EFTA00284089.pdf

Marvin Minsky Artificial Intelligence Laboratory Music, Mind, and Massachusetts Institute of Technology Cambridge, Massachusetts 02139 Meaning Why Do We Like Music? haps the myth persists because emotions (by their nature) draw attention, while the processes of rea- Why do we like music? Our culture immerses us in son (much more intricate and delicate) must be pri- it for hours each day, and everyone knows how it vate and work best alone. touches our emotions, but few think of how music The old distinctions among emotion, reason, and touches other kinds of thought. It is astonishing aesthetics are like the earth, air, and fire of an an- how little curiosity we have about so pervasive an cient alchemy. We will need much better concepts "environ-mental" influence. What might we dis- than these for a working psychic chemistry. cover if we were to study musical thinking? Much of what we now know of the mind emerged Have we the tools for such work? Years ago, in this century from other subjects once considered when science still feared meaning, the new field of just as personal and inaccessible but which were research called artificial intelligence (AI) started to explored, for example, by Freud in his work on supply new ideas about "representation of knowl- adults' dreams and jokes, and by Piaget in his work edge" that I'll use here. Are such ideas too alien for on children's thought and play. Why did such work anything so subjective and irrational, aesthetic, and have to wait for modem times? Before that, chil- emotional as music? Not at all. l think the prob- dren seemed too childish and humor much too hu- lems are the same and those distinctions wrongly morous for science to take them seriously. drawn: only the surface of reason is rational. I don't Why do we like music? We all are reluctant, with mean that understanding emotion is easy, only that regard to music and art, to examine our sources of understanding reason is probably harder. Our cul- pleasure or strength. In part we fear success itself— ture has a universal myth in which we see emotion we fear that understanding might spoil enjoyment. as more complex and obscure than intellect. In- Rightly so: art often loses power when its psycho- deed, emotion might be "deeper" in some sense of logical roots are exposed. No matter; when this prior evolution, but this need not make it harder to happens we will go on, as always, to seek more understand; in fact, I think today we actually know robust illusions! much more about emotion than about reason. I feel that music theory has gotten stuck by try- Certainly we know a bit about the obvious pro- ing too long to find universals. Of course, we would cesses of reason—the ways we organize and repre- like to study Mozart's music the way scientists sent ideas we get. But whence come those ideas analyze the spectrum of a distant star. Indeed, we that so conveniently fill these envelopes of order? A find some almost universal practices in every musi- poverty of language shows how little this concerns cal era. But we must view these with suspicion, for us: we "get" ideas; they "come" to us; we are "re- they might show no more than what composers minded of" them. I think this shows that ideas then felt should be universal. If so, the search for come from processes obscured from us and with truth in art becomes a travesty in which each era's which our surface thoughts are almost uninvolved. practice only parodies its predecessor's prejudice. Instead, we are entranced with our emotions, which (Imagine formulating "laws" for television screen- are so easily observed in others and ourselves. Per- plays, taking them for natural phenomenon unin- fluenced by custom or constraint of commerce.) The trouble with the search for universal laws This is a revised and updated version of A.1. Memo No. 616. The of thought is that both memory and thinking inter- earlier version will also appear in Music, Mind, and Brain: The Neuropsychology of Music edited by Manfred Clynes, and pub- act and grow together. We do not just learn about lished by Plenum, New York. things, we learn ways to think about things; then 0 1981 by Marvin Minsky we learn to think about thinking itself. Before long, 28 Computer Music Journal EFTA00284089 our ways of thinking become so complicated that presents the elements carefully, not introducing too we cannot expect to understand their details in many new ideas or developing them too far, for un- terms of their surface operation, but we might un- til the basics are learned the pupils cannot build on derstand the principles that guide their growth. In them. So, at first, the teacher repeats a lot. Sonatas, much of this article I will speculate about how lis- too, explain first one idea, then another, and then tening to music engages