Chapter 6 Chapter 4 The Gene Machine

Survival machines originally existed as reservoirs of genes.Their role is passive -- merely a wall of protection for genes against chemical warfare and accidental molecular attacks by their adversaries.In ancient times, the organic molecules present in abundance in the primordial soup were the "foodstuff" they depended on for survival.With the depletion of these organic foods, which have been bred for thousands of years under the powerful influence of sunlight, the once-free life of the survival machines is over.At this point, a large branch of them, now known as plants, began to use sunlight directly to build complex molecules from simple molecules, and to repeat the synthesis process that took place in the primordial soup at a much faster rate.Another branch, now known as animals, "discovered" how to use the fruits of plant chemical labor.Animals either eat plants or other animals.Over time, these two branches of survival machines developed increasingly subtle skills to enhance the efficacy of their way of life.At the same time, new ways of life emerge in endlessly, small branches and small branches are gradually formed, and each small branch is in a special aspect, such as in the ocean, on land, in the sky, underground, on trees, or in other living bodies. , Obtain superhuman survival skills.This process of continual formation of small branches has at last produced the rich variety of flora and fauna which so impresses mankind today.

Both animals and plants have evolved into multicellular bodies, with each cell receiving a complete copy of the full set of genes.When this evolutionary process began, why it happened, and completed in several separate stages, we have no way of knowing.Some people use "colony" as a metaphor for the bodies of animals and plants, and describe them as "colony" of cells.I would rather see the body as a group of genes, and the cells as the working units for the chemical industry of genes to operate. Although we may refer to bodies as groups of genes, each body does acquire its own unique personality in terms of its behaviour.An animal acts as an internally coordinated whole, a unit.I subjectively feel that I am a unit rather than a group.This is to be expected.The selection process favors those genes that cooperate with other genes.In order to compete for rare resources, to devour other survival machines and avoid being eaten by each other, survival machines are engaged in fierce and ruthless competition and struggle.

For all this competition and struggle, it must be far more advantageous to have a centrally coordinated system within a shared body than to have anarchy.Today, the staggered co-evolutionary processes that occur between genes have advanced to the point where the communal nature exhibited by individual survival machines is virtually unrecognizable.In fact, many biologists do not recognize the existence of such clusters, and therefore disagree with me. Fortunately, the disagreement is largely academic as far as the "reliability" (journalists' term) of the arguments presented in later chapters of this book is concerned.It would be tiresome, and unnecessary, to refer repeatedly to genes when we were talking about the behavior of survival machines, just as it would be inconvenient to refer to quantum and elementary particles when we were talking about the performance of automobiles.Indeed, it is generally convenient to refer to the individual as an agent "committed" to increasing the genetic population in future generations.And I will use simple language.Unless otherwise stated, "altruistic behavior" and "selfish behavior" both refer to the behavior of one animal individual towards another animal individual.

This chapter will deal with behavior, the kind of rapid motion widely exploited by the animal branch of survival machines.Animals have become active and aggressive gene delivery vehicles -- gene machines.In the vocabulary of biologists, behavior has a fast quality.Plants move too, but very slowly.Climbing plants look like active animals in fast-motion movies, but most plant activity is really limited to irreversible growth.Animals, on the other hand, have developed various modes of activity that are hundreds of thousands of times faster than plants.Furthermore, the animal's movements are reversible and can be repeated countless times.

The mechanisms that animals develop to perform rapid movements are muscles.Muscle is the engine, which, like a steam engine or an internal combustion engine, uses its stored chemical fuel as energy to generate mechanical motion.The difference: Muscles generate direct mechanical force in the form of tension, rather than air pressure like a steam or internal combustion engine.But muscles are like engines in that they typically exert their power by means of ropes and hinged levers.In the human body, the levers are the bones, the cords are the tendons, and the hinges are the joints.Much is known about how muscles work molecularly, but I find it more interesting to ask how we control when and how fast muscles contract.

Have you ever observed a complex man-made machine?For example, knitting or sewing machines, textile machines, automatic bottling machines or hay balers.These machines utilize a variety of prime movers, such as electric motors or tractors. But how to control the time and speed of these machines in operation is a more complicated problem.Valves open and close sequentially, steel hay balers deftly tie knots and extend knives at just the right moment to cut the string.The timing of many man-made machines is accomplished by means of cams.The invention of the cam was indeed a brilliant achievement.It uses eccentric or profiled wheels to transform simple movements into complex, rhythmic ones.

The principle of automatic playing musical instruments is similar to this.Other instruments, such as the steam organ, or the player piano, used paper scrolls or cards with holes punched in a pattern to produce the tones.In recent years, these simple mechanical timing devices have tended to be replaced by electronic timing devices.Digital computers are an example.They are large, versatile electronic devices that can be used to generate complex timed actions.The main components of modern electronic instruments like computers are semiconductors, and the transistors we are familiar with are a form of semiconductors.

Survival machines seem to have bypassed cams and punch cards.It uses a timing device that has more in common with an electronic computer, although strictly speaking, the basic operation of the two is different.The basic unit of a biological computer is the nerve cell, or neuron.As far as its internal workings are concerned, it is completely different from a transistor.The code that neurons use to communicate with each other is indeed a bit like a computer's pulse code, but a neuron is much more complex as a data processing unit than a crystal transistor.A neuron can communicate with other units through tens of thousands of wires, not just three.Neurons work slower than transistors, but when it comes to miniaturization, transistors are far inferior.Thus, miniaturization is a trend that has dominated the electronics industry for the past two decades.In this regard, the following fact is very telling: there are about ten billion neurons in our brains, and no more than a few hundred transistors can be packed into a single brain.

