A Little Background
It is a familiar enough scene, not just to the creatures of the Serengeti, but nowadays to anyone born within the sound of David Attenborough. In the shimmering heat of the African plain a herd of antelope graze. After a while, and for no discernible reason, one of them stops eating and gazes into the middle distance, his or her attention caught by some discrepancy in the movement of a clump of grass, the density of a shadow. While the rest of the herd continues to graze, more antelope join the first in scrutinising the immediate area. Suddenly, again for no apparent reason, one, more, and then all the antelope are in flight, pursued by the lions that have leapt from their concealment in the long grass. It’s a great scene, but because this is Africa there is no one to shout ‘Cut!’, and we know it will all end badly for one of them. I’ll spare you the details.
Still, gory though it be, it seems to us very natural, very much part of the order of things. Nonetheless, it is actually a puzzle, even a series of puzzles. If we go back to the beginning, past the lions, past the watchful antelope to the opening shot of the herd grazing, the question must naturally arise, “What are they doing?” “Eating grass,” I hear you reply, faintly puzzled at the turn this is taking. But how do they know it’s grass? There’s no question that they do know. They don’t go around randomly nibbling on rocks or each other in the faint hope that some of it will turn out to be nourishing; they go straight for the grass. How do they do that?
“Instinct, obviously”, I hear you cry, and it’s a good answer, but for the one tiny drawback that, despite dear Konrad’s1 best efforts, ‘instinct’ is more a category of ignorance than a body of knowledge. No one actually knows how ‘instinct’ works. For a start, it involves inheriting knowledge, which is tricky territory at best, as we shall see, plus I make it a personal rule never to explain a puzzle with a mystery. “All right,” you say, “They’ve seen it before.” Clearly you are getting the hang of this, except, of course, that they haven’t. Not this particular bit. They have seen other, similar grass, and probably eaten a lot of it, but that doesn’t explain how they know that it’s all the same stuff, or that any of it is grass, i.e. food.
Now, to be fair, a few years ago your answer would have been fine. In fact, both your answers would have been fine. However, since then a number of private laboratories and academic organisations like MIT have spent millions in state and corporate funds trying to get vast arrays of computers to recognise, for example, a cube when they see one, and until very recently, failing utterly, albeit very valuably. I should tell you that these were not trick cubes. In fact, they were not even different cubes, just a single ordinary cube. But change the lighting or the angle of it only slightly, and the computers instantly responded with the binary equivalent of, ”What’s that, now?” It turned out, to everyone’s surprise, that getting computers to recognise stuff was, in the trade’s pithy phrase, non-trivial.
“What about facial recognition software, then, eh?” is your natural riposte, and you are right; sort of. Faced with this unexpected situation, the software engineers’ pragmatic solution was to give it up entirely, and instead to get computers to search databases of all possible facial images and find the nearest match to a reference image. This was familiar territory; basically word search, albeit with very long words. The latest versions get around the “change the lighting” hurdle by storing the images in 3-D as well.
Useful though this brute-force approach actually is, it only goes to show that, when it comes to recognising the components of reality, artificial intelligence (see Cogito Ergo Rossum, below), no matter how well funded, is no match for the relatively cheap natural intelligence that comes with your average antelope.2 How can this be? How is it that what seems so natural to us is so “not trivial” to the folks at MIT and elsewhere?
This is obviously a good moment to see what is actually involved. It’s going to sound obvious, but if you want to recognise grass when you see it again, you have to be able to tell the difference between grass and not-grass when you see it the first time, and then remember it. The first thing the antelope has to be able to do, if it isn’t going to die of hunger right at the start, is to tell grass from not-grass, or to put it in its more general form, stuff from other stuff, grass from rocks being just one example. This is called the taxonomic function. Taxonomy is the science of sorting things into, hopefully, the correct piles. It can be as complicated or as simple as you like, depending on how fussy you are or what you do for a living. Take fish, for instance. Most of us can remember what was probably our first introduction to the difficulties of taxonomy, when we were told that dolphins and whales weren’t fish. Up until then, simple; after that point, a lot more complicated, but unless you are a professional with species quotas to worry about, you probably rely on the little labels at the fishmongers to keep you up to date. On the other hand, Stephen Jay Gould, the brilliant zoologist, once despairingly declared that there was no such thing as a fish, as he had managed to trace their origins to so many different phyla, families, genera, species, etc. that they did not seem to have anything in common, except the original ‘swims in water’ grouping familiar to you as a child.