the previously acquired recapitulate it all. (Music has many forms and there personal knowledge of the listener. are many ways to teach. I do not say that compos- It has become taboo for music theorists to ask ers consciously intend to teach at all, yet they are why we like what we like: our seekers have forgot- masters at inventing forms for exposition, including ten what they are searching for. To be sure, we those that swarm with more ideas and work our can't account for tastes, in general, because people minds much harder.) have various preferences. But this means only that Thus expositions show the basic stuff—the we have to find the causes of this diversity of atoms of impending chemistries and how some tastes, and this in turn means we must see that simple compounds can be made from those atoms. music theory is not only about music, but about Then, in developments, those now-familiar com- how people process it. To understand any art, we pounds, made from bits and threads of beat and must look below its surface into the psychological tone, can clash or merge, contrast or join together. details of its creation and absorption. We find things that do not fit into familiar frame- If explaining minds seems harder than explaining works hard to understand—such things seem songs, we should remember that sometimes enlarg- meaningless. I prefer to turn that around: a thing ing problems makes them simpler! The theory of has meaning only after we have learned some ways the roots of equations seemed hard for centuries to represent and process what it means, or to under- within its little world of real numbers, but it sud- stand its parts and how they are put together. denly seemed simple once Gauss exposed the larger What is the difference between merely knowing world of (so-called) complex numbers. Similarly, (or remembering, or memorizing) and understand- music should make more sense once seen through ing? We all agree that to understand something we listeners' minds. must know what it means, and that is about as far as we ever get. I think I know why that happens. A thing or idea seems meaningful only when we have Sonata as Teaching Machine several different ways to represent it—different per- spectives and different associations. Then we can Music makes things in our minds, but afterward turn it around in our minds, so to speak: however it most of them fade away. What remains? In one old seems at the moment, we can see it another way story about Mozart, the wonder child hears a and we never come to a full stop. In other words, lengthy contrapuntal mass and then writes down we can think about it. If there were only one way to the entire score. (I do not believe such tales, for his- represent this thing or idea, we would not call this tory documents so few of them that they seem to representation thinking. be mere legend, though by that argument Mozart So something has a "meaning" only when it has a also would seem to be legend.) Most people do not few; if we understood something just one way, we even remember the themes of an evening's concert. would not understand it at all. That is why the Yet, when the tunes arc played again, they are rec- seekers of the "real" meanings never find them. ognized. Something must remain in the mind to This holds true especially for words like under- cause this, and perhaps what we learn is not the stand. That is why sonatas start simply, as do the music itself but a way of hearing it. best of talks and texts. The basics are repeated sev- Compare a sonata to a teacher. The teacher gets eral times before anything larger or more complex the pupils' attention, either dramatically or by the is presented. No one remembers word for word all quiet trick of speaking softly. Next, the teacher that is said in a lecture or all notes that are played Minsky 29 EFTA00284090 Fig. 1. Introductory mea- sures of Ludwig van Beethoven's Symphony No. 5 in C Minor. -4014/ 2Allegro omAtorloOhasel Flutes Mote 4Ai Clarinets in Bb le ! ' I C) , Broca vvit - ir—trribt 1 sr. Fla= in Bib 4 it% Trumpets in C 4 Il Timpani in C.G 4a, I Allegro eoerio‘i • soe) Violin I Violin II (44 1 , 111 ' 171 C) -L 1111 ±j--1D -3— ,rrr I fr. --- L tttt r Viola J kJ C) Cello I in Crn 1 cr - r- Br -CT ? te r in a piece. Yet if we have understood the lecture or prefigures the second half—call it B? It is simple: piece once, we now "own" new networks of knowl- an agent A that recognizes A sends a message to edge about each theme and how it changes and re- another agent B, built to recognize B. That message lates to others. No one could remember all of Bee- serves to "lower B's threshold" so that after A hears thoven's Fifth Symphony from a single hearing, but A, B will react to smaller hints of B than it would neither could one ever again hear those first