readings> Benjamin Libet's half second 

If there is one thing that seems certain about consciousness it is that it is immediate. We are aware of life's passing parade of sensations — and of our own thoughts, feelings and impulses — at the instant they happen. Yet as soon as it is accepted that the mind is the product of processes taking place within the brain, we introduce the possibility of delay. It must take time for nerve traffic to travel from the sense organs to the mapping areas of the brain.

It must then take more time for thoughts and feelings about these messages to propagate through the brain's maze of circuitry. If the processing is complex — as it certainly must be in humans — then these delays ought to measurable, and even noticeable with careful introspection.

As it happens, the conduction speed of nerves was the very problem that got the new science of psychology off the ground in Germany in the mid-1800s. It had been thought that nerves would act more or less instantaneously. But in the 1840s, the brilliant young German physiologist, Hermann Helmholtz, showed that nerve impulses actually travel surprisingly slowly.

Helmholtz applied electric shocks at a series of points down the spinal nerve leading to a frog's hind leg. From the slightly faster reactions nearer to the muscles, he worked out that its nerves must conduct impulses at about 60 miles per hour. Turning to humans, Helmholtz carried out much the same experiment by asking subjects to push a buzzer as soon as they felt him touching their legs at different points. From the slight changes in reaction times, Helmholtz calculated that human leg nerves must carry signals at about two hundred miles per hour—fast but nowhere near instant.

Modern research has since shown that human nerves actually conduct at a whole range of speeds, the rate depending on the size of the axon and also the thickness of a fatty insulation material, known as myelin, wrapped around it. The nervous system is like a road network with a few fast motorways and many winding country lanes. Large, heavily myelinated nerves — such as the muscle and sensory nerves which must run the length of the body — transmit their impulses at up to 240 miles per hour. But the congested network of small unmyelinated nerves which make up the bulk of our brain, work much more slowly. Once inside our heads, impulses tend to crawl along at between two and 20 miles per hour.

What such conduction speeds mean is that while consciousness might be fast, it cannot be instant. It takes a minimum of 10 to 20 milliseconds (thousandths of a second) for any sensory message to reach the brain. After that, the brain must spend yet more time in evolving a response.

The question of exactly how long was tackled by Wilhelm Wundt, Helmholtz's assistant at the University of Heidelberg. Wundt carried out mental chronometry experiments in which he tried to measure the speed of thought processes using reaction times. By noting how fast subjects responded to a buzzer or flash of light, Wundt could get an idea of how long it took to form an impression of various kinds of sensation.

Reaction time tests could also be used to measure higher level processes. Differences in the speed with which a subject could find an object among a clutter of distractions, or name the capital city of a country, were used to gauge how rapidly people could shift their attention or recall a memory.

 From this work, Wundt developed a theory of the mind based on what he called perceptions and apperceptions. Perception was an early forming, pre-aware response to the world that allowed us to perform reflex actions like hitting tennis balls and driving cars. In this phase, we can react quickly and unthinkingly — or if there is thought, it is of an impulsive or creative nature. Then after perception comes the fuller, more reflective, consciousness of apperception. This is awareness with the mind sharply focused on the meaning of a moment and our response properly supervised and perhaps even a little ponderous.

Excited that science seemed able to get inside the fine-grain structure of an instant of consciousness, the field of psychology blossomed. One of the most important early discoveries was that even the process of forming a sensory picture was smeared out over about a tenth of a second. Experiments showed that the brain tended to fuse together events like two noises or two flashes of light if they followed in quick succession.

This could cause powerful illusions. For example, when a pair of bulbs are lit in alternation, an observer sees a single spot of light bouncing back and forth through the air. This impression of motion, christened the phi effect, is familiar from theatre-front billboards which use rows of blinking lights to create moving displays. It also makes films and TV pictures possible. When a movie is projected at the rate of 24 frames a second, the information in each frame blurs to form a smoothly flowing experience. Televisions and computer displays work on the same principle, using an electron gun to paint the screen with a rapid succession of stills.

The amount of detail in a film picture means that the illusion begins to break down and the motion become jerky if there is much more than about 50 milliseconds between each frame. But simple shapes or flashes of light can seem like one moving object even when they are separated by as much as 100 or 200 milliseconds.

And the phi effect is not just confined to the sense of vision. Fusing also happens with our sense of touch. If our arm is tapped at three or four points in quick succession—each tap following within 50 to 100 milliseconds — it feels as if a single object is being trailed down towards our hand.

The brain can fill in with other properties besides motion. In one famous experiment, subjects were asked to watch a pair of light bulbs, one red, the other green, placed an inch apart. The two lights flashed within 50 milliseconds of each other. Not only did it appear to subjects that there was a single spot of light bouncing back and forth between the two bulbs, but as near as they could judge, the spot also seemed to switch colour halfway across.

This meant that the imaginary bouncing light appeared to turn from green to red, or red to green, before the bulb on the other side had even come on! The brain was filling in with the appropriate colour in advance of the event.

Such illusions are revealing because they catch the brain in the act of covering up for its own lagging processing. If consciousness takes a certain time to build, it is inevitable that sensations will arrive almost on top of each other while a moment of awareness is still under construction. The brain's job is to fit the jumble of input into a coherent story.

This means choices have to be made. If two separate waves of sensation appear similar enough — like two light bulb flashes or frames of film — then it will make sense for the brain to read them as fragmentary glimpses of the same rapidly moving object or fast-changing scene. The motion mapping areas of the visual cortex would come to tag the stationary stimuli with a velocity and direction. And in everyday life, of course, the brain would be almost bound to be doing the right thing. It is only in the laboratory that the brain's habit of assuming a connection between two very close events is made to look foolish.

So with just a few simple experiments, Helmholtz, Wundt and their followers revealed more about consciousness in a few years than armchair philosophy had in centuries. Buoyed by this quick success, psychology took off. By 1900, more than a hundred laboratories had sprung up around the world. But almost as soon, the emphasis on the timing of mental events disappeared.

