Salience: the radiance of necessity

I. A trill in the bass

Eight bars into the last sonata Schubert wrote, after a theme so calm that it seems to have been singing long before the pianist sat down, the left hand descends to a low G flat and trills on it, pianissimo, for a little more than a bar. Measured by any instrument, it is the quietest event of the opening page. It is a murmur under the floor, a rumour of thunder from some country far away, and it lasts about as long as a held breath. Yet anyone who has lived with this music knows that the whole movement turns around that trill. The theme was beautiful before it. After it the theme is threatened, the calm becomes a calm that knows something, and every later return of the low G flat, in the development and in the great silences before the recapitulation, will be heard as the return of a guest who was never invited and never quite leaves. The pianist who plays the trill louder, so that the audience will notice it, has destroyed it. The pianist who plays it as a mere ornament has never heard it. Somewhere between these two failures the trill does what it has to do, and the ear of the listener is seized before the mind has decided anything.

The ordinary vocabulary of our time has a word for this seizure, and the word is salience. The trill is salient, we say, and we mean that it stands out, that it captures attention, that it wins the brief competition among the sounds of the page for a place in consciousness. The word arrives with a whole theory folded inside it. It presupposes a mind that scans, a field of stimuli that compete, and a selection that decides which stimulus will be lifted into awareness. Salience, in that theory, is the reward that attention pays to whatever is louder, brighter, newer, more surprising or more useful than its neighbours. Nothing shines until it is looked at, and the one who looks is the sun around whom the perceived world arranges itself.

The trill in the bass refutes this theory with a single pianissimo. It is the least loud, the least bright, the least new event on the page, and it is the one that gathers the whole movement around itself. Its power cannot come from the competition for attention, because it hardly competes. It comes from somewhere else, from the necessity with which that G flat belongs to this music, from the fact that the sonata needs that shadow in order to become the sonata it is. The trill calls, and attention answers. My whole argument in these pages unfolds from this inversion of order, and from the conviction that the inversion, followed all the way down, changes what we take music to be, and with music the human being who sings it.

I will defend a definition. Salience is the perceptible radiance of necessity. It is the mode in which a necessity that would otherwise remain silent and structural enters the senses and lays hold of a living body. To say that a note is salient is to say that here, at this point of the line, being calls more truly than elsewhere, and that the calling has become audible. The definition is short, and every word in it needs to be earned, beginning with the two that carry the most weight, salience itself, which our culture has borrowed from psychology without asking where it came from, and necessity, which our culture has learned to distrust.

II. What leaps

The English word comes from the Latin verb salire, to leap, to spring, to jump. Its present participle saliens described the dancer in the air and the fish breaking the surface of a river, and also, with particular insistence, water. The Romans called a jet of water that rises from the ground by its own force aqua saliens, and the engineers who built the fountains of the city spoke of salientes for the conduits that carried water upward. When Jerome translated the Gospel of John into Latin he gave to the promise made at the well of Sychar its most famous form. The water that Christ offers will become in the one who drinks it fons aquae salientis in vitam aeternam, a spring of water leaping up into eternal life. The Greek behind the Latin uses hállomai (ἅλλομαι), the same verb of leaping, so that the image survives the passage from one language to the other intact. The salient, in its oldest sense, is what springs up from itself.

Later the word acquired a military meaning. In the language of fortification, a salient is the part of a wall or a front that advances beyond the line, the angle that pushes outward toward the enemy. It is salient because it projects, and it projects because the shape of the defence requires it there. Only in the nineteenth century did the word migrate into the vocabulary of perception, and only in the twentieth did it become a technical term of psychology, the name for the property of a stimulus that draws the eye. In Spanish the same root gave saliente, the jutting edge of a rock, and sobresaliente, the highest mark a teacher can give, the work that rises above the others because it deserves to.

The history of the word therefore contains the argument of this essay before the argument begins. In its origin, the salient is what leaps up by its own force, like water from a spring, like the angle of a bastion that the shape of the whole demands. The leap belongs to the thing. The eye that notices the jet of water does not cause it to rise. The psychological use of the word kept the image of standing out and forgot the source of the leap, so that the force that once belonged to the spring was transferred to the gaze of the one who looks at it. To recover salience as an ontological concept is to give the force back to the spring.

III. What cannot yield

The second word is harder to rescue, because necessity has a bad reputation. It sounds like determinism, like the iron chain of causes that leaves no room for freedom, or like the dreary necessity of needs, the necessity of eating and paying rent. Applied to art it sounds worse still, because the modern artist was taught to prize contingency, chance, the gesture that could have been otherwise and was chosen precisely because it could. A philosophy of music that speaks of necessity risks sounding like a nostalgia for rules. The word must therefore be dug up from its roots before it can bear the weight I want to place on it.

Latin grammarians heard in necesse the negation ne joined to the verb cedere, to yield, to give way, to withdraw. Whether or not that etymology is historically exact, it captures the experience the word was meant to name. The necessary is what does not yield. It is what stays when everything that could leave has left, what remains standing when one tries to remove it. Greek had an older and darker word, anánkē (ἀνάγκη), which in Homer and the tragedians is the constraint that binds gods and men alike, the tight embrace from which no one escapes. Plato placed Ananke at the centre of the cosmos in the myth of Er, turning a spindle on her knees while the Sirens sing, each one a single note, and the eight notes together form one harmony. It is a remarkable image to find at the origin of our philosophy. At the axis of the world sits Necessity, and what she produces as the world turns is song.

