Draft — CIFAR Global Scholars

CIFAR GLOBAL SCHOLARS APPLICATION DRAFT

MOTIVATION LETTER

The Conjunctive Consolidation Threshold model, or CCT, is a tripartite pharmacological framework for preventing reward-memory encoding in addiction. It couples three axes, dopaminergic reward prediction error, NMDAR-dependent long-term potentiation, and affective contrast, into a single system of ordinary differential equations solved with RK45 and calibrated through Bayesian MCMC using PyMC's DEMetropolisZ sampler with 14 free parameters. The priors were elicited from a systematic screen of 1,847 records in the literature. All five pre-registered hypotheses, H1 through H5, were confirmed, with posterior super-additivity of 13 to 22 percentage points across model versions. Three sole-authored preprints are currently under review at peer-reviewed journals, and a co-authored paper is under review at Alcohol, an Elsevier journal.

CIFAR Global Scholars selects early-career researchers who demonstrate research excellence, originality, leadership potential, and interdisciplinary breadth. The CCT model is precisely this kind of work. It sits at the intersection of computational pharmacology, dynamical systems theory, and Bayesian statistics. It addresses a fundamental question in addiction neuroscience: can a memory trace be prevented from consolidating at the moment of reward encoding, rather than treated after it has formed? This is a mechanistic, theory-driven question, not an applied engineering task. It aligns directly with CIFAR's research programs in Brain, Mind and Consciousness and Learning in Machines and Brains.

The model has received endorsement from senior researchers whose work defines the field. Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU have all engaged with the work. These are researchers whose own contributions to reward processing, computational psychiatry, and reinforcement learning provide the intellectual scaffolding for the CCT framework. Their engagement signals that the model is taken seriously at the highest level of the field.

My path to this work is unconventional. I trained as a pharmacist at the University of Ibadan, graduating with a CGPA of 5.1 out of 7.0, a German equivalent of 1.9. I have worked as a clinical pharmacist and as a national product manager. I am currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute and the University of Potsdam. This trajectory has given me a practical grounding in pharmacology that purely computational researchers often lack, and a computational rigor that purely clinical researchers rarely develop. The CCT model could not have been built without both.

CIFAR's emphasis on global diversity and on researchers who can contribute to cross-disciplinary dialogue is a direct match for my situation. I am an independent researcher based in Nigeria, building models that require expertise in neuroscience, pharmacology, and applied mathematics. The CIFAR network would provide the collaborative environment that independent work cannot offer alone. I am applying to contribute to the program's fundamental research mission, and to bring a perspective shaped by African research contexts into that conversation.

RESEARCH STATEMENT

The CCT model addresses a specific gap in addiction research. Current pharmacological interventions for substance use disorder largely target the acute rewarding effects of drugs or manage withdrawal symptoms. Few approaches attempt to intervene at the level of memory consolidation, the process by which a drug-associated reward becomes encoded as a durable, cue-triggered memory that drives relapse. The CCT model proposes that three concurrent neurobiological processes must cross a conjunctive threshold for this encoding to occur: dopaminergic reward prediction error signaling, NMDAR-dependent long-term potentiation in reward circuitry, and affective contrast, the emotional salience differential between the drug state and the baseline state. If any one of these axes is suppressed below threshold during the critical consolidation window, the reward-memory trace fails to form.

The model is implemented as a coupled ODE system. The dopaminergic axis is modeled as a reward prediction error signal following a temporal difference formulation. The NMDAR axis is modeled as a calcium-dependent plasticity variable with a sigmoidal activation function. The affective contrast axis is modeled as a slow-moving homeostatic variable that tracks the difference between drug-induced affective state and baseline. These three axes are coupled through a conjunctive operator that gates the consolidation rate. The system is solved with RK45 and calibrated against published electrophysiological and behavioral data using Bayesian MCMC. The 14 free parameters are constrained by priors elicited from a structured screen of 1,847 records spanning pharmacology, electrophysiology, and behavioral neuroscience.

The central finding is that the three axes interact super-additively. Suppressing any single axis reduces consolidation probability, but suppressing two axes simultaneously produces a reduction 13 to 22 percentage points greater than the sum of the individual effects. This super-additivity is the mechanistic core of the model and the basis for its clinical implication: combination pharmacotherapy targeting two or three axes simultaneously may be dramatically more effective than single-target interventions, at lower doses of each individual drug. This is a testable prediction that can be evaluated in animal models and, eventually, in human trials.

