The great French mathematician and physicist Henri Poincaré (1854–1912) took a particular interest in the origins of his own astonishing creativity. His achievements were impressive: his work profoundly reshaped mathematics and physics – including laying crucial foundations for Einstein’s theory of relativity and the modern mathematical analysis of chaos. But he also had some influential speculations about where many of his brilliant ideas came from: unconscious thought.
Poincaré found that he would often struggle unsuccessfully with some mathematical problem, perhaps over days or weeks1 (to be fair, the problems he got stuck on were difficult, to say the least). Then, while not actually working on the problem at all, a possible solution would pop into his mind. And when he later checked carefully, the solution would almost always turn out to be correct.
How was this possible? Poincaré’s own suspicion was that his unconscious mind was churning through possible approaches to the problem ‘in the background’ – and when an approach seemed aesthetically ‘right’, it might burst through into consciousness. Poincaré believed that this ‘unconscious thought’ process was carried out by what might almost be a second self, prepared and energized by periods of conscious work, yet able to work away on the problem in hand entirely below the level of conscious awareness.
The notable twentieth-century German composer Paul Hindemith, in a well-known passage from his book, reports a similar belief, with a striking metaphor:
We all know the impression of a very heavy flash of lightning in the night. Within a second’s time we see a broad landscape, not only in its general outlines but with every detail. Although we could never describe each single component of the picture, we feel that not even the smallest leaf of grass escapes our attention. We experience a view, immensely comprehensive and at the same time immensely detailed, that we never could have under normal daylight conditions, and perhaps not during the night either, if our senses and nerves were not strained by the extraordinary suddenness of the event. Compositions must be conceived the same way. If we cannot, in the flash of a single moment, see a composition in its absolute entirety, with every pertinent detail in its proper place, we are not genuine creators.2
Taken literally, Hindemith’s claim would seem to imply that the entire process of composition is the work of the unconscious – the complete score is, it seems, worked out by subterfuge by unconscious processes, only to break forth into consciousness in a moment of spectacular incandescence. The unconscious work complete, the composer needs merely to go through the laborious process of transcribing the already finished work onto paper, a humdrum activity indeed, given that the creative labour has already been done. Hindemith’s conception of the composing of processing is all the more remarkable in the light of the extreme complexity and idiosyncrasy of the musical system governing his own pieces.3
Let us, for contrast, consider ‘insight’ of a much more prosaic kind, in our struggle to make sense of apparently baffling images. You may have seen one or both of the pictures in Figure 34 before. If you have, you will immediately know what they represent. If not, they will almost certainly seem to you to be nothing more than a baffling jumble of speckles, marks and smudges. If initially these make no sense, spend a minute or two inspecting them closely – if you are lucky, you may experience a rather delightful feeling when, suddenly, their interpretation ‘pops out’ (spoiler alert! don’t read on until you have finished examining the images in Figure 34). If you haven’t seen these pictures before, don’t give up too soon. You may find that, even after a minute or two of bafflement, you find the pictures suddenly make sense – and when they do, the pattern will seem so obvious that you may wonder how you could possibly not have spotted it right away. If you are still mystified after a couple of minutes, you can turn to Figure 35 (page 164), in which all is revealed.
Figure 34. Any idea what these are?4
The left image shows a Dalmatian dog sniffing the ground; the right-hand picture is a ‘portrait’ of a cow. Obvious once you see them – and once seen, they can’t be unseen. These images will be easy to make sense of, if you next see them years or decades in the future.
When the object does suddenly ‘pop out’, we have a sense of sudden insight, but no idea how to explain where it came from. Without warning, order emerges from chaos. We have no sense of getting ‘warmer’ or ‘colder’ before insight suddenly hits us – we have a sense of floundering aimlessly, followed, if we are lucky, by what feels like a ‘bolt from the blue’ of sudden understanding. The problem is solved not by a sequence of steps, getting ever closer to the answer. Quite the opposite: the cycle of thought churns on and on, exploring different possible organizations with no sign of progress until, suddenly, and within a single step, it chances upon the solution.