Aristotle gave the word its technical shape. In the fifth book of the Metaphysics he distinguishes several senses of the necessary, the compulsion that forces something against its tendency, the condition without which a good cannot be attained, and at the root of all of them the simple sense of what cannot be otherwise. More decisive for music is a distinction he draws in the second book of the Physics. There he separates absolute necessity, the necessity of what is always so, from what the commentators later called hypothetical necessity, the necessity that follows from an end. A saw must be made of iron, he writes, if it is to saw. The iron does not cause the sawing, the sawing requires the iron. The end, still absent, reaches back and determines what must be present now.

This is exactly the necessity of a melodic line. A note is necessary when the place toward which the phrase is going requires it, when the arrival that has not yet happened already claims this length, this weight, this silence before it. The low G flat in Schubert is necessary in that sense. Nothing in acoustics forces it. Everything in the movement that will follow requires it. And yet the musical case adds something that Aristotle's saw does not have, because the end of a phrase is not given in advance like the end of a tool. The sawing exists before the saw is made. The end of a phrase comes into existence only by being sung, so that the necessity of each note is discovered in the very act that brings the end into being. Musical necessity is teleological without a preexisting telos. It is the necessity of a direction that creates its goal by walking toward it, and that recognises, once there, that it could not have walked anywhere else.

I have written elsewhere that interior necessity is the border between the singular and the arbitrary. The arbitrary could have been otherwise and it would not matter. The merely general could not have been otherwise and it would not matter either, because it would be the same everywhere. The singular is what could not have been otherwise here, in this phrase, for this body, at this hour, and what matters precisely because it is unrepeatable. Necessity in this sense is the opposite of the iron chain. It is the shape that freedom takes when it is faithful. When a great interpreter plays the trill in the bass exactly as it has to be played, nobody hears a rule being obeyed. Everybody hears a freedom so complete that it has become indistinguishable from necessity, and the trill leaps up like water from a spring.

IV. The tradition that turned around first

It would be dishonest to present the inversion of attention and salience as a discovery made against a unanimous psychology. The phenomenological tradition turned around before me, and the place where it turned deserves to be examined with care, because it is the place where my own path must divide from it.

Husserl, in the lectures on passive synthesis that he gave in the twenties, described something he called affection. Before the ego turns toward an object, he observed, the object already exerts on the ego a kind of pull, an affective force, a Reiz, an allure that solicits the turning. The turning itself, the Zuwendung, is a response to a solicitation that came first. Within the field of what is passively given, some things press forward and others recede, and the ego, when it wakes to attention, follows the contours of this relief. This is already the reversal I am defending. The world solicits before the subject constitutes. But Husserl, faithful to the scientific habits of his age, made contrast the primary condition of affective force. A red patch on a grey ground affects more than a grey patch on a grey ground, a sudden sound more than a steady one. The affective relief of the field is, for him, in large part a relief of differences.

The trill in the bass contradicts this too. Its contrast with its surroundings is minimal, a slight tremor in a low register already shadowed by the pedal, and its affective force is immense. Husserl's analysis, for all its depth, still measures the pull of things by the quantity of their difference. The pull I am describing is measured by the depth of their belonging, by how much the whole requires them. A note can differ very little and belong immensely. Another can differ enormously, a sforzando in a quiet passage, and belong to nothing, so that it startles without calling. Contrast startles. Necessity calls. The first acts on the nervous system and the second on the person.

Bernhard Waldenfels, who died this year after a long life spent rethinking the phenomenology he had learned from Merleau-Ponty, carried the reversal further. In his phenomenology of attention he distinguished the moment in which something strikes us, the Auffallen, from the moment in which we attend to it, the Aufmerken, and insisted that the second is always a response to the first, never its cause. His whole philosophy transformed intentionality into responsivity. We do not aim at things, we answer their claims, their Anspruch, and every answer we give falls short of the claim it answers. Jean-Louis Chrétien, in a book I have loved for many years, made of this structure the very shape of speech, so that every word we utter responds to a call we did not hear begin.

I owe much to these philosophers of the call, and still I have to part from them on one decisive point. For them the call is first. For me something precedes the call, and that something is melos. The world does not begin by calling. It begins by singing, by unfolding itself as a living continuity that sustains itself in time while perishing, the way a melody does, and only within that song do certain points turn toward someone and begin to address him. A melody does not call at every instant. Most of the time it simply sings, and the listener is carried by it like a swimmer by a current. Then, at a certain note, the current turns and faces him. Salience is the name of that turning. It is the hinge at which melos becomes vocative, the point where the song that was singing to no one in particular begins to sing to me. If the philosophers of the call are right that attention answers, they have not seen what the call itself answers to, and what it emerges from. The call is already a moment of the song.

V. The loom of melos

To say where salience happens I need to describe the instrument with which I think about music, and which I call the loom of melos. The Greek word mélos (μέλος) meant both song and limb, a member of the living body, and the double meaning is a programme. Whatever appears by singing itself appears as a living continuity, articulated like a body into members that belong to one another, sustained in time while each moment dies into the next. A melody is the clearest case. Each of its notes perishes so that the following one can be, and yet nothing is lost, because each death becomes a step on a road. I call melos the mode in which being appears when it appears singing itself, and I believe that music is the place where this mode shows its structure with the fewest disguises.

