Theory, Metaphor, Narrative: How Science Creates the Stories it Tells
- Philip Ball
- Dec 8, 2025
- 20 min read
Philip Ball
Adapted from a presentation at ‘The Future of Truth?’
Wenner-Gren Center, Stockholm, Sweden
7-10 May 2025
“For most of the history of scientific methodology the assumption has been that the most important output of science is knowledge.” This claim by philosophers of science Brian Hepburn and Hanne Andersen would not, I suspect, be widely disputed. Science is expected to tell us things we can believe to be true about the world. How then is reliable scientific knowledge generated?
Richard Feynman offered an account of that in a 1964 lecture at Cornell University:
Now I’m going to discuss how we would look for a new law. In general, we look for a new law by the following process. First, we guess it – no, don’t laugh, that’s the truth. Then we compute the consequences of the guess, to see what, if this is right, if this law we guess is right, to see what it would imply and then we compare the computation results to nature or we say compare to experiment or experience, compare it directly with observations to see if it works. [And] If it disagrees with experiment, it’s wrong.
As the voluminous literature on the “scientific method” makes clear, this version of it is simplified to the point of caricature. Not the least of the caveats is that “disagreement with experiment” is a complicated matter – we cannot, for example, be sure the experiment has tested what we want it to test, or that it has been conducted properly. Another physics Nobel Laureate, Steven Weinberg, admitted to the truth of the matter: “We do not have a fixed scientific method to rally around and defend.” [1]
Much of this discussion about scientific methodology focuses on how we might abstract observations and experiments into theories – which are expected to have predictive and not just descriptive capability. Physicist and philosopher Pierre Duhem states that “A physical theory is… a system of mathematical propositions, deduced from a small number of principles, which aim to represent as simply, as completely, and as exactly as possible a set of experimental laws.”[2] There speaks the physical scientist; other disciplines, such as biology and the Earth sciences, are content to accept theories that are not expressed in abstract, mathematical terms – the theory of plate tectonics, for example – yet still serve to rationalize a set of observations and to make predictions about the world.
All the same, these efforts to distil what scientists do into a formula that generates knowledge tend to imply that the knowledge is ultimately expressed in theories. But that is not the same as providing an explanation for what we see. Duhem himself admits that “A physical theory is not an explanation”. His view of an explanation, however, is grand and abstract (as well as being framed in the antiquated sexualized manner): “To explain is to strip reality of the appearances covering it like a veil, in order to see the bare reality itself.”[3]
It doesn’t sound very much like what scientists actually do when they are explaining what they think their results have revealed. “This protein gets phosphorylated”, they might say, “which makes it able to bind this factor and start a signalling cascade.” Or: “The electrons in the sample are able to tunnel across the gap to the probe tip, creating a detectable current.” There are two things such explanations have in common. One is abundant use of metaphor: signalling, cascades, tunnelling. The other is that they tell a story: this happens, and it enables that to happen. The explanation is anything but “bare reality”: it is a narrative every bit as constructed and idealized as those we use to “explain” why our team lost the game or why economic crashes happen, albeit with the advantage that the narrative can feed back into the theory in ways that might allow us to probe its strengths and weaknesses. Or rather: to probe, perhaps, its details but not the validity of its concepts. The ideas of, say, signalling and molecular dialogues in cell biology are not at stake in that validation procedure, but just who speaks to whom.
Such narratives for a particular set of measurements or observations are not created de novo. They rest on meta-narratives about how this or that set of phenomena work. The activity of cells changes because signals they receive are conveyed to the genome and result in the regulation of genes. Quantum mechanics permits particles to traverse energy barriers in ways forbidden in classical mechanics.
Science progresses not just through an interplay of theory and experiment (as per Feynman) but through a delicate dance between those two milieux and the metaphors and narratives deployed to frame its explanations, as they are propagated not just from experts to lay audiences but also among specialists themselves.
What I want to suggest here is, first, that science is often not a competition between theories but a process of narrative creation in which the paradigm tends to be set by the most persuasive narrative—which is often the one that has the most intuitive and compelling metaphors.