Plants don't need neurons because they don't have to move to live.But most animal taxa have neurons.They may have "discovered" neurons early in the evolution of animals and then inherited them for all taxa; or they may have been rediscovered several times separately. Fundamentally, a neuron is nothing more than a type of cell.Like any other cell, it has a nucleus and chromosomes. Instead, its cell walls form elongated, thin, thread-like protrusions.Usually a neuron has a particularly long "wire", which we call an axon.An axon is so narrow in width that it can only be seen under a microscope, but it can be several feet long.Some axons are even as long as a giraffe's neck.Axons are usually bundled together in multiple strands to form the multi-core wires we call nerves.These axons run from one part of the body to the other, carrying messages like telephone trunks.Other kinds of neurons have short axons that are only found in the dense nervous tissue we call ganglia.If they are very large neurons, they are also present in the brain.In terms of function, we can think of brains and computers as being similar in that both types of machines emit complex patterns of output signals after analyzing complex patterns of input signals and referring to stored data.

The main way in which the brain actually contributes to the survival machine is by controlling and coordinating the contraction of muscles.To do this, they need wires, called motor nerves, that lead to the individual muscles.But effective preservation of genes is only possible if there is a relationship between the timing of muscle contractions and the timing of external events.The muscles in the upper and lower jaw have to wait until there is something worth chewing in the mouth to contract.Likewise, it only makes sense for the leg muscles to contract in a running pattern when there is something worth running for or something that must be avoided.For this reason, natural selection has favored animals equipped with sensory organs that translate various forms of physical events occurring in the external world into neuronal impulse codes.The brain is connected to the sense organs—eyes, ears, taste buds, etc.—by wires called sensory nerves.How sensory systems work is especially puzzling, since their highly sophisticated skill at recognizing images is far superior to that of the best and most expensive man-made machines.If this is not the case, typists will become redundant, because their work can be completely done by machines that recognize speech or handwriting.Typists will not be out of a job for decades to come.

At some point in the past, the sensory organs may have been connected directly to the muscles in some way, and in fact, today's anemones are not completely out of this state, because such a connection is valid for their way of life. But in order to establish a more complex and indirect connection between the timing of various external events and the timing of muscle contractions, some form of brain is required as an intermediary.A remarkable advance in the course of evolution was the "invention" of memory.With this memory, the timing of muscle contractions is influenced by events not only in the immediate past but also in the distant past.Memory, or storage, is also an essential component of digital computers.Computer memories are more reliable than ours, but they have smaller capacities and are far inferior to ours in information retrieval skills. The behavior of survival machines has one of the most prominent features, which is the obvious purpose.When I say this, I don't just mean that survival machines seem to be deliberate in helping animals survive genetically, even though they are.I am referring to the fact that the behavior of survival machines more closely resembles the purposeful behavior of humans.When we see animals "searching" for food, a mate, or a lost child, we can't help thinking that those animals are feeling some of the same feelings we ourselves experience when we are searching.These feelings may include a "desire" for an object, a "mental image" of the desired object, and a "purpose" that exists in the mind.Each of us knows from experience that at least one of the modern survival machines has undergone a process of evolution in which this purpose has gradually acquired the quality we call "consciousness."I am not well versed in philosophy, so I cannot delve into the implications of this fact.But this is fortunately irrelevant so far as we are concerned with the subject.For it is convenient for us to speak of the operation of a machine as if driven by some purpose, whether or not it is actually conscious.These machines are fundamentally