A loom crosses vertical threads, the warp, with horizontal passes, the weft, and the cloth comes into existence in the crossing without existing in any single thread. Plato wrote in the Republic that melos is made of three things, of word, of the fitting together of pitches, and of rhythm. From that sentence come the three threads of my loom. The first is léxis (λέξις), which the rhetoricians later reduced to diction or style, and which I take in an older and deeper sense, the saying before what is said, the fact that every sound that appears comes already turned toward someone, loaded with direction, intention and face, so that before it signifies anything it already addresses. The second is harmonía (ἁρμονία), which before it named the chords of the textbooks named the joint, the art of fitting together, the way that what sounds at once holds itself together in consonance or in tension, without melting and without breaking, a hospitality of sounds toward one another. The third is rhythmós (ῥυθμός), the shape of flowing, what makes time walk instead of merely passing, whose first analogues are in the body before they are in music, in the heartbeat, in the stride of someone walking, in breathing and in dance.

These three threads cross three strata, which I name with three humble words that I keep in Spanish because they sound in my ear as English cannot make them sound. The first is the Carne, the Flesh, the stratum of the face, of breathing, of weight, of the gifts that nobody chose and nobody can return, everything in us that appears before any decision. The second is the Inclinación, the Inclination, the stratum of intention, of gravitation, of gesture, and of the musical figure I love above all others, the appoggiatura, the note that leans on another and entrusts it with its weight before resolving into it. The third is the Hacia, the Toward, the stratum of direction, of continuity, of pulse, of the pure futurity of everything that, by sounding, promises to go on sounding. Three threads across three strata give nine crossings, and in those nine crossings any musical event can be read, along with many events that are not musical at all.

Now the definition can be given its place. Most of the time the threads of the loom run side by side, quietly, the way the threads of a cloth run unseen beneath the pattern. At certain points they cross with unusual force, and the diction of the line, the joining of its harmony and the pulse of its time fall together into a single knot. Salience happens at the knot. It is the point at which the weave gathers into a figure, the way the scattered threads of a tapestry suddenly gather into an eye that looks at you from the wall. The trill in the bass is such a knot. Its léxis is a warning addressed to no one yet, its harmonía is a dissonance against the key the theme had settled in, and its rhythmós is a tremor that unsettles the walking pulse. All three threads cross there, in the stratum of the Flesh, as weight and shadow, and the cloth looks back at the listener.

VI. A grammar of saliences

If salience is the perceptibility of necessity, it will be as differentiated as necessity is, and necessity inside melos is never uniform. No gesture weighs the same at every instant, no breath distributes its urgency evenly, no melodic line is a democracy of equal tones. To sing a phrase is already to have decided, before analysis and before harmony is named, that this syllable longs and that one carries, that this note is the summit and that one the descent that makes the summit legible. The loom allows these differences to be located, and once located they reveal that salience comes in kinds, a small grammar of the ways in which necessity becomes perceptible.

The salience of longing lives in the stratum of the Inclination. Its figure is the appoggiatura, the suspended note held under tension against a resolution it does not yet have, leaning with all its weight on the consonance it delays. Its radiance is the radiance of the incomplete, and it shines forward. When Wagner makes the opening of Tristan und Isolde lean for four hours on a resolution that arrives only in the final pages, he builds a whole drama out of this single salience, and the listener lives inside an inclination that has been prolonged until it becomes a world.

The salience of arrival belongs to the crossing of harmonía and the Toward. The cadence, the homecoming, the syllable that closes the sentence, shine backward. They gather everything that preceded them and give it at last a reason, illuminating the road by which they were reached. Its necessity is retrospective. Only when the phrase arrives do we understand that every earlier note was already walking here.

The salience of memory lives in a different crossing, where rhythmós folds back on itself. The theme that returns transformed, the motif recognised as one recognises a face grown older, shines with the necessity of the again. Valentin Silvestrov built an entire late style out of this salience, writing pieces he calls postludes, music that seems to arrive already as its own echo, as the afterglow of a work we were never allowed to hear, so that every phrase shines with the sadness of having already happened.

The salience of promise lives in the Toward. It belongs to the phrase that does not resolve and instead opens a door in the wall of the present, a modulation that breaks into a space we did not know was there. Its radiance points beyond the edge of what has sounded toward what has not yet come, and its necessity is the strangest of all, because it is the necessity of something that does not yet exist and already obliges.

The salience of consolation and the salience of address both live in the Flesh, near the face. Consolation belongs to the tone that answers a wound, the major third that arrives after a long darkness without triumph and with tenderness, the way a hand is placed on a shoulder. Address belongs to the phrase that suddenly turns and speaks to someone, that ceases to describe and begins to plead. In the slow movements of Mozart's piano concertos the solo voice does this again and again. It has been singing for itself, and at a certain bar it lifts its head and speaks to us.

And the salience of the ground, the least noticed of all, lives where harmonía meets the Flesh. In the chant of the Eastern churches a single held note, the ison (ἴσον), sustains the melody that walks above it. I have written that this drone is the audible Earth, the first analogue of melos, because it teaches a permanence that is the permanence of breath and not of marble, a permanence that costs, that must be maintained at every instant, and that will cease. The ison is almost never the thing we notice, and when it stops, the whole chant falls through the floor. Its salience is the salience of what holds, perceptible above all in its absence, and I will return to it, because it holds the key to the strangest saliences of all.