This modus operandi is well attested in the history of science,[4] and yet the apparent paradox is that science does make progress, rather than becoming merely a succession of changing stories, each hostage to the preconceptions of its times. I don’t propose here to explain that puzzle, long noted by historians and philosophers of science. But in threading the needle between relativism and social constructivism on the one hand and naïve realism on the other, I wish to draw attention to what I believe to be a neglected issue: we have no mechanism for assessing metaphors and narratives, for example to see whether they continue to match the empirical facts. Much of the important work done by metaphor and narrative in science happens, so to speak, out of sight: it tends to come in the form of traditions and customs that are inherited through the culture of science and, like many cultural norms, rarely questioned. I believe that this aspect of science cannot and should not be subjected to some kind of rigorous testing and scrutiny in the way that theories are; by their very nature, metaphors and narratives need to be afforded space for ambiguity, for open-endedness, and for plurality. Rather, the process needs to happen “in the open” and with recognition that these tools are not simply pedagogical but shape the way scientific questions are framed, priorities set, and resources allocated.
***
Let me begin by defining (without pretence of rigor) some terms.
By a scientific theory, I will be referring to a formalized description of relationships between observable phenomena or quantities, in a way that offers some mechanistic understanding of those phenomena. Often, as Duhem implies, this will be expressed mathematically. The theory of quantum mechanics can be expressed in terms of the Schrödinger equation, for example, which predicts observable outcomes of observations of quantum entities or events:
The theory of general relativity, meanwhile, relates an object’s mass to the curvature of spacetime it induces:
(Never mind the definitions of terms; that is not the point.)
Darwinian evolutionary theory is not quite like this. It is rather expressed in words, perhaps somewhat along these lines:
Species change over time and descend from a common ancestor. That process of change happens through random mutation and natural selection.
It is possible to use this idea to develop mathematical expressions for how, for example, changes in the frequencies of particular gene variants in a population depend on the fitness those variants award to the genotype carrying them.
There is a substantial literature on the relationship between a theory and a model, which I hope I will not too egregiously insult by saying that a model is generally regarded as an idealized description of a specific phenomenon that draws on theoretical ideas. For example, the Bohr model of the atom takes ideas from electromagnetic and early quantum theory to offer a simple picture of how atoms are composed of their constituent particles.
Metaphors make comparisons: they are figures of speech that allow us to talk or think about one thing in terms typically ascribed to another. We might, for example, use the metaphor of the solar system to talk about the Bohr atom, or the metaphor of a rubber sheet to talk about how gravity arises in general relativity. Andrew Reynolds has suggested that at first glance metaphors seem antithetical to science’s efforts to articulate clear, rigorously precise and objective statements of fact about reality.[5] But there is a long tradition, due especially to cognitive linguist George Lakoff and philosopher Mark Johnson,[6] and elaborated by Douglas Hofstadter and Emanuel Sander,[7] which argues that metaphors are the very stuff of thought: language is saturated in them (there is an example), and we can barely communicate anything without them.
Reynolds argues that we can identify three roles of metaphor in science. One is heuristic: by drawing attention to similarities and patterns, metaphors can guide model-building and suggest new hypotheses. The second is cognitive: metaphors help us to parse nature and to get a handle on how it works. The third is pedagogical or rhetorical, and this is how many scientists tend to think metaphor plays a role: it helps experts communicate their ideas without the technical jargon and abstract concepts. In that view, metaphors are sometimes regarded as an unfortunate necessity for speaking outside the ivory tower, but are no substitute for—literally or metaphorically—actually doing the maths.
There is a large literature on metaphors in science too, of course, in which I should mention in particular the work of philosophers Max Black[8] and Mary Hesse[9] in the 1960s. Hesse pointed out that metaphor is often important in the development of a new theory, where ideas may be carried over from one domain to another. For example, early work on molecular collision theory considered molecules to be like billiard balls colliding in a way that can be described using Newton’s laws—even though we had no idea what molecules actually are.
It is worth noting that science accumulates so-called dead metaphors—meaning not metaphors that are no longer useful or appropriate but ones that are no longer even considered metaphors. When a biologist, say, talks about the translation of mRNA to a protein, the fact that “translation” is obviously a metaphor taken from linguistics no longer seems relevant, because the process of translation in molecular biology is understood in considerable detail as a phenomenon in its own right and not a vague analogy with what literary translators do.