Now, imagine that, rather than allowing you to inspect these images continuously for many seconds or minutes, I had shown them briefly (perhaps just a few seconds at a time) once a week. Eventually, on one of these occasions, a Dalmatian would spring into view; on another, you would be confronted with the sad and steady gaze of the cow. These moments of sudden insight might seem to demand an explanation: you would ask ‘why does the image make sense now, when it made no sense before?’
A natural answer might suggest itself: ‘I must have been unconsciously working away on these images – and solved or partially solved the mystery without even knowing it. Then the answer “broke through” into consciousness, when I saw the image again.’ Yet this would be quite wrong – the same sudden ‘pop out’ occurs when we continuously contemplate the image, and there has been no opportunity for unconscious background pondering. The phenomenon of sudden insight stems not from unconscious thought, but from the nature of the problem: searching for a meaningful interpretation with few helpful and unambiguous clues.
These sudden flashes of ‘visual insight’, which could so easily be misattributed to unconscious thought, should make us sceptical of the unconscious origins of other flashes of insight in mathematics, science or music. Introspection, even the introspection of geniuses, is not to be taken at face value.
The brain is, as we have seen, a cooperative computing machine – large networks of neurons collectively piece together the solution to a single problem: the cycle of thought proceeds one step at a time. And the brain’s networks of neurons are highly interconnected, so there seems little scope for assigning different problems to different brain networks. Here, the contrast with Allport’s and Shaffer’s demonstrations of remarkable dual task performance, outlined in Chapter 8, is telling. Their studies suggested that people could do two things at once when distinct mental calculations – presumably associated with non-overlapping networks of neurons – were involved (e.g. for sight-reading music and taking dictation). And such specialized brain networks can sometimes be developed for highly practised and repetitive tasks. But solving difficult problems, whether mathematical, musical or of any other kind, is the very antithesis of a routine, specialized problem with a dedicated brain network: on the contrary, thinking about such problems will need to engage most of the brain. So the idea that profound unconscious thought can be ‘running in the background’ as we go about our everyday lives is fanciful indeed. Routine and highly practised activities aside, the cycle of thought can attend to, and make sense of, only one set of information at a time.
Poincaré and Hindemith cannot possibly be right. If they are spending their days actively thinking about other things, their brains are not unobtrusively solving deep mathematical problems or composing complex pieces of music, perhaps over days or weeks, only to reveal the results in a sudden flash. Yet, driven by the intuitive appeal of unconscious thought, psychologists have devoted a great deal of energy in searching for evidence for unconscious mental work. In these studies, they typically give people some tricky problems to solve (e.g. a list of anagrams); after a relatively short period of time, they might instruct participants to continue, to take a break, to do another similar or different mental task, or even get a night’s sleep, before resuming their problems. According to the ‘unconscious work’ perspective, resuming after a break should lead to a sudden improvement in performance, compared with people who just keep going with the task. Studies in this area are numerous and varied,5 but I think the conclusions are easily summarized. First, the effects of breaks of all kinds are either negligible or non-existent: if unconscious work takes place at all, it is sufficiently ineffectual to be barely detectable, despite a century of hopeful attempts. Second, many researchers have argued that the minor effects of taking a break – and indeed, Poincaré’s and Hindemith’s intuitions – have a much more natural explanation, which involves no unconscious thought at all.
The simplest version of the idea comes from thinking about why one gets stuck with a difficult problem in the first place. What is special about such problems is that you can’t solve them through a routine set of steps (in contrast, say, to adding up columns of numbers, which is laborious but routine) – you have to look at the problem in the ‘right way’ before you can make progress (e.g. with an anagram, you might need to focus on a few key letters; in deep mathematics or musical composition, the space of options might be large and varied). So ideally, the right approach would be to fluidly explore the range of possible ‘angles’ on the problem, until hitting on the right one. Yet this is not so easy: once we have been looking at the same problem for a while, we feel ourselves to be stuck or going round in circles. Indeed, the cooperative computational style of the brain makes this difficult to avoid.