Psychology came under the sway of behaviourism and the belief that people's claims about when or whether some event entered their awareness was much too subjective a form of evidence for the purposes of science.  The study of reaction times and perceptual responses never actually died out.

Yet as a field, psychophysics became a backwater. Research had a solidly practical feel and certainly psychophysicists did not try to use the detail they gleaned about sensory integration times or other aspects of mental processing as a launch pad for a more general assault on the problem of consciousness. Wundt's theories about awareness developing in distinct stages were left to gather dust.

Given this state of affairs, it is not hard to see why, when Benjamin Libet popped up in the 1960s to suggest consciousness might take as long as half a second to dawn, few people knew quite what to make of his claim.

stimulating the brain

Peering owlishly through thick glasses, a small, frail figure in an ancient pastel blue safari suit, Libet weighs his words with extreme care. An old school neuroscientist, he wants to talk only about established fact. Attempts to draw him into a speculative discussion about consciousness is met with the driest of chuckles. Yet for decades, Libet's own work has been the focus for puzzlement and even frank disbelief.

In the 1950s, Libet was a physiologist at the University of California School of Medicine in San Francisco doing thoroughly routine research on nerve transmission and neurotransmitters.  Then in 1958, a surgeon friend happened to set up neurosurgery unit at a nearby hospital. This doctor, Bertram Feinstein, was experimenting with a new operation to control the tremors, tics and spasms caused by degenerative brain conditions such as Parkinson's disease.

Feinstein's plan was to use electrodes to destroy small parts of the brain's motor system and so stop these tics at their source. Such lesioning was an extreme measure, but in an age before drug treatments became available, surgery was about the only option.

Feinstein's operation was a major procedure, but because the brain feels no pain, the surgery could be done under local anaesthetic with the patients fully conscious throughout. Feinstein and Libet realised that this gave them an almost unique opportunity to do experiments on a living human brain. With the consent of the patients, Libet could take a few minutes to stick in an extra electrode, give their brains a small tweak of current, and see what kind of experience the artificial stimulus would produce.

Libet was not actually the first to try this. In the 1930s and 1940s, Wilder Penfield, a Canadian neurosurgeon specialising in operations to treat epilepsy, had made a name for himself by probing the brains of his patients with an electrode. Penfield's original aim had been simply to locate the diseased regions acting as a focus for the seizures. Often before a fit, many epileptics experience an aura  — a characteristic sensation such as a clanging noise, flashing lights or a burning smell — that warns them an electrical storm is brewing. Penfield felt that if he could artificially induce such an aura, he would be able to pinpoint the area of brain needing removal.

But once he got started, Penfield became fascinated by the vivid responses he could provoke and began to probe the brains of his patients all over. When their primary sensory areas were being stimulated, the patients would report only something like a brief flash of colour or a chirruping noise. When the needle was pushed into higher level areas of the cortex, however, they often had sudden dream-like snatches of experience. One patient said he saw robbers coming at him with guns. Another heard a mother calling to a child.

Over the course of more than a thousand operations, Penfield gathered evidence that the cortex was much more organised than many people had believed. Indeed, several years before Hubel and Wiesel began their single cell recordings with cats, Penfield was already hinting that it had a hierarchical and topographical logic.

Although such experiments seemed an enviably direct way of tackling the connection between brains and conscious states, few other neurosurgeons felt inclined to follow Penfield's lead. The ethical issues aside, most simply did not have the background to become consciousness researchers. They were doctors not theorists. Nor were they willing to let some third party into their operating rooms, risking the health of their patients with a lot of fiddling about with electrodes. But fortunately for Libet, Feinstein felt the chance would be too good to miss.

The scope for experimentation was actually quite limited. Whereas Penfield had lifted the entire top of the skull off his patients, Feinstein's operation only needed a coin-sized hole over the motor centres. This meant there was little chance of exploring the higher levels of cortex function. But Libet still had access to the primary somatosensory cortex's mapping of touch sensations and also, by drilling a few inches down into the brain, the parts of the thalamus and brainstem which carried the nerves leading into the mapping.

Making a virtue of his restricted access, Libet planned a modest series of tests. A problem with Penfield's work had been that while it was wide-ranging, it had not asked some basic questions about exactly what form of electrical stimulus it took to produce a conscious reaction. Libet decided to examine this carefully.

Over the course of five years, Libet sat in on nearly a hundred operations. He quickly found that to get any sort of reaction from the patients, he had to use a pulsed current. A simple blast of electricity did nothing. But apart from that, the exact nature of the stimulation—the frequency of the pulses or the polarity of the current — appeared to matter surprisingly little. So long as the pulse train was persistent and gentle, the patients reported some remarkably realistic sensations, such as a furrowing of the skin, taps, jabs, brushes and occasionally a flush of warmth or coldness on some part of the body.

Occasionally, the sensations were very specific. Patients would exclaim that they could feel a drop of water trickling down the back of their hand or it felt as though they had a ball of cotton wool ball pinched between their fingers. But turning the current up above a few micro-amps did not make the sensations stronger. It just turned them into a harsh tingle, a feeling much like pins and needles.

As he continued his testing, Libet noticed something else. It normally took about half a second of pulses before the patients began to report anything. If the pulse train was stopped short at 460 or 470 milliseconds, then the patients did not seem to notice that the current had ever been turned on. But the instant Libet set the dial past the 500 millisecond mark, suddenly there was something to feel.

Of course, the result was not quite so clear cut. The half-second figure was only an average — for some patients, the time was characteristically either a little shorter or longer. Also, if a person happened not to be concentrating, the delay could stretch right out to a second or more.

And then the half second rule counted only for gentle current strengths. If the current was bumped up to the level where it would produce tingling rather than realistic sensations, the patients could notice a pulse train as short as a tenth of a second in duration. However Libet felt that he had stumbled upon something of fundamental significance. It was as if processing had to be kept going for a certain length of time to manufacture an organised conscious experience.