four otherwise. As a result, that mind "expects" to hear notes as just four notes! Once a tiny scrap of sound, B after A; that is, it will discern B, given fewer or these four notes have become a known thing—a more subtle cues, and might "complain" if it can- locus in the web of all the other things we know not. Yet that mind cannot reproduce either theme and whose meanings and significances depend on in any generative sense. The point is that interagent one another (Fig. I). messages need not be in surface music languages, Learning to recognize is not the same as memo- but can be in codes that influence certain other rizing. A mind might build an agent that can sense agents to behave in different ways. a certain stimulus, yet build no agent that can re- (Andor Kovach pointed out to me that composers produce it. How could such a mind learn that the do not dare use this simple, four-note motive any first half subject of Beethoven's Fifth—call it A— more. So memorable was Beethoven's treatment 30 Computer Music Journal EFTA00284091 "IA 14 rffr k t'k IR peas& 1 f Pp * if a '"S a 04 kr tRi I a O I .in LJ r / a 1 ,I77 4 aI ifprism. I ll - I - I - I - lyped Aillikl II - l-t - I - + - A , —eller —tai f---tLr 1r 4 C- 1 40L) 141 'a 4 J71 C) a, j WS r Irrr Cr rrr IIf Ji 11IT k Ji asik't r kr' I Cur f J j III CA • jjeJa J k 4 nJ 4 pin a i- AMEN &CO ' -rtetfi&v rr id tin (sr ifft_ mb, nO Acti -2 Jig! rte& r ■ -475- -4 I 'Ifs .1 4- I Minsky 31 EFTA00284092  'di git f ir MC? r I tf I I a ✓ 1 P ir yt--41 & 304 ettrtnit nittritt ii OFF tarit fr irr4 cm 0*-- e e e 4 4444 4 - 7 --- 137t4j 4, i MIL.7 • 4 JJ -3 gi JJJ f in CH • .147 d d ; i 1 1 i .1 1 ' 4 - .1.1l ✓ ✓ ✓ e v i di _. .,1 _.... 4 3 1 1 I 4Al _1 I i t -. ef f Ac-TiCrf 0 ? ar k — 1 1 IMP 1 f ; 4 t lints II II Is lig-I -I -1-1- 1 - I - I A €f 5trrkp i 40L *-kJ-4J- PP 4 32 Computer Music journal EFTA00284093  that now an accidental hint of it can wreck another to place each metric frame precisely on the after- piece by unintentionally distracting the listener.) image of the one before. If sonatas are lessons, what are the subjects of those lessons? The answer is in the question! One Cadence. Silence. Almost. Total. thing the Fifth Symphony taught us is how to hear those first four notes. The surface form is just de- Now it is the second subject-twin's turn to stand scending major third, first tone repeated thrice. At alone in time. The conductor must select a symme- first, that pattern can be heard two different ways: try: he or she can choose to answer prior cadence, (1) fifth and third in minor mode or (2) third and to start anew, or to close the brackets opened at the first, in major. But once we have heard the sym- very start. (Can the conductor do all at once and phony, the latter is unthinkable—a strange con- maintain the metric frame?) We hear a long, long straint to plant in all our heads! Let us see how it unison F (subdominant?) for, underneath that silent is taught. surface sound, we hear our minds rehearsing what The Fifth declares at once its subject, then its was heard. near-identical twin. First comes the theme. Pre- The next frame reveals the theme again, descend- sented in a stark orchestral unison, its minor mode ing now by thirds. (We see that it was the dominant location in tonality is not yet made explicit, nor is ninth, not subdominant at all. The music fooled us its metric frame yet clear: the subject stands alone that time, but never will again.) Then tour de force: in time. Next comes its twin. The score itself the subject climbs, sounding on every scale degree. leaves room to view this transposed counterpart This new perspective shows us how to see the four- as a complement or as a new beginning. Until note theme as an appogiatura. Then, as it descends now, fermatas have hidden the basic metric frame, on each tonic chord-note, we are made to see it as a a pair of twinned four-measure halves. So far we fragment of arpeggio. That last descent completes a have only learned to hear those halves as separate set of all four possibilities, harmonic and direc- wholes. tional. (Is this deliberate didactic thoroughness, or The next four-measure metric half-frame shows merely the accidental outcome of the other sym- three versions of the subject, one on each ascending metries?) Finally, the theme's melodic range is pitch of the tonic triad. (Now we are sure the key is squeezed to nothing, yet it survives and even gains minor.) This shows us how the subject can be made strength as single tone. It has always seemed to me to overlap itself, the three short notes packed per- a mystery of art, the impact of those moments in fectly inside the long tone's time-space. The second quartets when texture turns to single line and forte- half-frame does the same, with copies of the com- piano shames