Plenty has been said, then, about theories and metaphors in science. There is rather less to be found on narratives. The word, of course, connotes a story. And this, really, is how we communicate ideas, not as a series of isolated metaphors. Linguist Daniel Dor has argued that language itself arose not for the purpose of indication or command, but as a means of telling stories, of projecting the imagination between minds. [10]
Narratives in science could seem even more perilous than metaphors. They are essentially what Francis Bacon warned against as “Idols of the Marketplace.” The writer Francis Spufford has expressed Bacon’s concern nicely:
It was language that was the very biggest problem, with its seductive network of connections between things, and echoes, and metaphors, and even rhymes, all linking up the world obtrusively as you tried to see what was really there; all telling you loud stories about the world while you tried to listen to the quiet facts about objects as they were in themselves. [11]
Yet again, we can’t do without them. The most obvious examples of narratives in science are those involving evolution or growth: the Darwinian evolution of species in the great chain of being, the evolution of the universe from the Big Bang, the development of the organism from the fertilized egg.
Metaphors may suggest but not compel narratives. In the metaphor of the selfish gene, genes are metaphorically selfish because they replicate at the expense of other genes. (I am talking here about the metaphor, not the biological reality.) In the narrative that typically accompanies this picture, organisms are machines whose form and behaviour is dictated by their selfish genes.
Narratives are causal histories: not just one thing after another, but one thing because of a previous thing. And this, of course, is really how we make sense of the world: through causal narratives. As recent politics has taught us, it is not theories and not facts that capture attention and command persuasive power, nor even metaphors or analogies, but narratives. Some scientists have considered a kind of narrative causal ontology, and not just predictive equations, to be science’s ultimate goal. That was the view, at least, of James Clerk Maxwell, who felt that his equations of electromagnetism failed to meet that standard.
***
I want to illustrate these ideas with reference to two fields in particular: molecular biology and quantum mechanics.
These are two fields that have long made extensive use of metaphor, for different reasons. In quantum mechanics we are often forced to take recourse in metaphor because the phenomena being described are so unfamiliar; in molecular biology, meanwhile, the need for metaphor reflects the fact that our intuition is dumbfounded by complexity.
Quantum mechanics, as I said, is a theory. There is no unique or best way to formulate it, and indeed a hundred years ago it was mathematically formulated in two different ways by Werner Heisenberg and by Erwin Schrödinger, who showed that the two schemes are equivalent.
Yet a century later, physicists and philosophers are still undecided about what narrative we should tell about this theory. [12]
Here are some of the things we typically hear about it:
Quantum objects can be both waves and particles.
Quantum objects can be in two states or in two places at once.
We can’t simultaneously know two properties of a quantum object exactly.
Quantum objects can interact instantaneously via a ‘spooky action at a distance’.
You can’t observe quantum objects without disturbing them, and so the human observer plays a role: quantum mechanics is unavoidably subjective.
The theory of quantum mechanics, however, says none of this.[13] They are all metaphors for trying to convey quantum phenomena—to offer some interpretation of what the equations say. And all are misleading to some degree or another, but tellingly so—because all are seeking to express in language concepts for which language was not devised. We live in the classical world, and every intuition, every impulse of thought and expression, is geared to that. As Niels Bohr said, we are “suspended in language,” which is never going to be fit for purpose here.
That said, there is no substantial dispute among physicists that quantum theory works. There has never been a phenomenon observed in quantum physics that conflicts with what the theory predicts. It is one of the best tested, most reliable and indeed most accurate of all scientific theories.
So what is there to argue about?
The arguments are all about what story to tell. In Bohr’s view, called the Copenhagen interpretation, we must simply accept an incompatibility between the classical and the quantum regimes, and recognize the limits of what the theory permits us to say. Our instinct, the universal instinct of the scientist, is not just to do the maths, to “shut up and calculate” in one notorious formulation, but to give a casual account of what is actually going on. Bohr said that we must relinquish this impulse, because quantum mechanics undermines the very notion of causality.
Others find that deeply unsatisfactory. Einstein was one of them, and he and Bohr argued, good-naturedly, for as long as they both lived. Einstein believed there must be what became known as hidden variables that assigned definite values to properties that Bohr insisted are actually indeterminate until they are measured and thereby made classical.
