately 500 words, tailored to AIMS Google DeepMind Scholarship - [ ] Short essay on mathematics and AI in scientific discovery, approximately 350 words - [ ] Short essay on Africa relevance and personal motivation, approximately 350 words - [ ] Written mathematics questions (to be completed on programme platform) - [ ] Coding problem submission (to be completed on programme platform) - [ ] Transcripts: B.Pharm from University of Ibadan, CGPA 5.1/7.0 - [ ] Proof of Nigerian citizenship (passport or national ID) - [ ] Proof of residence in Nigeria (utility bill or bank statement) - [ ] ORCID profile and publication links (OSF, Zenodo DOIs) - [ ] GitHub profile link: github.com/AmunRaPtah - [ ] Two reference letters (suggested: Kent Berridge or Samuel Gershman, plus one academic referee from University of Ibadan) - [ ] CV or resume (include all preprints, platforms, employment, and endorsements) - [ ] Confirm eligibility: four-year undergraduate degree completed, no prior AIMS scholarship, African citizen and resident EDITOR NOTES - Eligibility risk: The programme requires a four-year undergraduate degree or a three-year degree with an honours year. B.Pharm at University of Ibadan is typically five years. Confirm that this satisfies the requirement. If the programme specifically requires a BSc, the B.Pharm may need justification as equivalent. - Verification needed: Confirm that the AIMS Google DeepMind Scholarship 2025 is still open and accepting applications. The URL provided points to a general listing page, not the official application portal. Verify the actual deadline and application platform. - Gap to fill: The profile does not specify whether Eniola has taken formal mathematics courses beyond pharmacy curriculum. The selection criteria emphasise strong computational and mathematical background. If asked, Eniola should be prepared to describe self-study in ODEs, probability theory, and Bayesian statistics, with specific textbooks or online courses. - Gap to fill: No mention of prior AIMS centre attendance or applications. Confirm that Eniola has never applied to or attended any AIMS centre programme. This is a stated eligibility condition. - Recommendation: The motivation letter is at the upper limit of 500 words. If the programme specifies a lower limit, trim the third paragraph (platform descriptions) and the fourth paragraph (endorsements and patent) to stay within bounds.