very simple, and the principle of mindless tracking of target states is frequently used in engineering science.A typical example is the Watt steam governor. The basic principle involved is what we call negative feedback, and negative feedback can take many forms.In general, he says, it works like this: this "purpose machine" that operates as if with a conscious purpose is equipped with some kind of measuring device, which can measure the difference between the existing state of things and the "desired" state. difference.The way the machine is structured allows it to go faster as the gap increases.In this way, the machine can automatically reduce the gap - which is why it is called negative feedback - and the machine can automatically stop when the "desired" state is achieved.A pair of balls mounted on a Watt governor are rotated by the impetus of a steam engine.The two balls are respectively installed on the tops of the two movably connected lever arms.As the rotational speed of the ball increases, the centrifugal force gradually counteracts the result of the gravitational force, making the lever arm more and more horizontal.As the lever arm is attached to the valve that supplies the machine with steam, the supply of steam is gradually reduced as the lever arm approaches horizontal.Therefore, if the machine is run too fast, the steam feed will be reduced, and the speed of the machine will be slowed down.Conversely, if the machine is running too slowly, the valve will automatically increase the steam feed, so that the speed of the machine will also increase.However, due to the relationship between overshoot or time lag, this type of destination machine often oscillates.To compensate for this deficiency, engineers always try to add some kind of device to reduce the amplitude of this oscillation. The "desired" state for a Watt governor is a certain rotational speed.Obviously, the machine itself does not consciously require this speed.The so-called "purpose" of a machine is nothing more than its tendency to return to that state.Modern purpose machines extend basic principles such as negative feedback to enable much more complex "lifelike" actions.for example, the missile appears to actively search for a target and pursue it once it is within range, while at the same time it also takes into account the various twists and turns of the target's evasive actions, and sometimes even "estimates" these actions Or "preemptive strike".These details are not discussed here.In short, they involve various kinds of negative feedback, "feed-forward," and other principles familiar to engineers.As far as we know, these principles are now widely used in the movement of living bodies.We need not necessarily think of a missile as anything approaching consciousness, but to an ordinary person its apparently deliberate, purposeful movements are hard to believe that the missile was not piloted by a pilot. directly controlled. A common misconception is that if a machine such as a missile is designed and built by a conscious being, then it must be under the direct control of a conscious being.Another variant of this misunderstanding is: "Computers can't really play chess because they can only listen to the people who operate the computer".We must understand the source of this misunderstanding as it affects our understanding of what it means to say that genes "control" behavior.Computer chess is a very telling example, so I want to touch on it briefly. Computer chess today is not at the level of a grand master, but it is comparable to a good amateur player.It is more accurate to say that a computer program is as good as a good amateur chess player, because the program itself is never demanding on which specific computer is used to perform its skills.So, what is the task of the programmer?First, he certainly isn't manipulating the computer 24/7 like a puppeteer.This is cheating.He programmed it, put it in the computer, and the computer operated on its own: no human intervention.In addition to letting the opponent press his move into the machine.Did the programmer anticipate all possible moves and thus compile a long list of clever moves for each situation?Of course not.Because in a chess game, there are as many possible moves as there are sands in the Ganges River, and even at the end of the world, it is impossible to compile a complete list.Also for the same reason, it is impossible for us to compile such a program for the computer, so that it can take all possible chess moves and all possible responses in the "computer" in advance, so as to seek a strategy to defeat the enemy.There are more different chess games than atoms in the galaxy.These are