S1-307 – The golden mountain

Why a thought can be perfectly clear and about nothing at all

Here is a sentence anyone can understand: the golden mountain is tall. It is not nonsense, the way a random string of syllables is nonsense. It has a definite meaning, we can picture what would make it true, and we can even say something true using its subject, as when we say that the golden mountain does not exist. Yet there is no golden mountain anywhere on the earth, and there never has been.

The natural way to explain such a sentence is to say that it does two jobs. First it points: the phrase at the front, the golden mountain, reaches out and picks an object, the way the name Mont Blanc reaches out and picks that particular mass of rock in the Alps. Then it describes: the words that follow, is tall, say something about the thing just picked. This picture feels exactly right, and for ordinary names it mostly is. The trouble is that it cannot be the whole story. If every meaningful subject had to point at something, then the golden mountain would have to be there in order to be pointed at, and one careful philosopher was driven by that very reasoning to conclude that in some fashion it is.

That conclusion is the warning sign, and this article is about what the golden mountain teaches once we refuse to swallow it. The meaning of a thought is one thing, and the existence of what it is about is another. We cannot tell the two apart from inside the thought itself. And the only way to learn whether our words touch the world is to leave the words and go look. This is the map and the territory pushed all the way down into a single sentence, and the surprise is how little of the reaching-out we imagine is really taking place.

Before the philosophers arrive, it helps to see that we already make this move, and not only about mountains. Someone searching for the one, the partner who fits perfectly, can hold a picture that is detailed, consistent, and deeply felt, and still be describing no one who exists. We can chase a true calling, or ache for closure after a loss, with a vivid and coherent idea in mind and no way to tell, from inside the idea, whether the world holds anything for it to be about. The picture can be flawless and answer to nothing, and staring at it harder never reveals the difference. That is the golden mountain in ordinary clothes, and it is why the puzzle earns its keep.

The pointing picture and its ghost

The philosopher was Alexius Meinong, and the pressure he felt was genuine. A sentence like the golden mountain does not exist is not only meaningful, it is true, and it seems to be true of something, the golden mountain, of which it then denies existence. How can we say something true about a thing while insisting there is no such thing? Meinong drew the consistent, and strange, conclusion. The golden mountain does not exist, he allowed, but it must in some thinner way be there, must have a kind of standing, or there would be nothing for our true sentence to be about. Objects like this do not exist, in his vocabulary, but they subsist.¹

Bertrand Russell, writing in 1905 and again in a clearer summary in 1919, thought this was a disaster dressed up as a discovery. Logic, he insisted, should admit a unicorn no more readily than zoology does. To say that unicorns exist in heraldry, or in literature, or in imagination, is a poor evasion, because what exists on the coat of arms is not an animal of flesh and blood but a drawing, and what exists in the play is not a prince named Hamlet but the words Shakespeare wrote and the thoughts we have in reading them. There is only one world, the real one; imagination is a part of it, and the thoughts of a writer are as real as his hands, but they do not conjure a second reality stocked with the objects they are about. If everyone stopped thinking about Hamlet, nothing of Hamlet would remain, because there was never anything to him beyond those thoughts.² The realm of subsisting objects, Russell saw, was invented for one purpose only, to give the pointing picture something to point at. Remove the assumption that every meaningful subject must point, and the ghost has no more work to do.

The hidden claim

Russell’s way of removing that assumption is the part worth carrying home, because it says something unsettling about our own sentences. He proposed that phrases like the golden mountain are not names at all, however much they dress like names. Underneath the grammar, they are disguised claims about what exists.

Take the plainer case first. A mountain is tall, on Russell’s reading, does not point at some particular mountain and describe it. It makes a broader and stranger assertion: there is something that is a mountain, and it is tall. The subject has quietly dissolved into a claim about existence. Adding the word the, Russell argued, does just one extra thing, it insists the thing is unique. So the golden mountain is tall says, underneath, that there is exactly one golden mountain and it is tall. What looked like a name holding an object up for description turns out to be a claim about what the world contains.³

The consequence for the golden mountain is exact. The sentence is not a failed act of pointing, a name reaching into the void and grasping nothing. It is a perfectly ordinary assertion that happens to be false, false because its existence claim, that there is a golden mountain, is false. Nothing mysterious survives. And notice what has happened to the reaching-out we were so sure of. It was never there. The felt sense that the phrase picks an object and holds it up for description is produced by the surface shape of the sentence, not by any actual contact with a mountain. The real content is a claim about existence, and a claim about existence is answerable only to the world.