To prove that the result was not just some sort of freak effect of stimulating the cortex directly, Libet went on to stimulate the other brain centres along the input path for touch sensations. By pushing the electrode a little deeper into the brain, Libet could reach the thalamus, a nut-shaped organ through which all sensory lines to the cortex must pass. Then deeper still, the electrode would penetrate the brainstem, the first thickening of the spinal cord as it enters the brain.

When he turned on the current, again Libet found that it needed half a second for the patients to begin to feel a realistic sensation. And this time, the cortex was reacting to messages arriving over its normal nerve pathways. So if there was a half second delay, it was likely to be due to the time it took the cortex to evolve its response rather than  because a naked current was an unnatural form of excitation.

some EEG results

To discover more about what might be going on at the magic half second mark, Libet decided to do some recording on top of his stimulation experiments. Using the newly fashionable technique of evoked response potentials (ERP) recording, he began to check if perhaps there was something tell-tale about the activity of neurons at the moment everything got settled and came together.

 As mentioned, the evoked potential method was an advance on plain EEG recording. The problem with EEG electrodes were that they were unselective, picking up every last crackle of activity. They heard the roar of the brain rather than tracking individual neural events. But by recording the brain responding to exactly the same stimulus several hundred times in a row, then averaging the signal, the background activity would eventually cancel itself out. Libet had the bonus that he would be laying his electrodes directly on the brain, so the result would be exceptionally sharp, with none of the usual distortion or damping by the bone of the scalp.

As a stimulus, Libet used a quick electrical shock to the back of a patient's hand. He then recorded the response from the somatosensory cortex over the course of some 500 trials and averaged the readings. The result was a rollercoaster set of squiggles marking the rises and falls in electrical potential as the cortex went about its job of representing the faint buzz on the skin. And as Libet had hoped, the squiggles kept on going for a surprisingly long time given the fact that the patients claimed to be conscious of the shock at almost the instant it was delivered.

The trace actually showed the brains of the patients making a first surge of response very early — within 10 to 20 milliseconds. This was precisely the time needed for the sensory messages to travel the distance from the hand to the brain, so it seemed there was no particular delay in the arrival of the necessary information in the somatosensory mapping area. Then over the next tenth of a second or so, the mapping area saw its most dramatic convulsion.

About 50 milliseconds into the processing of the moment, the recording needle showed a sudden and massive reversal in polarity from positive to negative. There was a plateau and then at about 150 milliseconds, the electrical activity slumped back down again. Next came the second, more gentle but persistent, phase of activity. At about 200 milliseconds, the trace rose slowly in a shallow arc that extended until at least half a second.

And while the first big bump at a tenth of a second seemed a strictly local response, the second wave of activity represented a much more diffuse reaction. Part of the reason for its relatively shallowness was that Libet was picking up the weak and variable echoes of activity taking place in far flung corners of the brain.

This seemed striking evidence of an evolving mental state in which an initially fierce bout of mapping in the somatosensory cortex eventually gave way to a feedback-tuned, whole brain state of focused attention. The trouble was that plenty of other interpretations could also fit the data.

Perhaps the patients really were conscious of a buzzing on the back of their hands as soon as the first blip of nerve signals hit the cortex. The long-winded response might be just the result of the patients having thoughts about that awareness. Or it might even be some kind of neural housekeeping activity, such as a rebalancing of the brain's circuitry after a burst of firing.

The truth was that Libet had no way of telling. Indeed, he did not even have reasonable grounds for speculation given that virtually nothing was known about the hierarchical and feedback-driven nature of brain processing when he was doing his experiments in the 1960s. However the one fact of which he could be sure was that it took quite some time for the brains of Feinstein's patients to show a global level of response to a hand stimulus — no matter how quickly they might otherwise have felt they were aware of the jolt.

backward masking

When Libet published these results in the Journal of Neurophysiology in 1964, they attracted great attention, particularly from John Eccles, a maverick Australian researcher who had just won the Nobel prize for his own work on nerve synapses.

Eccles was a controversial figure because not only was he one of the very few neuroscientists keen to talk about consciousness, but he was also a passionate Cartesian dualist, pushing the scientific heresy that the brain was merely an inanimate vehicle for the conscious soul. Eccles liked Libet's results because they appeared to drive a wedge between events in the brain and the dawning of awareness.

For most scientists, Libet was saying consciousness arrives late and the appearance of simultaneity with events must be the result of some kind of clever illusion. But Eccles turned the logic round to suggest consciousness really was instant and that brain processing followed in its wake. The soul saw and then instructed the circuitry of the brain to reflect the same sensory state.

This championing by Eccles both helped and hindered Libet. Almost immediately, it got him invited to speak at a rare Vatican-sponsored conference on consciousness where he found himself rubbing shoulders with luminaries of the field such as Wilder Penfield and Vernon Mountcastle. But it also made him look too much like Eccles's protege, so that mainstream neuroscientists who could not stomach Eccles's mysticism also tended to deride any suggestion of a half second processing delay. The general opinion was that Libet would have to come up with a rather more convincing demonstration of the fact before anyone took it too seriously.

Returning to the lab, Libet struck upon the idea of exploiting backward masking, a strange sensory phenomenon that had been troubling psychophysicists for many years. Backward masking is an illusion in which a strong stimulus can block awareness of a weaker stimulus occurring just beforehand. For example, if a dim flash of light is followed about a tenth of a second later by a brighter flash in the same place, then the first flash will go unseen.

It is as if the mental processing of the first flash is interrupted before it can run to completion. The effect is strongest if the two events are separated by about 100 milliseconds, but can work for a gap as long as 200 milliseconds. Complex visual sensations, like pictures and printed words, can be blotted out. And even other kinds of sensations like sounds or touches. A tap on the arm can be masked by a second harder tap.

Making the phenomenon still more intriguing, psychophysicists had found that a third stimulus could be used to mask the masker. If a further flash followed the second, then the very first flash now entered consciousness and it was the middle flash that became masked!