merely trivial problems that illustrate the problem of programming a chess-playing computer, which is in fact an extremely difficult problem to solve.It should come as no surprise that even the most well-thought-out programs are no match for chess grandmasters. The role of a programmer is in fact similar to that of a father who teaches his son how to play chess.Instead of telling it the various moves that apply to every opening, he tells the computer the main outlines of moves.He didn't say literally, "The elephant goes diagonally" in the language we use every day, but in the language of mathematics, "The new coordinates of the bishop come from the old coordinates, and the program is at the old coordinates x and the old coordinates y." plus the same constant, but not necessarily the same sign." The language actually used is of course more concise.Then he can program some "advice" into the program, using the same mathematical or logical language, which, if expressed in our everyday language, is nothing more than "don't expose your king to the enemy", Or some practical tricks, such as "dual use" of one knight, attacking the opponent's two sons at the same time.These specific chess moves are intriguing, but it would be too far from the topic to talk about them.The important thing is that after the computer has made the first move, it needs to operate independently, and cannot expect any further instructions from its master.All the programmer can do is deploy the computer to the best of his ability in advance, striking the right balance between the provision of specific knowledge and hints of strategy and tactics. Genes also control the behavior of their survival machines, but not directly, like fingers on puppets, but through indirect pathways, like computer programmers.What the genes can do is limited to the deployment in advance, and they can only stand by when the survival machines operate independently later.Why are genes so devoid of initiative?Why don't they hold the reins tightly and direct the behavior of survival machines at any time?This is because of the difficulty caused by time lag.There is a science fiction novel that illustrates this problem very cleverly by means of analogy.This gripping novel is A in Andromeda by Fred Hoyle and John Elliot.Like any good sci-fi, it has some interesting scientific arguments to back it up.Yet, strangely enough, the novel seems to deliberately avoid discussing one of its most important scientific points, leaving it to the reader to imagine.If I tell it all here, I think the two authors will not be offended. There is a civilized world in the Andromeda constellation two hundred light-years away from us.People there want to spread their culture to some distant worlds.What is the best way to do it?It is impossible to send someone directly to go once.In the universe, your maximum speed from one place to another cannot theoretically exceed the upper limit of the speed of light. What's more, due to the limitation of mechanical power, the maximum speed is much lower than the speed of light. In addition, in the universe, there may not be so many worlds worth visiting. Do you know which direction to go in to make this trip worthwhile?Radio waves are an ideal means of communicating with the rest of the universe because, if you have enough power to broadcast your radio signal in all directions instead of beaming it in one direction, there are a very large number of worlds that can receive your radio waves (the number of Proportional to the square of the distance traveled by the wave).Radio waves travel at the speed of light, which means that it takes two hundred years for a signal from Andromeda to reach Earth.Such a long distance makes it impossible to communicate between the two places.Even if every message sent from the earth was transmitted by twelve generations of people, it would be a waste of money and money to try to communicate with people so far away. This is a practical problem that we will soon face.Between the Earth and Mars, radio waves take about four minutes to travel.There is no doubt that astronauts must change their conversation habits in the future. They can no longer speak like you and me, but must use long monologues and talk to themselves.This type of call is not so much a conversation as a communication.As another example, Roger Payne points out that the acoustics of the ocean have certain peculiar properties which mean that the unusually loud "song" of humpback whales could theoretically be heard anywhere in the world if they It is swimming at a certain depth in the sea.Whether humpback whales actually talk to each other over long distances is unknown.If