Existence is not a feature to be read off an idea

Behind Russell’s technical result stands an older and simpler point, and it is the one we were already circling. Existence is not one more property that a thing can be described as having, on a par with being tall or being golden. Kant made the case against the medieval attempt to prove God real by definition alone. We cannot define something into being, he argued, because adding exists to a concept adds nothing to the concept. The idea of a golden mountain that exists contains not one feature more than the idea of a golden mountain; the word exists does not enrich the description, it makes a wholly different kind of claim, that somewhere in the world the description is answered.⁴

This is why no amount of clarity inside a thought can certify that the thought is about anything. Coherence is not contact. A description can be detailed, consistent, vivid, and free of any internal flaw, and answer to nothing whatever, and there is no further feature of the description, no extra brightness or solidity, that would betray the difference. The map can be beautifully drawn and correspond to no terrain, and studying the map more closely will never reveal which kind of map it is. To settle that, we have to do the one thing thinking alone cannot do for us, check the description against the world it claims to be about.

Why we cannot tell from the inside

Put the pieces together and we reach the thing that should genuinely unsettle us. The machinery that builds the golden mountain is tall is the very same machinery that builds the tallest mountain in the Alps is Mont Blanc. From the inside, in the act of thinking them, the two feel identical. Both present a subject that seems to point and a predicate that seems to describe. In one case the pointing lands on something and in the other it lands on nothing, and the thinking gives us no signal at all which case we are in. The sensation of referring, of having a definite object in mind, is generated by the grammar of the thought and not by any grip on the world, and it is exactly as available when there is nothing there.

We are, in other words, fluent producers of well-formed thoughts about nothing, and we cannot audit them from within. The prior article in this pillar, on what a word must do, showed how a shared word can lose its public grip on the world and drift into private meaning; this is the earlier and deeper version of the same lesson, that even a single thought, fully meaningful, tells us nothing by itself about whether its subject is real.⁵

The point generalizes past golden mountains, which no one believes in, to maps that people stake their lives on. A rich fiction is a structure of meaningful sentences answering to no object, and it is harmless because everyone knows it. An ideology and a conspiracy theory, in the parts where they have come loose from the world, have the same structure without the honesty: elaborate, internally consistent, gripping, and sealed off from the check that would test them. The golden mountain is their harmless laboratory specimen. It shows in miniature what a closed map is, a web of sense with no line running out to the territory, and it shows that from inside such a map everything can feel precisely as solid as a truth.⁶

Ordinary political speech runs the move constantly, and the cases below are not fringe curiosities. Take the one percent. As a statistical fact the phrase is real and checkable: some top slice of the distribution holds some measurable share of the wealth, and the number can be looked up and argued over. But the phrase is seldom used that way. It is used as the name of an agent, a thing that wants, decides, and acts to keep the rest of us in place. That agent is a golden mountain. The top percentile of a distribution is not a body that meets, forms an intention, and carries it out. It is a line drawn across millions of strangers who have coordinated nothing. The grammar supplies the shared will and the room where the plan is laid, and neither is there. What exists is a distribution. What we hear is a conspirator, and the gap between the two is the gap between the map and the territory.

The silent majority is the same failure fitted with a lock. The phrase names not the visible many but the unseen ones, the great mass who agree with the speaker and simply do not say so. It is the political cousin of the present king of France, an assertion about a body whose existence is presupposed rather than shown. But it is worse than the king, because the one property that defines this body, its silence, is the very property that seals it against any test. A majority that spoke could be counted, and could turn out to be a minority, or to disagree. A majority whose defining trait is that it does not speak can never be produced, never polled, never shown to be absent, because its silence explains away in advance every failure to find it. Nixon reached for the phrase in 1969, and it has been borrowed many times since and from more than one side, and its appeal is exactly its safety. It is a claim about a vast body of people that no observation is allowed to touch.

The elite runs the move a third way, and it belongs to no single side, since the populist left and the populist right both reach for it. As a description it can be made precise: the people with the most money, or the most credentials, or the most say over what gets decided, and used that way it is checkable. But the word is seldom used that way. It names a coordinating body, a class that meets, agrees, and acts to protect its own, and that body is a golden mountain. Money, credentials, and influence are spread across a huge and quarrelsome population who agree on little and coordinate almost nothing. The grammar supplies the private room and the shared plan, and neither is there. What exists is a scatter of powerful people with clashing interests. What we hear is a single schemer.