These kinds are recovered by singing, because the voice cannot help ranking them. The body knows the difference between longing and arrival the way it knows the difference between reaching and having reached. The score records their trace. The performer who cannot tell them apart will play them all in the same way, which is to say falsely, with a single gesture of emphasis applied indifferently to a longing, an arrival and a promise, as if necessity had only one face.

VII. The call and the decoy

Everything so far supposes that salience is received. The modern world, however, has built an immense machinery for manufacturing it, and here the concept acquires an edge.

Spanish keeps two words where English has one. The first is llamada, the call, the summons that comes from the thing itself, that I did not produce, that lays a claim on me because it bears necessity. The second is reclamo. The word means at once an advertisement and the decoy whistle with which a hunter imitates the cry of a bird to draw the living bird toward the net. Hunters also say señuelo, the lure, the stuffed or painted bird placed in the field so that real birds will land beside it. The reclamo is a counterfeit call. It has the acoustic shape of a summons and none of its truth, and it borrows the form of necessity in order to capture an attention that necessity has not earned. The language of Castilian hunters already contained a theory of advertising, and the theory is cruel and exact. The bird that answers the decoy believes it is answering one of its own.

Almost all manufactured salience is reclamo. The performer who makes an important note important by playing it louder, broadening the tempo around it, thickening it with vibrato, delaying it theatrically, has confused the two words. He has decided beforehand which note ought to matter, and then imposed that decision by force of emphasis. The note swells, attention obediently turns toward it, and nothing appears, because no necessity lies under the swelling, only the performer's wish that one should. In my own vocabulary this is the difference between attack and weight. To attack a note is to strike it and leave what follows to a resonance that dies by itself. To give it weight is to rest the whole body in it for as long as it lasts, to sustain it, and I have called the art of sustaining what cannot remain the ars sustinendi. Manufactured salience is always attack. Received salience is always weight.

The empirical study of performance has arrived, by its own roads, at a version of this distinction. Researchers who model expressive piano playing distinguish between immanent accents, those that grouping, metre, melody and harmony already place in the music, and performed accents, those that the pianist adds through timing, dynamics, articulation and pedal. When music theorists marked the immanent accents of the opening bars of a Chopin prelude, and the expressive rendering of the passage was then built around those accents and not around a generic phrasing curve, most listeners preferred the version that followed what the music already carried. The vocabulary is cautious and the experiment small, but the intuition beneath it is ancient. Some saliences belong to the music and the performer finds them, others belong to the performer and the music suffers them.

From inside the experience the difference can be described with precision. False salience arrests the gaze and sends it back to its source. It is opaque, it terminates in itself. When a performer manufactures a climax, the climax opens onto the performer, onto his effort and his desire to be found moving, and the ear strikes the surface and rebounds. Jean-Luc Marion described this structure as the structure of the idol, the visible thing that fills the gaze so completely that the gaze can go no further and sees in it only the measure of its own aim. True salience is transparent. The ear passes through it toward what summoned it. When the quietest tone of a whole movement turns out to be its centre of gravity, the listener does not stop at the tone. The tone is a window, necessity shines through it, and the listener is carried past the sound toward what the sound was faithful to. This is the structure of the icon, which conducts the gaze instead of detaining it. The performer disappears into the disclosure, and nobody is aware of him, though he has never worked harder.

Herbert Simon observed in 1971 that a wealth of information creates a poverty of attention. Half a century later the poverty has become an industry. The economy that lives on our attention is the industrial production of reclamo, an apparatus that generates salience severed from necessity, calls with no caller, decoy whistles at planetary scale. Its deepest effect is pedagogical. It teaches the ear that salience is emphasis, that what shines is what shouts, and it produces in the end a listener who can no longer receive a genuine call, because his capacity to be summoned has been exhausted by a lifetime of lures. Against all of it, the discipline of the artist is a strange and quiet resistance, the refusal to manufacture what can only be received.

VIII. When everything calls

The clinic offers a confirmation of all this that is too grave to be used lightly, and too precise to be ignored. In 2003 the psychiatrist Shitij Kapur proposed that psychosis is, in its first phase, a state of aberrant salience. In the healthy brain, the dopamine system helps to mark certain events as significant, as worth approaching or avoiding. In the prodrome of a psychotic illness this marking becomes chaotic, and events that would ordinarily pass unnoticed acquire an urgent, uncanny significance. A car parked across the street, a phrase overheard on a bus, the colour of a stranger's coat, all begin to mean something, though no one can say what. The delusions that come later, Kapur suggested, are the explanations a person constructs to account for a world that has started to call from every corner. Later studies found that people at high risk of psychosis do attribute motivational salience to irrelevant features of stimuli, and that the strength of this tendency tracks the severity of their delusional ideas.

Psychiatrists who have tried to join this account to the older phenomenology of madness recall the description that Karl Jaspers gave of the delusional atmosphere, the mood in which everything seems changed and is enveloped by a subtle, pervasive and strangely uncertain light. This suffering deserves the gravest respect, and only what it teaches is drawn from it here. It teaches that salience and necessity can be torn apart. A world can shine at every point without anything in it being necessary, and when that happens the result is the opposite of meaning. It is a light without a source, a call without a caller, the most painful form of the counterfeit. Whoever has accompanied a person through such a state knows that the cruelty of it lies precisely here, in a world that addresses you constantly and says nothing.

