Frances Egan’s deflationary account of mental representation charts a new course between realism and anti-realism — and changes what we should expect from a theory of content.
Frances Egan · Deflating Mental Representation · MIT Press, 2025
Consider a simple perceptual moment: you see a coffee cup on the table in front of you. Common sense and cognitive science both agree that something happens inside your head — a mental state arises that is, in some sense, about the cup. Philosophers call this property “intentionality” or “mental content.” The cup is the cup, your neural state is just neurons firing — so what, precisely, makes one stand in for the other? This deceptively simple question has occupied philosophers of mind for decades. In her book Deflating Mental Representation (MIT Press, 2025), Frances Egan argues that the question has been systematically misunderstood — and that a better framework is not only possible but already implicit in the best science of mind we have.
The Deadlock: Two Unsatisfying Views
Philosophers working on mental representation have long divided into two camps, each capturing something important while leaving something crucial unexplained.
Robust Realism
On one side stand the robust realists. For them, mental representations genuinely refer to things in the world, and this reference is an objective, determinate fact — independent of any observer, any explanatory purpose, any pragmatic interest. Your mental state when thinking about Paris is really about Paris, and this aboutness is a genuine relation between mind and world that a complete science of cognition must account for.
The realist tradition has produced ambitious and ingenious theories. Causal-informational accounts (Fodor, Dretske) try to ground content in reliable causal relationships between mental states and their worldly causes. Teleosemantic theories (Millikan, Neander) appeal to biological function — what a state was selected to track over evolutionary time. Each approach has admirers and critics, but all share the same deep ambition: to show that content is fixed by nature itself, without remainder.
The trouble is that no such theory has commanded consensus. The problem of misrepresentation is stubborn: if your CAT concept is caused by cats, what makes it wrong when a fox in dim light triggers it — rather than simply correct about a slightly unusual fox? And evolutionary function is a blunt instrument: it is unclear it can fix content finely enough to match the precision of our actual concepts. After forty years of sustained effort, as one call for papers on Egan’s book notes frankly, “no one account has achieved wide acceptance.”
Anti-Realism and Eliminativism
On the other side are the anti-realists, who conclude that the realists’ project is not merely difficult but misguided. Mental representations are, at best, useful fictions — ways of talking that serve predictive and practical purposes without carving the mind at its real joints. Dennett’s “intentional stance” is perhaps the best-known version: attributing beliefs and desires to a system is a strategy, not a discovery.
Anti-realism has the advantage of sidestepping the content-determination problem altogether, but it pays a steep price. If mental states don’t genuinely represent anything, it becomes puzzling why representational vocabulary is so indispensable — not just in folk psychology but in rigorous cognitive neuroscience. Scientists who study vision, memory, navigation, and language all freely speak of representations, and this talk seems to do real explanatory work, not mere rhetorical decoration.
Representation is not a substantive relation, nor is it an essential property of mental states — but this does not make it a fiction.
— Frances Egan, Deflating Mental Representation (2025)
The Deflationary Alternative
Egan’s central move is to reject the terms of the debate. Both camps, she argues, share a hidden assumption: that content, if real at all, must be a determinate, mind-independent feature of mental states — something that simply is what it is regardless of who is asking and why. Her deflationary account denies this assumption without thereby denying that content is real or explanatorily significant.
The account rests on three interconnected claims:
THREE CORE CLAIMS
1. No representation relation. Construing a mental or neural state as a representation does not presuppose a special, substantive relation holding between that state and what it is about. There is no deep metaphysical glue connecting your brain state to the coffee cup.
2. Content is not essential. The very same type of mental state could, in a different context or for different explanatory purposes, have been attributed a different content — or no content at all. Content is not a necessary feature of the state as such.
3. Content attribution is always pragmatically motivated. When scientists or ordinary people attribute content to a mental state, they do so in service of specific explanatory goals. The gloss they apply is justified not by a metaphysical relation but by whether it successfully serves those goals.
The key term here is gloss. Egan uses it deliberately: to gloss something is to provide an interpretation, a rendering, a reading that serves a purpose. It is not arbitrary — a good gloss must be grounded in, and accountable to, the underlying facts. But it is also not merely a readout of those facts. The gloss adds something: a framework, a perspective, a set of accuracy conditions that allow us to evaluate behavior as correct or mistaken.
Realism About Vehicles, Pragmatism About Content
A crucial and perhaps surprising feature of Egan’s view is that it is not globally anti-realist. She is a realist about what she calls representational vehicles — the physical states, computational structures, and neural mechanisms that cognitive science identifies as the causal underpinnings of behavior. These are real. They can be studied, modeled, and explained. What is pragmatically motivated is not their existence but the content we attribute to them.
One reviewer, Professor Oron Shagrir of Hebrew University, puts it aptly: the account “combines realism about the computational mechanisms of mental representations with anti-realism regarding their content.” This is a fine-grained distinction, but a crucial one. Egan is not saying that when a neuroscientist identifies a structure in early visual processing as “representing an edge,” there is nothing really there. There is a real computational structure, a real causal role. What the deflationary view insists is that calling it a representation of an edge — rather than of a luminance gradient, or of a surface discontinuity — reflects a decision about how to characterize the mechanism relative to the capacity being explained.
