What Is Eliminative Materialism?

Direct Answer

Eliminative materialism is one of the boldest positions in philosophy of mind. It says that the mental vocabulary we use every day — belief, desire, hope, fear, intention — doesn't describe real features of the mind at all. These terms belong to what philosophers call folk psychology: a prescientific framework for explaining behavior that, like other prescientific frameworks, will eventually be replaced by something better. And "replaced" here doesn't mean "translated." It means discarded.

The analogy that eliminativists love is historical. Ancient chemists thought fire was caused by the release of phlogiston, an invisible substance supposedly contained in combustible materials. When chemistry advanced, it turned out there was no phlogiston. The concept wasn't reduced to something more fundamental — it was eliminated, shown to refer to nothing real. Medieval medicine explained illness through an imbalance of four humors. Modern medicine didn't translate "excess black bile" into a more precise term. It showed that humors don't exist and replaced the entire framework. Eliminativists like Paul Churchland argue that beliefs and desires will suffer the same fate. Future neuroscience will give us a radically different picture of how the brain works, and our folk-psychological categories will look as quaint as phlogiston.

This is a controversial claim, to put it mildly. It challenges not just a philosophical theory but something most people take for granted: that they really do have beliefs, desires, and intentions. The eliminativist is saying that this deeply felt conviction is mistaken — that what you're experiencing when you "have a belief" isn't what you think it is, and that the categories you use to understand yourself will eventually be recognized as false.

Folk Psychology as a False Theory

The argument begins with a characterization of folk psychology. When you explain why someone went to the kitchen — "she believed there was beer in the fridge and she wanted a drink" — you're deploying a theory. It's not a formal scientific theory, but it has the structure of one: it posits unobservable states (beliefs, desires) that interact in lawlike ways (if you want X and believe action Y will get you X, you'll do Y) to produce observable outcomes (going to the kitchen). Folk psychology is a theory in the same broad sense that folk physics is a theory: it's a framework for prediction and explanation that everyone uses, that works reasonably well in ordinary cases, and that may or may not correspond to how the world actually works.

Churchland argues that folk psychology is a bad theory by scientific standards. It's stagnant — it hasn't improved significantly in thousands of years. It's explanatorily weak — it can't account for sleep, mental illness, memory failures, learning, or consciousness itself. It's full of predictive failures — we're constantly surprised by what people do, which suggests our theoretical framework is missing something fundamental. And it integrates poorly with the rest of science — nothing in physics, chemistry, or biology maps cleanly onto beliefs and desires.

If folk psychology were a scientific theory evaluated on its merits, Churchland argues, we'd have abandoned it long ago. We keep it only because it's woven into our language and our self-understanding, not because it's accurate. But a mature neuroscience won't be constrained by our linguistic habits. It will describe the brain in its own terms, and those terms may not include anything that maps onto "belief" or "desire."

Eliminativism Versus Reductionism

It's crucial to distinguish eliminative materialism from its more modest cousin, reductionism. The reductionist thinks that folk-psychological categories will survive the advance of neuroscience, but in translated form. Just as "water" turned out to be H2O — the same stuff, described more precisely — so "belief" will turn out to be some specific neuroscientific state. The old term isn't discarded; it's refined. Reduction preserves; elimination discards.

The eliminativist thinks this is wishful thinking. There's no guarantee that every useful concept will find a neat neuroscientific translation. The concept of caloric fluid — a hypothetical substance that flowed from hot objects to cold ones — was useful for a time. When thermodynamics replaced it, we didn't discover that caloric was really just molecular kinetic energy. We discovered that there was no caloric. The concept described nothing real, and the theory that employed it was simply wrong. Churchland thinks the same may be true of propositional attitudes. The brain may not contain anything that functions the way beliefs are supposed to function in folk psychology — discrete, semantically evaluable states that combine inferentially and cause behavior. The brain's actual computational architecture may be so different from the folk-psychological picture that no translation is possible.

Daniel Dennett, who's sometimes lumped with the eliminativists, actually occupies a more nuanced position. Dennett argues that beliefs are real enough as intentional stance patterns — they're useful predictions about how a system will behave. But he's skeptical that they correspond to discrete internal states. This puts him somewhere between reductionism and eliminativism: beliefs are real as explanatory tools but may not be real as internal structures. Dennett's position illustrates how the landscape is more complex than a simple reduction-versus-elimination dichotomy.

Objections to Eliminative Materialism

Eliminativism has drawn fierce criticism. The most basic objection is the self-refutation problem. The eliminativist says "beliefs don't exist." But isn't that statement itself a belief? If beliefs don't exist, then the eliminativist can't believe their own theory, and they certainly can't expect anyone else to believe it. The theory seems to undermine the very cognitive act required to assert it.