anything, they are in the same predicament as the astronauts on Mars.According to the speed of sound propagation in the water, it takes about two hours for the humpback whale's song to reach the shore of the Atlantic Ocean and then wait for the other party's song to come back.This, it seems to me, is the reason why a humpback whale's solo usually lasts eight minutes without repetition, and then repeats itself many times, each cycle lasting about eight minutes. The Andromedans in the novel do the same.They knew that there was no practical point in waiting for an echo from the other party, so they put together what they wanted to say, compiled it into a complete long message, and then broadcast it to space, each time lasting for several months, and then repeated it continuously.Their message, however, is very different from that of the whales.The Andromedan message was written in electrical code, instructing others how to build and program a giant computer.The telegram was, of course, not in human language.But to skilled cryptographers, almost any cipher can be broken; especially if the cipher's designers intended it to be easy to break.The telegram was first intercepted by Jodrell Bank's radio telescope, and the telegram was eventually translated.According to the instructions, the computer was finally built and its program was put into practice.The result was almost catastrophic for humanity, for the Andromedans did not have altruistic intentions toward everyone.This computer has pretty much brought the entire world under its dictatorship.In the end, the protagonist smashed the computer with a sharp ax at the critical moment. The interesting question, from our point of view, is in what sense we can say that the Andromedans were manipulating affairs on Earth.They have no direct control over what they do to the computer at any time.In fact, they didn't even know that the computer had been built, because it took two hundred years for the information to reach them.The computer makes decisions and takes actions completely independently.It can no longer even ask its master general strategic questions.Since the 200-year barrier is insurmountable, all instructions must be incorporated into the procedure in advance.In principle, this is roughly the same program as would be required for a computer to play chess, but with greater flexibility and adaptability to local conditions.This is because such a program would have to be specific not only to conditions on Earth, but also to worlds of all kinds with advanced technologies whose specific conditions the Andromedans had no idea. Just as the Andromedans had to have a computer on Earth to make their day-to-day decisions, our genes had to build a brain.But genes aren't just Andromedans giving out coded instructions, they're instructions themselves, and the reason they can't direct us puppets is the same: time lag.Genes function by controlling the synthesis of proteins.This is a powerful means of manipulating the world, but it will take time to bear fruit.Growing an embryo takes months of patiently manipulating proteins.On the other hand, the most important thing about behavior is the quickness of it.The unit of time used to measure behavior is not months but seconds or fractions of a second.Something happens in the outside world: an owl swoops overhead, the rustling grass exposes the prey, and in a split second the nervous system jerks, the muscles spring; and the prey narrowly escapes—or becomes. victim.Genes don't have reaction times like this.Like the Andromedans, the genes had to do what they could to pre-deploy everything, building themselves a fast executive computer.Make it grasp the laws of as many situations as possible that genes can "anticipate", and put forward "advice" for this.But life is as unpredictable as a chess game, and it is unrealistic to foresee everything in advance.Like the programmer of a chess game, the "instructions" of genes to a survival machine cannot be specific and subtle, but can only be general strategies and various tricks applicable to survival. As Young pointed out, genes must perform tasks like making predictions about the future.While the embryonic survival machine is being built, the dangers and problems it may encounter in its later life are unknown.Who can predict what carnivore will crouch in which bush and wait for an opportunity to attack it, or what fast-legged snack will suddenly appear in front of it and meander by?Humans can't predict these problems, and genes can't do anything about them.But certain generalities are foreseeable.Polar bear genes can safely know in advance that their unborn survival machines will have a frigid