Why do cognitive metaphors change based on available technology?
This explores why the metaphors we use to describe our own minds — memory as 'storage,' thinking as 'processing' — seem to borrow from whatever technology is dominant at the time, and what the corpus says about the mechanism behind that borrowing.
This reads the question as being about a specific cultural reflex: when a new information technology arrives, we start describing human cognition in its image. The corpus's sharpest material on this is the idea of 'LLMorphism' How does LLM vocabulary spread beliefs about human thinking?, which names the actual machinery of the transfer. Two mechanisms do the work. The first is analogical transfer — once you have a system that retrieves and recombines, it becomes easy to recast human memory as 'retrieval' and human creativity as 'recombination.' The second is metaphorical availability: the vocabulary of the dominant technology becomes the most psychologically reachable language, so it gets reached for. The striking claim is that this spreads without anyone explicitly arguing that minds are like LLMs — it propagates as a default, just by being the words within easy grasp.
Why does the available vocabulary win by default rather than the most accurate one? A clue sits in an unrelated-looking finding about word frequency Does word frequency correlate with semantic abstraction?: common words systematically express more abstract, general meanings, and systems with a frequency bias drift toward them, erasing specificity. The same pull operates on metaphor. The technology of the moment supplies a stock of high-availability terms, and reaching for the available term quietly swaps a precise account of cognition for whatever the salient analogy makes easy to say. The metaphor changes with the technology because the technology changes what's cheap to reach for.
There's a deeper reason the borrowing feels legitimate rather than arbitrary, and the corpus offers an unusually careful version of it Do humans and LLMs differ fundamentally or just superficially?. Borrowing Habermas's observer/participant distinction: viewed from outside as mechanisms, humans and LLMs are utterly different; but as participants in shared language, both draw on the same symbolic substrate. That shared substrate is exactly what makes a technology's metaphors feel apt for the mind — the resemblance isn't total, but it's real enough at the level of discourse to be persuasive, which is precisely the condition under which a metaphor spreads instead of being laughed off.
Worth noticing that metaphor here isn't decoration — it's load-bearing reasoning. One note reframes all figurative language as a single pragmatic task: recovering literal meaning from non-literal expression Can one model handle all types of figurative language?. If understanding a metaphor is an act of inference rather than lookup, then adopting a technological metaphor for cognition isn't a passive labeling — it actively reshapes how we infer what minds do. And the stakes aren't only conceptual: a four-month EEG study suggests that leaning on AI doesn't just change how we describe thinking but measurably changes the thinking itself, with brain connectivity scaling down under reliance Does AI assistance weaken our brain's ability to think independently?. So the loop closes uncomfortably — the technology supplies the metaphor for the mind, and then using the technology alters the mind the metaphor was describing.
Sources 5 notes
LLM features get projected onto humans through two mechanisms: analogical transfer (memory as retrieval, creativity as recombination) and metaphorical availability (LLM vocabulary becoming psychologically salient). This pattern propagates the bias without requiring explicit endorsement.
WordNet analysis shows hypernyms (general concepts) occur more frequently than hyponyms (specific ones). Combined with LLMs' frequency bias, this means preferring common paraphrases systematically drifts toward abstraction, erasing expert-level specificity.
Applied Habermas's observer/participant distinction to AI: from outside, humans and LLMs are utterly different; from within shared discourse, both draw on the same symbolic substrate, making the difference structural rather than absolute.
The Diplomat dataset (4,177 dialogues) reframes metaphors, idioms, and puns as one pragmatic task: recovering literal meaning from non-literal expression. This framing suggests LLMs need better semantic decoupling ability, not more category-specific training data.
A four-month EEG study of 54 participants found that brain connectivity systematically scaled down with AI reliance—LLM users showed weakest neural engagement, poorest memory retention, and impaired ability to recall their own recent work.