While a little unnerving, it is easy to see that backward masking is merely the reverse of the phi effect. If the brain needs time to sort out an impression of the world, then its interpretations of a jumble of sensory events can always go two ways. With the phi effect, the brain is deciding that two successive flashes of light are best read as a single dancing dot—the most probable explanation given the chances of two very similar events occurring in such close proximity.

By the same token, when a weak flash of light is quickly followed by a second more positive flash coming from the same place, the brain would be safest to read this as just one event. Indeed, the best demonstrations of backward masking come when the second stimulus is not only more powerful, but actively clashes with the first — for instance, if a black disc is followed by a black ring encircling the space where it is just been. As polar opposites, the disc and the ring cannot be merged into a single sensory event and so it is not surprising that one should block the representation of the other.

Of course, such explanations only made sense if it was already agreed that brain processing was smeared out in time at least a little. They said that the brain tarried, taking a good tenth of a second of so to gather the sensory input needed to produce a settled moment of conscious awareness. But Libet saw that he could use backward masking to prove that the real processing time was nearer half a second.

His experiment was straightforward. He would shock a subject on the hand and then a split second later, use his electrode to zap the same place on the somatosensory mapping. Libet's reasoning was that if it actually took half a second inside the brain for the original hand sensation to be fully processed, then the late blast of stimulation might be able to interfere with the still budding experience even quite deep into its processing cycle. And this was what he found. The cortex stimulus was able to mask a skin stimulus as much as 300 milliseconds later—three times longer than normal.

This seemed powerful evidence that consciousness takes a relatively long time to build and that any experience of it being instantaneous must be a backdated illusion. We only feel that we are there as events happen. But the critics were still not swayed. Their reply was that a much simpler explanation for Libet's results was that the late cortex blast merely wiped away the memory of the earlier hand stimulus. His subjects would have experienced the buzz almost as it occurred, but then this fleeting awareness would have been lost amid the noisy clamour of the electrode generated activity.

Libet had an answer to that. In a follow-up experiment, he showed he could not only produce a late masking, but also a late enhancement of a skin sensation. This time he would exploit the phi effect to "top up" a hand twinge with a little extra cortex stimulation.

For enhancement to work, the electrode jolt now had to be weak enough so as not to seem like a competing event. The experiment also had to be set up in a way that psychologically encouraged subjects to read any extra cortex excitation as part of the initial skin stimulus. Libet told the subjects that their task was to judge the relative strength of two hand sensations coming about five seconds apart when, unknown to them, both skin stimuli would actually be exactly the same — the only difference being that on some trials, Libet would be throwing in a secret burst of cortex stimulation.

As expected, even when the cortex jolt trailed a buzz on the hand by some 300 or 400 milliseconds, there was a blurring of the input. The hand sensation suddenly felt a bit stronger. And this time there could be no question of a memory for an earlier experience being erased.

Indeed, like the phi experiment in which a leaping dot apparently changed colour before the bulb on the other side had time to switch on, Libet's subjects seemed to be experiencing the effect of something before it physically occurred! Awareness of the strengthened stimulus was dated to a moment before Libet flicked the electrode switch. So either there were spooky goings on, or the brain really did take about half a second to reach a settled interpretation of the world and then back-date the results to foster the illusion of simultaneity.

the freewill experiment

The masking and enhancement experiments were reported in the mid-1970s and even though the enhancement results were not circulated so widely, Libet felt he had made his case. However he found that a half second processing time was still a result that did not fit. Too many researchers took the view that one day someone would spot the flaw in Libet's approach and the whole problem would simply evaporate like so many other scientific oddities. And then sadly, in 1977, Libet's work with Feinstein was brought to an abrupt halt when Feinstein fell ill and eventually died.

Libet was faced with a choice of either admitting defeat and letting the issue drop or discovering some new way of continuing his experiments that did not involve brain surgery. Casting around, Libet found a route. But it was to lead him to become embroiled in a confused, often bitter, philosophical row over the nature of human freewill.

Hearing Libet was seeking a fresh direction, Eccles drew his attention to some studies carried out by Hans Kornhuber, a German pioneer of ERP recordings. Kornhuber had discovered that before people make a voluntary movement, such as tapping a toe or reaching out to grab something, there is a surprisingly long build-up of neural potential in the parts of the cortex responsible for motor control. The brain begins to stir as much as a second ahead of time. For complex actions, such as when a pianist gets set to play a run of notes, the build-up is even longer. Kornhuber called this evidence of mental preparation the readiness potential (RP).

Clearly Kornhuber's results raised questions for those who believed that it is consciousness which commands the body and that we make decisions, such as choosing the moment to flex a wrist or bend a finger, through an instantaneous act of will. Typically, psychophysicists did not make a big fuss about Kornhuber's results and what they might imply for an understanding of consciousness. But Libet saw that the readiness potential offered him a new line of attack.

Already well versed in the art of EEG recordings, all he needed was some way of timing the exact moment when subjects felt they had willed some movement. This could be done with almost millisecond precision by asking the subjects to note the position of a rapidly circling dot of light on an oscilloscope as their impulse struck.

Six college students were recruited and sat in a chair. A trial would start with a "get ready" bleep from a buzzer. The students would then wait with arm outstretched for the sudden urge to lift a finger or their wrists to overtake them.

Because Libet wanted to catch the very first moment of entry into consciousness of the idea to act, the students were asked to report any trials in which they detected prior deliberation — inner speech thoughts along the lines of: "Shall I go now? OK, come on then." — so they would not get included in the final results. Only quick and spontaneous impulses were to count.

After several hundred trials, the EEG data was averaged. Timing themselves, the students reported they first became aware of an intention to act about 200 milliseconds before they moved their hands. Yet the ERP trace showed that even with such a simple and impulsive decision, their motor planning areas had begun to stir even earlier, showing flickers of activity around 500 milliseconds before any movement was made. In other words, the decision to flex a finger or wrist must have started out subconsciously and only emerged as a conscious wish about a third of a second into the processing cycle.