Anyway, back to antelopes. Their taxonomy is not academic, and probably starts with the two main groups for living creatures, namely: 1) Can I eat it? 2) Can it eat me? As you can imagine, very pertinent to life in the Serengeti and, in a modified form, Wall Street. Then there is a subtler category: Even if I can eat it, can it conceal something that can eat me? And we are back to Wall Street.
Baron Rothschild is said to have attributed his wealth to buying too late and selling too early. The antelope equivalent would arrive at the waterhole after the lions had fed, and would leave before they got hungry again. However, outside a zoo, lions don’t have a specific feeding time and they tend to be opportunistic, which would leave too small a window to feed the herd. Besides, the best grazing would have already gone to the bolder herds with better risk assessment and better monitoring. Furthermore, you can’t always be running. You need time to eat, and running takes energy which means you need even more time to restock. The most successful herds run the least. The Baron was right about one thing, though. No species on this planet is descended from an ancestor who waited until all the facts were in. We are all descended from the ones who hightailed it out of there at the first sign of trouble, made it to the rocks, and only then asked each other
Even with this scenario there are subtleties. You don’t want to be the first to run, because that will separate you from the herd and make you a prime target. Any High School kid will tell you that the best strategy is to belong to the herd, even if you don’t really belong in the herd. And for the same reason you don’t want to be the only one left standing there when the rest of the herd is just a cloud of dust on the horizon. So you do what all the others do, and they do what you do. Even the antelope that hasn’t really been paying attention will start to run the instant it becomes aware the others are hoofing it out of there.
Imitation of the herd is vital. Otherwise, if the first antelope just tiptoes off and tells no one, there would soon be no antelope. It is so ingrained that we even do it in the movies. When the hero ducks, we duck. If we’re ever in a panic our legs will start to run even before we do. They will start to dance while we are still sitting wondering how to approach the boy or girl across the dance floor. Just the sound of running water makes us want to pee. We run with the herd, even though we know that any eventual victims will come from the herd, which is a subject for later in this book. For now, let’s get back to the grass, especially the whole concealing of predators bit.
When an antelope stops eating, which is one of its favourite things to do, in order to stare intently at a distant patch of grass that appears no different from the patch it was just eating, it probably has something to do with its absolutely favourite thing to do, i.e. stay alive. We know that staying alive is an antelope’s absolutely favourite thing a), because it is not a male spider, and that is the only group I know of in general that has other priorities, and b), because it’s a tough old world out there, and any creature that is even a little ambivalent about staying alive would have been taken out of the game aeons ago. So, back to staring at the patch of grass. We know there is some point to it because an antelope has given up eating to do it, but there is a problem, and it is one we, the antelope and the guys at MIT share, so it is worth talking about.
The problem is this: we all have five senses, possibly more if you count that creepy feeling that you are being watched/followed, etc., and they are all operating all the time, 24/7/52, and so on. Even when we are asleep, people can still wake us by providing input to one of our senses; calling us, nudging us, or just banging about in the kitchen. Few people will sleep through the smell of a great breakfast, for instance, or maybe that’s just me, but you get the point. We are all full on all the time, even when we are asleep, and that is an enormous amount of information to be sifted through to get to the one or two rare bits that are actually useful to know. It is like being your own search engine.
But we know antelopes can do that, because one of them has stopped eating to stare at that distant patch of grass. What it must have is some kind of filter that can exclude all the ‘noise’, leaving only the information an antelope would find relevant. A first pass at this would be something along the lines of the noise-cancelling headphones people wear to cancel out aeroplane cabin noise, for instance. Unfortunately, they remove all the noise including the ‘Brace! Brace!’ command, but other information, from watching other passengers assuming the position for example, could, in large part, make up for that.3 However, imagine how much better it would be if you could filter input selectively, leave out the “Do you want milk with that?”, but leave in the “Brace! Brace!”. Selective filtering would give you such an evolutionary edge that it is worth checking to see if it is possible. Here is an experiment you can try yourselves. Find a friend (the hard part) and then have a conversation in a noisy environment, a railway carriage, for example. It will come as no shock to you to discover that this is possible. You can hear each other perfectly clearly over the noise of the train. However, take along an audio recorder and put it between you, thus closer to each of you than you are to each other, and record the conversation; amazingly, it will be largely inaudible
(If you have small children you do not need a train for this. Anywhere in the house will do).