As Egan writes in her Brains Blog précis: “Causal processes are modeled as mathematical processes, in other words, as computations. There is little point in asking whether the structures really represent specific vectors. The representational construal is simply a consequence of characterizing the mechanism as computing the specified function, and this computational characterization is justified insofar as it succeeds in characterizing the causal organization responsible for the target capacity.”
Content in Cognitive Science and Everyday Life
One of Egan’s most important contributions is showing that the deflationary account fits actual explanatory practice better than either rival view. Cognitive scientists do not, when they attribute content to internal states, first check whether a determinate representation relation holds. They ask: does this characterization of the mechanism help us explain and predict the capacities we are studying? The pragmatic criterion is built into the practice.
A similar point applies to commonsense psychology. When you explain your friend’s behavior by saying she believes the train leaves at noon, you are not reporting a bare metaphysical fact about her neural states. You are applying a framework — belief attribution — that serves the purpose of rendering her behavior intelligible and predictable. This does not make the attribution false or arbitrary; it makes it answerable to a different set of standards than the robust realist imagines.
WHY IT MATTERS
Egan’s account dissolves rather than solves the classical content-determination problem. The puzzle of how brute physical states acquire determinate, objective content presupposes that they must have such content independently of any attributive practice. Once that presupposition is dropped, the problem loses its grip — not because it has been answered but because it has been shown to rest on a mistaken picture of what content is.
Perception as Modeling
In a distinctive extension of the deflationary framework, Egan develops a novel account of perceptual experience. Rather than treating perception as a simple window onto the world — a view in which external facts are passively mirrored in inner experience — she proposes that perception involves a kind of modeling: aspects of external reality shape and structure our inner lives in systematic ways, and our experience of those inner states is itself partly a product of this modeling process.
This is a naturalistic, non-naive picture of experience. It acknowledges that what we perceive is not simply “what is out there” but is always shaped by the mechanisms through which the world enters the mind. The deflationary framework applies here too: the representational language we use to describe perceptual states — “she sees a red sphere,” “he represents the scene as three-dimensional” — is a gloss that serves specific explanatory purposes, grounded in real computational processes but not simply read off from them.
Egan in the Broader Debate
Situating Egan’s position in the contemporary landscape helps clarify both its novelty and its stakes. She is in conversation with a rich tradition that includes Fodor’s psychosemantics, Millikan’s teleosemantics, Dretske’s informational theory, and Dennett’s intentional stance. What distinguishes her is the precision with which she delineates the middle ground.
Unlike Dennett, she does not reduce representation to a merely predictive stance — there are real computational structures underlying the gloss, and the gloss is accountable to them. Unlike Fodor, she does not require that those structures have a single determinate content fixed by their causal history. Unlike Millikan, she does not appeal to evolutionary function to anchor content independently of current practice.
Critics have raised genuine questions. A review by Roberto Horácio de Sá Pereira and V. M. Barcellos in Manuscrito identifies “critical tensions regarding the individuation of mental states and the role of the external environment” — pressing issues about whether a purely pragmatic account can adequately handle the ways in which the world constrains content. These are live concerns that Egan’s view must address, and the emerging symposium literature on the book promises to sharpen them further.
Conclusion: Deflation Without Loss
The title of Egan’s book might suggest a defeatist project — as though to deflate mental representation is to puncture something important. But the ambition is quite the reverse. By lowering the metaphysical pressure that has made the content-determination problem seem intractable, she aims to show that representation can do all the explanatory work we actually need it to do — in cognitive science, in neuroscience, in commonsense psychology — without requiring a theory of content that nature has stubbornly refused to provide.
That is a significant philosophical achievement. The book, published as part of the prestigious Jean Nicod Lectures series and now available open access through MIT Press, arrives at a moment when debates about the nature of representation are intensifying across philosophy, cognitive science, and artificial intelligence. Egan’s deflationary framework will not settle those debates, but it reframes them in a way that is, as Professor Albert Newen of Ruhr University Bochum puts it, an overcoming of “the impasse between naive realism and radical rejections.”
For philosophers of mind, cognitive scientists, and anyone who has wondered what it really means for a brain state to be about something, Deflating Mental Representation is essential reading — not because it offers the answer, but because it changes what a good answer would have to look like.
FURTHER READING
— Frances Egan, Deflating Mental Representation, MIT Press, 2025 (open access)
— Frances Egan, “Brains Blog Précis of Deflating Mental Representation,” philosophyofbrains.com, January 2026
— Roberto Horácio de Sá Pereira & V. M. Barcellos, Review of Egan’s book, Manuscrito 48(4), 2025
— Jerry A. Fodor, Psychosemantics: The Problem of Meaning in the Philosophy of Mind, MIT Press, 1987
— Ruth Millikan, Language, Thought, and Other Biological Categories, MIT Press, 1984
— Daniel Dennett, The Intentional Stance, MIT Press, 1987
— Nicholas Shea, Representation in Cognitive Science, Oxford University Press, 2018
— Book Symposium CFP, Philosophy and the Mind Sciences, forthcoming 2026
