REVIEW 2 major objections 1 minor
Artificial Intelligence, conceptual metaphors and conceptual engineering: Are AI-based framings of human behaviour and cognition successful?
T0 review · 2 major / 1 minor · reviewed 2026-05-22 · grok-4.3
Pith's one-line read AI framings of human behavior and cognition risk the map-territory fallacy as metaphors but can enable conceptual engineering.
desk verdict The paper usefully separates metaphor risks from engineering potential in AI framings of cognition, but its central claims rest on an unshown premise that literal readings are common. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The contrast between conceptual metaphors, which introduce the map-territory fallacy and double metaphor, and conceptual engineering, which opens avenues for revising concept boundaries.
What would settle it
A controlled comparison in which participants reason about human decision-making using AI terms such as neural processing versus traditional psychological terms and check whether their inferences treat the AI model as literally true of the brain.
Extended reading notes
Core claim
When viewed as conceptual metaphors, the AI-framed descriptions risk committing the map-territory fallacy. The comparisons also contain a misleading double metaphor because of the metaphorical connection between human psychology and computation at the conceptual foundation of computation. If the challenges of conceptual ethics and reductionism are overcome, some AI-framings might enrich our epistemic and practical lives. At its worst the AI-framing leads us completely astray; at its best it prompts reflection on how the boundaries of our current concepts serve us and how they could be improved.
Load-bearing premise
That current AI-framings of human behavior are frequently intended or received as literal descriptions rather than loose analogies and that a foundational metaphorical link between psychology and computation creates an unavoidable double metaphor.
Editorial extensions
If this is right
- Treating AI framings as metaphors equates the computational model directly with human cognition and produces systematic errors in explanation.
- The double metaphor originates in the historical use of psychological concepts to define computation and then applying those concepts back to humans.
- Overcoming reductionism lets AI concepts refine rather than replace existing accounts of behavior and cognition.
- Conceptual ethics must be addressed before any redefinition of cognitive terms using AI language can proceed without harm.
Reading between the lines
- Researchers might develop explicit guidelines that force writers to declare whether an AI analogy is meant as loose metaphor or as a proposed conceptual revision.
- The same analysis could apply to other technological domains such as biology or physics when they supply framing concepts for psychology.
- Empirical tests could measure whether exposure to AI-derived terms produces measurable shifts in how people predict or intervene in cognitive tasks.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper examines the increasing use of AI concepts to describe human behavior, neuroscience, and psychology. It contrasts two interpretations of these 'AI-framings': as conceptual metaphors, which risk the map-territory fallacy and embed a misleading 'double metaphor' due to foundational links between psychology and computation; or as attempts at conceptual engineering, which may enrich epistemic and practical understanding if challenges of conceptual ethics and reductionism are addressed. The conclusion holds that such framings mislead at worst but can prompt reflection on conceptual boundaries at best.
Significance. If the central arguments hold, the paper offers a philosophically grounded framework for evaluating AI-based descriptions of cognition, distinguishing risks of literalism from potential conceptual benefits. This could inform interdisciplinary work in cognitive science and AI ethics by highlighting how framings affect concept use, though its impact would be strengthened by concrete cases. The analysis gives credit to the possibility of positive engineering outcomes when conditions are met.
major comments (2)
- [Abstract] Abstract (paragraphs on the two possible answers): The premise that scientists are 'increasingly tempted' to treat AI-framings as literal descriptions (rather than loose analogies) is load-bearing for the map-territory and double-metaphor warnings, yet the text provides no citations, frequency analysis, or specific examples from the literature to establish this prevalence or resulting epistemic harm.
- [Abstract] Abstract: The double-metaphor claim rests on an asserted 'metaphorical connection between human psychology and computation at the conceptual foundation of computation' without historical, conceptual, or referential support for why this link exists, is unavoidable, or necessarily produces a misleading layer; this is central to the metaphor interpretation's critique.
minor comments (1)
- [Abstract] The abstract could more explicitly delineate the transition between the metaphor and engineering views to improve readability of the two-answer structure.
Simulated Author's Rebuttal
We thank the referee for the constructive report and the recommendation for major revision. We address each major comment below, agreeing that the abstract requires strengthening to better support its central premises.
read point-by-point responses
-
Referee: [Abstract] Abstract (paragraphs on the two possible answers): The premise that scientists are 'increasingly tempted' to treat AI-framings as literal descriptions (rather than loose analogies) is load-bearing for the map-territory and double-metaphor warnings, yet the text provides no citations, frequency analysis, or specific examples from the literature to establish this prevalence or resulting epistemic harm.
