The Beginning of Infinity

by David Deutsch

  Such activities may seem to depend on explanation – on understanding how and why each action within the complex behaviour has to fit in with the other actions in order to achieve the overall purpose. But recent discoveries have revealed how apes are able to imitate such behaviours without ever creating any explanatory knowledge. In a remarkable series of observational and theoretical studies, the evolutionary psychologist and animal-behaviour researcher Richard Byrne has shown how they achieve this by a process that he calls behaviour parsing (which is analogous to the grammatical analysis or ‘parsing’ of human speech or computer programs).

  Humans and computers separate continuous streams of sounds or characters into individual elements such as words, and then interpret those elements as being connected by the logic of a larger sentence or program. Similarly, in behaviour parsing (which evolved millions of years before human language parsing), an ape parses a continuous stream of behaviour that it witnesses into individual elements, each of which it already knows – genetically – how to imitate. The individual elements can be inborn behaviours, such as biting; or behaviours learned by trial and error, such as grasping a nettle without being stung; or previously learned memes. As for connecting these elements together in the right way without knowing why, it turns out that, in every known case of complex behaviours in non-humans, the necessary information can be obtained merely by watching the behaviour many times and looking out for simple statistical patterns – such as which right-hand behaviour often goes with which left-hand behaviour, and which elements are often omitted. It is a very inefficient method, requiring a lot of watching of behaviours that a human could mimic almost immediately by understanding their purpose. Also, it allows only a few fixed options for connecting the behaviours together, so only relatively simple memes can be replicated. Apes can copy certain individual actions instantly – the ones of which they have pre-existing knowledge through their mirror-neuron system – but it takes them years to learn a repertoire of memes that involve combinations of actions. Yet those memes – trivially simple tricks by human standards – are enormously valuable: using them, apes have privileged access to sources of food that are closed to all other animals; and meme evolution gives them the ability to switch to other sources far faster than gene evolution would allow.

  So, an ape knows (inexplicitly) that another ape is ‘picking up a rock’, and not performing any of the countless other possible interpretations of the same actions, such as ‘picking up an object in a given relative position’, because picking up a rock is in its inborn repertoire of copiable behaviours while the other possibilities are not. Indeed, it may well be that apes cannot imitate the behaviour of ‘picking up an object in a given relative position’. Note, in this connection, that apes are unable to imitate sounds. They cannot even parrot sounds (repeat them blindly), despite having a complex inborn repertoire of calls that they can make, recognize and act upon in genetically predetermined ways. Their behaviour-parsing system simply did not evolve a predetermined translation mechanism from hearing sounds to uttering them, so they cannot ape them. Consequently there are no customized sounds in any of the apes’ memetically controlled behaviours.

  Thus, in the crucial respect that is relevant to meme replication, aping has the same logic as parroting: like the parrot, the ape avoids the infinite ambiguity of copying by already knowing, inexplicitly, the meaning of every action that it is capable of copying. And it is only capable of associating one meaning with each action that it can copy – one definition of how to perform the ‘same’ action under various circumstances. That is how ape memes can be replicated without the impossible step of literally copying knowledge from another ape. The recipient of the meme instantly recognizes the meaning of each element of the behaviour; and it relates the elements by statistical analysis, not by discovering how they support each other’s functioning.

  Human beings acquiring human memes are doing something profoundly different. When an audience is watching a lecture, or a child is learning language, their problem is almost the opposite of that of parroting or aping: the meaning of the behaviour that they are observing is precisely what they are striving to discover and do not know in advance. The actions themselves, and even the logic of how they are connected, are largely secondary and are often entirely forgotten afterwards. For example, as adults we remember few of the actual sentences from which we learned to speak. If a parrot had copied snatches of Popper’s voice at a lecture, it would certainly have copied them with his Austrian accent: parrots are incapable of copying an utterance without its accent. But a human student might well be unable to copy it with the accent. In fact a student might well acquire a complex meme at a lecture without being able to repeat a single sentence spoken by the lecturer, even immediately afterwards. In such a case the student has replicated the meaning – which is the whole content – of the meme without imitating any actions at all. As I said, imitation is not at the heart of human meme replication.

  Suppose that the lecturer had repeatedly returned to a certain key idea, and had expressed it with different words and gestures each time. The parrot’s (or ape’s) job would be that much harder than imitating only the first instance; the student’s much easier, because to a human observer each different way of putting the idea would convey additional knowledge. Alternatively, suppose that the lecturer had consistently misspoken in a way that altered the meaning, and had then made one correction at the end. The parrot would copy the wrong version. The student would not. Even if the lecturer never corrected the error at all, a human listener might still have a good chance of understanding the idea that was in the lecturer’s mind – and, again, without imitating any behaviour. If someone else reported the lecture but in a way that contained severe misconceptions, a human listener might still be able to detect what the lecturer meant, by explaining the reporter’s misconceptions as well as the lecturer’s intention – just as a conjuring expert might be able to detect what really happened during a trick given only a false account from the audience of what they saw.

