pith. sign in

arxiv: 2411.06812 · v2 · submitted 2024-11-11 · 💻 cs.AI · cs.CY· cs.LG

Generative midtended cognition and Artificial Intelligence. Thinging with thinging things

Xabier E. Barandiaran , Marta P\'erez-Verdugo This is my paper

Pith reviewed 2026-05-23 17:52 UTC · model grok-4.3

classification 💻 cs.AI cs.CYcs.LG
keywords generative midtended cognitionhybrid creativityextended cognitionsocial cognitionintentional processesgenerative AIlarge language modelscreative agency
0
0 comments X

The pith

Interventions by generative AI systems form a constitutive part of an agent's intentional creative processes.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces generative midtended cognition to capture how humans integrate iterative outputs from generative AI systems into their own creative intentions. This stands between purely internal directed creation and standard extended cognition that incorporates external processes. A sympathetic reader would care because current theories of mind and creativity may not adequately address the hybrid agency that arises with tools based on large language models. The authors supply an explicit definition that treats AI contributions as internal to the agent's intentional activity and introduce two dimensions of hybrid creativity: width for context sensitivity and depth for iteration granularity. They position this form of cognition as closer to social interaction than to classical extended paradigms while still requiring distinct analysis.

Core claim

Generative midtended cognition treats interventions by AI systems as constitutive of the agent's intentional creative processes. It occupies a position between traditional conceptions of intended creation understood as directed from within and extended processes that bring exo-biological elements into the creative process. The coupling between a human and generative technologies based on multimodal transformer architectures is closer but not equivalent to social cognition than to classical extended cognitive paradigms, yet it deserves specific treatment. Two dimensions of generative hybrid creativity are distinguished: width captures the sensitivity of the context of the generative process,

What carries the argument

generative midtended cognition, the hybrid process in which AI interventions constitute part of the agent's intentional creative activity, positioned between intended and extended cognition and resembling social cognition in its iterative structure.

If this is right

  • Authenticity of creative outputs becomes harder to attribute solely to the human agent when AI interventions count as constitutive.
  • Generative power asymmetry arises between users with and without access to advanced generative systems.
  • Widespread adoption may produce either creative boost through deeper iteration or atrophy through reduced independent generation.
  • Hybrid creativity varies systematically along the width dimension from narrow letter-level sensitivity to broad historical data context.
  • Hybrid creativity varies systematically along the depth dimension from full conversational exchanges down to finer iteration loops.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Legal standards for authorship and copyright might need adjustment if AI contributions are treated as internal to the creator's intentions.
  • Design of generative tools could prioritize features that enhance perceived depth of iteration to maximize hybrid benefits.
  • Empirical measures of creative ownership could be developed by testing whether users claim AI outputs as their own more readily than outputs from non-AI external sources.
  • The framework might extend to other iterative technologies beyond language models, such as image or code generators, to test similar constitutive effects.

Load-bearing premise

The type of cognitive activity in human-generative AI coupling is closer to social cognition than to classical extended cognitive paradigms and therefore needs its own category.

What would settle it

An experiment in which participants consistently report AI-generated suggestions as external aids rather than elements of their own intentions would undermine the constitutive claim.

Figures

Figures reproduced from arXiv: 2411.06812 by Marta P\'erez-Verdugo, Xabier E. Barandiaran.

Figure 1
Figure 1. Figure 1: Different agent-environment loops generate different aspects of cognition. Midtended cognition involves the [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Two main dimensions of generative hybrid space: depth and width. The first refers to the degree of reciprocal [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
read the original abstract