This skill is called ‘being able to concentrate’, and it is hugely valuable, but how can it possibly work? How can you hear your conversation over the noise of the train/children/television/neighbours, etc.? After all, you each have only one set of ears, a single auditory system. There’s no clever technology involved, no separate inputs to be manipulated; just you and your friend talking to each other. The same goes for the antelope, staring at that patch of grass. Obviously, something has caught its eye. It doesn’t
have to be a threat at this stage, just something out of the ordinary. And there, essentially, you have it: the nub of the problem. To be able to notice when something is out of the ordinary, you obviously have to have a pretty well-developed sense of what is ordinary, and that sense is what it takes to be able to pick out one voice, one conversation out of a cacophony of noise, or one suspicious patch of grass in an entire plain full of waving grass. All you have to do is compare any incoming information (noise, smell, sight, sensation) to the standard, ordinary version and, if it matches, ignore it. What you are left with is the important stuff you need to focus on, whether it’s local gossip in a conversation with a friend, or potentially a matter of life or death, as with the dubious patch of grass. In this way, reality becomes its own filter, or rather, because we are talking about perception here, a model of reality that, like noise-cancelling systems, can cancel out the “ordinary” world wherever it matches, leaving only the unexpected and the dangerous to interrupt what you’re doing. I call this the ontic model (OM), for two reasons: ontic means ‘having to do with existence and things that exist’, otherwise known as “the way things are”, so an ontic model is a model of existence or ‘things that exist’ that allows us to recognise things the way they are. Ontology is largely the study of ontic models (http://en.wikipedia.org/wiki/Ontology)4 The other main reason I use OM is that most of the good words are already taken. Even nice general words like ‘realistic’ or ‘existentialist’ have been copyrighted by one scientific discipline or other. My own field, which is concerned exclusively with the study of behaviour and its associated decision paths and processes, has to go undefined because all possible terms that include ‘behaviour’ have been attached to quite specific sub-genres: behaviourist, behaviouralist,5 etc. Even the almost generic Human Factors carries the ring of cockpit stress to this day. So, bags I ‘ontic’. There are no limits to the size or scope of OMs. You can have a model of the cosmos or an atom, of an ant or an elephant. You can have a model of a knee, and be aware that both ants and elephants have them. Nor does it have to be physical. You can have a model of beauty, and apply it with equal certainty to people, buildings, mountains and toenails. And others will understand you; perhaps even agree. But whatever the nature of their subject, all OMs have certain things in common that arise directly from their original and still primary function for most of the inhabitants of this planet – to keep us alive:
- OMs are universal models of the world and its aspects that allow us to recognise and categorise all inputs, internal and external, under all conditions and in all circumstances
- We do this by comparing inputs to expectations and assigning them to the closest model with the greatest predictive strength
- OMs can be triggered by partial input sets; i.e. they can spot a leopard in a tree at dusk. All successful predators are good at camouflage, and all successful prey see through it. The ability to categorise, distinguish and manipulate models from partial input sets is literally vital to survival
- They can be wrong; the predictive function that enables the sorting of inputs into appropriate categories also immediately flags failure of the model. Thus the grass OM’s prediction of allowable movement fails, and instantly other models are brought forward to explain the inputs. The whole lion thing is a major contender.6
all is calm, but with an option to panic. At a certain point, out of the corner of your eye, for some reason the traffic is not normal. The inputs that the ‘normal traffic’ OM has been unobtrusively dumping thus far are now not dumped, but as yet no satisfactory alternative OM has taken them on. You have “a bad feeling” largely caused by your limbic system trying to get
you ready to run. Being human you still haven’t entirely abandoned the ‘normal traffic’ OM because that would be a whole lot better than trying to ‘run’ on a major highway with a load of screaming kids in the back, but you are still ready to panic if a replacement OM can give you the go-ahead. Then you see the flashing blue lights on the other carriageway, and all is well. Or you see the truck coming towards you on your carriageway, and it isn’t. Either way you know what to do. If you were an antelope you would have taken the last exit.
