Dynamic Derived Intentionality in Large Language Models: From Implicit Beliefs to Semantic Parasitism

Abstract Large Language Models instantiate a distinctive category of intentionality, which I call dynamic derived intentionality: a productive, context-sensitive form of derived intentionality that goes well beyond the static cases of words on a page or symbols on a map, yet falls short of original intentionality. What separates the two is neither consciousness nor introspection nor anything in the biological substrate, but normative standing. Original intentionality is the having of what I call ownership of normative status, the standing of a subject who occupies the space of reasons and can be held answerable for its commitments; LLMs lack it. The analogy I draw with human implicit beliefs is diagnostic rather than probative: it holds fixed the representational vehicle the two share, sub-symbolic, distributed, statistically acquired, opaque, so that the one property they do not share, answerability, stands out. This is the answerability gap. I develop the notion of semantic parasitism to capture how these systems operate through a meaning they do not own, and read Reinforcement Learning from Human Feedback as calibration to norms that reside elsewhere. The account is defended against neo-Searlean, anthropomorphist, and deflationary objections of both a Dennettian and a McDowellian cast. The gap it identifies is architectural, not metaphysical: an architecturally discontinuous system could in principle come to occupy the space of reasons. Similar content being viewed by others Data Availability Not applicable. Notes For the philosophical debate about implicit beliefs see: Polanyi (1966), Lycan (1986), Devitt (2006), Matthews (2007), Zimmerman (2007), Gendler (2008), Schwitzgebel (2010), Levy (2015), Mandelbaum (2016), Brownstein and Saul (2016), Sullivan-Bissett (2019). On the dual-process architecture of implicit cognition see Kahneman (2011) and Evans and Stanovich (2013). The locus classicus for doxastic involuntarism is Alston (1989). The point required here is weaker than any voluntarist thesis: not that implicit beliefs can be revised at will, but that they fall, in principle, within the reach of a subject who can be held answerable for them. About this point see Shea (2023). Recently Schwitzgebel (2025) has defended a superficialist position about beliefs. Superficialism about property X holds that whether something has property X depends solely on surface-level facts rather than deep underlying facts. Applied to belief, this view claims that beliefs should be understood entirely through observable behaviors (both actual and potential), conscious experiences, and transitional cognitive states—where these transitional states are themselves understood through behavior and conscious experience. This superficialist approach to belief proves more compelling than theories based on deep cognitive or neural structures, both intuitively and practically, and it holds up just as well scientifically. In essence: this philosophical position argues that to determine whether someone holds a belief, we need not examine deep brain structures or internal cognitive mechanisms, but only how they behave (or would behave) and what they consciously experience. According to this position, it is possible to suggest that robots and Large Language Models already do, or will soon, believe. The foundational notion is Sellars’s logical space of reasons (Sellars 1956). I draw on Brandom (1994) for the account of conceptual content as articulated within the practice of giving and asking for reasons, and for its neutrality as to substrate, and on McDowell (1994) for the contrast between spontaneity and receptivity. The familiar divergence between Brandom and McDowell, over the priority of inference and the standing of experience, is orthogonal to the present argument, which needs only what they share: that original intentionality is a position within a normative practice rather than a property read off a functional profile. Coeckelbergh (2025, 1) characterizes epistemic agency as the control agents may exercise over their beliefs. The operative notion, for my purposes, is not voluntary control over belief revision, which doxastic involuntarism renders doubtful even for human agents, but answerability: the standing under which a subject is responsible for its commitments and can be called to vindicate or revise them. It is answerability, not control, that current systems lack. The original/derived distinction is drawn here in pragmatic terms. A constitution-theoretic gloss is available and compatible: in Voltolini’s (2024) terms one may say that the intentional object fails to be a constituent of the system’s states except parasitically. Nothing in the argument rests on it. It bears stressing, against a natural impression, that Voltolini’s theory does not tie intentionality to conscious states: for him intentionality is not the mark of the mental, and intentional mental states are merely its prototypical bearers, which is precisely what leaves room for the artefactual intentionality at issue here. The