Sapiens and Homo Deus: The E-Book Collection by Yuval Noah Harari
In the twenty-first century it sounds childish to compare the human psyche to a steam engine. Today we know of a far more sophisticated technology – the computer – so we explain the human psyche as if it were a computer processing data rather than a steam engine regulating pressure. But this new analogy may turn out to be just as naïve. After all, computers have no minds. They don’t crave anything even when they have a bug, and the Internet doesn’t feel pain even when authoritarian regimes sever entire countries from the Web. So why use computers as a model for understanding the mind?
Well, are we really sure that computers have no sensations or desires? And even if they haven’t got any at present, perhaps once they become complex enough they might develop consciousness? If that were to happen, how could we ascertain it? When computers replace our bus driver, our teacher and our shrink, how could we determine whether they have feelings or whether they are just a collection of mindless algorithms?
When it comes to humans, we are today capable of differentiating between conscious mental experiences and non-conscious brain activities. Though we are far from understanding consciousness, scientists have succeeded in identifying some of its electrochemical signatures. To do so the scientists started with the assumption that whenever humans report that they are conscious of something, they can be believed. Based on this assumption the scientists could then isolate specific brain patterns that appear every time humans report being conscious, but that never appear during unconscious states.
This has allowed the scientists to determine, for example, whether a seemingly vegetative stroke victim has completely lost consciousness, or has merely lost control of his body and speech. If the patient’s brain displays the telltale signatures of consciousness, he is probably conscious, even though he cannot move or speak. Indeed, doctors have recently managed to communicate with such patients using fMRI imaging. They ask the patients yes/no questions, telling them to imagine themselves playing tennis if the answer is yes, and to visualise the location of their home if the answer is no. The doctors can then observe how the motor cortex lights up when patients imagine playing tennis (meaning ‘yes’), whereas ‘no’ is indicated by the activation of brain areas responsible for spatial memory.7
This is all very well for humans, but what about computers? Since silicon-based computers have very different structures to carbon-based human neural networks, the human signatures of consciousness may not be relevant to them. We seem to be trapped in a vicious circle. Starting with the assumption that we can believe humans when they report that they are conscious, we can identify the signatures of human consciousness, and then use these signatures to ‘prove’ that humans are indeed conscious. But if an artificial intelligence self-reports that it is conscious, should we just believe it?
So far, we have no good answer to this problem. Already thousands of years ago philosophers realised that there is no way to prove conclusively that anyone other than oneself has a mind. Indeed, even in the case of other humans, we just assume they have consciousness – we cannot know that for certain. Perhaps I am the only being in the entire universe who feels anything, and all other humans and animals are just mindless robots? Perhaps I am dreaming, and everyone I meet is just a character in my dream? Perhaps I am trapped inside a virtual world, and all the beings I see are merely simulations?
According to current scientific dogma, everything I experience is the result of electrical activity in my brain, and it should therefore be theoretically feasible to simulate an entire virtual world that I could not possibly distinguish from the ‘real’ world. Some brain scientists believe that in the not too distant future, we shall actually do such things. Well, maybe it has already been done – to you? For all you know, the year might be 2216 and you are a bored teenager immersed inside a ‘virtual world’ game that simulates the primitive and exciting world of the early twenty-first century. Once you acknowledge the mere feasibility of this scenario, mathematics leads you to a very scary conclusion: since there is only one real world, whereas the number of potential virtual worlds is infinite, the probability that you happen to inhabit the sole real world is almost zero.
None of our scientific breakthroughs has managed to overcome this notorious Problem of Other Minds. The best test that scholars have so far come up with is called the Turing Test, but it examines only social conventions. According to the Turing Test, in order to determine whether a computer has a mind, you should communicate simultaneously both with that computer and with a real person, without knowing which is which. You can ask whatever questions you want, you can play games, argue, and even flirt with them. Take as much time as you like. Then you need to decide which is the computer, and which is the human. If you cannot make up your mind, or if you make a mistake, the computer has passed the Turing Test, and we should treat it as if it really has a mind. However, that won’t really be a proof, of course. Acknowledging the existence of other minds is merely a social and legal convention.
The Turing Test was invented in 1950 by the British mathematician Alan Turing, one of the fathers of the computer age. Turing was also a gay man in a period when homosexuality was illegal in Britain. In 1952 he was convicted of committing homosexual acts and forced to undergo chemical castration. Two years later he committed suicide. The Turing Test is simply a replication of a mundane test every gay man had to undergo in 1950s Britain: can you pass for a straight man? Turing knew from personal experience that it didn’t matter who you really were – it mattered only what others thought about you. According to Turing, in the future computers would be just like gay men in the 1950s. It won’t matter whether computers will actually be conscious or not. It will matter only what people think about it.
