About
I am a Professor of Philosophy at the University of Idaho. My work is situated at the intersections of cognitive science, philosophy, complex systems, and computational modeling. I am particularly interested in how scientists use diverse modeling techniques to understand the impacts of determinants of belief formation on complex biological and social phenomena.
My research explores questions about standards of evidence and their connections to opinion dynamics. I often use agent-based models and network theory as methodological tools, as well as survey studies. I am highly collaborative and interdisciplinary; my preference is to focus on questions that interest me rather than worry about disciplinary boundaries.
More recently, my work has expanded into analogy, metaphor, and narrative as windows into human cognition and AI limitations. The MARC project (Metaphor Abstraction and Reasoning Corpus) uses community-created puzzles to probe what humans can do that AI systems cannot, and doubles as a tool for teaching AI literacy and ethics. Its sibling, NARC (Narrative Augmented Reasoning Challenges), asks how stories transform visual reasoning — whether for humans or for machines. I have also been investigating standards of evidence across political and scientific domains, and the philosophical and ethical dimensions of large language models and generative AI.
Recent
Research Areas
Opinion Dynamics & Social Epistemology
Modeling how beliefs form and spread through populations. Work on the preference for belief, echo chambers, reflective equilibrium, and how network structure shapes collective knowledge production.
Disease-Behavior Models
Computational models linking behavioral responses to epidemic dynamics. Research on risk tolerance, vaccination attitudes, and how transient prophylaxis drives epidemic waves.
Standards of Evidence
Investigating how individuals assess and gather evidence across scientific and political domains, including the effects of ideology and cognitive reflection on evidence-seeking behavior.
Analogy, Metaphor, Narrative & AI Benchmarking
The MARC project explores how metaphorical reasoning reveals fundamental gaps between human and artificial intelligence. Its sibling, NARC, investigates how narrative reorganizes visual reasoning by pairing abstract grid sequences with short stories that disambiguate them. Together they connect philosophy of language, cognitive science, narratology, and AI evaluation.
AI Literacy & Ethics
Building frameworks and tools for understanding generative AI, including its limits, biases, and societal impacts. Developing pedagogical approaches that make AI literacy accessible to non-technical audiences.
MARC: Metaphor Abstraction and Reasoning Corpus
MARC extends François Chollet's Abstraction and Reasoning Corpus (ARC) by pairing visual grid-based puzzles with metaphorical natural language hints. The core insight: humans can leverage metaphors to solve novel puzzles with remarkable ease, while AI systems struggle—exposing a fundamental gap in machine intelligence around fluid reasoning, perspective-shifting, and genuine understanding.
What MARC enables
Research. MARC provides a platform for studying how metaphorical reasoning works in humans and where it breaks down in AI. Each puzzle probes intentionality, agency, and the capacity for analogical thinking—capacities that remain beyond current AI systems.
AI Literacy & Ethics. Because anyone can create a MARC puzzle, it democratizes AI benchmarking. Students and non-specialists can directly engage with the question of what makes human intelligence distinctive, without needing technical training. The project serves as a teaching tool for critical thinking about AI capabilities and limitations.
Community. MARC puzzles are community-created and community-driven. Campus competitions and open submission foster broad participation across disciplines.
NARC: Narrative Augmented Reasoning Challenges
NARC is a sibling project to MARC. Where MARC pairs visual puzzles with metaphorical hints, NARC pairs sequences of colored grids with short narratives. One or more grids in the sequence are hidden, and the goal is to reconstruct them pixel-perfectly. The grids alone are ambiguous — multiple completions look plausible. The narrative alone underdetermines the answer too. Only together do they pick out a unique solution.
I call this the NARC property: neither modality suffices, but their combination does. It offers a tractable setting for asking how narrative reorganizes perception and inference — and where humans and AI systems diverge in their ability to use a story to see what would otherwise be invisible.
What NARC enables
Research. NARC is designed to isolate the contribution of narrative to visual reasoning. The corpus draws on Erin James's Story Prism — a decomposition of narratives into Teller & Told, World, Events, Actors, and How It Feels — so puzzles can be tagged and varied along specific narrative facets. This makes it possible to ask which facets carry the disambiguating work, and to compare humans and AI systems facet by facet.
Connections. NARC sits at the intersection of ARC-AGI, econarratology, and AI evaluation. Puzzles span literary classics (Hemingway, Kafka, Shelley), scientific concepts (natural selection, entropy, plate tectonics), and philosophical thought experiments (the trolley problem, Plato's cave). The site reports one measured difficulty signal: on how many tested models a puzzle has the NARC property.
Corpus. 320 active puzzles with 976 narrative variants, 3–8 grids per puzzle, tested on 12 language models with masking, order, and keyword ablations. A separate set, NARC-tiny, holds 200 puzzles generated from TinyStories by the Narrative Hierarchy Model, a grammar in which every event rewrites into both a sentence and a grid transition. The corpus is browseable, solvable, and inspectable online, and anyone can try the puzzles.
Paper. “Only the Story Completes the Grid: Hand-Authored and Grammar-Generated Puzzles for Narrative Augmented Reasoning” is under review at ICLR 2027.
Selected Presentations
Most recent first.
Selected Publications
For a complete list, see my CV or Google Scholar.
Preprints & Under Review
Refereed
Data & Code
Public datasets and software from my projects. More on GitHub and Hugging Face.
Datasets
Software
Teaching
Regular Undergraduate Courses
- PHIL 2010: Critical Thinking
- PHIL 2020: Introduction to Symbolic Logic
- Rotating: Decision Theory, Phil Language, Theory of Knowledge, Metaphysics
Seminars & Special Topics
- PHIL 361: Professional Ethics — Generative AI
- PHIL 4040: Rational Choice and Strategic Interactions
Teaching Tools
Course materials and syllabi are available for current students on the university's learning management system. Prospective students interested in course content can contact me directly via email.