The distinction I have been drawing between call and decoy is therefore no aesthete's refinement. It names a real fracture in human experience, one that the illness reveals in its extreme form and that our culture of manufactured stimuli reproduces in a diluted and chronic one. A civilisation that makes everything salient in order to sell it does not make anyone ill by that alone, and I would never say it does. But it trains a whole population to live in a faint version of the uncanny light, in a world that calls from every surface and has nothing to say, and it should not surprise us that so many people in that world describe themselves as exhausted, as unable to rest, as unable to hear.

IX. Against the theory of surprise

The strongest rival account deserves its strongest form, and its strongest form is no longer the one I met at the beginning, the salience of the loud and the bright. That simpler account still exists, and it has been formalised with real elegance. In the model of visual attention that Laurent Itti and Christof Koch published in 2000, a scene is decomposed into maps of colour, intensity and orientation, each location is compared with its surroundings, and the differences are summed into a single map of saliency whose peaks predict where the eye will go. The model works remarkably well for the first glance at a photograph. It explains why a red coat in a grey crowd catches the eye. It has nothing to say about the trill in the bass, because the trill is not a peak of difference.

The serious rival is predictive. In its contemporary form, the mind is a machine for anticipating its own inputs, constantly generating predictions about what it will perceive and correcting them by the errors that perception returns. What becomes salient, on this view, is what the model failed to predict, the deviation that forces a revision. Salience becomes surprise, mathematically defined as the improbability of an event given the expectations of the listener. Music psychology has built an impressive science on this foundation. And it would be easy, and false, to answer it by pointing out that the most salient events in music are often the least surprising, the cadence one could have sung four bars before it arrived, the return home that arrives exactly where necessity required. The best theorists of musical expectation saw this long ago.

David Huron, in the book that gave this science its classic form, distinguished several responses that an expected event produces, and among them what he called the prediction effect, the pleasure the mind takes in having predicted correctly. The fulfilled cadence is rewarded because it was foreseen. Neuroscience then showed that the pleasure of music is divided in time. In a study that became famous, Valorie Salimpoor and her colleagues found that dopamine is released in one region of the striatum during the anticipation of a passage that gives listeners chills, and in another region at the moment of the chills themselves. The waiting is itself a reward. Others have insisted on separating two kinds of prediction error that are easily confused, the error about what the next chord will be, and the error about how much one will like it. And the most refined versions of the theory no longer identify salience with surprise at all. They speak of precision, of the weight the brain assigns to each prediction error according to how reliable it judges the signal to be. Salience, in this vocabulary, becomes the gain on a channel, and relevance becomes a function of priors, precision and the actions an agent is preparing.

This is a formidable account, and it understands the fulfilled cadence. Where it fails is elsewhere, in a place that the refinements make clearer instead of hiding. Every quantity in the theory is a property of the model. Surprise is improbability relative to the listener's expectations. Precision is the confidence the brain places in its own channels. Relevance is computed from the agent's priors and goals. Nothing in the theory belongs to the music. The whole weight of salience is still carried by the subject, now a subject made of probabilities instead of acts, and the music is reduced to the occasion on which the model tests itself. The theory can explain why the cadence pleases me. It cannot say why the cadence is right, why it had to be this chord, why a thousand listeners formed on different repertoires, in different centuries, recognise its rightness with the same certainty. The rightness is the thing to be explained, and the theory has placed it outside its own field of vision.

What the predictive account calls a fulfilled expectation I would describe as a necessity recognised. The difference is more than a change of words. A fulfilled expectation is a successful act of my model. A recognised necessity is the moment in which something that does not belong to me comes forth and is seen to be what it had to be. The joy that accompanies it is the joy of recognition, of a truth so deeply prepared that when it arrives one feels one had always known it. That joy is the opposite of surprise and something more than prediction. It is homecoming, and nobody comes home to his own model.

Here the argument touches the phenomenology of givenness at its most daring point, and must part from it. Marion described the saturated phenomenon as the phenomenon that gives more than any concept can receive, and he insisted that such a phenomenon arrives without preparation, as an event no horizon could have foreseen. His examples are overwhelming, the event that tears through history, the face that no gaze can hold. And still the musical case shows another kind of saturation, which Marion's account does not describe. The cadence of the Arietta in Beethoven's last sonata, the final B major of Tristan, the return of the theme at the end of the Goldberg Variations, all are perfectly foreseen, and all saturate. They give more than the mind can hold precisely because they arrive where they were expected, carrying with them the full weight of everything that prepared them. I call this saturation by fulfilment. It is the excess that belongs to the inevitable, the overflowing of a promise kept. The word yes, in the mouth of someone long awaited, is the least surprising word in the language, and it can make a life tremble.

X. The salience of absence

The strangest saliences have nothing in them. A rest can be the loudest event of a phrase. A general pause can carry more necessity than any sounding chord. In the finale of Haydn's String Quartet in E flat, op. 33 no. 2, the music stops so many times before its real ending, with silences of such calculated length, that the audience never knows whether to applaud, and the joke works only because each silence is salient, because each one is heard as an event. In Bruckner the general pauses that separate the blocks of a symphony are the moments in which the cathedral of sound shows that it is built on air. And the Mazurka in A minor op. 17 no. 4 by Chopin ends on the same uncertain harmony with which it began, a sonority that refuses to declare its key, so that the dance seems to dissolve back into the mist from which it came, and the absence of a cadence becomes the most memorable cadence of the piece.