The Depressing Lives of Laboratory Rats
Having acquainted ourselves with the mind – and with how little we really know about it – we can return to the question of whether other animals have minds. Some animals, such as dogs, certainly pass a modified version of the Turing Test. When humans try to determine whether an entity is conscious, what we usually look for is not mathematical aptitude or good memory, but rather the ability to create emotional relationships with us. People sometimes develop deep emotional attachments to fetishes like weapons, cars and even underwear, but these attachments are one-sided and never develop into relationships. The fact that dogs can be party to emotional relationships with humans convinces most dog owners that dogs are not mindless automata.
This, however, won’t satisfy sceptics, who point out that emotions are algorithms, and that no known algorithm requires consciousness in order to function. Whenever an animal displays complex emotional behaviour, we cannot prove that this is not the result of some very sophisticated but non-conscious algorithm. This argument, of course, can be applied to humans too. Everything a human does – including reporting on allegedly conscious states – might in theory be the work of non-conscious algorithms.
In the case of humans, we nevertheless assume that whenever someone reports that he or she is conscious, we can take their word for it. Based on this minimal assumption, we can today identify the brain signatures of consciousness, which can then be used systematically to differentiate conscious from non-conscious states in humans. Yet since animal brains share many features with human brains, as our understanding of the signatures of consciousness deepens, we might be able to use them to determine if and when other animals are conscious. If a canine brain shows similar patterns to those of a conscious human brain, this will provide strong evidence that dogs are conscious.
Initial tests on monkeys and mice indicate that at least monkey and mice brains indeed display the signatures of consciousness.8 However, given the differences between animal brains and human brains, and given that we are still far from deciphering all the secrets of consciousness, developing decisive tests that will satisfy the sceptics might take decades. Who should carry the burden of proof in the meantime? Do we consider dogs to be mindless machines until proven otherwise, or do we treat dogs as
On 7 July 2012 leading experts in neurobiology and the cognitive sciences gathered at the University of Cambridge, and signed the Cambridge Declaration on Consciousness, which says that ‘Convergent evidence indicates that non-human animals have the neuroanatomical, neurochemical and neurophysiological substrates of conscious states along with the capacity to exhibit intentional behaviours. Consequently, the weight of evidence indicates that humans are not unique in possessing the neurological substrates that generate consciousness. Nonhuman animals, including all mammals and birds, and many other creatures, including octopuses, also possess these neurological substrates.’9 This declaration stops short of saying that other animals are conscious, because we still lack the smoking gun. But it does shift the burden of proof to those who think otherwise.
Responding to the shifting winds of the scientific community, in May 2015 New Zealand became the first country in the world to legally recognise animals as sentient beings, when the New Zealand parliament passed the Animal Welfare Amendment Act. The Act stipulates that it is now obligatory to recognise animals as sentient, and hence attend properly to their welfare in contexts such as animal husbandry. In a country with far more sheep than humans (30 million vs 4.5 million), that is a very significant statement. The Canadian province of Quebec has since passed a similar Act, and other countries are likely to follow suit.
Many business corporations also recognise animals as sentient beings, though paradoxically, this often exposes the animals to rather unpleasant laboratory tests. For example, pharmaceutical companies routinely use rats as experimental subjects in the development of antidepressants. According to one widely used protocol, you take a hundred rats (for statistical reliability) and place each rat inside a glass tube filled with water. The rats struggle again and again to climb out of the tubes, without success. After fifteen minutes most give up and stop moving. They just float in the tube, apathetic to their surroundings.
You now take another hundred rats, throw them in, but fish them out of the tube after fourteen minutes, just before they are about to despair. You dry them, feed them, give them a little rest – and then throw them back in. The second time, most rats struggle for twenty minutes before calling it quits. Why the extra six minutes? Because the memory of past success triggers the release of some biochemical in the brain that gives the rats hope and delays the advent of despair. If we could only isolate this biochemical, we might use it as an antidepressant for humans. But numerous chemicals flood a rat’s brain at any given moment. How can we pinpoint the right one?
For this you take more groups of rats, who have never participated in the test before. You inject each group with a particular chemical, which you suspect to be the hoped-for antidepressant. You throw the rats into the water. If rats injected with chemical A struggle for only fifteen minutes before becoming depressed, you can cross out A on your list. If rats injected with chemical B go on thrashing for twenty minutes, you can tell the CEO and the shareholders that you might have just hit the jackpot.
16. Left: A hopeful rat struggling to escape the glass tube. Right: An apathetic rat floating in the glass tube, having lost all hope.
16.Adapted from Weiss, J.M., Cierpial, M.A. & West, C.H., ‘Selective breeding of rats for high and low motor activity in a swim test: toward a new animal model of depression’, Pharmacology, Biochemistry and Behavior 61:49–66 (1998).
