# Philosophy of Intelligence - Oxford HT26 **Instructors:** [[Raphaël Millière]] & [[Carlotta Pavese]] **When:** Tuesdays 16-18, Hilary Term 2026 **Where:** Lecture Theatre 10100 L1, Schwarzman Center, Oxford ## Course Overview This seminar examines philosophical questions regarding the nature, measurement, and attribution of intelligence across humans, non-human animals, and artificial systems. ## Lecture Schedule ### Lecture 1: What is Intelligence? Conceptual Foundations *Instructor: [[Carlotta Pavese]]* This opening lecture offers an overview of the seminar and then goes on to address the fundamental question of how intelligence should be defined and whether it constitutes a coherent scientific category. We examine competing approaches: folk psychological conceptions that vary across cultures, behavioral characterizations designed for scientific integration, and questions about whether intelligence is a natural kind amenable to scientific investigation. **Core Questions:** - What do ordinary people mean when they attribute intelligence? - Can we provide a scientifically useful characterization of intelligence that is species-neutral and origin-neutral? - Is intelligence a natural kind, a homeostatic property cluster, or something else entirely? **Primary Readings:** - Curry, D.S. (2021). Street smarts. *Synthese*. https://doi.org/10.1007/s11229-020-02641-z - Coelho Mollo, D. (2022). Intelligent Behaviour. *Erkenntnis*. https://doi.org/10.1007/s10670-022-00552-8 **Secondary Readings:** - Serpico, D. (2017). What Kind of Kind is Intelligence? *Philosophical Psychology*. https://doi.org/10.1080/09515089.2017.140170 - Hand, M. (2007). The concept of intelligence. *London Review of Education*, 5(1). - Ryle, G. (1949). Chapter 2 of *The Concept of Mind*. - Ryle, G. (1974). "Intelligence and the Logic of the Nature-Nurture Issue. Reply to JP White." *Journal of Philosophy of Education*, 8(1): 52–60. ### Lecture 2: Behaviorism, Functionalism, and Internal Processes *Instructor: [[Raphaël Millière]]* This lecture examines whether intelligence can be characterized purely in terms of behavioral capacities or whether the internal processes generating behavior are essential to intelligence. We consider Turing's influential proposal for an operational test of machine intelligence and Block's argument that behavioral equivalence is insufficient—that genuine intelligence depends on the character of internal information processing. **Core Questions:** - Does the Turing Test adequately capture intelligence? - Can two systems be behaviorally identical yet differ in intelligence? - What role do internal computational processes play in constituting intelligence? **Primary Readings:** - Turing, A.M. (1950). Computing Machinery and Intelligence. *Mind*. https://www.jstor.org/stable/2251299 - Block, N. (1981). Psychologism and Behaviorism. *The Philosophical Review*. https://doi.org/10.2307/2184371 **Secondary Readings:** - Kipper, J. (2021). Intuition, intelligence, data compression. *Synthese*. https://doi.org/10.1007/s11229-019-02118-8 - Dennett, D. (1996). Cow-sharks, magnets, and swampman. *Mind and Language*, 11, 76-77. ### Lecture 3: The Philosophy of Psychometrics *Instructor: [[Carlotta Pavese]]* This lecture critically reviews psychometric approaches to human intelligence. We examine the operationalist foundations of IQ testing, the question of what IQ tests actually measure, and whether correlational evidence can validate the claim that IQ tests measure intelligence. We then turn to heritability research, clarifying what heritability estimates do and do not tell us, addressing common misinterpretations, and considering the ethical responsibilities of researchers investigating sensitive questions. **Core Questions:** - What philosophical assumptions underlie IQ testing? - Can correlations between IQ and life outcomes validate IQ as a measure of intelligence? - What do IQ tests actually measure, if not (primarily) intelligence? - What does "heritability" mean, and what can we infer from heritability estimates? - Can within-group heritability tell us anything about between-group differences? - What are the ethical responsibilities of researchers investigating sensitive questions? **Primary Readings:** - Block, N.J. & Dworkin, G. (1974). IQ: Heritability and Inequality, Part 