This was Libet's simplest experiment, involving no brain surgery and just a month or two of work. Yet because the experiment questioned some basic philosophical assumptions about human freewill, it created easily the most controversy when published during the early 1980s. Libet's results seemed to imply we are not in charge of our own minds. By the time we know we want to do something even as trivial as lifting a finger, a decision will already have been made by low level brain processes outside our control. It was as if awareness was just there for the ride.

In an attempt to head off the worst of the confusion, Libet pointed out that although awareness for the gathering impulse may have kicked in with only 200 milliseconds to go, this still seemed to give the mind time to block the movement. The conscious self might not have initiated the decision to start flexing the hand — which surely, after all, was the definition of an impulsive action — but it was there in time to exercise a veto, so preserving the cherished notion of freewill.

But for many commentators, the feeling that consciousness was an instant phenomenon was just too insistent. It seemed impossible to accept that each moment of awareness might come with a hidden inner structure—that even states of thought and intention might have to evolve and that this evolution could then be concealed from our experiencing selves.

Ironically, the problem was quite the opposite for some of Libet's more informed critics. They could see that even a straight computational model of brain processing said that consciousness had to emerge by stages. But the stumbling block for them was the suddenness with which he seemed to be saying that consciousness clicked on.

If a state of awareness had to evolve over half a second — although most would bet that a tenth of a second was a more reasonable estimate of the processing time — then the dawning of subjective experience should be a gradual thing. The students doing the experiment should have progressed from being a little bit aware of a rising impulse to being very aware. So Libet's insistence that they look at a clock and mark "a time of awareness" was probably creating an artefact. He was not measuring when consciousness emerged, just when consciousness became sufficiently certain to be reported with some confidence.

The contrasting reactions showed that some people were happy with the idea of a sudden clicking on of awareness, some with a slow dawning of awareness, but no one liked Libet's combination of a slow yet sudden crystallisation of a state of subjective experience. Or to put it in terms of the battle between the digital and the dynamic, consciousness could be a binary state or it could be an evolving state, but how could you have an evolving system that suddenly went through some kind of late binary transition? In the 1980s, before complexity theory, mind scientists and philosophers just did not have an intellectual framework for thinking about such a state of affairs and their confusion showed.

preconscious processing

The truth was that the results of Libet's freewill experiment were never as unsettling as they sounded. The very first psychologists — mental chronometers like Wundt — had been quite at home with the idea that consciousness might come gradually and also by discernible stages. Indeed, Wundt's theory of perception followed by apperception seemed to fit Libet's work pretty neatly.

If sensing and acting were brain processes—states of information that had to evolve — then there would always be a time course, an arc of development. And it also seem logical that the early stages in the history of a moment would lack in strength, organisation, selectivity of focus, or some other critical quality, to make them conscious. Therefore they would be subconscious. Or rather, preconscious — a wave of activity on its way to becoming a settled state of consciousness.

But being preconscious would not mean that the early stages could do no profitable work in the economy of the brain. In fact there was every reason to believe that much of our mental lives are lived in a twilight world of not properly conscious impulses, inklings, automatisms and reflexive action.

The standard example of an intelligent, yet heavily automated, mental process is driving a car. When people are first learning to drive, each gear change or cornering manoeuvre demands their full attention. Their awareness can be so absorbed in the sheer mechanics of the task that they hardly have time to take in the road ahead, let alone notice the passing scenery. Yet after some months or years of driving, the skills become ingrained enough to be second nature.

People manage to drive home from work while chatting to a friend, listening to the news on the radio, day-dreaming, or planning a shopping list. Not only are they no longer conscious of the details of gear changes and steering, but they may become so caught up in non-driving thoughts that they are barely aware of having turned at the lights or made a tricky lane change.

This automation of some quite dangerous decision-making and skilled action is not the exception but the norm. Our brains seem designed so they will handle as much as possible at a subconscious level of awareness and execution, leaving focal consciousness to deal with tasks which are either particularly difficult or novel. Not surprisingly, when cognitive scientists thought about the distinction between these two levels of brain activity, they found it natural to view them as two separate modes of processing  and so likely to be inhabiting different pathways in the brain.

But if the brain evolves its states of processing, then automatic behaviour and peripheral levels of awareness would fit into story in quite a different way. There would be no specialised pathways but instead a whole brain reaction to the moment that began with many pockets of localised response and escalated to some more globally focused reaction. The brain would begin with a preliminary shake-down of all arriving information to try to deal with the routine business of the moment — stuff that is easy or habitual, like changing gears.

This would then clear the way for attention to be paid to whatever else was proving especially difficult or interesting. So there might be the appearance of two modes of processing, but really it would be two different depths of processing carried out across the same brain structure.

Already at the time of Libet's freewill experiment, a number of psychologists were beginning to have thoughts in this direction. In particular, Bernard Baars of the Wright Institute at the University of California, Berkeley, had developed what he called his global workspace model of consciousness.

A cuddly, bearded, bear of a man with honey eyes, Baars felt that the idea of an escalation from some local and preconscious level of processing to a broad-scale, whole brain, attack on a problem, could explain an awful lot about the mind. So Baars was one of the few who always believed that Libet's results made perfect sense.

However Baars could also see that there was a deceptive simplicity to Libet's finger-lifting task that was to blame for much of the controversy. People were taking the experiment to mean that lower level brain processes generate our thoughts and the conscious level mind arrives too late in the day to do more than supervise the results. Yet that was not the case at all.

Two points had to be remembered. Firstly, the students went into the experiment knowing what kind of action they were expected to produce. If they were at all uncertain, they would check with Libet, flexing a finger and asking, "You mean, go like that?”.  So they were primed with a conscious level image of what was meant to happen and had even warmed up their motor centres with rehearsals of the movement they might make. Much of the thinking needed to deal with what was in truth a fairly novel situation — taking part in an experiment with electrodes stuck to their heads and making an otherwise pointless motion — was done well in advance.

Then even as they were doing the experiment, they had to be alert to how well they were doing what was asked. Were they keeping their minds sufficiently blank to allow the required impulse to just "happen"? As Baars put it, there was always a conscious level context in place, framing whatever occurred.