Sceptics could object that this entire description needlessly humanises rats. Rats experience neither hope nor despair. Sometimes rats move quickly and sometimes they stand still, but they never feel anything. They are driven only by non-conscious algorithms. Yet if so, what’s the point of all these experiments? Psychiatric drugs are aimed to induce changes not just in human behaviour, but above all in human feeling. When customers go to a psychiatrist and say, ‘Doctor, give me something that will lift me out of this depression,’ they don’t want a mechanical stimulant that will cause them to flail about while still feeling blue. They want to feel cheerful. Conducting experiments on rats can help corporations develop such a magic pill only if they presuppose that rat behaviour is accompanied by human-like emotions. And indeed, this is a common presupposition in psychiatric laboratories.10
The Self-Conscious Chimpanzee
Another attempt to enshrine human superiority accepts that rats, dogs and other animals have consciousness, but argues that, unlike humans, they lack self-consciousness. They may feel depressed, happy, hungry or satiated, but they have no notion of self, and they are not aware that the depression or hunger they feel belongs to a unique entity called ‘I’.
This idea is as common as it is opaque. Obviously, when a dog feels hungry, he grabs a piece of meat for himself rather than serve food to another dog. Let a dog sniff a tree watered by the neighbourhood dogs, and he will immediately know whether it smells of his own urine, of the neighbour’s cute Labrador’s or of some stranger’s. Dogs react very differently to their own odour and to the odours of potential mates and rivals.11 So what does it mean that they lack self-consciousness?
A more sophisticated version of the argument says that there are different levels of self-consciousness. Only humans understand themselves as an enduring self that has a past and a future, perhaps because only humans can use language in order to contemplate their past experiences and future actions. Other animals exist in an eternal present. Even when they seem to remember the past or plan for the future, they are in fact reacting only to present stimuli and momentary urges.12 For instance, a squirrel hiding nuts for the winter doesn’t really remember the hunger he felt last winter, nor is he thinking about the future. He just follows a momentary urge, oblivious to the origins and purpose of this urge. That’s why even very young squirrels, who haven’t yet lived through a winter and hence cannot remember winter, nevertheless cache nuts during the summer.
Yet it is unclear why language should be a necessary condition for being aware of past or future events. The fact that humans use language to do so is hardly a proof. Humans also use language to express their love or their fear, but other animals may well experience and even express love and fear non-verbally. Indeed, humans themselves are often aware of past and future events without verbalising them. Especially in dream states, we can be aware of entire non-verbal narratives – which upon waking we struggle to describe in words.
Various experiments indicate that at least some animals – including birds such as parrots and scrub jays – do remember individual incidents and consciously plan for future eventualities.13 However, it is impossible to prove this beyond doubt, because no matter how sophisticated a behaviour an animal exhibits, sceptics can always claim that it results from unconscious algorithms in its brain rather than from conscious images in its mind.
To illustrate this problem consider the case of Santino, a male chimpanzee from the Furuvik Zoo in Sweden. To relieve the boredom in his compound Santino developed an exciting hobby: throwing stones at visitors to the zoo. In itself, this is hardly unique. Angry chimpanzees often throw stones, sticks and even excrement. However, Santino was planning his moves in advance. During the early morning, long before the zoo opened for visitors, Santino collected projectiles and placed them in a heap, without showing any visible signs of anger. Guides and visitors soon learned to be wary of Santino, especially when he was standing near his pile of stones, hence he had increasing difficulties in finding targets.
In May 2010, Santino responded with a new strategy. In the early morning he took bales of straw from his sleeping quarters and placed them close to the compound’s wall, where visitors usually gather to watch the chimps. He then collected stones and hid them under the straw. An hour or so later, when the first visitors approached, Santino kept his cool, showing no signs of irritation or aggression. Only when his victims were within range did Santino suddenly grab the stones from their hiding place and bombard the frightened humans, who would scuttle in all directions. In
Yet even Santino doesn’t satisfy the sceptics. How can we be certain that at 7 a.m., when Santino goes about secreting stones here and there, he is imagining how fun it will be to pelt the visiting humans at noon? Maybe Santino is driven by some non-conscious algorithm, just like a young squirrel hiding nuts ‘for winter’ even though he has never experienced winter?14
Similarly, say the sceptics, a male chimpanzee attacking a rival who hurt him weeks earlier isn’t really avenging the old insult. He is just reacting to a momentary feeling of anger, the cause of which is beyond him. When a mother elephant sees a lion threatening her calf, she rushes forward and risks her life not because she remembers that this is her beloved offspring whom she has been nurturing for months; rather, she is impelled by some unfathomable sense of hostility towards the lion. And when a dog jumps for joy when his owner comes home, the dog isn’t recognising the man who fed and cuddled him from infancy. He is simply overwhelmed by an unexplained ecstasy.15
We cannot prove or disprove any of these claims, because they are in fact variations on the Problem of Other Minds. Since we aren’t familiar with any algorithm that requires consciousness, anything an animal does can be seen as the product of non-conscious algorithms rather than of conscious memories and plans. So in Santino’s case too, the real question concerns the burden of proof. What is the most likely explanation for Santino’s behaviour? Should we assume that he is consciously planning for the future, and anyone who disagrees should provide some counter-evidence? Or is it more reasonable to think that the chimpanzee is driven by a non-conscious algorithm, and all he consciously feels is a mysterious urge to place stones under bales of straw?