1. *Philosophy & Public Affairs*. https://www.jstor.org/stable/2264953 - Block, N.J. & Dworkin, G. (1974). IQ, Heritability and Inequality, Part 2. *Philosophy & Public Affairs*. https://www.jstor.org/stable/2265104 **Secondary Readings:** - Curry, D. S. (2021). G as Bridge Model. *Philosophy of Science*, 88(5), 1067–1078. https://doi.org/10.1086/714879 - De Boeck, P., Robert Gore, L., Gonzalez, T., & San Martin, E. (2020). An Alternative View on the Measurement of Intelligence and Its History. In R. J. Sternberg (Ed.), *The Cambridge Handbook of Intelligence*. Cambridge University Press. https://doi.org/10.1017/9781108770422 - Curry, D. S. (2025). On IQ and other sciencey descriptions of minds. *Philosophers' Imprint*. - Sternberg, R. J. (2015). Successful intelligence: A model for testing intelligence beyond IQ tests. *European Journal of Education and Psychology*, 8(2), 76-84. - Richardson, K. (2002). What IQ tests test. *Theory & Psychology*, 12(3), 283-314. - Gardner, H. (1987). The theory of multiple intelligences. *Annals of dyslexia*, 19-35. ### Lecture 4: Animal Intelligence *Instructor: [[Raphaël Millière]]* This lecture examines intelligence in non-human animals, addressing both methodological foundations and substantive questions about animal minds. We examine Morgan's Canon—the principle that animal behavior should not be explained by appeal to higher faculties if explicable by lower ones—alongside the complementary danger of "anthropofabulation". We then consider what we can reasonably infer about animal cognition given the underdetermination problem, and examine the evolutionary history of intelligence. **Core Questions:** - What justifies Morgan's Canon, and how should we understand "higher" and "lower" faculties? - What is anthropofabulation and how does it distort comparative research? - How can we overcome underdetermination in attributing cognition to animals? - Do animals reason about unobservable variables like mental states and causal forces? - What does evolutionary reconstruction tell us about animal intelligence? **Primary Readings:** - Sober, E. (1998). Morgan's Canon. *Proceedings of the Aristotelian Society*. - Buckner, C. (2013). Morgan's Canon, meet Hume's Dictum: avoiding anthropofabulation in cross-species comparisons. *Biology & Philosophy*. https://doi.org/10.1007/s10539-013-9376-0 **Secondary Readings:** - Halina, M. (2024). Animal Minds. *Cambridge Elements*. https://doi.org/10.1017/9781009438636 - Andrews, K. & Monsó, S. (2021). Animal Cognition. *Stanford Encyclopedia of Philosophy*. https://plato.stanford.edu/archives/spr2021/entries/cognition-animal/ - Bates, L.A. & Byrne, R.W. (2020). The Evolution of Intelligence. In Sternberg (Ed.), *Cambridge Handbook of Intelligence*. ### Lecture 5: Human Intelligence *Instructor: [[Carlotta Pavese]]* What makes human intelligence unique, if anything? This lecture examines the relationship between learning, cognitive development, and the distinctiveness of human cognition. We start with the proposal that learning serves as the fundamental criterion of intelligence. We review developmental evidence from infancy and childhood showing that humans are remarkable learners from the earliest stages of life. We also discuss competing explanations of the uniqueness of human intelligence: is it due from a qualitative change introduced by language, or to quantitative increases in information-processing capacity over time? What does it mean to say that human behavior is flexible or especially so? What kind of flexibility is, if at all, a mark of intelligence? **Core Questions:** - Is learning the fundamental criterion of intelligence? - What do infant and child cognition reveal about the foundations of intelligence? - Does language qualitatively transform human cognition, or is human uniqueness a matter of degree? - Can quantitative differences in information-processing capacity explain the full range of human cognitive achievements? - Is flexibility fundamental for intelligent behavior? How should we understand the flexibility of intelligent behavior? **Primary Readings:** - Fridland, E. (2015). Learning Our Way to Intelligence: Reflections on Dennett and Appropriateness. https://doi.org/10.1007/978-3-319-17374-0_8 - Dennett, D.C. (1994). The Role of Language in Intelligence. In Khalfa (Ed.), *What is Intelligence?