The second point was that Libet was asking the students consciously to notice something that in ordinary circumstances their brains would ignore. The very essence of automatic behaviour is that we are confident enough about its execution to let it go unremarked. We do not have to note the fine detail of a gear change. The context will tell us when to change and habit will make the action smooth. But Libet wanted to make clear at exactly what point we can become globally aware of a bubbling up impulse. He was asking for the escalation of a minor brain event precisely because he wanted to expose the time course of such a processing step.

subliminal effects

With the freewill experiment, Libet was starting to get right inside the moment, seeing the cycle of development that leads up to a state of settled consciousness. But by now fast approaching retirement age, he only had time for one more experiment. What he ought to do was obvious.

The general reaction to his work had made it clear that most people still took his half second finding to mean that no worthwhile brain activity could be completed in under half a second, not that there might be successive stages of preconscious then conscious processing. So the next logical step would be to demonstrate that the brain could react in some subconscious way  to a lesser period of electrode stimulation — that while it might take half a second's worth of pulses to sustain the activity needed to produce the phantom sensation of a furrowing in the skin or water dripping down the back of the hand, perhaps less than half a second might still have some quiet effect on the brain.

This final test was difficult to set up as it meant going back to brain stimulation and Feinstein was not around to provide a ready supply of subjects. However a University of California neurosurgeon, Yoshio Hosobuchi, happened to be experimenting with implanted electrodes as a novel method of pain relief.

Certain kinds of severe pain — usually caused by damage to the spinal nerves — do not respond to drug treatment, but can be blocked by electrical stimulation of the part of the thalamus bringing in the pain messages. Patients have the electrodes permanently fixed in place then strap a battery pack to their belts so they can give themselves a quick jolt whenever the pain gets too bad. The fact that the electrodes were implanted allowed Libet to do an experiment without even going into the operating theatre.

Libet planned a variation on a standard subliminal perception test. In this experiment, a sub-threshold stimulus — one too weak to be consciously seen, such as the fleeting projection of some word or picture on a screen — is presented during one of two time intervals. Because the image flashes past much too rapidly to be noticed, the subjects will feel that they saw nothing during either of the test periods. But if told not to worry and just have a guess in which interval something might have occurred, more often than not, people will guess the right answer.

The success rate is certainly not high. Hits usually range between 60 and 70 percent — which is not much more than the 50 percent they would score due to chance anyway. Yet still, something must be happening at a subconscious level to tip the balance by even this much.

For the purposes of  Libet's experiment, the stimulus became a jolt delivered through the pain-control electrodes buried deep in each patient's thalamus. The subjects were sat down with a pair of lights to mark the two time intervals. Then during one of the intervals Libet would switch on the electrode for either longer or shorter than half a second.

When the pulse train lasted for more than half a second, all the subjects were quite definite they had experienced something, and when it was less, they were equally sure nothing had happened. But as Libet predicted, if asked simply to have a wild guess, they turned out to be right about 65 percent of the time. In some cases, as little as 150 milliseconds worth of thalamic stimulation could make a difference.

Plainly, the idea of a subliminal effect itself raises some thorny issues. There is the problem of how a slight eddy of sensory input—one too faint to be positively detected — can still cause enough disturbance in the brain occasionally to tilt a verbal response in the right direction. However, Libet's experiment at least confirmed that lesser periods of electrode stimulation had some slight impact. His half second result could not be taken to say that the brain stood idle and incapacitated while it waited for a state of consciousness to snap belatedly into place.

the problem of the gap

Libet's last paper, published in Brain in 1991, received a much more sympathetic hearing because the neuroscience world was changing rapidly. If nothing else, it helped simply that researchers were talking about consciousness again. For as soon as consciousness came out of the closet, a lot more attention started being paid to shades of consciousness. The difference between focal level processing and more automatic or preconscious levels of processing became a fashionable topic.

But the emergence of a dynamic approach to the mind also made the general tenor of Libet's results seem increasingly acceptable. As with Desimone's evidence of attention effects, a finding that was heresy one year became almost scientific orthodoxy the next. All it took was a subtle shift in the prevailing point of view.

Yet still, there was a grave sticking point with Libet's half second work. The trouble was that awareness simply does not feel delayed and Libet's talk about the final picture including some back-dating illusion sounded a little weak. There had to be a better explanation of how the brain papers over its processing gaps.  The answer, it turns out, lies in anticipation.

other evidence for a half second delay

Half second measures time to make a global shift: Two further lines of evidence for this come from the psychophysical phenomena of iconic memory and the attentional blink.

Iconic memory can be interpreted as a delayed "shrink" of the perceptual field. Instead of creating an escalated focus as fast as possible, the escalation is deliberately delayed, so telling something about the natural cycle time of the process. In a typical experiment, a slide containing three rows of four letters is flashed up on a screen for just 50 milliseconds. Subjects have to avoid reading any of the letters until a tone sounds to tell them which column they are mentally suppose to turn to and report.

As long as the signal is not delayed more than a third to half a second, their performance is almost perfect. They can inspect a still lingering iconic image and read off the chosen column. But this act of looking then wipes out all memory for the other two columns, as if the top-down act of focusing irrevocably sculpts what had been a flat and even spread of preconscious mapping.

See "The information available in brief visual presentations," G Sperling, Psychological Monographs 74:11 (1960). Note that it has been suggested that the cortex grabs the information direct from lingering retinal activity. See "Locus of short-term visual storage," B Sakitt, Science 190, p1318-1319 (1975).

The attentional blink is a more recently discovered effect which reveals the brain needs about half a second to recover from being "pinched up" to catch a perceptual event—it cannot focus sharply on a second event until it has had time to realign its anticipatory state.

In a typical experiment, subjects are shown a series of alphabet letters and asked not only to report a sighting of any x's, but also the occasional insertion of the number 4. When asked to spot x's or 4's alone, people can catch them all even at a presentation rate of eight a second. But with two clashing targets to report, a half second mental blindspot appears.