Churchland's response is to bite the bullet. He argues that what we now call "believing that p" will eventually be described in neuroscientific terms — as a particular pattern of activation in a neural network, say. The statement "beliefs don't exist" should be understood as saying "the states that folk psychology calls beliefs don't exist," not as denying that there are any neuroscientific states that play a roughly analogous role. This is coherent in principle, but critics argue it concedes too much. If there are neuroscientific states that play a roughly analogous role, then perhaps the folk-psychological term "belief" picks out something real after all, even if imprecisely — which looks more like reduction than elimination.

A second objection targets the claim that folk psychology is a bad theory. Critics point out that folk psychology is spectacularly successful by any reasonable standard. We use it to navigate social life, coordinate complex activities, raise children, write novels, and predict each other's behavior with remarkable accuracy. A theory that fails as often as Churchland suggests wouldn't support civilization. The fact that it doesn't explain sleep or mental illness is beside the point — it's not designed to. You might as well say that Newtonian mechanics is a false theory because it doesn't explain genetics.

This connects to a third objection: the plurality of explanatory levels. Philosophy of science recognizes that different sciences operate at different levels of description, and that higher-level descriptions aren't automatically false just because lower-level descriptions exist. Economics explains market behavior in terms of supply, demand, and rational choice. The fact that these aren't neuroscience terms doesn't mean they're false or eliminable. They're the right vocabulary for their level of explanation. Similarly, folk psychology may be the right vocabulary for certain explanatory purposes even if neuroscience provides a more fundamental account of brain function. Eliminating it would be like eliminating the concept of "weather" because atmospheric physics is more precise.

A fourth objection comes from phenomenology. When I introspect, I seem to directly encounter my beliefs and desires. They don't feel like theoretical posits — they feel like given data. The eliminativist asks us to distrust this introspective access, but that's a steep price. How can we be so wrong about our own mental lives? Churchland's answer is that introspection is itself a fallible, theory-laden process. We don't have transparent access to our mental states; we interpret them through the lens of folk psychology. If the lens is distorting, our introspective reports will be distorted too. This is plausible in some cases — cognitive science has shown that people are often wrong about why they did something — but critics question whether it can be extended to the wholesale elimination of propositional attitudes.

Neuroscience and the Future

The eliminativist's bet is that neuroscience will vindicate their position. As we learn more about how the brain actually processes information, we'll discover that its architecture doesn't match the folk-psychological picture. The brain doesn't store discrete beliefs in discrete locations. It doesn't compute by combining propositional attitudes through logical inference. Its representational structures are distributed, probabilistic, and deeply unlike the sentential framework that folk psychology presupposes.

There's something to this. Connectionist models of cognition — where knowledge is stored as patterns of activation across networks rather than as explicit symbolic representations — do challenge the classical picture of mental states as sentence-like structures. If the brain turns out to be fundamentally connectionist, the argument goes, then the concept of a "belief" as a discrete propositional state may not map onto anything real.

But this cuts both ways. Even connectionist networks have representational states that play causal roles in producing behavior. They can be wrong about the world, they can be updated by experience, and they can combine to produce novel outputs. If that's not a belief, it's at least belief-like enough to make the elimination look like a difference of description rather than a difference of ontology. The neuroscience may force us to revise our concept of belief, but revision isn't elimination.

The deepest question the eliminativist raises isn't really about neuroscience. It's about what we're committed to when we use psychological language. Are we describing internal states, making predictions, constructing narratives, or doing all three at once? If folk psychology is primarily a predictive and interpretive tool rather than a theory of internal states, then its replacement by neuroscience isn't an elimination of false entities but a shift in explanatory purpose. The concept of "belief" might survive not because it names a neuroscientific state but because it names a pattern in behavior that remains useful to track even after we understand its neural substrate.

Eliminative materialism may ultimately prove too strong. But its value isn't just in being right. It's in forcing us to ask hard questions about the relationship between our self-understanding and our scientific knowledge — questions that more comfortable positions can too easily avoid.

Further Learning

1. Stanford Encyclopedia of Philosophy, Eliminative Materialism.
2. Paul Churchland, Matter and Consciousness (MIT Press, 1984).
3. Paul Churchland, "Eliminative Materialism and the Propositional Attitudes," The Journal of Philosophy 78 (1981): 67-90.
4. Lynne Rudder Baker, Saving Belief (Princeton University Press, 1987).
5. Daniel Dennett, The Intentional Stance (MIT Press, 1987).

To explore related concepts, consider reading about the philosophy of mind, philosophy of science, and the work of Daniel Dennett.