None of this makes sense in the theory of stimulus, because an absent stimulus cannot compete for attention. All of it makes sense if salience belongs to necessity. Necessity has a shape, and the shape can require something that then fails to appear, and the failure appears in its place. The listener whose whole body leans into the expectation of a return, and who is met with silence where the theme should be, hears the silence as an event, because the silence is the perceptible form of a necessity denied. The absent note casts a shadow exactly its own shape. This is negative salience, the shining of what should be and is not, made perceptible by the ache of its lack.

Breath teaches this before analysis can. The inhalation before a phrase, far from being empty time to be minimised, is the salience of preparation, the gathering in which the coming necessity is already present as tension, as readiness, as the not yet that shapes the now. A singer who does not honour the breath begins every phrase from nothing, and so every phrase begins from nothing and arrives from nowhere. The rest is breathing written into the score. It is where the piece breathes, and where the piece breathes it is most alive, because breathing is the proof that a body is not a machine.

A difficult question waits here, and it is the question this essay has not been able to close. If a silence shines because a necessity is denied, something must keep that necessity alive while nothing sounds. The silence does not remember by itself. I believe the answer lies in the two forms of continuity that the loom places at its edges, in the Toward, the pure futurity that keeps the line walking even when it is mute, and in the ground that the ison makes audible, the held note that sustains the chant and that, in silence, goes on being held by the body of the listener. A general pause in Bruckner is heard as an event because the pulse has not stopped in us, and because the harmonic ground of the symphony is still being sustained inside the ear like a drone that no instrument plays. Negative salience is the salience of the Toward and of the ground, made perceptible at the very moment they are left alone. What exactly the listener is holding during such a silence, and with what part of himself he holds it, is something I have not yet finished thinking.

XI. The face of the work

I have spoken for years of something I call the face of the work, el rostro de la obra, and salience is the mode in which a work has a face at all. To face someone is to turn toward him, to present oneself, to appear before him. Spanish has a word for this that English lacks, comparecer, which the courts use for the act of appearing before a judge, and which I have made the centre of my philosophy. Comparecencia is a double structure. The one who appears comes forth, presents himself, steps into the open, and at the same time the one before whom he appears is summoned to answer, placed under the obligation of a response. To appear before is always, at once, to come forth and to lay a claim.

Salience is comparecencia in the register of the senses. When a phrase becomes salient, two things happen inseparably. The phrase steps out of the background into presence, and I am turned toward it and placed under a demand to answer with attention, with fidelity, with care. This is why the human face is the truest image of salience and much more than one example among others. A face can be perfectly still, perfectly quiet, and utterly commanding. Its salience owes nothing to contrast. It faces me, and by facing me it says, before any word, that I am answerable to it. The salience of the face is already an ethics, and it arrives as a claim.

The work of art has a face in exactly this sense. When it is truly present it does not lie before me like an object to be scanned and catalogued. It turns toward me, it comparece, and its salient points are the places where it faces me most fully, where its necessity is most concentrated and its demand most acute. To play a piece truly is to let it acquire a face. To play it falsely is to leave it faceless, a surface of correct notes that never turns toward anyone, that asks nothing and lets everyone pass without claim. The faceless performance is worse than immoral. It is unaddressed. Nothing appeared, because nothing faced.

Fidelity, in this sense, has nothing to do with obedience to a text. The performer stands before the work as before a face, and the face asks him to discover what it truly asks, and to refrain from imposing what he would prefer it to ask. To manufacture salience is to refuse the face, to speak over it, to answer a question it never posed. The faithful performer is the one who has learned to let the work look at him, and to answer the face it actually has, which is always more surprising and more demanding than the one he wished it wore.

XII. The culminating point

Among the musicians of the last century, the one who came closest to a philosophy of salience was a conductor. Sergiu Celibidache taught, in his lectures on musical phenomenology and in thousands of hours of rehearsal, that every phrase has a single culminating point, a summit toward which everything rises and from which everything descends, and that the right tempo of a piece is the tempo at which all its culminations can be perceived as parts of one whole. A tempo too fast leaves the culminations unperceived, and a tempo too slow lets them fall apart into separate events. In his teaching the summit is the place where the music becomes one, and the task of the conductor is to make that unity audible.

I learned more from this doctrine than I can say, and I part from it on two points that matter for everything I have written here. The first is plurality. Celibidache knew one salience, the summit, and he organised the whole of a work as a hierarchy of summits converging on a single peak. The grammar of saliences I have described is richer. The appoggiatura that longs, the cadence that arrives, the motif that remembers, the modulation that promises, the third that consoles, the held note that grounds, are kinds of salience, and they differ in kind, so that a phrase may have its summit in one place and its deepest necessity in another. The trill in the bass of Schubert is no summit. It is a shadow, and the movement turns around it all the same.

The second point is the body. Celibidache rejected intentionality altogether, convinced that any intention of the performer interposes itself between the sound and its appearing, and the reduction to unity he sought was a reduction carried out in consciousness, by an ear trained to the point of austerity. I accept intentionality in the sense of address, of the turning of the line toward someone, and I believe the reduction to unity happens first in the body, in the breath that cannot give equal weight to every note and must decide, in the arm that sustains, in the face that turns. My first teacher, Salvador Chuliá, used to stop a student who had played every note correctly and say that there were no mistakes and still it did not sing. What was missing in such a performance was never a summit. It was the whole grammar, the living differentiation of necessities that a singing body produces without thinking and that a playing body must learn to recover.