Mental cul-de-sacs of this kind occur when our brains fail to find a satisfactory analysis or interpretation. Conscious attempts to clear the cul-de-sac can, of course, often be successful: we jettison some information, and focus instead on slightly different information. We focus on different parts of a cryptic crossword clue (perhaps the word ‘jumble’ means this is some kind of anagram). We actively dredge up different pieces of knowledge that we think might help us. (‘Oh – this looks like a geometry problem with circles and angles; I learnt about circle theorems at school; what were they all about?’) All too often, though, such deliberate attacks on a problem fail. Indeed, we can find ourselves endlessly going down the same mental cul-de-sacs, such as when I try to think of the word artichoke, for which my typical internal dialogue is something like: ‘No, not avocado ! Not asparagus ! Not aubergine! And certainly not aspidistra ! Oh, this is ridiculous! Help!’
Breaking out of mental cul-de-sacs is precisely what a break will give us. Starting afresh with a relatively clear mind is more likely to succeed than a mind filled with partial solutions and suggestions which, after an increasingly frustrating struggle, have clearly failed. And, by sheer chance, we might even get a clue that helps us. But probably the most important aspect of setting a problem aside is that, when we return to it, we see it afresh unencumbered by our previous failed attempts. Often, our new perspective will be no more successful than the old, but, now and again, the answer will suddenly snap into place.
But what is not happening, despite our intuitions to the contrary, is hidden background thought mulling over our problem beneath the level of conscious awareness, and taking no part in the cycle of thought. Unconscious problem-solving, and unconscious thought of all kinds, is a myth.
Poincaré’s description of his particular method of solving mathematical problems suggests why he was particularly susceptible to brilliant flashes of insight. His normal strategy for solving a problem was to work out the outlines of the solution entirely in its head, without pen and paper; and only then to, somewhat laboriously, translate his intuitions into the symbolic language of mathematics, to be checked and verified. But what is crucial is that, for Poincaré, mathematical problems were transformed into perceptual problems: and with the right perceptual intuition, the process of creating a ‘proof’ acceptable to fellow mathematicians would be relatively routine, if slow. A perceptual problem is just the kind of problem that can be solved in a single cycle of thought – provided that we happen to lock onto just the right information and ‘see’ the pattern in that information in just the right way like the Dalmatian and doleful cow of Figure 35.
From this point of view, Poincaré’s mathematical brainwave is of precisely the same type as the ‘insight’ we experience when we glance back at our Dalmatian or cow images to find that, finally, for some reason, order has mysteriously emerged from chaos. Crucially, in neither case is the sudden revelation the product of hours or days of unconscious thought. Instead, the solution is found in a single cycle of thought when we contemplate the problem again. Having broken free of our previous and incorrect analysis, by happy chance our brains alight upon the correct solution. The mental fragments are recombined in just the right way, click delightfully together, and the problem is solved.
This viewpoint is nicely illustrated by one of the most celebrated stories of scientific insight: the discovery of the structure of benzene by the great nineteenth-century chemist August Kekulé. The brainwave struck as he was having a daydream in which a snake began to swallow its own tail. It suddenly struck Kekulé that benzene might itself have a circular structure (the kind of metaphorical leap that our elastic minds are prone to, as we shall see later); and before long he had worked out his detailed analysis of the chemical structure of the benzene ring.
We might wonder, of course, how it is that the right perceptual interpretation happens to come to mind, helping to solve what seemed to be an intractable problem, almost in the moment. Could it be, perhaps, that the unconscious has been working away furiously on the details of the solution for hours or days,6 and finally decides to communicate with the ‘conscious mind’, not simply by telling it the answer, but by conjuring up a cryptic image? Despite the charm of this story, it isn’t very plausible. There is no mystery about how just the right perceptual image comes to mind to lead to the starling scientific insight – because almost all the perceptual images that flash through our minds are not the right kind of images to spark a brilliant insight. Only the extraordinarily rare cases in which, by happy accident, the appropriate image or collision of images happens to unlock some important discovery turn into stories told to the next generation of mathematicians and scientists.
Thus, insight may indeed occur in a flash (i.e. in a single cycle of thought) when our minds lock onto a problem afresh and see it from a subtly different perspective. But there is no reason to believe that it bursts through from the deliberations of a second, unconscious self.