distinction between original (or intrinsic) and derived intentionality has been central to debates about mental representation since at least the 1980s. This distinction bears directly on the question of whether artificial systems can possess genuine intentionality. Searle (1983) articulated the distinction clearly: original intentionality is possessed by mental states in virtue of their intrinsic features, while derived intentionality is possessed by things like words, maps, and pictures only in virtue of the original intentionality of the agents who use and interpret them. For Searle, only biological systems with the right causal powers can have original intentionality; everything else has at most derived intentionality. Dennett (1989) challenged this sharp distinction, arguing that even human intentionality is, in a sense, derived from evolutionary processes, cultural learning, and social practices. On his view, the original/derived distinction is a matter of degree rather than kind, with human intentionality being no more “original” in any metaphysically significant sense than the intentionality we might attribute to sophisticated artifacts. John Haugeland (1990) offered a different perspective, distinguishing between original, derived, and what he called “derivative” intentionality. For Haugeland, derivative intentionality is more than mere derived intentionality; it involves systems that can be held responsible for getting things right or wrong according to constitutive standards. This normative dimension suggests a richer category between mere derived intentionality and full original intentionality. Subsequent work has complicated the distinction in ways orthogonal to the present argument: the extended-mind thesis (Clark and Chalmers 1998), collective-intentionality theories (Gilbert 1989; Tollefsen 2002), and the enactive approach (Thompson 2007; Varela, Thompson, and Rosch 1991) each press, in different ways, against the assumption that original intentionality must be individual, biological, and intrinsic. References Alston, W. P. (1989). Epistemic justification: Essays in the theory of knowledge. Cornell University Press. Bender, E.M., Koller, A. (2020). Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data, «Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics», July 2020 Block, N. (1998). Semantics, conceptual role. Routledge Encyclopedia of Philosophy., 8, 652–657. Brandom, R. B. (1994). Making it explicit: Reasoning, representing, and discursive commitment. Harvard University Press. Brentano, F. (1874). Psychology from an empirical standpoint, Routledge and Kegan Paul, London (1973) Brownstein, M., & Saul, J. M. (2016). Implicit bias and philosophy (Vol. 2). Oxford University Press. Chandna, B., Bashir, Z., & Sen, P. (2025). Dissecting bias in LLMs: A mechanistic interpretability perspective. Arxiv. https://doi.org/10.48550/arXiv.2506.05166 Chang, Z., et al. (2025). A Survey of LLM alignment instruction understanding intention reasoning, and reliable generation. arXiv. https://doi.org/10.48550/arXiv.2502.09101 Christiano, P. et. al. (2017). Deep reinforcement learning from human preferences. Advances in Neural Information Processing Systems», 30 Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58, 7–19. Coeckelbergh, M. (2025). AI and epistemic agency: How AI influences belief revision and its normative implications. Social Epistemology. https://doi.org/10.1080/02691728.2025.2466164 Coelho Mollo, D., & Millière, R. (2023). The vector grounding problem. Arxiv. https://doi.org/10.48550/arXiv.2304.01481 Dennett, D. (1989). The intentional stance. MIT Press. Devitt, M. (2006). Ignorance of Language, Clarendon Press, Oxford. Dretske, F. I. (1981). Knowledge and the flow of information. MIT Press. Dretske, F. I. (1995). Naturalizing the mind. The MIT Press. Evans, JSt. B. T., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science., 8(3), 223–241. Fodor, J. A. (1987). Psychosemantics. The MIT Press. Fodor, J. A. (1990). A theory of content and other essays. The MIT Press. Fodor, J. A., & Lepore, E. (1992). Holism: A shopper’s guide. Blackwell. Gendler, T. S. (2008). Alief and belief. Journal of Philosophy, 105(10), 634–663. Gilbert, M. (1989). On social facts. Princeton University Press. Grice, P. (1957). Meaning. The Philosophical Review, 66(3), 377–388. Grindrod, J. (2024). Large language models and linguistic intentionality. Synthese, 204, 71. Harnad, S. (1990). The symbol grounding problem. Physica D, Nonlinear Phenomena, 42, 335–346. Haugeland, J. (1990). The intentionality all-stars. Philosophical Perspectives, 4, 383–427. Horgan, T. (2024). Why virtual worlds aren’t real: how phenomenal intentionality constrains mental reference, in U. Kriegel (Ed.), Oxford Studies in Philosophy of Mind, vol. 4, pp. 461–482 Horgan, T., & Tienson, J. (2002). The intentionality of phenomenology and the phenomenology of intentionality. In D. J. Chalmers (Ed.), Philosophy of mind: Classical and contemporary readings (pp. 520–532). Oxford University Press. Horgan, T., & Tienson, J. (2023). Phenomenal intentionality and intentionality holism. Protosociology, 40, 60–75. Horgan, T., Tienson, J., & Graham, G. (2005). Phenomenal intentionality and the brain in a vat. In R. Schantz (Ed.), The externalist challenge: New studies in cognition and intentionality (pp. 297–317). De Gruyter. Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux. Kriegel, U. (2011a). Subjective consciousness: A self-representational theory. Oxford University Press. Kriegel, U. (2011b). The sources of intentionality. Oxford University Press. Kriegel, U. (2018). The varieties of consciousness. Oxford University Press. Lycan, W. G. (1986). Tacit Belief, in R. Bogdan (ed.), Belief: Form, Content, and Function, Oxford University Press, Oxford, pp. 61-82. Levy, N. (2015). Neither fish nor fowl: Implicit attitudes as patchy endorsements. Noûs, 49(4), 800–823. Li, K. et al. (2023). Emergent world representations: exploring a sequence model trained on a synthetic task. ICLR 2023 https://doi.org/10.48550/arXiv.2210.13382 Lyre, H. (2024). “Understanding AI”: semantic Grounding in large language models. Arxiv. https://doi.org/10.48550/arXiv.2402.10992 Mandelbaum, E. (2016). Attitude, inference, association: On the propositional structure of implicit bias. Noûs, 50(3), 629–658. Matthews, R. (2007). The measure of mind. Oxford University Press. McDowell, J. (1994). Mind and world. Harvard University Press. Millière, R. (2025). Normative conflicts and shallow AI alignment. Philosophical Studies. https://doi.org/10.1007/s11098-025-02347-3 Millikan, R. G. (1984). Language, thought, and other biological categories: New foundations for realism. MIT Press. Nanda, N. et al. (2023). Emergent linear representations in world models of self-supervised sequence models https://doi.org/10.48550/arXiv.2309.00941 Neander, K. (2017). A mark of the mental. The MIT Press. Olsson, et al. (2022). In-context learning and induction heads. Arxiv. https://doi.org/10.48550/arXiv.2209.11895 Packard, N. H., et al. (1980). Geometry from a time series. Physical Review Letters, 45(9), 712–716. Piantadosi, S. T., & Hill, F. (2022). Meaning without reference in large language models. Arxiv. https://doi.org/10.48550/arXiv.2208.02957 Polanyi, M. (1966). The tacit dimension. Doubleday. Romera-Paredes, B., et al. (2024). Mathematical discoveries from program search with large language models. Nature, 625, 468–475. https://doi.org/10.1038/s41586-023-06924-6 Ryle, G. (1949). The Concept of Mind. Hutchinson. Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274, 1926–1928. https://doi.org/10.1126/science.274.5294.1926 Schwitzgebel, E. (2010). Acting contrary to our professed beliefs, or the gulf between occurrent judgment and dispositional belief. Pacific Philosophical Quarterly, 91, 531–553. Schwitzgebel, E. (2025). Superficialism about belief, and how we will decide that robots believe. Semiotic Studies. https://doi.org/10.26333/sts.xxxix1.11 Searle, J. (1980). Minds, brains and programs. Behavioral and Brain Sciences, 3(3), 417–457. Searle, J. (1983). Intentionality: An essay in the philosophy of mind. Cambridge University Press. Searle, J. (1992). The rediscovery of the mind. MIT Press. Sellars, W. (1956). Empiricism and the Philosophy of Mind. In H. Feigl & M. Scriven (Eds.), Minnesota Studies in the Philosophy of Science (Vol. I, pp. 253–329). University of Minnesota Press. Shea, N. (2023). Organized representations forming a computationally useful processing structure. Synthese. https://doi.org/10.1007/s11229-023-04373-2 Stiennon, N., et al. (2020). Learning to summarize with human feedback. Advances in Neural Information Processing Systems., 33, 3008. Sullivan-Bissett, E. (2019). Biased by Our Imaginings. Mind & Language, 34(5), pp. 627–647. https://doi.org/10.1111/mila.12225 Takens, F. (1981). Detecting strange attractors in turbulence. In D. Rand & L. S. Young (Eds.), Dynamical systems and turbulence (pp. 366–381). Springer. Thompson, E. (2007). Mind in life. Biology, phenomenology, and the sciences of mind. Harvard University Press. Tollefsen, D. P. (2002). Collective intentionality and the social sciences. Philosophy of the Social Sciences, 32, 25–50. Varela, F. J., Thompson, E., & Rosch, E. (1991). The embodied mind: cognitive science and human experience. The MIT Press. Voltolini, A. (2024). Intentionality as constitution. Routledge. Wei, J., et al. (2022). Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems. https://doi.org/10.48550/arXiv.2201.11903 Williams, J. G. R. (2020). The Metaphysics of Representation. Oxford University Press. Funding No funding was received for conducting this study. Author information Authors and Affiliations Corresponding author Ethics declarations Conflict of interest The author declares no competing interests. Additional information Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rights and permissions Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. About this article Cite this article Tortoreto, A. Dynamic Derived Intentionality in Large Language Models: From Implicit Beliefs to Semantic Parasitism. Minds & Machines 36, 44 (2026). https://doi.org/10.1007/s11023-026-09800-0 Received: Accepted: Published: Version of record: DOI: https://doi.org/10.1007/s11023-026-09800-0