sforzando in perceived intensity. But plement ascending the dominant seventh chord. such acts, which on the surface only cause the This fits the halves together in that single, most structure or intensity to disappear, must make the familiar, frame of harmony. In rhythm, too, the largest difference underneath. Shortly, I will pro- halves are so precisely congruent that there is no pose a scheme in which a sudden, searching change room to wonder how to match them—and attach awakes a lot of mental difference-finders. This very them—into one eight-measure unit. change wakes yet more difference-finders, and this The next eight-measure frame explains some awakening wakes still more. That is how sudden more melodic points: how to smooth the figure's silence makes the whole mind come alive. firmness with passing tones and how to counter- We are "told" all this in just one minute of the poise the subject's own inversion inside the long lesson and I have touched but one dimension of note. (I think that this evokes a sort of sinusoidal its rhetoric. Besides explaining, teachers beg and motion-frame idea that is later used to represent threaten, calm and scare; use gesture, timbre, the second subject.) It also illustrates compression quaver, and sometimes even silence. This is vital in of harmonic time; seen earlier, this would obscure music, too. Indeed, in the Fifth, it is the start of the the larger rhythmic unit, but now we know enough subject! Such "lessons" must teach us as much Minsky 33 EFTA00284094 about triads and triplets as mathematicians have the bad thoughts. Perhaps the music that some call learned about angles and sides! Think how much background music can tranquilize by turning un- we can learn about minor second intervals from der-thoughts from bad to neutral, leaving the sur- Beethoven's Grosse Fuge in &fiat, Opus 133. face thoughts free of affect by diverting the uncon- scious. The structures we assemble in that detached kind of listening might be wholly solipsistic webs What Use Is Music? of meaninglike cross-references that nowhere touch Why on earth should anyone want to learn such "reality" In such a self-constructed world, we things? Geometry is practical—for building pyra• would need no truth or falsehood, good or evil, pain mids, for instance—but of what use is musical or joy. Music, in this unpleasant view, would serve knowledge? Here is one idea. Each child spends as a fine escape from tiresome thoughts. endless days in curious ways; we call this play. A child stacks and packs all kinds of blocks and boxes, lines them up, and knocks them down. What Syntactic Theories of Music is that all about? Clearly, the child is learning about space! But how on earth does one learn about time? Contrast two answers to the question, Why do we Can one time fit inside another? Can two of them like certain tunes? go side by side? In music, we find out! It is often said that mathematicians are unusually involved Because they have certain structural features. in music, but that musicians are not involved in Because they resemble other tunes we like. mathematics. Perhaps both mathematicians and The first answer has to do with the laws and musicians like to make simple things more compli- rules that make tunes pleasant. In language, we cated, but mathematics may be too constrained to know some laws for sentences; that is, we know satisfy that want entirely, while music can be rig- the forms sentences must have to be syntactically orous or free. The way the mathematics game is acceptable, if not the things they must have to played, most variations lie outside the rules, while make them sensible or even pleasant to the ear. As music can insist on perfect canon or tolerate a ca- to melody, it seems, we only know some features sual accompaniment. So mathematicians might that can help—we know of no absolutely essen- need music, but musicians might not need mathe- tial features. I do not expect much more to come matics. A simpler theory is that since music en- of a search for a compact set of rules for musi- gages us at earlier ages, some mathematicians are cal phrases. (The point is not so much what we those missing mathematical musicians. mean by rule, as how large a body of knowledge is Most adults have some childlike fascination for involved.) making and arranging larger structures out of The second answer has to do with significance smaller ones. One kind of musical understanding outside the tune itself, in the same way that asking, involves building large mental structures out of Which sentences are meaningful? takes us outside smaller, musical parts. Perhaps the drive to build shared linguistic practice and forces us to look those mental music structures is the same one that upon each person's private tangled webs of thought. makes us try to