Hindemith’s claim that entire pieces of music come to their creators in their entirety should similarly not be taken at face value. His potent image of the sense of grasping the full detail of a night-time scene lit up by a flash of lightning is itself telling. After all, in Part One, we found that our sense of a detailed and vivid visual world is itself illusory: we have the illusion that the information that we can create on demand already lies, pre-formed and within reach (e.g. with a flick of our eyes and/or a shift of attention). And surely the same is true for musical composition. What Hindemith surely means is that, after the flash of inspiration has occurred, creating the piece (and writing it out in musical notation) it proceeds with fluency. Indeed, the piece flows out of the composer in a way that feels inevitable and predetermined. Hindemith explains:
This does not mean that any f sharp in the six hundred and twelfth measure of the final piece would have been determined in the very first flash of cognition. If the seer should in this first flash concentrate his attention on any particular detail of the whole, he would never conceive the totality, but if the conception of this totality strikes his mind like lightning, this f sharp and all the other thousands of notes and other means of expression will fall into line almost without his knowing it.7
So the flash of insight is not the sense of the entirety of the piece, written out in some inner brain code by an inner ‘unconscious composer’; instead, in composing as in mathematics and science, the flash of inspiration represents no more than the revelation of a new and promising direction to explore – it is the starting point for a long period of creation, perhaps even creative struggle. Of course, if the struggle ultimately proves successful – the symphony is written, the mathematical proof is found – it is easy to imagine that all this subsequent work was an inevitable consequence, a mere ‘working out’ of that initial moment of revelation. But this is just loose talk: no more true than the thought that the entire Western philosophical tradition is merely the ‘working out’ of the details of Plato and Aristotle; or that more than half a century of rock music is merely the inevitable consequence of the first accentuated backbeat rhythm or the inspired idea to attach an electrical ‘pick-up’ to amplify a guitar.
Imagine you are driving through crowded city streets, chatting to a friend over the music from the car radio. This might seem to you a miracle of multitasking. Surely, you are aware of the road and adjusting the steering and applying the brakes when needed; you must be aware of any conversation you are actively engaged in; and you are aware of the music (otherwise, why bother to play it?). And it seems that you must simultaneously be making sense of the road, conversation and music, to juggle the three activities successfully.
Yet remember the grand illusion. You ‘feel’ you are aware, simultaneously, of the car in front of you, the buildings flowing past, the markings on the road, the trees and sky. But, as we saw in Part One, you very definitely are not. Of course, just as you can flick your eyes across the visual scene to answer any query that occurs to you, so you are able rapidly to shift your attention from the conversation to the radio, or to your surroundings, conveying the impression that each is ever present. But if a lorry pulls out in front of you unexpectedly, and you brake, honk or swerve, the flow of conversation will be abruptly halted. Shaken, both driver and passenger may, indeed, be left with no idea even of what they were talking about.
Could it be that we are not multitasking at all? Might we, instead, merely be hopping from one task to the next? Is multitasking a myth?
Hal Pashler, whose work we discussed in Chapter 3, provided a fascinating insight into the severe limits of multitasking while driving with his co-authors Jonathan Levy and Erwin Boer.8 They asked people to engage in a simple simulated driving task – rather like driving in a video game. The participants’ main objective was simply to follow the car in front along a moderately winding road, using a steering wheel and the right foot to control the accelerator and brake pedals (as in normal driving). In addition, though, they had to carry out what we’ll call a ‘detection’ task.9 Now and again, easy-to-spot perceptual events would occur. Participants had to detect either one or two beeps; or whether the colour of the rear window of the car in front changed once or twice (that is, they had to respond to auditory or visual stimuli). If either of these events occurred, the driver had to report whether the event occurred once or twice – and these reports could also take one of two forms: pressing a button or simply saying the words ‘one’ or ‘two’ (call these the manual or vocal responses respectively). They had, of course, to do this while continuing to ‘drive’ safely – following the car in front, and braking where necessary.
One might imagine that experienced drivers would be more or less unaffected by carrying out such a simple additional task – after all, intuitively we often feel that we are driving on autopilot, and when we do react, we feel that we have done so almost as a matter of reflex, and certainly without prior conscious deliberation.