XIII. Speech already sings

If salience is prior to analysis, it should be found before music, in the voice that has not yet begun to sing. Leoš Janáček spent three decades proving that it is. From around 1897 until his death he carried a notebook and wrote down, in musical notation, the little tunes that phrases acquire in living mouths, a market woman calling her prices, a child complaining, a dying man asking for water. He called them nápěvky mluvy, speech melodies, and he insisted that they were anything but arbitrary. The shape a word takes in the mouth of a grieving woman, he believed, is dictated by the situation, by the necessity of the moment pressing on a body, so that the melody of speech is a precise record of a soul at a given instant. In one of his essays he transcribed the voice of Bedřich Smetana's daughter, and readers have felt that through her intonation they could almost hear the voice of the dead composer himself, a fragment of life resurrected after time had passed. Milan Kundera saw in this obsession the attempt to capture the present itself, the concrete moment that every novelist and every composer loses as soon as he begins to work.

For my argument the speech melodies are decisive. They show that the salience of an utterance is present before any music is composed, because the situation has already shaped the melody of the words. Léxis and rhythmós are not yet separable there, diction is already song, and the salience of the phrase is the perceptibility of the necessity that the situation imposed on the speaking body. The grieving voice does not deviate from a statistical model. It obeys the necessity of grief, and we hear the obedience as salience. Janáček's whole art demonstrates that melos precedes music, that necessity is audible in speech before it is organised into song, and that salience is prior to composition, prior to harmony and prior to form. It is in the voice, because it is in the situation, because it is in being.

Children confirm this in their own way. In a series of experiments, untrained adults and children as young as three were allowed to move through sequences of chords at their own pace, and they paused longer on the chords that closed phrases, reproducing the phrase final lengthening that expert performers produce. Nobody taught them. Their bodies heard where the necessity of the line gathered and lingered there. Other studies have traced a chain that runs from the predictability of a melody to the timing of the performer and from the timing to the tension the listener feels, and have found that expressive timing is the strongest link in that chain. What reaches the listener as tension passes through the performing body, through the breath and the hand that decide how long a note can be held. Salience is born in the body before the intellect recognises it.

XIV. The discipline of reception

If salience is received and never made, the central labour of the artist is preparation, the long and unglamorous work of becoming a body to which necessity can appear. This is an áskēsis (ἄσκησις) in the old sense of the word, which meant the training of the athlete before it meant the discipline of the monk, and like every real discipline it is chiefly negative. Nobody learns to create salience. One learns to stop obstructing it.

The obstruction is nearly always the self. The performer who wants to be found moving, the interpreter who has a story to tell and imposes it on the piece, the pianist who cannot bear to let a note simply be necessary because he needs it also to be his, are one obstruction under different names. They fill the space where the call would arrive with their own noise, and then, hearing only themselves, mistake the echo for a summons. The discipline of reception is the slow subtraction of this interfering self. It does not annihilate the self, which would be impossible and would leave no one present to receive. It quiets it, makes it transparent, lets it disappear into the disclosure so that the work can face the listener through an unobstructed medium.

Simone Weil wrote, in a letter to Joë Bousquet in the last year of her life, that attention is the rarest and purest form of generosity. In her essay on school studies she described attention as a waiting, a suspension of one's own thought so that the object can enter it, and she warned that most of what we call attention is in fact a contraction of the muscles, a frown, an effort that blocks the very thing it seeks. Iris Murdoch, who read Weil closely, gave this discipline its most famous image. Looking out of a window in an anxious and resentful state of mind, she suddenly sees a kestrel hovering, and in a moment everything is altered, the brooding self has disappeared and nothing remains but the kestrel. She called it unselfing. Both of them understood something that the cognitive psychology of attention cannot formulate, that the highest attention is a receptivity and that the self is its main obstacle.

Singing is the primary training, and now it can be seen why. The voice cannot manufacture what the fingers can. A pianist can fake a hierarchy with the pedal and the wrist, can impose weight where there is none and get away with it for a while. The singing body cannot lie about necessity, because breath is finite and forces a real economy. One cannot give equal weight to everything. One must decide, in the body, what carries and what connects, and the decision is made below the level of theory, in the flesh, where it cannot be faked. To sing a phrase is to be compelled by one's own mortality, by the finite length of one's breath, to establish a true hierarchy of necessity. The ear formed by singing therefore hears a salience that the pedagogy of listening to oneself cannot teach. That pedagogy produces a performer obsessed with the sound he is making. Singing produces a performer obsessed with the necessity he is serving. The first hears himself. The second hears the call. Celibidache, who distrusted almost everything, sang the lines to his orchestras in rehearsal, as anyone can see in the films of his work, and his singing was more eloquent than any of his explanations.

Beyond technique lies a harder discipline that has no method. It is made of patience, which waits for the call instead of filling the silence with a lure, of humility, which suspects its own preferences of being decoys, and of love, which is the only faculty able to receive a face as a face and to answer it as a claim. María Zambrano gave the name of poetic reason to a way of thinking that does not seize its object and instead opens a space in which the object can give itself, a reason that knows by consenting. These are not acquired once and for all. They are practised and lost and practised again, for as long as one plays.