This was one idea that is amenable to experimental testing, and so far all experiments have been consistent with Bohr’s view and not Einstein’s. But there are other ways of restoring definite properties to definite quantum particles that remain consistent with experiments, so long as the particles are moved around under the influence of some mysterious quantum field. That’s another narrative.
Probably the narrative we hear most about is the Many Worlds interpretation devised by Hugh Everett in 1956, in which all possible outcomes of a quantum measurement—a particle here, but also the same particle there— are actually realised, but in separate worlds that split from one another when the measurement is made. In yet another view, measurement induces an actual physical process that collapses all the indistinctness of a quantum state into a definite outcome. And this is not by any means a complete list.
The problem is that, by definition, all these narratives have to be consistent with what the theory itself predicts – which has always fitted with what we see. So it’s not obvious that there are, even in principle, ways to adjudicate experimentally which narrative is right. As a result, the debate is largely a sociological one. Which narrative you adopt tends to be determined by personal preference, by who your mentors and collaborators are, and so forth. The popular appeal of the Many Worlds narrative is obvious, and it’s no coincidence that it has found more favour among physics popularizers than among actual quantum physicists. Nonetheless, it has some very well informed and influential supporters. Some have argued that Bohr’s view dominated the field for decades mostly because he and his Copenhagen circle were very effective and assertive in proselytizing for it. At any rate, this is an excellent case study for how the social dynamics of science influence the dominant narratives.
Practically speaking, one can’t say there is much at stake here. We have managed to make quantum computers without any consensus about what quantum mechanics means. In fact, it’s arguably beneficial to have many competing narratives, since each may suggest different research directions – David Deutsch, who did much of the foundational work in quantum computing, was inspired by his belief in the Many Worlds interpretation.
But still we would like to know who is right, surely? To know what is in fact true about quantum mechanics – which is to say, true about reality itself?
***
Now I want to turn to a field in which the implications of competing narratives might matter a great deal: biology.
Biology has made abundant use of metaphor; for example:
The gene as selfish.
The genome as a blueprint or instruction book – and one that can be “edited”.
The genomic sequence as a language – and one that gets transcribed and translated.
The protein as a machine.
Epigenetics as annotation of a script.
The cell as a factory.
The cancer cell as a rogue agent.
The brain (and the cell) as a computer.
Evolution as a tree of life.
And, as with the old metaphors evoked by Descartes and others in which the body is a system of pumps, levers and pulleys, it has often been ambiguous to what extent these comparisons are meant to be interpreted as analogical or literal.
Ecologists Christoff Kueffer and Brendon Larson have commented on the potential danger of such metaphorical talk:
The problem… is not so much that a metaphor is wrong but that it is misleading: It encourages the interpretation of a partial view as the whole truth or the attribution of too much importance to the view provided by one metaphor as opposed to the different insights provided by a plurality of them.[14]
The selfish gene is perhaps the most notorious of these. Physiologist Denis Noble has pointed out the problem with it.[15] He quotes this extract from Richard Dawkins’ famous 1976 book:
Now they swarm in huge colonies, safe inside gigantic lumbering robots, sealed off from the outside world, communicating with it by tortuous indirect routes, manipulating it by remote control. They are in you and me; they created us, body and mind; and their preservation is the ultimate rationale for our existence. [16]
Noble then systematically reverses this statement:
Now they are trapped in huge colonies, locked inside highly intelligent beings, moulded by the outside world, communicating with it by complex processes, through which, blindly, as if by magic, function emerges. They are in you and me; wea re the system that allows their code to be read; and their preservation is totally dependent on the joy we experience in reproducing ourselves. We are the ultimate rationale for their existence.
He points out that no one seems able to think of an experiment that would detect an empirical difference between these two statements[17]—rather as in the case of the interpretations of quantum mechanics.
Dawkins himself recognizes the limitations of his metaphor, because he has said that he could with equal justification have called his most famous book The Cooperative Gene. Can a metaphor be true and useful if the same can be said for its opposite? Here It is tempting to quote Niels Bohr again, who said that the opposite of a profound truth is also a profound truth. But that seems like an unreliable rule of thumb.