It would be reassuring if this were true – but the reverse is the case. In particular, when the person has to detect and respond to a ‘signal’ (colour changes or beeps) at the same time as the car in front begins to slow down, braking is badly affected. Indeed, the average increase in braking time when there is a signal to respond to, compared to when there isn’t, is about one sixth of a second: an amount of extra time that could, of course, be all too significant if we were braking for real (e.g. the time it takes a car travelling at 60mph to travel roughly an additional 15 feet before the brake is applied).
It seems reasonable to suspect, too, that the different variants (manual versus vocal; auditory versus visual) of the additional task would have different effects. Surely, speaking one’s answer aloud should interfere less with a desperate scramble to the brake pedal than pressing a button with the hand. Perhaps a signal directed at different limbs (moving the foot versus the hand) might get confused, but surely not signals between the leg and the machinery of speech – the lips, tongue and voice box. And wouldn’t one expect that reporting two beeps would interfere less with braking than reporting two flashes, on the reasonable grounds that the braking task involves visual analysis to see when the vehicle in front is slowing. In fact, all these different combinations slow down braking to the same degree.
Moreover, the negative effect of even a simple additional task is very difficult to eliminate. For example, in a further study, Levy and Pashler checked what would happen if people were expressly told to focus on braking as fast and as safely as possible and simply to abandon any additional task if they happened to be in the middle of it.10 Often people did indeed abandon the additional task, but their braking was still significantly slowed.
This should make you worry about having a conversation while you are driving. Of course, driving while holding a mobile phone to your ear is a bad idea – apart from anything else, you have just one hand on the wheel. But experiments have consistently shown that ‘hands-free’ phone conversations turn out to be almost as dangerous – the flow of conversation and the flow of driving interfere with each other much more severely than one might expect. Although we imagine that we can ‘see’ everything around us when we are driving, that we could brake or swerve if we needed to, independently of our chatter, these intuitions are entirely wrong. We ‘see’ only a minute fraction of the road flowing past us (recall the pilots obliviously landing ‘through’ a plane taxiing across the runway; and the invisible lady with the umbrella) – and it requires active vigilance to direct our limited window of attention where it is most needed (to scan the next junction, to track a pedestrian who might step out into the road). And worse, our driving actions (and reactions) can get badly entangled with other actions, as Pashler and colleagues’ studies show.
Conversation with a passenger, as well as by phone, has many of the same dangers. Thankfully, though, passengers and drivers tend to slow or stop their conversation when road conditions get dangerous – driving simply takes precedence over speaking. A particular danger of speaking by mobile phone is that this doesn’t happen, because the person on the other end of the line has no idea what hazardous manoeuvres may be demanding the attention of the driver, and the driver feels socially obliged to maintain a flow of conversation if at all possible. So the driver is all too likely to keep focusing on the conversation, unaware that she is significantly increasing the likelihood of an accident.
Could it be, though, that while we may be unable actively to pay attention to more than one thing at a time, our brains might be able unconsciously to search our mental archives, pulling out, as it were, useful files for later use? If this is right, then Poincaré’s unconscious could perhaps have been running through potentially relevant bits of higher mathematics, stored over a lifetime of study. Then, when Poincaré returned to a problem, some of the vital clues to the solution might have been ready to hand – and a flash of insight would result. According to this viewpoint, the brain might not be able to solve a problem unconsciously, but unconscious activation of relevant memories might prepare the ground for finding the solution.
So, can we find evidence for unconscious memory search? With my colleagues Elizabeth Maylor and Greg Jones at the University of Warwick, I carried out an experiment some years ago that tested whether unconscious memory searches can help out the conscious mind.11
Rather than choose deep mathematical reasoning, we chose the simplest possible task: retrieving familiar words from memory. Suppose, for example, that I ask you to name as many foods as you can. Despite the vast range of food vocabulary at your disposal, you will almost certainly find yourself slowing down surprisingly quickly, with flurries of fruits, bursts of baked goods, and surges of seasonings, punctuated by surprising, and ever longer, silences. Suppose, instead, I ask you to name as many countries as you can. Although there are 200 or so countries recognized by the United Nations, most of which will be familiar to you, you will, again, find yourself struggling sooner than you might expect.