None of the three phrases is empty the way the golden mountain is empty. There are rich people, there are citizens who hold their tongue, and there are powerful people. The emptiness is finer than that, and more useful to the person who reaches for it. In each case the sentence hands us a single object, an agent with a will, a majority with a view, a cabal with a plan, that the words have manufactured and the world does not contain, and in each case the solidity we feel in that object is made by the grammar and reports nothing beyond it. The first asserts a unit that is only a spread. The second asserts a body that its own silence forbids us ever to find. The third asserts a schemer where there is only a scatter that coordinates nothing.

There is a further turn, and it is what makes these phrases more dangerous than the golden mountain rather than merely like it. The golden mountain stays inert because no one troubles to furnish it. A political phrase is furnished at once, and by the listener. Because the words hand over an object but no public way to check it, each hearer fills the empty outline privately, and what fills it is a caricature: for the one percent, a figure in a top hat rather than a percentile; for the elite, a shadowy committee rather than a scatter of rivals. Walter Lippmann described this a century ago as the distance between the world outside and the pictures in our heads, and it was he who gave the ready-made picture its modern name, the stereotype, the image we bring to a thing in place of looking at it.⁷ The phrase works like a stencil: the outline is shared and the filling is private, and the trap is that the private filling feels shared, so a room that agrees on no single person can leave it certain that everyone named the same enemy.

This is the failure the article on what a word must do described, now turned to use rather than merely suffered. A word with no public test of correct use drifts into private meaning, and a political phrase keeps that test loose on purpose, because looseness is what lets one word carry a thousand private meanings while a divided crowd feels like one. Two of the three examples buy the most with it. The silent majority is unfalsifiable by design: any claim at all can be hung on it, and its defining silence forbids the check that would test the claim. The elite is its outward-facing twin, an out-group with no fixed membership, against which resentment can be stirred without ever naming a person who could be produced and examined. Neither needs the speaker to be a cynic; sometimes the vagueness is deliberate, sometimes the speaker’s own map is that crude, and often it is both, and the mechanism works either way. And once the picture is in place it seals itself, in exactly the sense this series gives a closed map: the generous rich man and the squabbling elite do not disturb the caricature, they are folded in as the exception, or the disguise.

The dispute Russell did not settle

It would be false to the spirit of this series to present Russell’s analysis as the last word, because it is not, and the ways it has been challenged make the underlying point stronger rather than weaker. Peter Strawson argued in 1950 that Russell had mishandled the empty cases. Say the present king of France is bald when there is no king of France, and Strawson thought the right verdict is not that we have said something false but that we have failed to say anything true or false at all. The sentence presupposes that there is such a king rather than asserting it, and when the presupposition fails the question of truth does not arise; the utterance misfires. That is a different diagnosis from Russell’s, and for our purposes an even sharper one, because it describes a way for a thought to fail to engage the world without announcing its failure as falsehood.⁸

A larger challenge came from Saul Kripke, who denied that ordinary names work the way Russell’s account implies at all. Names, Kripke argued, are not disguised bundles of description; they are fixed to their bearers by a historical chain of use running back to some first naming, so that a name can keep pointing at the right person even when almost everything the speaker believes about them is wrong. Whether reference is secured by description, as Russell held, or by a causal chain, as Kripke held, remains open, and Meinong too has had able modern defenders. But every party to the dispute agrees on the one thing we needed. Meaning does not guarantee existence. What is contested is the machinery by which a word, when it does reach the world, manages to reach it, and not whether a perfectly meaningful word can reach nothing at all.

Why a mind can do this at all

So far the golden mountain has been treated as a puzzle about language and then as a hazard. One plainer question sits underneath both and has not yet been asked: why can we think what is not real at all? A mind built only to register the world as it is would never produce the sentence the golden mountain is tall. Ours produces such sentences without effort, and that ease is a clue to what the machinery is for.

David Hume reached for the very same example. “When we think of a golden mountain,” he wrote, “we only join two consistent ideas, gold, and mountain, with which we were formerly acquainted.” All the creative power of the mind, he argued, amounts to no more than “compounding, transposing, augmenting, or diminishing” the materials that experience supplies.⁹ Hume meant this as a deflation. Imagination invents nothing new; it rearranges. Read with a later question in mind, the rearranging turns out to be the whole point.

Consider what a mind needs in order to act ahead of events. The future does not exist yet. Nothing in it can be seen, touched, or remembered. Any system that prepares for it must therefore build states of the world it has never observed, by taking pieces of what it has observed and putting them together in arrangements that have not occurred. In 1943 the psychologist Kenneth Craik described the payoff. An organism that carries “a small-scale model of external reality and of its own possible actions within its head,” he wrote, “is able to try out various alternatives, conclude which is the best of them, react to future situations before they arise.” Memory research has since found the machinery Craik could only guess at. Imagining a possible future draws on much of the same brain network as remembering the past, and on this account memory is less an archive than a store of parts from which possible futures are assembled.¹⁰ The article on the purpose of thinking (S1-602) follows that argument at length.