* - Frensch, P. A., & Sternberg, R. J. (2014). Expertise and intelligent thinking: When is it worse to know better?. In *Advances in the psychology of human intelligence* (pp. 157-188). Psychology Press. **Secondary Readings:** - Bornstein, M.H. (2020). Intelligence in Infancy. In Sternberg (Ed.), *Cambridge Handbook of Intelligence*. - Gelman, S.A. & DeJesus, J.M. (2020). Intelligence in Childhood. In Sternberg (Ed.), *Cambridge Handbook of Intelligence*. https://doi.org/10.1017/9781108770422 - Cantlon, J.F. & Piantadosi, S.T. (2024). Uniquely human intelligence arose from expanded information capacity. *Nature Reviews Psychology*. https://doi.org/10.1038/s44159-024-00283-3 - Gopnik, A., O'Grady, S., Lucas, C. G., Griffiths, T. L., Wente, A., Bridgers, S., ... & Dahl, R. E. (2017). Changes in cognitive flexibility and hypothesis search across human life history from childhood to adolescence to adulthood. *Proceedings of the National Academy of Sciences*, 114(30), 7892-7899. - Kilov, D. (2021). The brittleness of expertise and why it matters. *Synthese*, 199(1), 3431-3455. - Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: what is it, who has it, and how did it evolve?. *Science*, 298(5598), 1569-1579. ### Lecture 6: Comparative Cognition Across Biological Minds and AI *Instructor: [[Raphaël Millière]]* This lecture examines methodological challenges that arise when comparing intelligence across humans, animals, and artificial systems. **Core Questions:** - How can behavioral evidence constrain inferences about underlying cognitive mechanisms across biological and artificial intelligence? - How can the signature testing approach apply to artificial systems? - How do auxiliary task demands affect performance independently of competence? - What forms of anthropocentric biases affect comparisons between biological and artificial intelligence? **Primary Readings:** - Taylor, A.H. et al. (2022). The signature-testing approach to mapping biological and artificial intelligences. *Trends in Cognitive Sciences*. https://doi.org/10.1016/j.tics.2022.06.002 - Millière, R. & Rathkopf, C. (2025). Anthropocentric bias in language model evaluation. *Computational Linguistics*. https://doi.org/10.1162/COLI.a.582 - Harding, J., & Sharadin, N. (2024). What is It for a Machine Learning Model to Have a Capability? *British Journal for the Philosophy of Science*. **Secondary Readings:** - Halina, M. (2023). Methods in Comparative Cognition. *Stanford Encyclopedia of Philosophy*. https://plato.stanford.edu/archives/fall2023/entries/comparative-cognition/ - Firestone, C. (2020). Performance vs. Competence in human–machine comparisons. *Proceedings of the National Academy of Sciences*, 117(43), 26562–26571. https://doi.org/10.1073/pnas.1905334117 - Frank, M. C. (2023). Baby steps in evaluating the capacities of large language models. *Nature Reviews Psychology*, 2(8), Article 8. https://doi.org/10.1038/s44159-023-00211-x - Hu, J. & Frank, M.C. (2024). Auxiliary task demands mask the capabilities of smaller language models. *OpenReview*. https://openreview.net/forum?id=U5BUzSn4tD - Lampinen, A. (2024). Can Language Models Handle Recursively Nested Grammatical Structures? A Case Study on Comparing Models and Humans. *Computational Linguistics*, 50(4), 1441–1476. https://doi.org/10.1162/coli_a_00525 - Boyle, A. (2024). Disagreement & classification in comparative cognitive science. *Noûs*. https://doi.org/10.1111/nous.12480 ### Lecture 7: The Jagged Frontier of AI *Instructor: [[Raphaël Millière]]* This lecture addresses questions specific to artificial intelligence, with particular attention to the puzzling capability profile of state-of-the-art AI systems. Current AI models achieve or exceed human-level performance on an impressive range of benchmarks yet exhibit striking weaknesses on comparatively simple tasks that are trivial for humans. Given this "jagged frontier" of capabilities, should we conclude that these systems lack intelligence altogether, or that they occupy a new and previously unexplored region within a broader "intelligence space"? We also revisit insights from previous weeks by contrasting crystallized skill with adaptive intelligence, and evaluating whether AI models exhibit genuine generalization to new problems or merely sophisticated statistical