  Rather than imitating behaviour, a human being tries to explain it – to understand the ideas that caused it – which is a special case of the general human objective of explaining the world. When we succeed in explaining someone’s behaviour, and we approve of the underlying intention, we may subsequently behave ‘like’ that person in the relevant sense. But if we disapprove, we might behave unlike that person. Since creating explanations is second nature (or, rather, first nature) to us, we can easily misconstrue the process of acquiring a meme as ‘imitating what we see’. Using our explanations, we ‘see’ right through the behaviour to the meaning. Parrots copy distinctive sounds; apes copy purposeful movements of a certain limited class. But humans do not especially copy any behaviour. They use conjecture, criticism and experiment to create good explanations of the meaning of things – other people’s behaviour, their own, and that of the world in general. That is what creativity does. And if we end up behaving like other people, it is because we have rediscovered the same idea.

  That is why the audience at a lecture, when striving to assimilate the lecturer’s memes, are not tempted to face the rear wall of the lecture room, or to imitate the lecturer in any one of infinitely many other ways. They reject such interpretations of what is worth copying about the lecturer not because they are genetically incapable of conceiving of them, as other animals are, but because they are bad explanations of what the lecturer is doing, and bad ideas by the audience’s own values.

  Both puzzles have the same solution

  In this chapter I have presented two puzzles. The first is why human creativity was evolutionarily advantageous at a time when there was almost no innovation. The second is how human memes can possibly be replicated, given that they have content that the recipient never observes.

  I think that both those puzzles have the same solution: what replicates human memes is creativity; and creativity was used, while it was evolving, to replicate memes. In other
words, it was used to acquire existing knowledge, not to create new knowledge. But the mechanism to do both things is identical, and so in acquiring the ability to do the former, we automatically became able to do the latter. It was a momentous example of reach, which made possible everything that is uniquely human.

  A person acquiring a meme faces the same logical challenge as a scientist. Both must discover a hidden explanation. For the former, it is an idea in the minds of other people; for the latter, a regularity or law of nature. Neither person has direct access to this explanation. But both have access to evidence with which explanations can be tested: the observed behaviour of people who hold the meme, and physical phenomena conforming to the law.

  The puzzle of how one can possibly translate behaviour back into a theory that contains its meaning is therefore the same puzzle as where scientific knowledge comes from. And the idea that memes are copied by imitating their holders’ behaviour is the same mistake as empiricism or inductivism or Lamarckism. They all depend on there being a way of automatically translating problems (like the problem of planetary motions, or of how to reach leaves on tall trees or to be invisible to one’s prey) into their solutions. In other words, they assume that the environment (in the form of an observed phenomenon, or a tall tree, say) can ‘instruct’ minds or genomes in how to meet its challenges. Popper wrote:

  The inductivist or Lamarckian approach operates with the idea of instruction from without, or from the environment. But the critical or Darwinian approach only allows instruction from within – from within the structure itself . . .

  I contend that there is no such thing as instruction from without the structure. We do not discover new facts or new effects by copying them, or by inferring them inductively from observation, or by any other method of instruction by the environment. We use, rather, the method of trial and the elimination of error. As Ernst Gombrich says, ‘making comes before matching’: the active production of a new trial structure comes before its exposure to eliminating tests.

  The Myth of the Framework

  Popper could just as well have written, ‘We do not acquire new memes by copying them, or by inferring them inductively from observation, or by any other method of imitation of, or instruction by, the environment.’ The transmission of human-type memes – memes whose meaning is not mostly predefined within the receiver – cannot be other than a creative activity on the part of the receiver.

  Memes, like scientific theories, are not derived from anything. They are created afresh by the recipient. They are conjectural explanations, which are then subjected to criticism and testing before being tentatively adopted.

  This same pattern of creative conjecture, criticism and testing generates inexplicit as well as explicit ideas. In fact all creativity does, for no idea can be represented entirely explicitly. When we make an explicit conjecture, it has an inexplicit component whether we are aware of it or not. And so does all criticism.

  Thus, as has so often happened in the history of universality, the human capacity for universal explanation did not evolve to have a universal function. It evolved simply to increase the volume of memetic information that our ancestors could acquire, and the speed and accuracy which they could acquire it. But since the easiest way for evolution to do that was to give us a universal ability to explain, through creativity, that is what it did. This epistemological fact provides not only the solution of the two puzzles I mentioned, but also the reason for the evolution of human creativity – and therefore the human species – in the first place.