environment.This prediction is not the result of genetic thinking.They never think: they just prepare their coats in advance, as they have always done in some previous bodies.That's why they still exist in the gene pool.They also foresee that the earth will be covered with snow, and this foresight is reflected in the color of their fur.The gene makes the fur white, thus achieving camouflage.If the climate in the Arctic changes so drastically that polar bear cubs find themselves born in tropical deserts, the genetic predictions will be wrong.They will pay for it.The cubs die, and with them the genes in their bodies. In a complex world, predicting the future is risky.Every decision a survival machine makes is a gamble, and it is the responsibility of genes to program the brain in advance so that the decisions it makes will more than likely lead to positive outcomes.In the casino of evolution, the chips at play are survival, strictly speaking, the survival of genes.But in general, and as a reasonable approximation, it can also be said to be the survival of the individual.If you go down to the waterhole to drink, the risk of being eaten by the predators waiting at the waterhole increases.If you don't go, you will eventually die of thirst.Whether you go or not, risks exist.You have to make decisions that give your genes the best chance of survival.Perhaps the best thing to do is to hold back until you absolutely have to and go down for a quick drink so that you don't need to drink water for a long time.This way, you reduce the number of trips to the watering hole, but when you finally have to drink, you have to put your head down and drink for a long time.Another risky way is to drink less and run more, that is, run to have a drink or two, and then run back immediately, so that a few more runs can also solve the problem.Which risk-taking strategy is the best depends on various complex circumstances, among which the hunting habits of carnivores are also an important factor.In order to achieve maximum effect, carnivores are also constantly improving their hunting habits.Therefore, some form of trade-off between the pros and cons of the various possibilities is necessary.But we certainly don't necessarily think of these animals as consciously weighing pros and cons.We just have to believe that if the genes of those animals built the sharp brains that make them more likely to win the bets; then, as a direct consequence, the animals are more likely to survive and the genes reproduce. We can take the betting metaphor a little further.A gambler must consider three main quantities: stake, chance, and winnings.Gamblers are willing to place big bets if the winnings are huge.An all-or-nothing gambler has the opportunity to make a lot of money.He could lose it all, of course, but on average, a person who bets big has little or no advantage over someone who bets small for small wins.There are also similarities between speculators who buy short and sell short on the exchanges and investors who play it safe.In some respects, the metaphor of a stock exchange is more apt than a casino, where wins and losses are rigged and the house always wins in the end (technically, this means that those who bet big lose more than those who bet small). more, and those who play small bets are poorer than those who do not bet. But in a sense the case of not betting is inappropriate for the present topic).That aside, there seem to be reasons for both big and small bets.Are there any animals that make big bets, or are there animals that are more conservative?As we shall see in Chapter 9, one can often think of male animals as high-stakes, high-risk gamblers, and females as steady investors, especially when males are mates with each other. Among species in which males and females compete.The naturalist who reads this book can think of some species that might be called high-stakes, high-risk, and others that are more conservative.This is where I turn to the more general topic of how genes make predictions about the future. In some unpredictable environments, how genes can predict the future is a difficult problem. One way to solve this problem is to give survival machines a learning ability in advance.To do this, genes can be programmed in the form of instructions to their survival machinery in the form of: "The following will bring benefits: sweet taste in the mouth, arousal, moderate temperature, smiling children, etc. And the following will bring Unhappiness: various pains, nausea, empty bellies, crying children, etc. If you happen to do something and then it happens, don't do it again; on the other hand, do it repeatedly Anything that benefits you." A program