- They are hierarchical, from ‘animal’ to ‘mammal’ to ‘feline quadruped’ to ‘Timmy’; but they still allow us to recognise Timmy under all conditions and circumstances7probable explanation is that “cat” is a paradigm like all the others”
- They are adaptable, allowing us to assign inputs never before experienced to existing or even new categories, and then adjust to recognise subsequent instances under conditions and circumstances never before experienced
In short, what I am saying, as any scientist will have grasped by now, is that we and antelope have theories of things, among them grass. They see the world pretty much as we do. When I say ‘theory’ I don’t mean it in the everyday sense of an interesting idea about something, advanced just as a possible explanation. I mean the real thing. Unlikely as it sounds, the antelope’s theory of grass meets all the criteria for a physical theory in the scientific sense.8 just as scientific theories are analytical tools for understanding, explaining, and making predictions about a given subject matter, the antelope’s OM faculty, whatever we choose to call it, can classify everything of which it is aware, understand it and explain it, and it does this by making predictions for its immediate behaviour. Just as in science, any failure in its predictive strength causes the model to fail and be replaced. The antelope’s theory of grass has to function in such a way that any member of the herd is in a position to understand, verify and challenge or falsify it – which the lions can also do, of course. The theory acts as a filter, too, because it determines the relevance of input taxonomically, and thus decides not only what input is to be evaluated (grass from not-grass) and what is relevant in terms of survival, but also at which hierarchical level the input should be treated, whether the grass is behaving as potential food (single OM) or as evidence of threat (multiple possible OMs). For the antelope, and all other creatures, the fundamental unit of reality, the basic ontic model, is the sum of all possible histories of, in this instance, grass. We will discuss how scientific theories also filter and validate themselves later.
We know that the antelope’s model of the world is valid because they are still here. Over 98% of documented species are now extinct.9 The one thing that distinguishes the theory of grass from a standard scientific theory is that its sole criterion of value is survival, and survival (outside the academic environment) is not guaranteed by knowledge, but by behaviour. Everybody who runs away in time gets to survive, including the stupid ones (See ‘imitation’ above). It’s a physical theory, but not academic in any sense. Lives depend on it. Like war, it doesn’t determine who’s right, but who’s left.
Life, evolution, is not about the survival of the fittest; it’s just the survival of the survivors, tautological though that may be. Everybody knows that scientists reject the idea of teleological evolution, the thought that species are being refined towards some divine ideal of perfection, but even scientists believe that evolution selects for the variants that are best adapted to their environment, the “fittest” in that sense. However, the reality (I use the word advisedly) is that being the “fittest” often consists solely in not being there when the big rock falls. Or the mud slides. If you haven’t already done so you must read Stephen Jay Gould’s brilliant book, “Wonderful Life”, on the discoveries of the Burgess Shale. 505 myagot (million years ago, give or take) a series of mudslides fell at the base of a 500ft cliff of the Cathedral formation in what is now Alberta, Canada, trapping and killing, at the last count, some 65,000 creatures on the ocean bed.10
We know this because they became fossilised in the mud, beautifully preserving them, and then the whole bang shoot was whisked away to become part of the Rockies where they were almost instantly11 discovered by a palaeontologist called Charles Walcott, who counted them. Among these creatures were Hallucigeniae, Anomolocares, Nectocares and Opabiniae, this last having five eyes and a snout like a vacuum cleaner attachment. The names alone give you some idea of the surprise with which their appearance was greeted by their discoverers. There is continuing debate as to whether these beasts can be classified into modern phyla (phylum is the next category down from Animal Kingdom, so very big, but no room for these guys), but whether or not common ancestors can be found, there were no descendants. Whatever evolutionary advantage that fifth eye would have brought with it, or the fins on the crab-like Nectocaris, or the literally back-to-back vertebrae of Hallucigenia, there’s no question that for these species and the vast majority of their friends and neighbours, the muck stopped there. And this was just one, admittedly huge, deposit. We’ll talk about mass extinctions in a different context later, but