This paper introduces the concept of ``generative midtended cognition'', exploring the integration of generative AI with human cognition. The term "generative" reflects AI's ability to iteratively produce structured outputs, while "midtended" captures the potential hybrid (human-AI) nature of the process. It stands between traditional conceptions of intended creation, understood directed from within, and extended processes that bring exo-biological processes into the creative process. We examine current generative technologies (based on multimodal transformer architectures typical of large language models like ChatGPT), to explain how they can transform human cognitive agency beyond what standard theories of extended cognition can capture. We suggest that the type of cognitive activity typical of the coupling between a human and generative technologies is closer (but not equivalent) to social cognition than to classical extended cognitive paradigms. Yet, it deserves a specific treatment. We provide an explicit definition of generative midtended cognition in which we treat interventions by AI systems as constitutive of the agent's intentional creative processes. Furthermore, we distinguish two dimensions of generative hybrid creativity: 1. Width: captures the sensitivity of the context of the generative process (from the single letter to the whole historical and surrounding data), 2. Depth: captures the granularity of iteration loops involved in the process. Generative midtended cognition stands in the middle depth between conversational forms of cognition in which complete utterances or creative units are exchanged, and micro-cognitive (e.g. neural) subpersonal processes. Finally, the paper discusses the potential risks and benefits of widespread generative AI adoption, including the challenges of authenticity, generative power asymmetry, and creative boost or atrophy.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper introduces the concept of 'generative midtended cognition' to characterize hybrid human-generative AI creative processes (e.g., with multimodal transformer models). It supplies an explicit definition in which AI interventions are treated as constitutive of the agent's intentional creative processes, positions the concept between traditional intended creation and extended cognition while claiming closer affinity (but not equivalence) to social cognition, introduces two dimensions of generative hybrid creativity (width: context sensitivity from letter to historical data; depth: granularity of iteration loops, standing between conversational and subpersonal processes), and discusses risks/benefits including authenticity, power asymmetry, and creative boost or atrophy.

Significance. If the distinctions and definition prove useful, the paper supplies a new conceptual vocabulary for analyzing human-AI creative coupling in cognitive science and AI ethics. The explicit definition and the width/depth dimensions constitute a clear taxonomic contribution; however, the work remains definitional and contrastive rather than deriving predictions, models, or empirical tests.

major comments (1)
  1. [Abstract] Abstract: the premise that the human-AI coupling 'is closer (but not equivalent) to social cognition than to classical extended cognitive paradigms' is asserted without any comparative analysis, counterexamples, or argument showing why this affinity necessitates a distinct concept; this premise directly motivates the call for 'specific treatment' and the new term.
minor comments (2)
  1. [Abstract] Abstract: the dimensions of width and depth are introduced but the text does not indicate how they are operationalized or related to the listed risks/benefits (authenticity, asymmetry, atrophy); a brief mapping in the full manuscript would strengthen the framework.
  2. [Abstract] Abstract: the neologism 'midtended' is presented without reference to its etymology or prior usage in the literature, which may hinder immediate comprehension for readers outside the immediate subfield.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review and the recommendation of minor revision. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the premise that the human-AI coupling 'is closer (but not equivalent) to social cognition than to classical extended cognitive paradigms' is asserted without any comparative analysis, counterexamples, or argument showing why this affinity necessitates a distinct concept; this premise directly motivates the call for 'specific treatment' and the new term.

    Authors: We acknowledge that the abstract states the claim concisely. The manuscript grounds the affinity in the explicit definition, which treats AI interventions as constitutive of the agent's intentional creative processes (a feature shared with social cognition, where contributions from others constitutively shape intentionality, but unlike classical extended cognition, which typically incorporates passive, non-constitutive tools). The width and depth dimensions further mark the hybrid process as standing between conversational social exchanges and subpersonal mechanisms. To make this motivation explicit in the abstract itself, we will revise it to include a brief clause referencing these constitutive and dimensional distinctions. revision: yes

Circularity Check

0 steps flagged

No significant circularity: explicit conceptual definition without reduction to inputs

full rationale

The paper introduces 'generative midtended cognition' via an explicit definition and distinguishes two descriptive dimensions (width, depth) of hybrid creativity as conceptual framing. No equations, fitted parameters, derivations, or predictions appear anywhere in the text. The distinction from extended and social cognition is asserted as a premise for needing specific treatment but is not used to derive further results. No self-citations function as load-bearing justifications for any claim, and the contribution remains self-contained as definitional organization rather than any reduction of outputs to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The framework rests on domain assumptions about the constitution of agency and introduces a new conceptual entity without independent empirical handles.

axioms (2)
  • ad hoc to paper Interventions by generative AI systems can be constitutive of an agent's intentional creative processes.
    Directly invoked in the explicit definition of generative midtended cognition.
  • domain assumption Human-AI generative coupling is structurally closer to social cognition than to classical extended cognition.
    Used to justify the need for a distinct category.
invented entities (1)
  • generative midtended cognition no independent evidence
    purpose: To name and analyze the hybrid constitutive process between humans and generative AI.
    New term and framework introduced by definition; no external falsifiable prediction or measurement supplied in the abstract.