This suggests that the more rapid attentional performance is only possible because a single anticipatory framework remains in place. Once subjects have to swap between goal states, the full cycle time is exposed.

See "Temporary suppression of visual processing with an RSVP task: an attentional blink?" J Raymond, K Shapiro and KM Arnell, Journal of Experimental Psychology: Human Perception and Performance 18, p849-860 (1992).

For evidence that information is still being handled at a preconscious level during the blink, see "Word meanings can be accessed but not reported during the attentional blink," SJ Luck, EK Vogel and KL Shapiro, Nature 383, p616-618 (1996).

The dynamic nature of this effect is evident in the fact that lesser levels of goal conflict lead to apparently faster switching. See Perception and Communication by Donald Broadbent (London: Pergamon Press, 1958) for early discussions about attentional bottlenecks.

Note further that the question of attentional dwell time or the brain's processing frame rate has been a source of great confusion in cognitive psychology because it is assumed that as a computational device, the brain should have a single, fixed, processing cycle speed. Attention shifts should take a set time. The idea that the brain is always representing, and that sharp attention is the pinching up of this surface, helps explain the widely variable performance that is actually observed.

For the evolutionary nature of attentional state, see "Integrated field theory of consciousness," M Kinsbourne, in Consciousness in Contemporary Science, edited by Anthony Marcel and Edoardo Bisiach (Oxford: Oxford University Press, 1988).


Research by Helmholz and Wundt: For a general account, see Pioneers of Psychology by Raymond Fancher (New York: Norton, 1990). See also Selected Writings of Hermann von Helmholtz, edited by Russell Kahl (Middletown, Connecticut: Wesleyan University Press, 1971), Wilhelm Wundt and the Making of a Scientific Psychology, edited by Robert Rieber (New York: Plenum, 1980), and "On the speed of mental processes," FC Donders, Acta Psychologia 30, p412-431 (1969).

Nerves conduct at varying speeds: For a theoretical analysis, see "Effect of geometrical irregularities on propagation delay in axonal trees," Y Manor, C Koch and I Segev, Biophysical Journal 60, p1424-1437 (1991).

The phi effect: Originally noted by Sigmund Exner in Wundt's laboratory, phi or the apparent motion effect was explored systematically by the founder of Gestalt psychology, Max Wertheimer—"Experimentelle Studien über das Sehen von Bewegung," M Wertheimer, Zeitschrift für Psychologie 61, p161 (1912). For review, see "Apparent movement," SM Anstis, Handbook of Sensory Physiology 8, p655-673 (1978).

Fusing also happens with touch: Sensory Saltation by Frank Gelard and Carl Sherrick (New York: Lawrence Erlbaum Associates, 1975). Also "Anticipated stimuli across skin," MP Kilgard and MM Merzenich, Nature 373, p663 (1995).

Lights change colour halfway across: "Shape and colour in apparent motion," PA Kolers and M von Grünau, Vision Research 16, p329-335 (1976).

Libet's puzzling results: For a comprehensive collection of Libet's timing research, see Neurophysiology of Consciousness: Selected Papers and New Essays by Benjamin Libet (Boston, Massachusetts: Birkhäuser, 1993).

Penfield made name stimulating the brain: The Cerebral Cortex of Man: A Clinical Study of Localization of Function by Wilder Penfield and Theodore Rasmussen, (New York: Macmillan, 1950), and The Excitable Cortex in Conscious Man by Wilder Penfield (Springfield, Illinois: Thomas, 1958).

Libet's first results: "Production of threshold levels of conscious sensation by electrical stimulation of human somatosensory cortex," B Libet et al, Journal of Neurophysiology 27, p546-578 (1964).

Eccles a controversial figure: Eccles eventually came to feel that quantum effects might rule synapse activity—with the soul acting at each nerve ending to tip the balance of quantum probabilities in the direction it wanted. See The Principles of Design and Operation of the Brain, edited by John Eccles and Otto Creutzfeldt (Berlin: Springer, 1990), and Evolution of the Brain: Creation of the Self by John Eccles (London: Routledge, 1989).

Libet invited to speak at Vatican conference: Reported in Brain and Conscious Experience, edited by John Eccles (New York: Springer, 1966).

Libet seemed like Eccles's protege: Libet's actual feelings about the meaning of his own work were always rather cryptic. His standard position was that he talked experimental fact and did not indulge in empty theorising. Yet Libet often seemed to be lending to support to an Eccles-like interpretation of the half second lag by insisting that the onset of awareness was digitally sudden. Despite the results of his own subliminal experiments, Libet did not believe in preconsciousness or slow degrees of awareness. Consciousness was something that—when it finally arrived—was absolute. All neural activity beforehand was unconscious and qualia-free (personal communication). It was only with the publication of his collected papers in 1993—and at 77, too old to do further experiments himself—that Libet ventured into print to outline his own guess. Aligning himself more with Sperry ("Neurology and the mind-brain problem," RW Sperry, American Scientist 40, p291-312, 1952) than Eccles, Libet put forward a "mental force field" type theory. In an epilogue written for Neurophysiology of Consciousness: Selected Papers and New Essays (Libet, Birkhäuser), he suggested that brain circuits worked in an unconscious fashion. But through some unknown mechanism, their mass activity must lead to the emergence of a mental field—a unified subjective state. This mental field would appear after half a second and then could act in a general way to will the brain circuitry into carrying out specified actions. There would be an interaction at a global level, but not at the level of individual synapses as argued by Eccles. True to form, Libet said he was only willing to put forward this theory because he also had a test that could disprove it. The experiment would involve isolating a chunk of cortex (using a fine wire to undercut the white matter connections while leaving the blood supply intact), then stimulating the chunk with an electrode to see if it could "broadcast" the resulting sensation it was feeling to the rest of the brain. Also published as "A testable field theory of mind-brain interaction," B Libet, Journal of Consciousness Studies 1, p119-126 (

Phenomenon of backward masking: For review, see Visual Masking: An Integrative Approach by Bruno Breitmeyer (Oxford: Oxford University Press, 1984), "Backward masking," D Raab, Psychological Bulletin 60, p118-129 (1963), and "Semantic activation without conscious identification in dichotic listening, parafoveal vision and visual masking: a survey and appraisal," D Holender, Behavioral and Brain Sciences 9, p1-66 (1986).

