I thought I was done with this series about the human side of science, drawn from interviews with 27 cosmologists. Earlier posts explored, how this project started, the method, the role of wonder and beauty, the craft, how scientists think, what keeps them going and how they deal with not knowing, what shaped these scientific lives from the outside, and a personal coda.

Then a student I was chatting with asked me if I had looked at the metaphors the scientist were using… I realized that I had not. So I went back to Claude Cowork redid the analysis with this in mind, and this post emerged. In some ways this is a companion to the fifth essay in the series, regarding the internal landscape of scientist’s cognition. This one focuses on the metaphors they use when they’re not talking about how they think.

“I was hooked on astronomy.” —Vera Rubin
“I’m a data merchant, that’s one of my stocks in trade.” —John Huchra
“A bakery filled with delicious goods… nose pressed against the glass.” —Edwin Turner

Three ways of talking about the same enterprise, and each one draws from a different sensory world: Rubin reaches for the body, Huchra for the marketplace, Turner for the bakery window. None of them is answering a question about metaphor or cognition; they’re just talking, which is precisely what makes it interesting.

The idea that metaphors are more than decoration has been around since Lakoff and Johnson’s Metaphors We Live By in 1980, where they argued that figurative language isn’t something we drape over pre-formed thoughts but is itself constitutive, that we think in metaphors before we think in propositions. If that’s right, then the unprompted figurative language scientists use when describing their work should leave traces of how their minds actually operate. And as I worked through these transcripts, I found that it does.

Three scientists in this set all describe themselves as visual thinkers, a label that turns out to conceal more than it reveals. Roger Penrose, who thinks in static geometric forms, talks about tensors “in terms of blobs, with arms and legs,” and his language stays consistently spatial, locked into a fixed point of view as if he’s looking at a diagram that doesn’t move. Sandra Faber inhabits another visual world: she “can’t evaluate a model unless I can generate a mental movie of what’s happening and what’s moving where,” so that where Penrose sees frozen geometry, Faber sees cinema. And Philip Peebles reaches for something else when he describes the distribution of galaxies as “much like the distribution of surface height in water in a choppy pond,” building shapes and terrains in his mind, physical models inherited from a father’s workshop. Three visual thinkers, three different visual worlds, and a reminder that the categories we use to sort minds are almost always too coarse for what they contain.

The pattern holds beyond the visual, and in some cases the contrast between thinkers becomes sharper still. Joseph Silk, an algebraic thinker, reaches for mechanisms and processes (things “gel together,” they “bombard,” they operate on “time scales”), and when he does address the question of vivid language directly he describes it as something you “dress up” ideas with, ornamentation laid over the real work. Set that against Peebles, the only scientist in the set who explicitly credits a metaphor with driving his scientific thinking: “It was strictly a metaphor that drove me to think about galaxy space distributions.” What Silk treats as clothing, Peebles treats as the skeleton.

Alan Guth organizes his intellectual world around a different axis: cleanliness. Particle physics is “a much cleaner subject”; cosmology feels “messy”; he seeks “the cleanest, most mathematically well-formed problems,” and when he does reach for images they tend to be topological, knots and twists in fields, forms defined by their mathematical properties rather than their sensory feel. Stephen Hawking, constrained to roughly 500 words across the entire interview, produces almost no figurative language at all, and that absence is itself diagnostic: when every word costs enormous physical effort, what survives is the architecture of the thought and nothing else. Compare that to Huchra‘s data merchant, or to Allan Sandage, who describes cosmology as entering a “cathedral” and moves from childhood “magic” through a “crushing blow” of disenchantment to a deeper “mysticism” grounded in equations, an entire arc of feeling compressed into three metaphors.

Fred Hoyle has the widest metaphorical range in the corpus, stretching from the “clatter of machinery” in his Yorkshire engineering valley to mathematics as “a huge ocean” where “you feel like if you are thrown into it, there is an infinite distance to swim and you still get nowhere.” Edwin Turner gives us perhaps the most vivid single image: the universe as a “bakery filled with delicious goods” where we stand with our “nose pressed against the glass,” a cosmologist’s existential longing compressed into a child at a shop window.

When I stepped back from individual cases, I was struck by how consistently the metaphor register tracks the cognitive style identified through other evidence. Two independent passes through these transcripts (one conducted with prior knowledge of the scientists’ cognitive profiles, one blind) converged on the same finding: the figurative language lines up. Penrose’s geometric metaphors, Rubin’s embodied ones, Huchra’s commercial ones, Guth’s purity-oriented ones. The metaphor analysis doesn’t discover a cognitive style that the other methods missed; rather, it confirms what they found through a different window, one that doesn’t depend on self-report or introspective accuracy, and that carries particular force because the scientists weren’t trying to tell us anything about how they think.

I’ve spent years thinking about cognitive diversity in educational contexts (the TPACK framework, the work on creativity, the ambigrams that live at the boundary of visual and verbal), and all of it circles around the same conviction: that minds are more varied than our categories acknowledge, that “visual thinker” is not one thing, that “empiricist” is not one thing. What these cosmologists’ metaphors reveal, quietly and without being asked, is that the internal experience of doing science is plural in ways our categories barely register. Penrose and Faber and Peebles all see, but they see in different media. Rubin and Huchra both trust observation over theory, but Rubin’s empiricism lives in the body while Huchra’s lives in the ledger.

What is fascinating of course is now oblivious each of these scientists are about the diversity of cognitive processes, best captured by Peebles’ question: “How else do you think, besides in images?” He was actually puzzled, and that invisible assumption (that everyone’s cognitive world resembles your own) may be the most important thing the metaphors expose. These figures of speech carry the weight of the thinking itself, revealing what kind of thing the universe is for each thinker: machine, organism, painting, puzzle, cathedral, bakery, ocean, or equation.

Which nudges us to ask, as educators, how do the students in our class think? How would we know?