pith-pipeline@v0.9.0 · 5841 in / 1453 out tokens · 27075 ms · 2026-05-23T17:52:16.804581+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

9 extracted references · 9 canonical work pages

  1. [1]

    Akhtar, Z. B. (2024). Unveiling the evolution of generative AI (GAI): A comprehensive and investigative analysis toward LLM models (2021–2024) and beyond. Journal of Electrical Systems and Information Technology, 11(1),

    work page 2024

  2. [2]

    https://doi.org/10.1186/s43067-024-00145-1 Barandiaran, X. E. (2008). Mental Life: A naturalized approach to the autonomy of cognitive agents. [PhD Thesis, University of the Basque Country (UPV-EHU)]. https://xabier.barandiaran.net/phdthesis/ Barandiaran, X. E. (2017). Autonomy and Enactivism: Towards a Theory of Sensorimotor Autonomous Agency. Topoi, 36(...

    work page doi:10.1186/s43067-024-00145-1 2008

  3. [3]

    https://doi.org/10.1007/s13347-022-00560-0 14 Generative midtended cognition and AI Clark, A. (1998). Being there: Putting brain, body, and world together again. MIT Press. Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(03), 181–204. https://doi.org/10.1017/S01405...

    work page doi:10.1007/s13347-022-00560-0 1998

  4. [4]

    Clowes, R. (2020). The Internet Extended Person: Exoself or Doppelganger? LÍMITE Revista Interdisciplinaria de Filosofía y Psicología, 15(22). Cuffari, E. C., Di Paolo, E., & De Jaegher, H. (2015). From participatory sense-making to language: There and back again. Phenomenology and the Cognitive Sciences, 14(4), 1089–1125. https://doi.org/10.1007/s11097-0...

    work page doi:10.1007/s11097-014-9404-9 2020

  5. [5]

    https://doi.org/10.1007/s13347-023-00632-9 Farina, M., & Lavazza, A. (2022). Incorporation, Transparency and Cognitive Extension: Why the Distinction Between Embedded and Extended Might Be More Important to Ethics Than to Metaphysics. Philosophy & Technology, 35(1),

    work page doi:10.1007/s13347-023-00632-9 2022

  6. [6]

    https://doi.org/10.1007/s13347-022-00508-4 Gallagher, S. (2023). Embodied and Enactive Approaches to Cognition (1st ed.). Cambridge University Press. https://doi.org/10.1017/9781009209793 González de la Torre, P., Pérez-Verdugo, M., & Barandiaran, X. E. (2024). Attention is all they need: Cognitive science and the (techno)political economy of attention in...

    work page doi:10.1007/s13347-022-00508-4 2023

  7. [7]

    https://doi.org/10.1007/s13347-024-00696-1 Haraway, D. (1991). Simians, Cyborgs, and Women: The Reinvention of Nature. Routledge. Harvey, I. (2024). Motivations for Artificial Intelligence, for Deep Learning, for ALife: Mortality and Existential Risk. Artificial Life, 30(1), 48–64. https://doi.org/10.1162/artl_a_00427 Heersmink, R. (2015). Dimensions of i...

    work page doi:10.1007/s13347-024-00696-1 1991

  8. [8]

    (2023, February 2)

    https://www.frontiersin.org/articles/10.3389/fnbot.2023.1149303 Hu, K., & Hu, K. (2023, February 2). ChatGPT sets record for fastest-growing user base—Analyst note. Reuters. https://www.reuters.com/technology/chatgpt-sets-record-fastest-growing-user-base-analyst-note-2023-02-01/ Hurley, S. (2010). The Varieties of Externalism. In R. Menary, The Extended M...

    work page doi:10.3389/fnbot.2023.1149303 2023

  9. [9]

    https://proceedings.neurips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html Verbeek, P.-P. (2008). Cyborg intentionality: Rethinking the phenomenology of human–technology relations. Phe- nomenology and the Cognitive Sciences, 7(3), 387–395. Wheeler, M. (2019). The reappearing tool: Transparency, smart technology, and the extended mind. A...

    work page doi:10.1007/s00146-018-0824-x 2017