The model has been validated through pre-registered hypothesis testing. All five hypotheses, H1 through H5, were specified before data analysis and all five were confirmed. The preprints are under review at three journals: International Addiction Research and Therapy, Progress in Neuro-Psychopharmacology and Biological Psychiatry, and Neuroscience and Biobehavioral Reviews. A co-authored paper extending the framework is under review at Alcohol.

The CCT model is not the only research line I pursue. I have conducted a pre-registered, powered replication study on the use of bipartite persistent homology for predicting hERG cardiotoxicity, which found that topological features do not beat a plain descriptor baseline, AUROC 0.8426 versus 0.8782. I have extended this topological approach to drug-resistance prediction and found that interface geometry carries almost no signal for that task, AUROC 0.425 and 0.485 on the Platinum benchmark. These negative results led me to develop TOPOLOGIX, a sequence-based approach using ESM-2 protein language model delta-embeddings and Morgan fingerprints with a Random Forest classifier, which achieves AUROC 0.804 plus or minus 0.025 on the Platinum benchmark and covers 100 percent of mutations versus roughly 18 percent for structure-limited tools. I have also built neurocascade, a receptor-to-behavior brain-circuit simulation engine, and ergofluids, a Koopman-operator method for modeling drug transport in tumor tissue, the latter of which is honestly reported as not yet meeting its primary real-data validation criterion.

The CCT model is the line of work I am putting forward for CIFAR Global Scholars because it is the most mature, the most directly aligned with CIFAR's Brain, Mind and Consciousness program, and the one with the clearest trajectory from fundamental mechanism to societal impact in addiction treatment. The negative results in topology and the honest reporting of ergofluids' validation status demonstrate the rigor with which I conduct research. The CCT model is the culmination of that rigor applied to a problem of genuine clinical urgency.

SHORT ANSWER ESSAY: INTERDISCIPLINARY APPROACH

The CCT model is interdisciplinary by construction, not by decoration. It requires simultaneous fluency in three domains that rarely speak to each other. The first is molecular pharmacology: the NMDAR axis is grounded in the specific kinetics of glutamate receptor subtypes, the dopaminergic axis in the phasic firing properties of ventral tegmental area neurons, and the affective contrast axis in the neurobiology of opponent processes. The second is dynamical systems theory: the model is a coupled ODE system whose behavior depends on timescale separation, fixed points, and bifurcation structure, not on simple input-output mapping. The third is Bayesian statistics: the model's parameters are not fitted by point estimation but inferred as posterior distributions, with uncertainty propagated through every prediction.

This combination is rare. Pharmacologists rarely build ODE models with Bayesian calibration. Dynamical systems theorists rarely engage with the specifics of NMDA receptor subunit composition. Bayesian statisticians rarely work with clinical pharmacology data. The CCT model forces all three together because the scientific question demands it. The question, can reward-memory encoding be prevented by multi-target pharmacological intervention, cannot be answered by any single discipline alone.

The model's development process reflects this interdisciplinary commitment. The priors were elicited from a structured literature screen of 1,847 records spanning pharmacology, electrophysiology, and behavioral neuroscience. The model was implemented in Python using scipy and PyMC, with the ODE solver and the MCMC sampler integrated into a single pipeline. The pre-registration specified all five hypotheses before any analysis was run. The negative results from my topology studies, which I reported directly rather than reframing, demonstrate that I apply the same rigor to falsifying my own hypotheses as to confirming them.

CIFAR's selection criteria emphasize interdisciplinary breadth and the ability to contribute to cross-disciplinary dialogue. The CCT model is evidence of both. It is a contribution to addiction neuroscience that could not have been made without computational pharmacology, and a contribution to computational pharmacology that could not have been made without clinical training. I am applying to CIFAR Global Scholars because the program's structure, which brings researchers from different disciplines into sustained conversation, is the environment in which this kind of work thrives.