Rather, we need to interrogate the metaphor more closely. There is, after all, no agreed definition of what we mean by a gene, and indeed we will inevitably mean different things by it in different contexts—for example, in evolutionary and developmental biology. Dawkins regularly uses “gene” to mean several different things, without warning or perhaps even recognition. Genes in the sense of pieces of DNA that encode proteins don’t compete in any meaningful sense; the metaphor of cooperation, to produce the organism during development, is far better to describe what they do. Yet it is meaningful to talk about competition, even selfish competition, between different variants of the same gene within a population, as happens in evolution. Different metaphors and different narratives are thus suited to different phenomena, and to different theories. Perhaps we might indeed view them as complementary, in Bohr’s sense.
Dawkins’ view is intimately bound up with the idea of the genome as a blueprint or instruction book: if genes are in control, they must dictate the whole nature of the organism. This is still a mainstream view, at least in genetics, as these examples show. The metaphor underpinned the Human Genome Project, in which regard Andrew Reynolds has aptly described it as a “promotional metaphor.” That is certainly how it functions for mail-order genome-sequencing companies like 23AndMe. Many scientists, especially geneticists and scientists not working in biology, are simply baffled by any suggestion that the metaphor is not a good one.
Yet biological science over the past two or three decades has made it clear that the blueprint metaphor—and the accompanying narrative that the organism is made by a systematic readout of the instructions in the genome—is no longer appropriate.[18]
This notion of a genomic plan is in fact simply a part of the “DNA mystique” critiqued by Dorothy Nelkin and Susan Lindee,[19] and from a different perspective by Evelyn Fox Keller.20 Keller explains how it, and the associated notion of “gene action”, functioned as what I would call an “enabling metaphor”: a manner of speaking that allowed research to continue even while there was a great deal of ignorance about what we were speaking about. And Keller reminds us of how this happens:
What then do I mean when I say that the discourse of gene action- now augmented with metaphors of information and instruction – exerted a critical force on the course of biological research? Can words have force in and of themselves? Of course not. They acquire force only through their influence on human actors. Through their influence on scientists, administrators, and funding agencies, they provide powerful rationales and incentives for mobilizing resources, for identifying particular research agendas, for focusing our scientific energies and attention [and funding!] in particular directions.[21]
And in some ways, says Keller, it has worked well. But at what cost? It wouldn’t be fair to blame geneticists for the way their work has been distorted by “race scientists” who promote the racist and groundless idea that races are fundamentally and irrevocably different in characteristics like intelligence, but nonetheless it is true that the genetic determinism easily read into the blueprint picture is central to that view. The same is true for eugenic arguments, which are certainly not just confined to fringe pseudoscience. The idea that genetic engineering can systematically and predictably alter complex polygenic traits in individuals, or reduce or eradicate complex polygenic diseases in populations is inconsistent what we now know about the relationship between genotype and phenotype. Beyond this, the notion that we are machines made by genes can have a corrosive effect on our perception of human agency.
Yet dislodging this narrative is extremely hard. You can see the appeal: it offers a simple view of a very complex issue. Genetic scientists often complain about the simplistic notions of genetic determinism – the “gene for” picture – portrayed in the media and accepted by the public, but often without acknowledging the role that they and their colleagues have played in ingraining that view, or indeed the way it still informs a great deal of science and professional discourse.
***
My point, then, is not simply concerned with how we might improve science communication by avoiding bad or obsolete narratives. I want to highlight and interrogate how narratives and metaphors appear in science and then define the conceptual paradigm of a field in a way that seems almost immune to the empiricism of the scientific method. While acknowledging that anecdotes are not data, my experience has been that in biology (at least) there is sometimes a reluctance to interrogate metaphorical thinking, particularly when it involves agential language. It is considered enough to say “Of course, when I am talking about genes being selfish or cells trying to make a body, I am speaking metaphorically: there is no conscious intent or decision involved.” But in a context where genuine agency inheres—living organisms really do have goals and reasons, which do not demand complex cognition—this lack of deeper discussion blurs the distinctions: metaphorical agency can end up being required to do real explanatory work, and genuine agency at one level can be denied by displacing it to another level where it can be given the plausible deniability of metaphorical talk. One result is a lack of clarity about what counts as an explanation.