Put Hume and Craik side by side and the golden mountain looks different. It is not a defect of language that philosophers had to patch. It is the visible byproduct of foresight. A mind that can assemble the bridge before it is built, the conversation before it happens, and the winter before it arrives will also assemble a mountain of gold, a perfect partner who is no one in particular, and a one percent that meets in a back room, because the assembling does not check whether its products exist. It cannot check, since its most valuable products, the futures, do not exist yet by definition. The power to think what is not real and the power to plan are one power, seen from two sides.

This also explains the finding that unsettled us earlier, that from the inside we cannot tell whether a thought is about anything. One distinction keeps the claim honest. We usually know when we are daydreaming, and the mind keeps a rough and fallible record of which pictures came from the senses and which it made itself. What it does not keep is a record of whether a described thing exists out in the world, because the simulator was never built to ask that question. And a simulation is useful only if it runs as though it were real, following its own logic to see where it leads. What follows is this article’s reading rather than an established result: the felt solidity of an imagined object may be part of what makes simulation work at all, since a rehearsal that felt hollow would teach us less. If so, the vividness that makes the golden mountain persuasive is not a malfunction. It is the machinery doing what it was built to do.

A line often credited to Alfred North Whitehead puts the purpose of all this in a single sentence: the purpose of thinking is to let the ideas die instead of us dying. Whatever its origin, and the documented form of the thought belongs to Karl Popper, it names exactly what the simulation buys.¹¹ A plan that would have failed can fail in the head first. A danger rehearsed need not be met unprepared. The idea dies, and we walk on.

But the bargain carries a condition, and the golden mountain shows what happens when the condition goes unmet. An idea can die in our place only if something kills it, and what kills ideas is the check against the world. A simulation that is never tested is not a rehearsal. It is a belief that has skipped its examination. The schemer called the one percent, the silent majority that agrees, the partner who fits perfectly: each is a product of the same machinery that lets us plan, and each, held without the check, stops dying for us and starts steering us instead. That is why the look this article ends on is not a warning bolted onto an otherwise healthy faculty. It is the faculty’s other half. A mind that builds what is not there needs the habit of going to look, or the ideas that were meant to die for us go on living at our expense.

The look

The practice this series recommends takes a natural, and uncomfortable, form here. The diagnostic question the series keeps returning to, what would it take to change your mind, becomes, for any thought we are tempted to treat as being about something real, a harder cousin: what would the world have to contain for this to be about anything, and have we checked, or only felt the pull of the grammar? The feeling that our words are gripping an object is not a report from the world. It is a product of the sentence, issued just as confidently when the object is missing. So the felt solidity of an idea is no evidence of its reference, and the surest sign that a thought has quietly slipped its mooring is that it never risks the check, that no possible observation is allowed to bear on whether its central terms name anything at all. Everything meaningful can be thought. Only some of it is about anything. And the difference, the whole difference, lies not in the thought but in the territory it has, or has not, been brought to face.

Further reading

Bertrand Russell, “On Denoting” (Mind, 1905), is the seminal paper and among the most admired in the history of analytic philosophy. It is dense and its notation has aged, but the core moves are all here, and even the opening pages repay the effort.

Bertrand Russell, Introduction to Mathematical Philosophy (1919), especially the chapter titled “Descriptions,” is where Russell restates the theory in plain prose for the general reader, and is by far the easier door into the argument.

P. F. Strawson, “On Referring” (Mind, 1950), is the classic reply and the source of the idea that an empty description produces not falsehood but a failure of presupposition. It is short, readable, and quietly reframes the whole question.

Saul Kripke, Naming and Necessity (1980), delivered as lectures and preserving their spoken clarity, is the work that overturned the descriptive theory of names, and it remains one of the most influential and accessible books in modern philosophy.

Notes

¹ Alexius Meinong’s theory of objects (Über Gegenstandstheorie, 1904) is the target of Russell’s criticism. The distinction between existing and merely subsisting objects is Meinong’s attempt to honor the evident meaningfulness of talk about nonexistent things. Whether he was as easily refuted as Russell suggests is disputed; the modern revival of nonexistent objects, in the work of Terence Parsons and Graham Priest, shows the question is still live.

² Russell’s account appears first in “On Denoting” (Mind, 1905) and in the more approachable summary of Introduction to Mathematical Philosophy (1919), from which the phrasings quoted here are drawn. The example of the golden mountain, and the insistence that logic must admit a unicorn no more readily than zoology can, together with the treatment of Hamlet and the heraldic unicorn as drawings and thoughts rather than objects, are Russell’s own.