understand the world. (Or perhaps Those private webs feed upon themselves, as in all that drive is just an accidental mutant variant of it; spheres involving preference: we tend to like things evolution often copies needless extra stuff, and that remind us of the other things we like. For ex- minds so new as ours must contain a lot of that.) ample, some of us like music that resembles the Sometimes, though, we use music as a trick to songs, carols, rhymes, and hymns we liked in child- misdirect our understanding of the world. When hood. All this begs this question: If we like new thoughts are painful we have no way to make them tunes that are similar to those we already like, stop. We can attempt to turn our minds to other where does our liking for music start? I will come matters, but doing this (some claim) just submerges back to this later. 34 Computer Music Journal EFTA00284095  The term resemble begs a question also: What such things might mean or represent, we have to are the rules of musical resemblance? I am sure speak of how they are made. that this depends a lot on how melodies are "repre- We cannot describe how the mind is made with- sented" in each individual mind. In each single out having good ways to describe complicated pro- mind, some different "mind parts" do this different cesses. Before computers, no languages were good ways: the same tune seems tat different times) to for that. Piaget tried algebra and Freud tried dia- change its rhythm, mode, or harmony. Beyond that, grams; other psychologists used Markov chains and individuals differ even more. Some listeners squirm matrices, but none came to much. Behaviorists, to symmetries and shapes that others scarcely hear quite properly, had ceased to speak at all. Linguists at all and some fine fugue subjects seem banal to flocked to formal syntax, and made progress for a those who sense only a single line. My guess is that time but reached a limit: transformational grammar our contrapuntal sensors harmonize each fading shows the contents of the registers (so to speak), memory with others that might yet be played; per- but has no way to describe what controls them. haps Bach's mind could do this several ways at This makes it hard to say how surface speech re once. Even one such process might suffice to help lates to underlying designation and intent—a baby- an improviser plan what to try to play next. (To try and-bath-water situation. The reason I like ideas is sufficient since improvisers, like stage magicians, from AI research is that there we tend to seek pro- know enough "vamps" or "ways out" to keep the cedural description first, which seems more appro- music going when bold experiments fail.) priate for mental matters. How is it possible to improvise or comprehend a I do not see why so many theorists find this ap- complex contrapuntal piece? Simple statistical ex- proach disturbing. It is true that the new power de- planations cannot begin to describe such processes. rived from this approach has a price: we can say Much better are the generative and transforma- more, with computational description, but prove tional (e.g., neo-Schenkerian) methods of syntactic less. Yet less is lost than many think, for mathe- analysis, but only for the simplest analytic uses. At matics never could prove much about such com- best, the very aim of syntax-oriented music theo- plicated things. Theorems often tell us complex ries is misdirected because they aspire to describe truths about the simple things, but only rarely tell the sentences that minds produce without attempt- us simple truths about the complex ones. To be- ing to describe how the sentences are produced. lieve otherwise is wishful thinking or "mathemat- Meaning is much more than sentence structure. We ics envy" Many musical problems that resist for- cannot expect to be able to describe the anatomy of mal solutions may turn out to be tractable anyway, the mind unless we understand its embryology. in future simulations that grow artificial musical And so (as with most any other very complicated semantic networks, perhaps by "raising" simulated matter), science must start with surface systems of infants in traditional musical cultures. It will be ex- description. But this surface taxonomy, however citing when one of these infants first shows a hint elegant and comprehensive in itself, must yield of real "talent." in the end to a deeper, causal explanation. To un- derstand how memory and process merge in "lis- tening," we will have to learn to use much more Space and Tune "procedural" descriptions, such as programs that describe how processes proceed. When we enter a room, we seem to see it all at In science, we always first explain things in once; we are not permitted this illusion when lis- terms of what can be observed (earth, water, fire, tening to a symphony. "Of course," one might de- air). Yet things that come from complicated pro- clare, for hearing has to thread a serial path through cesses do not necessarily show their natures on the time, while sight embraces a space all at once. Ac- surface. (The steady pressure of a gas conceals those tually, it takes time to see new scenes, though we countless, abrupt microimpacts.) To speak of what are not usually aware of this. That totally compel- Minsky 35 EFTA00284096  ling sense that we are conscious of seeing every- mind regards itself. When speaking of illusion, we thing in the room instantly and immediately is assume that someone is being fooled. "I know certainly the strangest of our "optical" illusions. those lines are straight," I say, "but they look bent Music, too, immerses us in seemingly stable to me." Who are the different l's and me's? We are worlds! How can this be, when there is so little of it all convinced that somewhere in each person struts present at each moment? I will try to explain this a single, central self; atomic, indivisible. (And se- by (1) arguing that hearing music is like viewing cretly we hope that it is also indestructible.) scenery and (2) by asserting that when we hear good I believe, instead, that inside each mind work music our minds react in very much the same way many different agents. (The idea of societies of they do when we see things.' And make no mis- agents (Minsky 1977; 1980a; 19806) originated in take: I meant to say "good" music! This little the- my work with Seymour Papert.) All we really need ory is not meant to work for any senseless bag of to know about agents is this: each agent knows musical tricks, but only for those certain kinds of what happens to some others, but little of what music that, in their cultural times and places, com- happens to the rest. It means little to say, "Eloise mand attention and approval. was unaware of X" unless we say more about which To see the problem in a slightly different way, of her mind-agents were uninvolved with X. Think- consider cinema. Contrast a novice's clumsy patched ing consists of making mind-agents work together; and pasted reels of film with those that transport us the very core of fruitful thought is breaking prob- to other worlds so artfully composed that our own lems into different kinds of parts and then assign- worlds seem shoddy and malformed. What "hides ing the parts to the agents that handle them best. the seams" to make great films so much less than (Among our most important agents are those that the sum of their parts—so that we do not see them manage these assignments, for they are the agents as mere sequences of scenes? What makes us feel that embody what each person knows about what that we are there and part of it when we are in fact he or she knows. Without these agents we would be immobile in our chairs, helpless to deflect an atom helpless, for we would not know what our knowing of the projected pattern's predetermined destiny? I is for.) will follow this idea a little further, then try to ex- In that division of labor we call seeing, I will sup- plain why good music is both more and less than pose that a certain mind-agent called feature-finder sequences of notes. sends messages (about features it finds on the ret- Our eyes are always flashing sudden flicks of dif- inal to another agent, scene-analyzer. Scene- ferent pictures to our brains, yet none of that sac- analyzer draws conclusions from the messages it cadic action leads to any sense of change or motion gets and sends its own, in turn, to other mind- in the world; each thing reposes calmly in its parts. For instance, feature-finder finds and tells "place"! What makes those objects stay so still about some scraps of edge and texture; then scene- while images jump and jerk so? What makes us analyzer finds and tells that these might fit some such innate Copernicans? I will first propose how bit of shape. this illusion works in vision, then in music. Perhaps those features come from glimpses of a We will find the answer deep within the way the certain real table leg. But knowing such a thing is I. Edward Fredkin suggested to me the theory that listening to not for agents at this level; scene-analyzer does not music might exercise some innate map-making mechanism in know of any such specific things. All it can do is the brain. When I mentioned the puzzle of music's repetitious- broadcast something about shape to hosts of other ness, he compared it to the way rodents explore new places: first agents who specialize in recognizing special they go one way a little, then back to home. They do it again a things. few times, then go a little farther. They try small digressions, (Since special things—lik e tables, words, or dogs— but frequently return to base. Both people and mice explore new must be involved with memory and learning, there territories that way, making mental maps lest they get lost. Mu- is at least one such agent for every kind of thing sk might portray this building process, or even exercise those this mind has learned to recognize.