MOTIVATION LETTER The Wissenschaftskolleg Early Career Fellowship 2027/2028 offers something rare: a year of unstructured intellectual freedom within a deliberately interdisciplinary community. My research sits at the intersection of computational methods, philosophy of mind, and qualitative social science, and it requires exactly this kind of environment to mature from a working prototype into a defensible scholarly contribution. I am a Nigerian pharmacist and computational researcher, currently enrolled in the M.Sc. Digital Health programme at the Hasso Plattner Institute / University of Potsdam. My work spans addiction neuroscience, protein machine learning, and dynamical systems, but the project I propose for this fellowship is different in kind. It is called psyche-twin, and it is a typed, multi-scale knowledge-graph architecture for self-modeling. The system fuses multiple independent evidence streams, including LLM-derived statements, psychological assessment data, behavioral logs, and document-derived facts, into a single append-only event log. When these streams disagree, the disagreement is stored as an explicit graph edge rather than averaged away. A first-person conversational interface then generates responses grounded in the graph's current state. The philosophical and psychological questions embedded in this architecture are the ones I want to pursue at the Wissenschaftskolleg. What does it mean for a model of a person to be faithful to that person's lived experience? How should a system represent contradictory self-reports, and what does the choice of representation reveal about underlying theories of identity? Can a computational artifact support genuine self-inquiry, or does it merely flatten the subject into a database? These questions belong to the humanities and social sciences, and they cannot be answered by engineering alone. My technical background gives me credibility to build the system, but the Wissenschaftskolleg's selection criteria emphasize innovative research agendas and contributions to an interdisciplinary scholarly community. I bring both. The graph's treatment of disagreement as explicit structure is a methodological contribution to digital humanities and qualitative social science, where conflicting accounts are often treated as noise rather than data. I want to develop this idea in dialogue with philosophers, sociologists, and cognitive scientists who approach identity and memory from traditions I have not yet absorbed. The fellowship's freedom-oriented ethos matches my working style. I have completed five pre-registered computational studies as an independent researcher, including a Bayesian-calibrated pharmacological model of reward-memory encoding with all five pre-specified hypotheses confirmed, and a powered replication that falsified a published claim about topological descriptors in cardiotoxicity prediction. I know how to design, execute, and report rigorous research without institutional supervision. What I lack is sustained conversation with scholars outside my immediate computational circle. The Wissenschaftskolleg provides that. I request consideration for the 2027/2028 cohort. My M.Sc. coursework will be complete by then, and I will be positioned to devote the fellowship year fully to the psyche-twin project and its philosophical foundations. RESEARCH STATEMENT Project title: Psyche-twin: A Knowledge-Graph Architecture for Self-Modeling and the Problem of Conflicting Self-Evidence Research problem Contemporary digital tools for self-tracking and self-reflection collect vast amounts of data about individuals, yet they share a common flaw: they treat conflicting evidence as noise. A user who reports anxiety on a questionnaire but behaves calmly in behavioral logs, or who tells an AI assistant one story about their childhood and writes a different one in a journal, is typically smoothed into a single averaged profile. This smoothing destroys precisely the information that matters most for understanding a person. Contradictions between self-reports, behaviors, and documents are not measurement errors; they are the raw material of identity. My project builds a computational architecture that preserves these contradictions as first-class structures. The system, called psyche-twin, is a typed, multi-scale knowledge graph with an append-only event log. Evidence streams are never merged or reconciled automatically. Instead, each stream contributes typed nodes and edges, and disagreements between streams become explicit graph edges labeled as conflicts. A conversational interface generates responses grounded in the graph's current state, which means the system can say, truthfully, "your questionnaire and your behavior log disagree on this point," rather than presenting a false consensus. Research questions Three questions structure the project. First, what formal representations best capture the relationship between conflicting self-evidence? I will compare typed conflict edges against alternative representations, including probabilistic belief models and dialectical argumentation frameworks, and evaluate each for expressiveness and computational tractability. Second, what does the choice of representation imply about underlying theories of identity? A system that stores contradictions as stable graph structures embodies a different philosophical commitment than one that treats them as temporary states to be resolved. I will articulate these commitments explicitly and connect them to existing literature in philosophy of mind and narrative identity theory. Third, can a first-person interface grounded in a conflict-preserving graph support genuine self-inquiry? I will conduct qualitative user studies with a small cohort of participants, examining whether engagement with the system produces new self-understanding or merely reinforces existing self-narratives. Methodology The technical architecture is already prototyped. The system uses a typed graph database with an append-only event log, multiple independent ingest pipelines for LLM-derived, assessment-derived, behavioral, and document-derived evidence, and a conversational interface built on top of the graph state. The methodological innovation is the conflict edge: a typed relationship that records disagreement between evidence streams without resolving it. I will extend this prototype in three directions during the fellowship year. First, I will formalize the conflict-edge ontology and publish a specification document suitable for peer review. Second, I will build a reference implementation with documented APIs so other researchers can adopt the architecture. Third, I will design and run a qualitative study with human participants, using think-aloud protocols and semi-structured interviews to examine how people interact with a system that refuses to reconcile their contradictions. Contribution to the field The project makes three contributions. Methodologically, it offers a concrete, implementable alternative to the averaging logic that dominates digital self-tracking. The conflict edge is a small idea with large consequences: once disagreements are stored as explicit structure, they become available for analysis, visualization, and reflection. Theoretically, the project connects computational design choices to philosophical commitments about identity, memory, and self-knowledge. A graph that preserves contradictions embodies a relational, non-unitary theory of the self, and I will make that theory explicit. Practically, the project produces an open-source tool that researchers in psychology, digital humanities, and human-computer interaction can adapt for their own studies. Feasibility and timeline The prototype exists and passes 62 automated tests across its core modules. The fellowship year will proceed in three phases. Months one through four: formalize the conflict-edge ontology, publish the specification, and conduct a systematic review of relevant philosophical and psychological literature on self-contradiction. Months five through eight: build the reference implementation with documented APIs and run a small pilot study with five participants to test the qualitative protocol. Months nine through twelve: conduct the full qualitative study with fifteen to twenty participants, analyze results, and draft two manuscripts for submission to peer-reviewed journals in digital humanities and philosophy of mind. Why the Wissenschaftskolleg The project requires sustained attention and interdisciplinary dialogue. It is not a natural fit for a conventional laboratory, where the pressure to produce incremental results would push toward resolving conflicts rather than preserving them. The Wissenschaftskolleg's structure, a year of freedom within a community of scholars from different disciplines, is precisely the environment this work needs. I am specifically seeking conversation with philosophers working on personal identity, sociologists studying narrative and self-presentation, and digital humanists developing methods for analyzing unstructured personal data. The fellowship's emphasis on collaborative intellectual life, rather than output metrics, aligns with the project's slow, reflective character.