CHECKLIST

- [ ] Verify current CIFAR Global Scholars eligibility criteria on the program website, specifically the nomination requirement and whether independent researchers without institutional affiliation are eligible
- [ ] Confirm the application deadline and submission portal on the CIFAR website
- [ ] Identify and contact a CIFAR program director or current CIFAR Fellow who can serve as nominator, ideally within the Brain, Mind and Consciousness or Learning in Machines and Brains programs
- [ ] Prepare a current CV formatted to CIFAR specifications, including all publications, preprints, and pre-registrations
- [ ] Compile the three CCT preprints and the co-authored Alcohol paper as supporting documents
- [ ] Obtain letters of support from Kent Berridge, Samuel Gershman, Nathaniel Daw, and Marcelo Mattar, confirming their willingness to endorse the CCT model and the application
- [ ] Draft a one-page research summary of the CCT model suitable for a non-specialist reviewer
- [ ] Prepare a statement of institutional support or affiliation, if required, given independent researcher status
- [ ] Confirm the M.Sc. enrollment at Hasso Plattner Institute and University of Potsdam is reflected in the application as current institutional affiliation
- [ ] Review all materials for consistency with the pre-registration documents and the actual results reported in the preprints

EDITOR NOTES

- Eligibility risk: CIFAR Global Scholars typically requires nomination by a CIFAR program director or current Fellow, and self-nomination is not accepted. The applicant must secure a nominator before the deadline, and this is the single largest risk to the application. The endorsements from Berridge, Gershman, Daw, and Mattar are strong but do not guarantee any of them is a current CIFAR Fellow or program director.
- The applicant's independent researcher status may be a barrier if CIFAR requires affiliation with a research institution. The M.Sc. enrollment at HPI/Potsdam may satisfy this requirement, but this must be verified with the program directly.
- The CCT model is the correct research line to lead with for this program, but the applicant should be prepared to explain why the negative topology results and the honest reporting of ergofluids' validation status are strengths, not weaknesses, in the context of CIFAR's emphasis on fundamental research rigor.
- The applicant must insert personal details not present in this profile: specific dates of pre-registration, the exact journal submission status of each preprint, and any conference presentations or invited talks related to the CCT model.
- The word counts for the motivation letter and research statement should be verified against the actual CIFAR application form, as the program may have specific limits that differ from the fallback limits used here.

CIFAR GLOBAL SCHOLARS APPLICATION DRAFT

MOTIVATION LETTER

The Conjunctive Consolidation Threshold model, or CCT, is a tripartite pharmacological framework for preventing reward-memory encoding in addiction. It couples three axes, dopaminergic reward prediction error, NMDAR-dependent long-term potentiation, and affective contrast, into a single system of ordinary differential equations solved with RK45 and calibrated through Bayesian MCMC using PyMC's DEMetropolisZ sampler with 14 free parameters. The priors were elicited from a systematic screen of 1,847 records in the literature. All five pre-registered hypotheses, H1 through H5, were confirmed, with posterior super-additivity of 13 to 22 percentage points across model versions. Three sole-authored preprints are currently under review at peer-reviewed journals, and a co-authored paper is under review at Alcohol, an Elsevier journal.

CIFAR Global Scholars selects early-career researchers who demonstrate research excellence, originality, leadership potential, and interdisciplinary breadth. The CCT model is precisely this kind of work. It sits at the intersection of computational pharmacology, dynamical systems theory, and Bayesian statistics. It addresses a fundamental question in addiction neuroscience: can a memory trace be prevented from consolidating at the moment of reward encoding, rather than treated after it has formed? This is a mechanistic, theory-driven question, not an applied engineering task. It aligns directly with CIFAR's research programs in Brain, Mind and Consciousness and Learning in Machines and Brains.

The model has received endorsement from senior researchers whose work defines the field. Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU have all engaged with the work. These are researchers whose own contributions to reward processing, computational psychiatry, and reinforcement learning provide the intellectual scaffolding for the CCT framework. Their engagement signals that the model is taken seriously at the highest level of the field.

My path to this work is unconventional. I trained as a pharmacist at the University of Ibadan, graduating with a CGPA of 5.1 out of 7.0, a German equivalent of 1.9. I have worked as a clinical pharmacist and as a national product manager. I am currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute and the University of Potsdam. This trajectory has given me a practical grounding in pharmacology that purely computational researchers often lack, and a computational rigor that purely clinical researchers rarely develop. The CCT model could not have been built without both.