Third stimulus could mask the masker: "Recovery of masked visual targets by inhibition of the masking stimulus," WN Dember and DG Purcell, Science 157, p1335-1337 (1967)

Libet's backward masking experiment: "Cortical and thalamic activation in conscious sensory experience," B Libet et al, in Neurophysiology Studied in Man, edited by George Somjen (Amsterdam: Excerpta Medica, 1972).

Libet's enhancement experiment: "Neuronal vs subjective timing for a conscious sensory experience," B Libet, in Cerebral Correlates of Conscious Experience, edited by Pierre Buser and Arlette Rougeul-Buser (Amsterdam: Elsevier/North-Holland Biomedical Press, 1978). These results were mentioned only in passing because Libet felt there were not quite enough subjects for them to be statistically bullet-proof. His intention was to add more subjects, but then his collaborator, Bertram Feinstein, died. Eventually Libet was persuaded the results could be published in full. See "Retroactive enhancement of a skin sensation by a delayed cortical stimulus in man: evidence for delay of a conscious sensory experience," Libet et al, Consciousness and Cognition 1, p367-375 (1992).

Kornhuber's readiness potential experiments: "Hirnpotentialänderungen bie Wilkürbewegungen und passiven Bewegungen des Menschen: Beritschaftspotential und reafferente Potentiale," HH Kornhuber and L Deeke, Pfügers Archiv für Gesamte Physiologie 284, p1-17 (1965). See also Psychophysiology: Human Behavior and Physiological Response by John Andreassi (Hove, England: Lawrence Erlbaum Associates, 1989). 

Libet's freewill experiment: "Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential): the unconscious initiation of a freely voluntary act," B Libet et al, Brain 106, p623-642 (1983). Experiment described again and its implications debated in: "Unconscious cerebral initiative and the role of conscious will in voluntary action," B Libet, Behavioral and Brain Sciences 8, p529-566 (1985).

Critics felt timing an urge created an artifact: See, for example, "Toward a psychophysics of intention," LE Marks, Behavioral and Brain Sciences 8, p547 (1985) and "Timing volition: questions of what and when about W," JL Ringo, Behavioral and Brain Sciences 8, p550 (1985).

Preconscious and automatic modes of processing: The term preconscious is chosen in preference to others, such as subconscious, subliminal, implicit—or worst of all, unconscious—because it has the useful connotation of something that is on its way to being conscious, or which could become escalated to consciousness in the right circumstances. The implication is that all processing shares a common path and what differs is the depth of processing. Preattentive processing is perhaps a more standard term, but sounds as if it applies only to the sensory half of the equation and not to the full arc of processing that allows us to execute acts like changing gears, or opening doors, habitually.
         For an excellent review of how the concept of preconsciousness has been dealt with in mind science, see "Is human information processing conscious?" M Velmans, Behavioral and Brain Sciences 14, p651-725 (1991). See also Preconscious Processing by Norman Dixon (Chichester, England: John Wiley, 1981), "The psychological unconscious and the self," JF Kihlstrom, in Experimental and Theoretical Studies of Consciousness: CIBA Foundation Symposium 174, edited by Gregory Bock and Joan March (Chichester, England: John Wiley, 1993), "Discrimination and learning without awareness:a methodological survey and evaluation," CW Eriksen, Psychological Review 67, p279-300 (1960), and Perception Without Awareness: Cognitive, Clinical and Social Perspectives, edited by Robert Bornstein and Thane Pittman (New York: Guilford Press, 1992).

Natural to see conscious and preconscious as separate modes: The reductionist hunt for some sharp dividing line between conscious and non-conscious brain processes has muddied the water tremendously, leading to much confusion in the literature. For recent examples of experiments that "prove" a brain mechanism difference, see "Classical conditioning and brain systems: the role of awareness," RE Clark and LR Squire, Science 280, p77-81 (1998), "Brain regions responsive to novelty in the absence of awareness," GS Berns, JD Cohen and MA Mintun, Science 276, p1272-1275 (1997), and "Dissociation of the neural correlates of implicit and explicit memory," MD Rugg et al, Nature 392, p595-598 (1998). For a review, see "Characteristics of dissociable human learning systems," DR Shanks and MF St John, Behavioral and Brain Sciences 17, p367-447 (1994)
Baars's global workspace model: See A Cognitive Theory of Consciousness by Bernard Baars (Cambridge: Cambridge University Press, 1988) and In The Theater of Consciousness by Bernard Baars (New York: Oxford University Press, 1998). For others with similar models of an escalation of initially preconscious processing, see Cognition and Reality: Principles and Implications of Cognitive Psychology by Ulric Neisser (New York: WH Freeman, 1976), Chronometric Explorations of Mind by Michael Posner (Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1978), and "Integrated cortical field model of consciousness," M Kinsbourne, in Experimental and Theoretical Studies of Consciousness: CIBA Foundation Symposium 174 (Bock and March, John Wiley).
Libet's subjects knew what action to produce: The point that the students went into the freewill experiment with a consciously-held context was obvious to many commentators. See, for example, "Problems with the psychophysics of intention," BG Breitimeyer, Behavioral and Brain Sciences 8, p539-540 (1985), and "Conscious intention is a mental fiat," E Scheerer, Behavioral and Brain Sciences 8, p552-553 (1985).

Libet's experiment showing subconscious reactions: "Control of the transition from sensory detection to sensory awareness in man by the duration of a thalamic stimulus: the cerebral "time-on" factor," B Libet et al, Brain 114, p1731-1757 (1991).

Variation on a standard subliminal test: "Conscious and unconscious perception: experiments on visual masking and word recognition," AJ Marcel, Cognitive Psychology 15, p197-237 (1983).

home> back to readings