But what if I asked you to name as many food items or countries as possible? The only way to do this is to focus on foods for a while, and then move over to countries when foods are getting tricky, and then back to foods again when you are running out of countries – and so on. This is interesting in itself – perhaps indicating that our memories are organized so that foods are linked to other foods, and countries are linked to other countries. But this switching strategy is also interesting for another reason: it provides a way of finding out how far we are able to continue to search for the category we are not currently generating.
According to the cycle of thought perspective, any unconscious racing around our mental archives is entirely ruled out. That is, if we are scouring our memories for foods, we are not able to simultaneously search for countries, and vice versa. If so, we should generate foods or countries more rapidly than we can generate one or the other alone, although not by much.
Suppose, instead, that while focusing our conscious minds on generating foods, unconscious mental search processes can work away, in the background, unearthing a string of countries. Then, when we switch to countries, we should be able rapidly to ‘download’ these – they would not need to be found afresh, because the unconscious search process would have identified them already. If it is indeed possible to search for foods or countries simultaneously (even though we can consciously report the results of only one search at a time), then the rate at which we generate answers in both categories should be substantially greater than the rate at which we can generate answers from either category alone.
Across a wide range of test stimuli, the results were unequivocal: there is absolutely no sign that we can search for xs when we are currently thinking about ys; or search for ys, when we have been thinking about xs. As soon as we switch from searching one category to searching another, all search processes for that first category appear to cease abruptly. While it would be hugely advantageous for an unconscious process to keep running in the background, there is absolutely no evidence that this occurs. This is particularly striking, when we consider how useful such ability would be to us in daily life. We are continually faced with a welter of tasks that somehow we have to interleave: keeping track of the current conversation, reading the newspaper, planning what we’re going to do next/tomorrow/with our entire lives, pondering tricky philosophical questions … How useful it would be if, while focusing on one task, our unconscious mind could be making progress with other tasks by bringing to light relevant pieces of information! Sadly, though, when our conscious mind is focusing on problem A, ‘research’ on problems B, C, D and so on seems to come to a complete standstill.
Now and again, thoughts do ‘pop into our minds’ – names we had struggled to remember, things we have forgotten to do, and occasionally even insights into tricky problems with which we have been struggling. But this isn’t the product of unconscious, background thought. It arises when we flip back to thinking about an old problem for a moment and, now free of the unhelpful mental loops which got us stuck in the first place, we see a solution that had evaded us before – or, in some cases, dimly suspect where such a solution might lie.
The distinction between suspecting a solution and finding a solution is a rather slippery one, and leads to considerable overestimation of the evidence for unconscious mental processes. Recall Kekulé’s daydream in which a snake began to swallow its own tail. His momentary insight was surely a suspicion that the structure of benzene might be a ring or circle; and surely he must have followed endless false trails before alighting on the correct answer. Indeed, Kekulé only knew he had the correct answer after carefully piecing together the detailed structure of the benzene ring, and checking that it worked. So the ‘flash of inspiration’ is perhaps better termed a ‘flash of suspicion’. On those rare occasions when the flash of suspicion turns out to be justified, it is so easy to have the illusion that one’s brain had somehow worked out the complete answer, and checked it in detail, before ‘suggesting’ it to the conscious mind in the first place. And if that were true, this chain of events would, of course, require unconscious thought, and lots of it. But in reality the checking and analysis comes after the momentary mental flash, not before.
On reflection, this is just yet another variation on the grand illusion, and the trickery that underpins it. Just as we have the sense that the whole perceptual world is loaded into our minds, because it is available whenever we need it, so it is easy to imagine that the entire solution to a problem is loaded into our minds (in the moment of inspiration) just because we find the solution easy to grasp. If the ‘flash of suspicion’ turns out to be the key to solving our problem, then the checking will flow easily; each question we ask will readily have an answer and all the pieces of the intellectual puzzle will start to fall into place.