CIFAR's emphasis on global diversity and on researchers who can contribute to cross-disciplinary dialogue is a direct match for my situation. I am an independent researcher based in Nigeria, building models that require expertise in neuroscience, pharmacology, and applied mathematics. The CIFAR network would provide the collaborative environment that independent work cannot offer alone. I am applying to contribute to the program's fundamental research mission, and to bring a perspective shaped by African research contexts into that conversation.

RESEARCH STATEMENT

The CCT model addresses a specific gap in addiction research. Current pharmacological interventions for substance use disorder largely target the acute rewarding effects of drugs or manage withdrawal symptoms. Few approaches attempt to intervene at the level of memory consolidation, the process by which a drug-associated reward becomes encoded as a durable, cue-triggered memory that drives relapse. The CCT model proposes that three concurrent neurobiological processes must cross a conjunctive threshold for this encoding to occur: dopaminergic reward prediction error signaling, NMDAR-dependent long-term potentiation in reward circuitry, and affective contrast, the emotional salience differential between the drug state and the baseline state. If any one of these axes is suppressed below threshold during the critical consolidation window, the reward-memory trace fails to form.

The model is implemented as a coupled ODE system. The dopaminergic axis is modeled as a reward prediction error signal following a temporal difference formulation. The NMDAR axis is modeled as a calcium-dependent plasticity variable with a sigmoidal activation function. The affective contrast axis is modeled as a slow-moving homeostatic variable that tracks the difference between drug-induced affective state and baseline. These three axes are coupled through a conjunctive operator that gates the consolidation rate. The system is solved with RK45 and calibrated against published electrophysiological and behavioral data using Bayesian MCMC. The 14 free parameters are constrained by priors elicited from a structured screen of 1,847 records spanning pharmacology, electrophysiology, and behavioral neuroscience.

The central finding is that the three axes interact super-additively. Suppressing any single axis reduces consolidation probability, but suppressing two axes simultaneously produces a reduction 13 to 22 percentage points greater than the sum of the individual effects. This super-additivity is the mechanistic core of the model and the basis for its clinical implication: combination pharmacotherapy targeting two or three axes simultaneously may be dramatically more effective than single-target interventions, at lower doses of each individual drug. This is a testable prediction that can be evaluated in animal models and, eventually, in human trials.

The model has been validated through pre-registered hypothesis testing. All five hypotheses, H1 through H5, were specified before data analysis and all five were confirmed. The preprints are under review at three journals: International Addiction Research and Therapy, Progress in Neuro-Psychopharmacology and Biological Psychiatry, and Neuroscience and Biobehavioral Reviews. A co-authored paper extending the framework is under review at Alcohol.

The CCT model is not the only research line I pursue. I have conducted a pre-registered, powered replication study on the use of bipartite persistent homology for predicting hERG cardiotoxicity, which found that topological features do not beat a plain descriptor baseline, AUROC 0.8426 versus 0.8782. I have extended this topological approach to drug-resistance prediction and found that interface geometry carries almost no signal for that task, AUROC 0.425 and 0.485 on the Platinum benchmark. These negative results led me to develop TOPOLOGIX, a sequence-based approach using ESM-2 protein language model delta-embeddings and Morgan fingerprints with a Random Forest classifier, which achieves AUROC 0.804 plus or minus 0.025 on the Platinum benchmark and covers 100 percent of mutations versus roughly 18 percent for structure-limited tools. I have also built neurocascade, a receptor-to-behavior brain-circuit simulation engine, and ergofluids, a Koopman-operator method for modeling drug transport in tumor tissue, the latter of which is honestly reported as not yet meeting its primary real-data validation criterion.

The CCT model is the line of work I am putting forward for CIFAR Global Scholars because it is the most mature, the most directly aligned with CIFAR's Brain, Mind and Consciousness program, and the one with the clearest trajectory from fundamental mechanism to societal impact in addiction treatment. The negative results in topology and the honest reporting of ergofluids' validation status demonstrate the rigor with which I conduct research. The CCT model is the culmination of that rigor applied to a problem of genuine clinical urgency.