( Thus, we can very parts of the mind. hope, this message reaches table-maker, an agent 38 Computer Music Journal EFTA00284097 specialized to recognize evidence that a table is in ing our eyes or raising our heads provide no cause the field of view. After many such stages, descen- to exorcise that floor inside our current spatial dants of such messages finally reach space-builder, model of the room. My paper on frame-systems an agent that tries to tell of real things in real (Minsky 1974) says more about these concepts. space. Here we only need these few details. Now we can see one reason why perception Now, back to our illusions. While feature-finder seems so effortless: while messages from scene- is not instantaneous, it is very, very fast and a analyzer to table-maker are based on evidence that highly parallel pattern matcher. Whatever scene- feature-finder supplied, the messages themselves analyzer asks, feature-finder answers in an eye need not say what feature-finder itself did, or how it flick, a mere tenth of a second (or less if we have did it. Partly this is because it would take scene- image buffers). More speed comes from the way in analyzer too long to explain all that. In any case, which space-builder can often tell itself, via its own the recipients could make no use of all that infor- high-speed model memory, about what has been mation since they are not engineers or psychol- seen before. I argue that all this speed is another ogists, but just little specialized nerve nets. root of our illusion: if answers seem to come as Only in the past few centuries have painters soon as questions are asked, they will seem to learned enough technique and trickery to simulate have been there all along. reality. (Once so informed, they often now choose The illusion is enhanced in yet another way by different goals.) Thus space-builder, like an ordi- "expectation" or "default." Those agents know good nary person, knows nothing of how vision works, ways to lie and bluff! Aroused by only partial evi- perspective, foveae, or blind spots. We only learn dence that a table is in view, table-maker supplies such things in school: millennia of introspection space-builder with fictitious details about some never led to their suspicion, nor did meditation, "typical table" while its servants find out more transcendental or mundane. The mind holds tightly about the real one! Once so informed, space-builder to its secrets not from stinginess or shame, but can quickly move and plan ahead, taking some simply because it does not know them. risks but ready to make corrections later. This only Messages, in this scheme, go various ways. Each works, of course, when prototypes arc good and are motion of the eye or head or body makes feature- rightly activated—that is what intelligence is all finder start anew, and such motions are responses about. (by muscle-moving agents) to messages that scene- As for "awareness" of how all such things are analyzer sends when it needs more details to re- done, there simply is not room for that. Space- solve ambiguities. Scene-analyzer itself responds to builder is too remote and different to understand messages from "higher up." For instance, space- how feature-finder does its work of eye fixation. builder may have asked, "Is that a table?" of table- Each part of the mind is unaware of almost all that maker, which replies (to itself), "Perhaps, but it happens in the others. (That is why we need psy- should have another leg—there," so it asks scene- chologists; we think we know what happens in our analyzer to verify this, and scene-analyzer gets the minds because those agents are so facile with "de- job done by making eye-mover look down and to faults," but we arc almost always wrong.) True, the left. Nor is scene-understander autonomous: each agent needs to know which of its servants can its questions to scene-analyzer are responses to re- do what, but as to how, that information has no quests from others. There need be no first cause in place or use inside those tiny minds inside our such a network. minds. When we look up, we are never afraid that the How do both music and vision build things ground has disappeared, though it certainly has in our minds? Eye motions show us real objects; "dis-appeared." This is because space-builder re- phrases show us musical objects. We "learn" a members all the answers to its questions and never room with bodily motions; large musical sections changes any of those answers without reason; mov- show