interpolation within the boundaries of their training data. Finally, we consider new developments in embodied AI systems in light of our previous discussion of the relationship between embodied skills and intelligence. **Core Questions:** - What should we conclude from the "jagged" capabilities of AI systems: striking performance on some tasks alongside brittle failure on others? - How should we operationalize and measure generalization, in a way that supports fair comparisons across systems with radically different training histories? - Is "general intelligence" (and especially "AGI") a coherent scientific target, or a moving label shaped by shifting definitions, incentives, and benchmark selection? - What is the relationship between embodiment and intelligence in AI? **Primary Readings:** - Mollo, D. C. (2025). AI-as-exploration: Navigating intelligence space. *Theoria. an International Journal for Theory, History and Foundations of Science*. https://doi.org/10.1387/theoria.25837 - Millière, R. & Buckner, C. (forthcoming). Generative Artificial Intelligence, Chapter 2 ("Generation and Generalization"), Cambridge University Press. **Secondary Readings:** - Chollet, F. (2019). On the Measure of Intelligence. *arXiv*. https://arxiv.org/abs/1911.01547 - Dretske, F. (1993). Can Intelligence Be Artificial? *Philosophical Studies*. https://www.jstor.org/stable/4320430 - Jones, C. R., Rathi, I., Taylor, S., & Bergen, B. K. (2025). People cannot distinguish GPT-4 from a human in a Turing test. *Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency*, 1615–1639. https://doi.org/10.1145/3715275.3732108 - Mitchell, M. (2024). The Turing Test and our shifting conceptions of intelligence. *Science*, 385(6710), eadq9356. https://doi.org/10.1126/science.adq9356 - Mitchell, M. (2024). Debates on the nature of artificial general intelligence. *Science*, 383(6689), eado7069. https://doi.org/10.1126/science.ado7069 - Mitchell, M. (2025). Artificial intelligence learns to reason. *Science*, 387(6740), eadw5211. https://doi.org/10.1126/science.adw5211 - Jin, S., Xu, J., Lei, Y., & Zhang, L. (2024). Reasoning grasping via multimodal large language model. *arXiv preprint* arXiv:2402.06798. - Brohan, A., Chebotar, Y., Finn, C., Hausman, K., Herzog, A., Ho, D., ... & Fu, C. K. (2023, March). Do as i can, not as i say: Grounding language in robotic affordances. In *Conference on robot learning* (pp. 287-318). PMLR. ### Lecture 8: Intelligence and Skills *Instructor: [[Carlotta Pavese]]* This lecture discusses the relation between skills and intelligence. Should we understand intelligence in terms of skills and of intelligent behavior in terms of skillful behavior? What difficulties stand in the way of this reduction? Is skillful behavior uniformly intelligent? Or should we only consider some kind of skillful behavior as properly speaking intelligent? What is Moravec's paradox and what does it tell us about intelligence? We discuss the view that intellectual skills have a privileged connection to intelligence over practical and embodied skills, and the role of the distinction between different kinds of knowledge in accounts of intelligence. **Core Questions:** - Is there a principled distinction between "intellectual" and "practical" intelligence? - Are theoretical skills more central to intelligence than embodied skills? - Can reflexes be intelligent, and what does this reveal about the nature of intelligence? - Does the automatization of skill through practice preserve or eliminate intelligence? - What is the relation between skills and other intelligent capacities, such as intelligent reflex and intelligent habits? Are skills just well trained habits? Or should we think of skills and habits as different kinds of capacities? **Primary Readings:** - Pavese, C. (2026). Intelligence Socialism. *Oxford Studies in Philosophy of Mind*. - Krakauer, J.W. (2019). The Intelligent Reflex. *Philosophical Psychology*. https://doi.org/10.1080/09515089.2019.1607281 **Secondary Readings:** - Christensen, W., Sutton, J., & McIlwain, D. J. (2016). "Cognition in skilled action: Meshed control and the varieties of skill experience." *Mind & Language*, 31(1), 37-66. - Pavese, C. (forthcoming) Procedural Memory and Know-how. *Handbook of the Philosophy of Memory*. - Ryle (1949) Chapter 2 of *The Concept of Mind*.