written in this way has the advantage of greatly reducing the number of elaborate rules that had to be incorporated into the original program, while allowing it to cope with changes in circumstances whose details were not foreseen in advance.On the other hand, it is still necessary to make some predictions.In our example, a sweet taste in the mouth and an arousal are "good" in the sense that the genes figured that eating sugar and mating might be good for the genes' survival.But according to this example, they cannot foresee that saccharine and self-abuse may also bring them satisfaction.Nor can they foresee the dangers of sugar overeating in our somewhat unnaturally plentiful sugar environment. Learning strategies has been used in some programs for computers playing chess.These programs do improve over time as computers play against humans or against other computers.Although they have a library of rules and tactics, they also have a small random tendency pre-programmed into their decisions.They record past decisions, and each time they win a game, they slightly increase the weight of the tactic that led to victory in this game, so that the computer is more likely to use the same tactic again next time. One of the most interesting ways to predict the future is through simulations.A general who wants to know whether a certain military plan is superior to other alternatives faces the problem of making predictions.The weather, troop morale, and possible enemy countermeasures are all unknowns.One way to know if the plan is feasible is to try it out and see how it works.However, it is not advisable to try every imaginable plan, because after all, the number of young people who are willing to devote themselves "for the motherland" is limited, and there are so many possible plans.Exercises against imaginary enemies can also test the practicality of various plans, which is better than doing it with real swords and guns.The exercise can take the form of an all-out battle between the "North Country" and the "South Country", using empty shells.But even that takes a lot of time and material.A less expensive way is to use toy soldiers and tanks to move around on the large map for maneuvers. In recent years, computers have taken over most of the simulation functions, not only in military strategy, but in all fields such as economics, ecology, sociology, etc., where predictions about the future must be made.It uses the technique of building a model of something in the world inside a computer.That doesn't mean that, if you lift the computer's lid off, you can see a miniature imitation of the same thing as the simulated object.In a chess-playing computer, there is no "image" in the memory that is recognizable as a chessboard with knights and pawns in their positions.有的只是代表棋盘以及各种棋子位置的一行行的电子编码。对我们来说，地图是世界某一部分的平面缩影。在计算机里面，地图通常是以一系列城镇和其他地点的名字来代表的。每个地点附有两个数字--它的经度和纬度。计算机的电脑实际上如何容纳它这个世界的模型是无关紧要的。重要的是容纳的形式允许它操纵这个模型，进行操作和试验，并以计算机操作员能够理解的语言汇报运算的结果。通过模拟技术，以模型进行的战役可以得出胜负，模拟的班机可以飞行或坠毁，经济政策可以带来繁荣或崩溃。无论模拟什么，计算机的整个运算过程只需实际生活中极小的一部分时间。当然，这些反映世界的模型也有好坏之分，而且即使是上好的模型也只能是近似的。不管模拟得如何逼真也不能预测到将要发生的全部实际情况，但好的模拟肯定远胜于盲目的试验和误差。我们本来可以把模拟称为代替性的"试验和误差"，不幸的是，这个术语早为研究老鼠心理的心理学家所优先占用了。 如果模拟是这样一个好办法，我们可以设想生存机器本该是首先发现这个办法的，早在地球上出现人类以前，生存机器毕竟已经发明了人类工程学的许多其他方面的技术：聚焦透镜和抛物面反射镜、声波的频谱分析、伺服控制系统、声纳、输入信息的缓冲存储器以及其他不胜枚举的东西，它们都有长长的名字，其具体细节这里不必细说。模拟到底是怎么一回事呢？我说，如果你自己要作出一个困难的决定，而这个决定牵涉到一些将来的未知量，你也会进行某种形式的模拟。你设想在你采取各种可供选择的步骤之后将会出现的情况。你在脑子里树立一个模型，这个模型并不是世上万物的缩影，它仅仅反映出依你看来是有关的范围有限的一组实体。你可以在心目中看到这些事物的生动形象，或者你可以看到并操纵它们已经概念化了的形象。无论怎样，不会在你的脑子里出现一个实际上占据空间的、反映你设想的事物的模型。但和计算机一样，你的脑子怎样表现这个模型的细节并不太重要，重要的是你的脑子可以利用这个模型来预测可能发生的事物。那些能够模拟未来事物的生存机器，比只会在明显的试验和误差的基础上积累经验的生存机器要棋高一着。问题是明显的试验既费时又费精力，明显的误差常常带来致命的后果。模拟则既安全又迅速。 模拟能力的演化似乎终于导致了主观意识的产生。其所以如此，在我看来，是当代生物学所面临的最不可思议的奥秘。没有理由认为电子计算机在模拟时是具有意识的，尽管我们必须承认，有朝一日它们可能具有意识。意识之产生也许是由于脑子对世界事物的模拟已达到如此完美无缺的程度，以致把它自己的模型也包括在内。显然，一个生存机器的肢体必然是构成它所模拟的世界的一个重要部分；可以假定，为了同样理由，模拟本身也可以视为是被模拟的世界的一个组成部分。事实上，"自我意识"可能是另外一种说法，但我总觉得这种说法用以解释意识的演化是不能十分令人满意的，部分原因是它牵涉到一个无穷尽的复归问题--如果一个模型可以有一个模型，那么为什么一个模型的模型不可以有一个模型呢……？不管意识引起了哪些哲学问题，就本书的论题而言，我们可以把意识视为一个进化趋向的终点，也就是说，生存机器最终从主宰它们的主人即基因那里解放出来，变成有执行能力的决策者。脑子不仅负责管理生存机器的日常事务，它也取得了预测未来并作出相应安排的能力。它甚至有能力拒不服从基因的命令，例如拒绝生育它们的生育能力所容许的全部后代。但就这一点而言，人类的情况是非常特殊的，我们在下面将谈到这个问题。 这一切和利他行为和自私行为有什么关系呢？我力图阐明的观点是，动物的行为，不管是利他的或自私的，都在基因控制之下。这种控制尽管只是间接的，但仍然是十分强有力的。基因通过支配生存机器和它们的神经系统的建造方式而对行为施加其最终的影响。但此后怎么办，则由神经系统随时作出决定。基因是主要的策略制定者；脑子则是执行者。但随着脑子的日趋高度发达，它实际上接管了越来越多的决策机能，而在这样做的过程中运用诸如学习和模拟的技巧。这个趋势在逻辑上的必然结果将会是，基因给予生存机器一个全面的策略性指示：请采取任何你认为是最适当的行动以保证我们的存在。但迄今为止还没有一个物种达到这样的水平。 和计算机类比以及和人类如何作出决定进行类比确实很有意思。但我们必须回到现实中来，而且要记注，事实上进化是一步一步通过基因库内基因的差别性生存来实现的。因此，为使某种行为模式--利他的或自私的--能够演化，基因库内"操纵"那种行为的基因必须比"操纵"另外某种行为的、与之匹敌的基因或等位基因有更大的存活可能性。一个操纵利他行为的基因，指的是对神经系统的发展施加影响，使之有可能表现出利他行为的任何基因。