us musical "places." Walks and climbs move Minsky 37 EFTA00284098 us from room to room; so do transitions between teach new things, or tear down prior arts. For some musical sections. Looking back in vision is like re- such purposes composers must use the known capitulation in music; both give us time, at certain forms and frames or else expect misunderstanding. points, to reconfirm or change our conceptions of Of course, when expectations are confirmed too the whole. often the style may seem dull; this is our concern Hearing a theme is like seeing a thing in a room, in the next section. Yet, just as in language, one a section or movement is like a room, and a whole often best explains a new idea by using older ones, sonata is like an entire building. I do not mean to avoiding jargon or too much lexical innovation. If say that music builds the sorts of things that space- readers cannot understand the words themselves, builder does. (That is too naive a comparison of the sentences may "be Greek to them." sound and place.) I do mean to say that composers This is not a matter of a simple hierarchy, in stimulate coherency by engaging the same sorts of which each meaning stands on lower-level ones, for interagent coordinations that vision uses to produce example, word, phrase, sentence, paragraph, and its illusion of a stable world using, of course, dif- chapter. Things never really work that way, and jab- ferent agents. I think the same is true of talk or berwocky shows how sense comes through though writing, the way these very paragraphs make sense— many words are new. In every era some contempo- or sense of sense—if any. rary music changes basic elements yet exploits es- tablished larger forms, but innovations that violate too drastically the expectations of the culture can- Composing and Conducting not meet certain kinds of goals. Of course this will not apply to works whose goals include confusion In seeing, we can move our eyes; lookers can and revolt, or when composers try to create things choose where they shall look, and when. In music that hide or expurgate their own intentionality, we must listen here; that is, to the part being but in these instances it may be hard to hold the played now. It is simply no use asking music-finder audience. to look there because it is not then, now. Each musical artist must forecast and predirect If composer and conductor choose what part we the listener's fixations to draw attention here and hear, does not this ruin our analogy? When music- distract it from there—to force the hearer (again, analyzer asks its questions, how can music-finder like a magician) to ask only the questions that the answer them unless, miraculously, the music hap- composition is about to answer. Only by establish- pens to be playing what music-finder wants at just ing such preestablished harmony can music make that very instant? If so, then how can music paint it seem that something is there. its scenes unless composers know exactly what the listeners will ask at every moment? How to en- sure—when music-analyzer wants it now—that Rhythm and Redundancy precisely that "something" will be playing now? That is the secret of music; of writing it, playing, A popular song has 100 measures, 1000 beats. What and conducting! Music need not, of course, confirm must the martians imagine we mean by those mea- each listener's every expectation; each plot de- sures and beats, measures and beats! The words mands some novelty. Whatever the intent, control themselves reveal an awesome repetitiousness. is required or novelty will turn to nonsense. If al- Why isn't music boring? lowed to think too much themselves, the listeners Is hearing so like seeing that we need a hundred will find unanswered questions in any score; about glances to build each musical image? Some repeti- accidents of form and figure, voice and line, tem- tive musical textures might serve to remind us of perament and difference-tone. things that persist through time like wind and Composers can have different goals: to calm and stream. But many sounds occur only once: we must soothe, surprise and shock, tell tales, stage scenes, hear a pin drop now or seek and search for it; that is 38 Computer Music Journal EFTA00284099 why we have no "ear-lids." Poetry drops pins, or Measure-takers notice certain patterns of time- says each thing once or not at all. So does some events like 3/4, 4/4, 6/8. music. Difference-finders observe that the figure Then why do we tolerate music's relentless here is same as that one there, except a per- rhythmic pulse or other repetitive architectural fea- fect fifth above. tures? There is no one answer, for we hear in dif- Structure-builders perceive that three ferent ways, on different scales. Some of those ways phrases form an almost regular portray