this was a mini-extinction, unique to us because the shale managed to preserve the soft body parts for us to study, not just the bony bits, but there’s no reason to suppose it was unique at the time. Given enough mud and falling rocks, which there surely were, life as we know it starts to look scarily arbitrary. For example, it has been calculated that, after the volcanic winter caused by the eruption of Mount Toba 70,000 odd years ago, Mankind was down to around 1,000 breeding pairs. Whether or not Mount Toba was the cause, it is generally agreed that there was a population bottleneck at around this time, creating more of a gene puddle than a pool. Richard Dawkins has an even more intriguing theory. He believes that Adam and Eve were real people, although they probably never met, she having died about 80,000 years before he was born. Anyway, at around the time of the Mount Toba eruption, Adam and one or more direct descendants of Eve mated, and from his loins sprang all modern humans. Forget a thousand pairs; we’re down to just one guy, and recent at that. So, Hi there, cousin!12that there wouldn’t be enough time for Eve to have all the offspring that would be needed to populate the planet. On the other hand, if there are thousands of genetic Eves, all Adam has to do is put it about a bit, something his male descendants have been trying to emulate ever since
On the other hand, life itself is pretty recent. As far as we know, for the first ten billion years of its existence there was no life at all in the universe. There certainly wasn’t any on this planet, and even here there has only ever been one successful life form, and that’s DNA. And DNA isn’t even technically alive.13 All the vast diversity of life that we are so proud of on this planet is just the result of minor variations in the sequence of identical nucleotides along identical polymers. People are always very keen to point out that our genes are 98% the same as chimpanzees’, but forget that we are around 80% the same as horses, and 50% or so the same as grass. That’s why we can eat each other; all those lovely proteins and starches. It’s also why we can’t eat rocks.
While we are here in the primordial ooze, it’s worth taking a moment to see what it would take to set out on the path to sentient life. It actually doesn’t take much. You need the capacity for just three things: sensation, preference and movement. First, you need to be able to sense in some way what is going on. Initially it doesn’t have to be very sophisticated: light vs. dark, wet vs. dry, nutrition vs. whatever the opposite of nutrition is, and ultimately life vs. death. You have to be able to prefer one over the other, and you need some level of movement to be able to act on that preference. Without movement, behaviour, preferences have no consequences, and thus no evolutionary benefit. There exists today a marine invertebrate, endearingly known as the sea squirt, that spends the first part of its life floating around, seeking a suitable place to stay. As soon as it finds one it likes, it anchors itself permanently to it and promptly proceeds to ingest its own brain. It no longer needs it, as it will never move again, so its preferences will mean nothing. In fact, in its situation, it is probably best not to care. To my knowledge, only this little creature and, of course, bureaucrats, have this ability, but it proves my point.
I should emphasise that preference does not necessarily entail intelligence. It can be purely automatic, hard-wired into a primitive organism so that it always seeks the light or the dark, the wet or the dry. It can also be wrong. A primitive polycellular organism that innately prefers the wet to the dry may easily be also completely unsuited to life in the water, and so drown the very first time it gets to exercise that choice. A sad fate, but you can see that it wouldn’t take a very long time for all the little creatures whose preferences were wrong to die out, leaving procreation only to those whose preferences were conducive to survival. You can also see that, as the preference mechanism became more complex over time, the organisms would be able to respond to ever more complex options, even becoming ultimately quite picky, without the intelligence required for genuine freedom of choice ever becoming necessary.
Problems begin to arise quite swiftly, though, as decisions begin to be dictated by circumstances, not predilection. You will readily appreciate that a fixed preference for turning right as opposed to left is never going to prove hugely successful as a strategy for escape. Predators would just sit and wait for their prey to come round again.14male to carry on chasing, while they nip back to wait for the desperate deer
Being able to change your mind in response to what is actually happening is obviously crucial, and the evolutionary edge is going to go to the individuals with the most realistic grasp of how things really are. Back to the OM.