我们有没有通过实验取得的证据，表明利他行为是可遗传的呢？No.但这也是不足为奇的，因为到目前为止，很少有人对任何行为进行遗传学方面的研究。还是让我告诉你们一个研究行为模式的实例吧！这个模式碰巧并不带有明显的利他性，但它相当复杂，足以引起人们的兴趣。这是一个说明如何继承利他行为的典型例子。 蜜蜂有一种叫腐臭病（foul brood）的传染病。这种传染病侵袭巢室内的幼虫。 养蜂人驯养的品种中有些品种比其他的品种更易于感染这种病，而且至少在某些情况下各品系之间的差异证明是由于它们行为上的不同。有些俗称卫生品系的蜜蜂能够找到受感染的幼虫，把它们从巢室里拉出来并丢出蜂房，从而迅速地扑灭流行病。那些易感染的品系之所以易于染病正是因为它们没有这种杀害病婴的卫生习惯。实际上这种卫生行为是相当复杂的。工蜂必须找到每一患病幼虫所居住的巢室，把上面的蜡盖揭开，拉出幼虫，把它拖出蜂房门，并弃之于垃圾堆上。 由于各种理由，用蜜蜂做遗传学实验可以说是一件相当复杂的事情。工蜂自己一般不繁殖，因此你必须以一个品系的蜂后和另外一个品系的雄蜂杂交，然后观察养育出来的子代工蜂的行为。罗森比勒（WCRothenbunler）所作的实验就是这样进行的。他发现第一代子代杂交种的所有蜂群都是不卫生的：它们亲代的卫生行为似乎已经消失，尽管事实上卫生的基因仍然存在，但这些基因已变成隐性基因了，象人类的遗传蓝眼睛的基因一样。罗森比勒后来以第一代的杂交种和纯粹的卫生品系进行"回交"（当然也是用蜂后和雄蜂），这一次他得到了绝妙的结果。子代蜂群分成三类：第一类表现出彻底的卫生行为，第二类完全没有卫生行为，而第三类则是折衷的。这一类蜜蜂能够找到染病的幼虫，揭开它们的蜡蜂巢的盖子，但只到此为止，它们并不扔掉幼虫。据罗森比勒的猜测，可能存在两种基因，一种是进行揭盖的，另一种是扔幼虫的。正常的卫生品系两者兼备，而易受感染的品系则具有这两种基因的等位基因--它们的竞争对手。那些在卫生行为方面表现为折衷的杂交种，大概仅仅具有揭盖的基因（其数量是原来的两倍）而不具有扔幼虫的基因。罗森比勒推断，他在实验中所培育出来的，显然完全是不卫生的蜂群里可能隐藏着一个具有扔幼虫的基因的亚群，只是由于缺乏揭盖子基因而无能为力罢了。他以非常巧妙的方式证实了他的推断：他自己动手把蜂巢的盖子揭开。果然，蜡盖揭开之后，那些看起来是不卫生的蜜蜂中有一半马上表现出完全正常的把幼虫扔掉的行为。 这段描述说明了前面一章提到的若干重要论点。它表明，即使我们对把基因和行为连接起来的各种胚胎因素中的化学连接一无所知，我们照样可以恰如其分地说"操纵某种行为的基因"。事实上，这一系列化学连接可以证明甚至包括学习过程。例如，揭蜡盖基因之所以能发挥作用，可能是因为它首先让蜜蜂尝到受感染的蜂蜡的味道。就是说，蜂群会发觉把遮盖病仔的蜡盖吃掉是有好处的，因此往往一遍又一遍地这样做。即使基因果真是这样发挥作用的，只要具有这种基因的蜜蜂在其他条件不变的情况下终于进行揭盖活动，而不具有这种基因的蜜蜂不这样做，那么，我们还是可以把这种基因称为"揭盖子"的基因。 第二，这段描述也说明了一个事实，就是基因在对它们共有的生存机器施加影响时是"合作的"。扔幼虫的基因如果没有揭盖基因的配合是无能为力的，反之亦然。不过遗传学的实验同样清楚地表明，在贯串世世代代的旅程中，这两种基因基本上是相互独立的。就它们的有益工作而言，你尽可以把它们视为一个单一的合作单位；但作为复制基因，它们是两个自由的、独立的行为者。 为了进行论证，我们有必要设想一下"操纵"各种不大可能的行为的基因。譬如我说有一种假设的"操纵向溺水的同伴伸出援手的行为"的基因，而你却认为这是一种荒诞的概念，那就请你回忆一下上面提到的卫生蜜蜂的情况吧。要记住，在援救溺水者所涉及的动作中，如一切复杂的肌肉收缩，感觉整合，甚至有意识的决定等等，我们并不认为基因是唯一的一个前提因素。关于学习、经验以及环境影响等等是否与行为的形成有关这个问题我们没有表示意见。你只要承认这一点就行了：在其他条件不变的情况下，同时在许多其他的主要基因在场，以及各种环境因素发挥作用的情况下，一个基因，凭其本身的力量比它的等位基因有更大的可能促使一个个体援救溺水者。这两种基因的差别归根结底可能只是某种数量变数的差异。有关胚胎发育过程的一些细节尽管饶有风趣，但它们与进化的种种因素无关。洛伦茨明确地阐明了这一点。 基因是优秀的程序编制者，它们为本身的存在而编制程序。生活为它们的生存机器带来种种艰难险阻，在对付这一切艰难险阻时这个程序能够取得多大的成功就是判定这些基因优劣的根据。这种判断是冷酷无情的，关系到基因的生死存亡。 下面我们将要谈到以表面的利他行为促进基因生存的方式。但生存机器最感关切的显然是个体的生存和繁殖，为生存机器作出各种决定的脑子也是如此。属同一"群体"的所有基因都会同意将生存和繁殖放在首位。因此各种动物总是竭尽全力去寻找并捕获食物，设法避免自己被抓住或吃掉；避免罹病或遭受意外；在不利的天气条件下保护自己；寻找异性伴侣并说服它们同意交配；并以一些和它们享受的相似的优越条件赋予它们的后代。我不打算举出很多例子--如果你需要一个例证，那就请你下次仔细观察一下你看到的野兽吧。但我却很想在这里提一下一种特殊的行为，因为我们在下面谈到利他行为与自私行为时必须再次涉及这种行为。我们可以把这种行为概括地称为联络（communication）。 我们可以这样说，一个生存机器对另一个生存机器的行为或其神经系统的状态施加影响的时候，前者就是在和后者进行联络。这并不是一个我打算坚持为之辩护的定义，但对我们目前正在探讨的一些问题来说，这个定义是能够说明问题的。 我所讲的影响是指直接的、偶然的影响。联络的例子很多：鸟、蛙和蟋蟀的鸣唱；狗的摇动尾巴和竖起长颈毛；黑猩猩的"露齿而笑"；人类的手势和语言等。许许多多生存机器的行动，通过影响其他生存机器的行为的间接途径，来促进其自身基因的利益。各种动物千方百计地使这种联络方式取得成效。鸟儿的鸣唱使人们世世代代感到陶醉和迷惘。我上面讲过的弓背鲸的歌声表达出更其高超的意境，同时也更迷人。它的音量宏大无比，可以传到极其遥远的地方，音域广阔，从人类听觉能够听到的亚音速的低沉的隆隆声直到超音速的、短促的刺耳声。蝼蛄之所以能发出宏亮的歌声，这是因为它们在泥土中精心挖成双指数角状扩音器一样的土穴，在里面歌唱，唱出的歌声自然得到扩大。在黑暗中翩翩起舞的蜂群能够为其他觅食的蜂群准确地指出前进的方向以及食物在多远的地方可以找到。这种巧妙的联络方法只有人类的语言可以与之比美。 动物行为学家的传统说法是，联络信号之逐步完善对发出信号者和接收信号者都有益。譬如说，雏鸡在迷途或受冻时发出的尖叫声可以影响母鸡的行为。母鸡听到这种吱吱啁啁的叫声后通常会应声而来，把小鸡领回鸡群。我们可以说，这种行为的形成是由于它为双方都带来好处；自然选择有利于迷途后会吱吱啁啁叫的雏鸡，也有利于听到这种叫声后随即作出适当反应的母鸡。 如果我们愿意的话（其实无此必要），我们可以认为雏鸡叫声之类的信号具有某种意义或传达某种信息。在这个例子里，这种呼唤声相当于"我迷路了！"我在第一章中提到的小鸟发出的报警声传递了"老鹰来了！"这一信息。那些收到这种信息并随即作出反应的动物无疑会得到好处。因此，这个信息可以说是真实的。可是动物会发出假的信息吗？它们会扯谎吗？说动物说谎这种概念可能会令人发生误解，因此我必须设法防止这种误解的产生。我记得出席过一次比阿特丽斯（Beatrice）和加德纳（Alan Gardner）主讲的一次讲座，内容是关于他们所训练的遐迩闻名的"会说话的"黑猩猩华舒（她以美国手势语表达思想。对学习语言的学者来说，她的成就可能引起广泛的兴趣）。听众中有一些哲学家，在讲座结束后举行的讨论会上，对于华舒是否会说谎这个问题他们费了一番脑筋。我猜想，加德纳夫妇一定有些纳闷，为什么不谈谈其他更有趣的问题呢？I'm feeling it too.在本书中，我所使用的"欺骗"、"说谎"等字眼只有直截了当的含义，远不如哲学家们使用的那么复杂。他们感兴趣的是有意识的欺骗。而我讲的仅仅是在功能效果上相当于欺骗的行为。如果一只小鸟在没有老鹰出现的情况下使用"鹰来了"这个信号，从而把它的同伴都吓跑，让它有机会留下来把食物全都吃掉，我们可以说它是说了谎的。我们并不是说它有意识地去欺骗。我们所指的只不过是，说谎者在牺牲其同伴的利益的情况下取得食物。其他的小鸟之所以飞走，这是因为它们在听到说谎者报警时作出在真的有鹰出现的情况下那种正常反应而已。 许多可供食用的昆虫，如前一章提到的蝴蝶，为了保护自己而模拟其他味道恶劣的或带刺的昆虫的外貌。我们自己也经常受骗，以为有黄黑条纹相间的食蚜蝇就是胡蜂。有些苍蝇在模拟蜜蜂时更是惟妙惟肖，肉食动物也会说谎。琵琶鱼在海底耐着性子等待，将自己隐蔽在周围环境中，唯一触目的部分是一块象虫一样蠕动着的肌肉，它挂在鱼头上突出的一条长长的"钓鱼竿"末端。小鱼游近时，琵琶鱼会在小鱼面前抖动它那象虫一样的钩饵，把小鱼引到自己的隐而不见的咀巴旁。大咀突然张开，小鱼被囫囵吞下。琵琶鱼也在说谎。它利用小鱼喜欢游近象虫一样蠕动着的东西这种习性。它在说，"这里有虫"，任何"受骗上当"的小鱼都难逃被吞掉的命运。 有些生存机器会利用其他生存机器的性欲。蜂兰（bee orchid）会引诱蜜蜂去和它的花朵交配，因为这种兰花活象雌蜂。兰花必须从这种欺骗行为中得到的好处是传播花粉，因为一只分别受到两朵兰花之骗的蜜蜂必然会把其中一朵兰花的花粉带给另外一朵。萤火虫（实际上是甲虫）向配偶发出闪光来吸引它们。每一物种都有其独特的莫尔斯电码一样的闪光方式，这样，不同物种之间不会发生混淆不清的现象，从而避免有害的杂交。正象海员期待发现某些灯塔发出的独特的闪光模式一样，萤火虫会寻找同一物种发出的密码闪光模式。Photuris属的萤火虫雌虫"发现"如果它们模拟Photinus属的萤火虫雌虫的闪光密码，它们就能把Photinus属的萤火虫雄虫引入壳中。 Photuris属的雌虫就这样做了。当一只Photinus属的雄虫受骗接近时，雌虫就不客气地把它吃掉。说到这里，我们自然会想起与此相似的有关塞王（Siren）和洛勒莱（Lorelei）的故事，但英国西南部的康瓦耳（Cornwall）人却会追忆昔日那些为行劫而使船只失事的歹徒，他们用灯笼诱船触礁，然后劫掠从沉船中散落出来的货物。 每当一个联络系统逐渐形成时，这样的风险总会出现：即某些生物利用这个系统来为自己谋私利。由于我们一直受到"物种利益"这个进化观点的影响，因此我们自然首先认为说谎者和欺骗者是属于不同的物种的：捕食的动物，被捕食的动物，寄生虫等等。然而，每当不同个体的基因之间发生利害冲突时，不可避免地会出现说谎、欺骗等行为以及联络手段用于自私的目的的情况。这包括属于同一物种的不同个体。我们将会看到，甚至子女也要欺骗父母，丈夫也要欺骗妻子，兄弟俩也要相互欺骗。 有些人相信，动物的联络信号原来是为了促进相互的利益而发展的，只是后来为坏分子所利用。这种想法毕竟是过于天真。实际的情况很可能是：从一开始，一切的动物联络行为就合有某种欺诈的成分，因为所有的动物在相互交往时至少要牵涉到某种利害冲突。我打算在下面一章介绍一个强有力的观点，这个观点是从进化的角度来看待各种利害冲突的。