SHORT ANSWER ESSAY: INTERDISCIPLINARY APPROACH

The CCT model is interdisciplinary by construction, not by decoration. It requires simultaneous fluency in three domains that rarely speak to each other. The first is molecular pharmacology: the NMDAR axis is grounded in the specific kinetics of glutamate receptor subtypes, the dopaminergic axis in the phasic firing properties of ventral tegmental area neurons, and the affective contrast axis in the neurobiology of opponent processes. The second is dynamical systems theory: the model is a coupled ODE system whose behavior depends on timescale separation, fixed points, and bifurcation structure, not on simple input-output mapping. The third is Bayesian statistics: the model's parameters are not fitted by point estimation but inferred as posterior distributions, with uncertainty propagated through every prediction.

This combination is rare. Pharmacologists rarely build ODE models with Bayesian calibration. Dynamical systems theorists rarely engage with the specifics of NMDA receptor subunit composition. Bayesian statisticians rarely work with clinical pharmacology data. The CCT model forces all three together because the scientific question demands it. The question, can reward-memory encoding be prevented by multi-target pharmacological intervention, cannot be answered by any single discipline alone.

The model's development process reflects this interdisciplinary commitment. The priors were elicited from a structured literature screen of 1,847 records spanning pharmacology, electrophysiology, and behavioral neuroscience. The model was implemented in Python using scipy and PyMC, with the ODE solver and the MCMC sampler integrated into a single pipeline. The pre-registration specified all five hypotheses before any analysis was run. The negative results from my topology studies, which I reported directly rather than reframing, demonstrate that I apply the same rigor to falsifying my own hypotheses as to confirming them.

CIFAR's selection criteria emphasize interdisciplinary breadth and the ability to contribute to cross-disciplinary dialogue. The CCT model is evidence of both. It is a contribution to addiction neuroscience that could not have been made without computational pharmacology, and a contribution to computational pharmacology that could not have been made without clinical training. I am applying to CIFAR Global Scholars because the program's structure, which brings researchers from different disciplines into sustained conversation, is the environment in which this kind of work thrives.

CHECKLIST

- [ ] Verify current CIFAR Global Scholars eligibility criteria on the program website, specifically the nomination requirement and whether independent researchers without institutional affiliation are eligible
- [ ] Confirm the application deadline and submission portal on the CIFAR website
- [ ] Identify and contact a CIFAR program director or current CIFAR Fellow who can serve as nominator, ideally within the Brain, Mind and Consciousness or Learning in Machines and Brains programs
- [ ] Prepare a current CV formatted to CIFAR specifications, including all publications, preprints, and pre-registrations
- [ ] Compile the three CCT preprints and the co-authored Alcohol paper as supporting documents
- [ ] Obtain letters of support from Kent Berridge, Samuel Gershman, Nathaniel Daw, and Marcelo Mattar, confirming their willingness to endorse the CCT model and the application
- [ ] Draft a one-page research summary of the CCT model suitable for a non-specialist reviewer
- [ ] Prepare a statement of institutional support or affiliation, if required, given independent researcher status
- [ ] Confirm the M.Sc. enrollment at Hasso Plattner Institute and University of Potsdam is reflected in the application as current institutional affiliation
- [ ] Review all materials for consistency with the pre-registration documents and the actual results reported in the preprints

EDITOR NOTES

- Eligibility risk: CIFAR Global Scholars typically requires nomination by a CIFAR program director or current Fellow, and self-nomination is not accepted. The applicant must secure a nominator before the deadline, and this is the single largest risk to the application. The endorsements from Berridge, Gershman, Daw, and Mattar are strong but do not guarantee any of them is a current CIFAR Fellow or program director.
- The applicant's independent researcher status may be a barrier if CIFAR requires affiliation with a research institution. The M.Sc. enrollment at HPI/Potsdam may satisfy this requirement, but this must be verified with the program directly.
- The CCT model is the correct research line to lead with for this program, but the applicant should be prepared to explain why the negative topology results and the honest reporting of ergofluids' validation status are strengths, not weaknesses, in the context of CIFAR's emphasis on fundamental research rigor.
- The applicant must insert personal details not present in this profile: specific dates of pre-registration, the exact journal submission status of each preprint, and any conference presentations or invited talks related to the CCT model.
- The word counts for the motivation letter and research statement should be verified against the actual CIFAR application form, as the program may have specific limits that differ from the fallback limits used here.

AI Draft — CIFAR Global Scholars