Authors: We agree that the abstract would be strengthened by greater evidentiary support for this premise. The full manuscript discusses observed trends in the AI and cognitive science literature, but to address the concern directly we will revise the abstract to incorporate specific examples from recent publications that apply AI concepts literally to human cognition, along with citations, thereby clarifying the basis for the map-territory and double-metaphor concerns. revision: yes
-
Referee: [Abstract] Abstract: The double-metaphor claim rests on an asserted 'metaphorical connection between human psychology and computation at the conceptual foundation of computation' without historical, conceptual, or referential support for why this link exists, is unavoidable, or necessarily produces a misleading layer; this is central to the metaphor interpretation's critique.
Authors: The double-metaphor argument receives detailed historical and conceptual development in the body of the paper, referencing the foundational role of computational metaphors in early cognitive science and the work of figures such as Turing and von Neumann. Nevertheless, we accept that the abstract presents the claim without sufficient indication of its grounding. We will revise the abstract to include a brief pointer to this supporting analysis or a concise justification of the link and its implications. revision: yes
Circularity Check
No significant circularity in the conceptual analysis.
full rationale
The paper presents a philosophical argument distinguishing AI-framings of human behavior as either conceptual metaphors (risking map-territory fallacy and double metaphor) or conceptual engineering (potentially enriching if ethical and reductionist challenges are met). It relies on established distinctions from metaphor theory and conceptual ethics without any fitted parameters, self-citation chains for uniqueness theorems, or derivations that reduce by construction to the paper's own inputs. The central claims are interpretive and conditional rather than predictive or definitional loops; the assumption about literal temptations is asserted but does not create a self-referential reduction of the sort enumerated in the analysis patterns. The derivation chain is self-contained as normative reasoning drawing on external philosophical resources.
Assumptions & free parameters
assumptions (3)
- domain assumption AI concepts applied to human cognition can function as conceptual metaphors that risk the map-territory fallacy
- domain assumption Computation theory rests on a metaphorical connection to human psychology, creating a double metaphor
- domain assumption Conceptual engineering can enrich epistemic and practical lives if ethics and reductionism challenges are overcome
Cite this review
Pith. "Pith review of Artificial Intelligence, conceptual metaphors and conceptual engineering: Are AI-based framings of human behaviour and cognition successful?." pith.science (2026). https://pith.science/paper/BYPZBYHI
@misc{pith2026250407756,
author = {Pith},
title = {Pith review of: Artificial Intelligence, conceptual metaphors and conceptual engineering: Are AI-based framings of human behaviour and cognition successful?},
year = {2026},
howpublished = {\url{https://pith.science/paper/BYPZBYHI}},
note = {Machine review of arXiv:2504.07756}
}
read the original abstract
Understanding human behaviour, neuroscience and psychology using concepts from the domain of AI is increasing in popularity. Given the massive integration of AI technologies into our daily lives, AI-related concepts are being used to compare AI systems with human behaviour, brain functions, and cognitive abilities like language acquisition. But scientists and philosophers are also increasingly tempted to take the AI-framing of the human conceptual domain as a literal one. This paper investigates the epistemic and practical success of these 'AI-framings': What does it mean to apply the conceptual constellation of AI to the human conceptual domain? We consider and compare two possible answers: either these examples are conceptual metaphors, or they are attempts at conceptual engineering. Firstly, we argue that when viewed as conceptual metaphors, the AI-framed descriptions risk committing the ''map-territory fallacy''. Secondly, we argue the comparisons also contain a misleading 'double metaphor' because of the metaphorical connection between human psychology and computation at the conceptual foundation of computation. But we also argue that there is a possible semantic catch to the AI-framing, which is captured by the conceptual engineering view. This is that the AI-framings point towards avenues for forms of conceptual engineering. If the challenges of conceptual ethics and reductionism are overcome, some AI-framings might enrich our epistemic and practical lives. So, at its worst - as implicit conceptual metaphor - the AI-framing leads us completely astray; at its best, it prompts us to reflect anew on how the boundaries of our current concepts serve us and how they could be improved.
Reviewed May 22, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.