  It must have happened something like this. In early pre-human societies, there were only very simple memes – the kind that apes now have, though perhaps with a wider repertoire of copiable elementary behaviours. Those memes were about practical things like how to get food that was otherwise inaccessible. The value of such knowledge must have been high, so this created a ready-made niche for any adaptation that would reduce the effort required to replicate memes. Creativity was the ultimate adaptation to fill that niche. As it increased, further adaptations co-evolved, such as an increase in memory capacity (to store more memes), finer motor control, and specialized brain structures for dealing with language. As a result, the meme band-width (the amount of memetic information that could be passed from each generation to the next) increased too. Memes also became more complex and sophisticated.

  This is why and how our species evolved, and why it evolved rapidly – at first. Memes gradually came to dominate our ancestors’ behaviour. Meme evolution took place, and, like all evolution, this was always in the direction of greater faithfulness. This meant becoming ever more anti-rational. At some point, meme evolution achieved static societies – presumably they were tribes. Consequently, all those increases in creativity never produced streams of innovations. Innovation remained imperceptibly slow, even as the capacity for it was increasing rapidly.

  Even in a static society, memes still evolve, due to imperceptible errors of replication. They just evolve more slowly than anyone can notice: imperceptible errors cannot be suppressed. They would generally evolve towards greater fidelity of replication, as usual with evolution, and hence to greater staticity of the society.

  Status in such a society is reduced by transgressing people’s expectations of proper behaviour, and is improved by meeting them. There would have been the expectations of parents, priests, chiefs and potential mates (or whoever controlled mating in that society) – who were themselves conforming to the wishes and expectations of the society at large. Those people’s opinions would determine one’s ability to eat, thrive and reproduce, and hence the fate of one’s genes.

  But how does one discover the wishes and expectations of other people? They might issue commands, but they could never specify every detail of what they expected, let alone every detail of how to achieve it. When one is commanded to do something (or expected to, as a condition for being considered worthy of food or mating, for instance), one might remember seeing an already-respected person doing the same thing, and one might try to emulate that person. To do that effectively, one would have to understand what the point of doing it was, and to try to achieve that as best one could. One would impress one’s chief, priest, parent or potential mate by replicating, and following, their standards of what one should strive for. One would impress the tribe as a whole by replicating their idea (or the ideas of the most influential among them) of what was worthy, and acting accordingly.

  Hence, paradoxically, it requires creativity to thrive in a static society – creativity that enables one to be less innovative than other people. And that is how primitive, static societies, which contained pitifully little knowledge and existed only by suppressing innovation, constituted environments that strongly favoured the evolution of an ever-greater ability to innovate.

  From the perspective of those hypothetical extraterrestrials observing our ancestors, a community of advanced apes with memes before the evolution of creativity began would have looked superficially similar to their descendants after the jump to universality. The latter would merely have had many more memes. But the mechanism keeping those memes replicating faithfully would have changed profoundly. The animals of the earlier community would have been relying on their lack of creativity to replicate their memes; the people, despite living in a static society, would be relying entirely on their creativity.

  As with all jumps to universality, the way in which the jump emerged out of gradual changes is interesting to think about. Creativity is a property of software. As I said, we could be running AI programs on our laptop computers today if we knew how to write (or evolve) such programs. Like all software, it would require the computer to have certain hardware specifications in order to be able to process the required amount of data in the required time. It so happened that the hardware specifications that would make creativity practicable were included in those that were being heavily favoured for pre-creative meme replication. The principal one would have been memory capacity: the more one co
uld remember, the more memes one could enact, and the more accurately one could enact them. But there may also have been hardware abilities such as mirror neurons for imitating a wider range of elementary actions than apes could ape – for instance, the elementary sounds of a language. It would have been natural for such hardware assistance for language abilities to be evolving at the same time as the increased meme bandwidth. So, by the time creativity was evolving, there would already have been significant co-evolution between genes and memes: genes evolving hardware to support more and better memes, and memes evolving to take over ever more of what had previously been genetic functions such as choice of mate, and methods of eating, fighting and so on. Therefore, my speculation is that the creativity program is not entirely inborn. It is a combination of genes and memes. The hardware of the human brain would have been capable of being creative (and sentient, conscious and all those other things) long before any creative program existed. Considering a sequence of brains during this period, the earliest ones capable of supporting creativity would have required very ingenious programming to fit the capacity into the barely suitable hardware. As the hardware improved, creativity could have been programmed more easily, until the moment when it became easy enough actually to be done by evolution. We do not know what was being gradually increased in that approach to a universal explainer. If we did, we could program one tomorrow.