³ The unpacking of an indefinite description as a claim about existence (there is something that is a mountain, and it is tall), and of a definite description as the same claim with a uniqueness clause added, is the heart of the theory of descriptions. Russell’s worked example is “the author of Waverley was Scott,” analyzed as the conjunction of at least one, at most one, and whoever wrote it was Scott. The result that a sentence containing an empty definite description comes out plainly false, rather than meaningless or unanswerable, is the payoff, and the point on which Strawson later joined issue.

⁴ Kant’s claim that being “is not a real predicate,” that existence adds nothing to the concept of a thing, appears in the Critique of Pure Reason (1781) in the course of dismantling the ontological argument for the existence of God. The connection drawn here, that we cannot read reference off the internal features of a thought because existence is never one of those features, is the series’ use of Kant’s point rather than Kant’s own subject, but the logic is his.

⁵ The prior article in this pillar, on what a word must do (S1-303), argued that a word needs a publicly accessible criterion of correct use. The claim here is one step earlier: that meaningfulness itself, before any question of shared use, does not secure reference. The two together mark off two distinct ways a map can fail, by losing its public grip, and by never having had a referent to grip.

⁶ The wider point, that a coherent and gripping map can be wholly sealed from the world, connects this article to the treatment of grand explanatory schemes that explain everything and so can be refuted by nothing (000d), and to the conspiracy pattern (S1-803), where an internally consistent story is protected from every possible check. The map-territory distinction itself, and Korzybski’s phrase for it, are set out in S1-203.

⁷ The point that an empty or loosely defined phrase is completed by the listener with a ready-made picture is Walter Lippmann’s, in Public Opinion (1922), whose opening chapter is titled “The World Outside and the Pictures in Our Heads” and whose central chapters introduce the modern sense of the stereotype. That the very looseness of such a term can be its political strength, letting one word gather many separate grievances and name a shared enemy, is argued by Ernesto Laclau, “Why do Empty Signifiers Matter to Politics?”, in Emancipation(s) (1996), pp. 36-46. Both refine the article’s point rather than replacing it: the caricature the phrase installs is a picture in the head answering to nothing fixed in the world.

⁸ P. F. Strawson, “On Referring” (Mind, 1950), distinguishes a sentence from a use of a sentence and argues that “the present king of France is bald,” in the absence of such a king, is neither true nor false but a case of presupposition failure. Keith Donnellan, “Reference and Definite Descriptions” (Philosophical Review, 1966), adds a further complication, that a description can be used to refer successfully to a person even when it misdescribes them, so that reference runs partly through the speaker’s intention and the context and not through the descriptive words alone. Saul Kripke’s Naming and Necessity (1980) is the source of the rigid-designation account of names summarized in the main text. All three refine, rather than overturn, the lesson that meaning and existence come apart.

⁹ David Hume, An Enquiry Concerning Human Understanding (1748), Section II, “Of the Origin of Ideas,” where the golden mountain appears beside a second example, the virtuous horse, assembled from the idea of virtue and the familiar shape of a horse. Hume’s purpose was to show that every idea derives from earlier impressions. The reading of his recombination as the working part of foresight is this article’s, not Hume’s.

¹⁰ Kenneth Craik, The Nature of Explanation (1943). The finding that imagining the future depends on much of the same neural machinery as remembering the past, and the proposal that a central function of memory is to supply material for simulating possible futures, are set out in Daniel L. Schacter, Donna Rose Addis and Randy L. Buckner, “Remembering the past to imagine the future: the prospective brain,” Nature Reviews Neuroscience 8 (2007): 657-661, and are known as the constructive episodic simulation hypothesis. That people normally track whether a memory came from perception or from imagination, and sometimes get it wrong, is the reality-monitoring work of Marcia K. Johnson and Carol L. Raye, “Reality Monitoring,” Psychological Review 88 (1981). The inference that the simulator cannot check existence because its most important products, futures, do not yet exist is this article’s own.

¹¹ The sentence circulates widely with Whitehead’s name attached, but the quotation collections that carry it give no source, and a search for this article found it in none of his published works. The documented version of the idea is Karl Popper’s, from his Darwin Lecture, “Natural Selection and the Emergence of Mind,” given at Darwin College, Cambridge, on 8 November 1977: “Let our conjectures, our theories, die in our stead! We may still learn to kill our theories instead of killing each other.” A quotation that carries a famous name and no traceable source is a small golden mountain of its own: it feels as if it points at something, and the only way to find out is to go and look.