Our antelope is what Stephen Hawking would call a Model Dependent Realist. There is a debate going on, especially among physicists and other natural philosophers, as to what constitutes reality. Is ours the only universe or are there multiple universes, even though we can never see them or have any evidence for them? Even in this universe, since we only have the evidence of our senses, is there indeed anything out there? If a tree falls in the forest, … etc.? Each of these views has its own model. The Model Dependent Realist does not engage in the debate; the MDR simply assumes that what he or she thinks is going on is real, and behaves accordingly. We know little of the inner life of the antelope, but I think it is safe to assume, given the consistency of their behaviour in the face of events, that they believe in the reality of what they see, probably because, in their experience, antelope really can die and really can be eaten if they don’t run away really fast.15ended up as Professor and Director of Clinical Psychology at the Ohio State University. He developed a theory called Personal Construct Psychology which states that people, specifically psychiatric patients, do what they do because of what they think is going on. I know what you’re thinking: “I bet he got a grant for that.” But to be fair, at the time most psychiatrists believed that patients did what they did because of what the psychiatrists thought was going on, so it was a major advance. Anyway, like George, I think antelope do what they do because of what they think is going on. Back to the story
This is where the limbic system I mentioned earlier kicks in. I assume that most of you already know what the limbic system is and does, except, of course, those of you who are actually working on it. You probably have no idea what I am talking about, if indeed it exists at all.16 For the rest of you, when I say limbic system, I am not being specific about what parts of the brain are, as they say, “involved” or, better yet, “implicated”. We will look in greater depth at how our knowledge of the brain is progressing in Chapter XX, “The Mind’s ‘I’”, but neuroscience is one of those areas of science where the closer we look, the less we see.17discuss later, one of the things they know least about is how we know anything at all The certainty with which Broca could identify the Limbic Node and its operations has given way to a very proper hesitation to make any claims at all. Fortunately for me, and perhaps you, the limbic system I am referring to is wherever the Palaeomammalian18 brain currently resides, and where the drives that psychologists rather indelicately refer to as the Four Fs (Fight, Flight, Feeding and Mating) are handled.
I say this because, wherever they may be currently processed, there is no doubt that they do exist. We can all get frightened, and want to run away when attacked, and our bodies are not neutral about these things. The sympathetic nervous system gets involved, adrenalin (epinephrine) flows, pupils dilate and a host of other responses follow, each designed to help us run if we can, or fight if we have to.19 A flood of hormones and neurotransmitters pump us up, helping suppress fear and dull pain, often accompanied by feelings of rage and aggression, followed by elation if the outcome is successful. This final stage of elation is probably why people frighten themselves for fun, everything from horror movies to strapping two bits of slippery wood to your feet, standing on sloping ice and dropping as fast as you can towards what looks like certain death.20spent many a merry morning watching from the local bar as the blood wagon bounces down the mountainside with the broken bodies of thrill-seekers Or there’s bungee jumping, which is supposed to be actually safe but your body doesn’t believe that; or parachuting, or driving too fast and aggressively.
While we’re on the subject, road rage is always assumed to be the logical outcome of the aggression shown in the driving, and has been a constant puzzle to most people, especially the family members who are in the car when the attack comes. “He’s always so calm and proper at home, but put him behind a wheel … “ In fact, both the aggression of the driving and the final outburst when some other driver transgresses our hero’s view of his rights, are the product of fear. Like the parent on the school run, he is simply frightened of being out there where bad things can happen. Obviously this can be a thrill, enhanced by driving too fast and too close, but he, or she, sees themself as skilled and properly equipped, like the skier above. At the same time, it can easily become real fear if other users don’t obey the rules that our guy sees as vital to his survival, his model of the way things ought to be. As such incidents accumulate, the fight-or-flight fear response permeates the sympathetic nervous system, building up to a point where it is triggered by what is probably just the last straw, and our hero blows up. Unfortunately, however, the flight option is usually denied him at that point since if he could get away he probably would have, and he’ll keep driving like that until for whatever reason he finds himself trapped. Then it is like putting a nervous dog on a lead; aggression is the only option left.
Of course, life would be infinitely easier if he wouldn’t insist on trying to enforce his view of the rules of the road. He’d still get to drive as fast as he wanted when there wasn’t anyone in the way, and just have to be a little patient when there was. It would add five minutes to the journey, and take away all the stress. In fact, one may well ask, and if you are a backseat driver, I’m sure you have, why is it so important to be right all the time? Especially when it ultimately makes the consequences of being wrong so bad? I shall now reveal to you a central truth about Mankind: most people would rather be right than happy. If this weren’t true, couples wouldn’t argue all the time, and the divorce rate would be a fraction of what it is.21sentence; the guilty are eligible for parole. So why not pretend to be guilty? See above We’ll return to this subject when we discuss the Inquisition, as we surely will.