In their search for a kind of truth, scientists are trained to hold their theories lightly, knowing they must stand or fall in the arena of Popperian testing. They don’t always resist developing a strong attachment to a theory—that is only human, especially if the theory is your own—but they recognize the principle. But metaphors and narratives are another matter. In more than 30 years of covering science, I have noticed how tenaciously scientists cling to them, even if it requires an unacknowledged reframing, distortion or reinterpretation of the original narrative. A narrative can in fact signal an affiliation to an ontology or a philosophy. The narrative of the selfish gene warns us away from a rosy-eyed view of a nurturing nature. The machine metaphor for living organisms is considered by some to be indispensable even though organisms clearly do not work like any machine humans have ever made,[22] because—the story goes—the alternative would be to concede a kind of mystic vitalism. The metaphor, in other words, does not here do the pedagogical work of metaphor—to promote understanding by comparing the unfamiliar to the familiar—but serves a totemic function.
Philosopher Brigitte Nerlich has said that “public understanding of science is, at least in part, a struggle over metaphors”[23]— a statement that I think remains true if we omit “public understanding of.” Metaphors and narratives, not equations, are how we express ideas: science always comes with a story attached. But as we know in this age of conspiracy theory, the most alluring stories are not always those closest to what is true.
Philip Ball is a scientist, writer, and a former editor at the journal Nature. He has won numerous awards and has published more than twenty-five books, most recently How Life Works: A User’s Guide to the New Biology; The Book of Minds: How to Understand Ourselves and Other Beings, From Animals to Aliens; and The Modern Myths: Adventures in the Machinery of the Popular Imagination. He writes on science for many magazines and journals internationally and is the Marginalia Review of Books' Editor for Science. Follow @philipcball.bsky.social
[1] Weinberg, S., 1995, “The methods of science… and those by which we live”, Academic Questions, 8(2): 7–13.
[2] P. Duhem, The Aim and Structure of Physical Theory, p. 19. Princeton University Press, 1991.
[3] Ibid, p. 7.
[4] See, for example, S. Shapin & S. Schaffer, Leviathan and the Air-Pump. Princeton University Press, 1985.
[5] A. S. Reynolds, Understanding Metaphors in the Life Sciences. Cambridge University Press, 2022.
[6] G. Lakoff & M. Johnson, Metaphors We Live By. University of Chicago Press, 1980.
[7] D. Hofstadter & E. Sander, Surfaces and Essences. Basic Books, 2013.
[8] M. Black, Models and Metaphors: Studies in Language and Philosophy. Cornell University Press, 1962.
[9] M. Hesse, “The Explanatory Function of Metaphor”, Congress of the Int. Union for the Logic, Methodology and Philosophy of Science, Jerusalem, March 1964.
[10] D. Dor, The Instruction of Imagination: Language as a Social Communication Technology. Oxford University Press, 2015.
[11] F. Spufford, True Stories & Other Essays. Yale University Press, 2017.
[12] E. Gibney, “Physicists disagree wildly on what quantum mechanics says about reality, Nature survey shows”. Nature 643, 1175-1179 (2025).
[13] P. Ball, Beyond Weird. Bodley Head, 2018.
[14] C. Kueffer & B. M. H. Larson, “Responsible use of language in scientific writing and science communication”. BioScience 64, 719-724 (2014).
[15] D. Noble, The Music of Life: Biology Beyond the Genome. Oxford University Press, 2006.
[16] R. Dawkins, The Selfish Gene. Oxford University Press, 1976.
[17] In this case I’m personally not so sure about that. The difference, it seems to me, is in the location of agency: Dawkins puts it in the genes, Noble in the organism. But agency is not simply a metaphor: it can be defined, and we can look for entities that exhibit agential capabilities. And by all meaningful definitions of agency, it is something that complex entities like organisms, with internal degrees of freedoms and representations of their environment, can have, but pieces of DNA or short strings of information cannot.
[18] P. Ball, How Life Works: A User’s Guide to the New Biology. University of Chicago Press, 2023.
[19] D. Nelkin & M. S. Lindee, The DNA Mystique: The Gene as a Cultural Icon. W. H. Freeman,1995.
[20] E. F. Keller, Refiguring Life: Metaphors of Twentieth-Century Biology. Columbia University Press, 1995.
[21] Keller, ibid.
[22] D. Nicholson, “Organisms ≠ Machines”. Stud. Hist. Philos. Biol. Biomed. Sci. 44, 669-678 (2013).
[23] In Reynolds, op. cit. p. 167.