XV. Four faces of one event

If salience is the perceptibility of necessity, it takes its place beside three older words that philosophy has always tried to keep apart and that my ontology asks to think together. I have defined truth as the event of embodied appearing, singular and necessary, the moment in which reality manages to sing itself. Meaning, beauty and salience can then be seen as further faces of that same event, each one turned toward a different power of the person who receives it.

Truth is necessity appearing faithfully, coming forth as what it is, without disguise and without loss. Meaning is necessity becoming intelligible, entering the order of what can be understood, related, said again in other words. Salience is necessity becoming perceptible, entering the senses and laying hold of a body in time. Beauty is necessity becoming radiant, shining with a light that seems to come from within the thing and that awakens love. The four are one event seen from four sides, and their unity explains why each of them, cut off from the others, decays into a caricature. Truth without salience becomes the correctness of a proposition that nobody feels, the dead accuracy of a performance without mistakes that does not sing. Meaning without salience becomes interpretation that no body has lived. Beauty without truth becomes ornament. And salience without the other three becomes exactly what the economy of attention produces and what the clinic of psychosis reveals in its extreme form, a radiance with no source, a call with nothing to say.

The ancient philosophers knew a version of this unity when they spoke of the transcendentals, of being as one, true and good, and later as beautiful. They placed the unity in being, as properties that being possesses whether or not anyone receives them. My version places it in appearing, in the event in which being comes before someone, and so it adds a fourth term that the old list did not need, because the old list did not need a body. Salience is the transcendental of the flesh. It is how the unity of the true, the meaningful and the beautiful reaches a creature who has ears, a breath and a limited time. I believe this tetrad could organise a whole philosophy of art, and I leave it here as one leaves a seed in the ground.

XVI. Proof in time

The whole account rests on a distinction between discovering salience and imposing it, between call and decoy, icon and idol, necessity received and emphasis manufactured. In the moment of performance, from inside the playing body, this distinction cannot be certified. The performer who imposes a false climax experiences himself as discovering. The counterfeiter of salience is rarely a cynic. He is usually a sincere man who has mistaken his preference for a necessity and his emphasis for a call, and he feels exactly the conviction that the discoverer feels.

The test of salience is therefore public and temporal, and this is one of its glories. Necessity is certified by comparecencia, by whether the thing itself comes forth for others, whether the work acquires a face that faces strangers, whether the salience survives the withdrawal of the ego that proposed it. The manufactured climax convinces its author and dies at the edge of the stage. The discovered necessity outlives the performer, is recognised by people who never knew him, returns in other hands and continues to face listeners across generations. Necessity is public in a way that preference can never be. It can be received by anyone properly prepared, because it belongs to being and not to the one who found it. The long history of interpretation is the court in which this test is carried out, and its verdicts are slow and never final.

In the instant of playing the performer acts without proof, answerable to a face whose demand he can misread and sometimes does. This is the shape of his freedom and the dignity of his art. A method that would relieve him of the risk would also relieve the work of its face and reduce the call to a procedure, and music would become what our age keeps trying to make it, a technique that can be verified. The performer receives a responsibility in place of a guarantee. He plays without proof and answers for what he plays, and this is what it means to stand before a face.

XVII. What cannot be kept

The deepest and least noticed feature of salience is that it cannot be stored. Information can be recorded, data archived, stimuli reproduced at will. Salience is the perceptibility of necessity to a living body in lived time, and lived time does not return. The trill that seized you the first time you heard the B flat sonata will not seize you in the same way again. The next hearing will be another event, richer perhaps, or duller, because you are no longer the same and the hour is no longer the same. A child's first word happens once. The hand held at the bedside of someone dying is salient beyond all measure because it will not be held again. The most salient moments of a life are precisely the ones we cannot possess, and they shine because they are passing.

This is why the economy of attention lies at the level of being and not only at the level of morals. It promises a salience that can be summoned at will, replayed, delivered on demand and held. Nothing that can be held on demand is salient in the deep sense, because salience is the shining of the necessary within the finite, and the finite does not repeat. What can be summoned at will is decoy by definition, a call detached from the mortality that makes a real call precious. The counterfeit can be recognised in the end by its very availability.

To receive salience is therefore to consent to loss. It is to let oneself be seized by what one cannot keep, to be present to a radiance whose whole nature is to pass, and to answer it fully, without the reservation that would come from knowing it must end. This is the tragic form of fidelity. The decoy kills the bird it attracts, and possession kills the call it grasps. Fidelity attends to the living thing in the full knowledge that attending will not preserve it. The performer who has understood this plays every phrase as a farewell, out of ontology and not out of sentiment, because every phrase is a farewell, a necessity made perceptible for the length of a breath and then given to the irreversible. Music is the art that makes this condition audible, because its material is time and time is the medium of loss. The radiance and the grief are one light.

And still the trill in the bass returns. Every pianist who learns the sonata will meet it again at the eighth bar, the same low G flat, the same pianissimo, and every time it will have to be found as if for the first time, because nothing in the score can keep it, and every time it will either leap up or remain a mark on paper. The necessity it carries is permanent, and its salience is not. Between the two lies the whole art of the performer, and perhaps the whole condition of a creature who lives in time, called again and again by what does not change, and able to hear the call only in the hours that pass. What remains to be understood is why the same call, which cannot be kept, can nonetheless be answered more truly the second time than the first.