What Is Consciousness?
Consciousness is everything you experience. It’s the melody stuck in your head, the sweet taste of chocolate mousse, the throbbing pain of a toothache, the boundless love for your child, and the bitter knowledge that, in the end, all feelings will fade away.
Since ancient times, the origin and nature of these experiences—sometimes called qualia—have been shrouded in mystery. Many modern analytic philosophers of mind, most notably Daniel Dennett of Tufts University, consider the existence of consciousness to be such an unbearable affront to their view of a meaningless material universe that they declare it an illusion. In their view, qualia either do not exist or are inaccessible to scientific investigation.
If this were true, this essay would be very short. I would only need to explain why you, I, and most other people are firmly convinced that we have feelings. However, when you have a tooth abscess, no argument, no matter how sophisticated, in favor of your pain being an illusion will lessen your suffering in the slightest. Since I find such desperate attempts to solve the mind-body problem unconvincing, let’s move on.
Most scientists accept consciousness as a given and seek to understand its relationship to the objective world described by science. Over a quarter-century ago, Francis Crick and I decided to set aside philosophical debates about consciousness (which have been ongoing since at least Aristotle’s time) and instead search for its physical traces. What happens in the highly sensitive part of the brain that gives rise to consciousness? By understanding this, we may get closer to solving a more fundamental problem.
Neural Correlates of Consciousness
We are searching for the neural correlates of consciousness (NCC), defined as the minimal neural mechanisms jointly sufficient for any specific conscious experience. For example, what must happen in your brain for you to feel a toothache? Do certain nerve cells need to vibrate at a magical frequency? Or is the activation of special “consciousness neurons” required? Where in the brain are these cells located?
The word “minimal” is key in defining NCC. After all, the entire brain could be considered a neural correlate of consciousness, since it constantly generates experiences. But we can be more precise about where consciousness resides. Take the spinal cord—a flexible tube of nervous tissue about a foot and a half long, running through the spine and containing about a billion nerve cells. An injury to the neck that severs the spinal cord from the brain paralyzes the limbs and torso, making them unable to feel or move, and deprives the person of control over their bowels and bladder. Yet, the full richness of life’s experiences remains for these tetraplegics: they see, hear, smell, react emotionally, and remember just as well as before the tragic incident that changed their lives.
Now consider the cerebellum—the “little brain” at the lower back of the brain, one of its most ancient autonomous regions. The cerebellum is involved in controlling movement, posture, and gait, and is essential for the smooth execution of complex motor sequences. Playing piano, typing, ice dancing, and rock climbing all involve the cerebellum. It contains Purkinje cells—the brain’s most remarkable neurons, with their fan-like dendrites and complex electrical dynamics. The cerebellum has about 69 billion neurons (most of them granule cells), four times more than the rest of the brain combined.
What happens to consciousness when the cerebellum is damaged by stroke or surgery? Very little! Patients with cerebellar damage report losing some skills, like fluid finger movements when playing piano or typing, but never complain of losing the ability to be conscious. Their vision, hearing, and other senses work fine, they retain self-control, remember the past, and plan for the future. Even being born without a cerebellum does not significantly affect consciousness.
The entire vast cerebellar apparatus is not involved in subjective experience. Why? Its wiring offers important clues. The cerebellum is characterized by uniformity and parallelism (much like batteries connected in parallel). Essentially, it’s a feedforward circuit: one group of neurons affects another, which in turn affects a third. There’s no complex feedback system responding to back-and-forth electrical impulses. (Given the time it takes for consciousness to form a sensation, most theorists conclude that feedback loops in the brain’s convoluted circuits must be involved.) Moreover, the cerebellum is functionally divided into hundreds or even thousands of independent computational modules, each working in parallel with different, non-overlapping inputs and outputs, controlling different motor or cognitive systems. These modules barely interact, and such interaction is necessary for consciousness.
The key lesson from our review of the spinal cord and cerebellum is that not all nervous tissue, when excited, creates sensations. The “genie” of consciousness has its own special lamp. As it turns out, the magic lamp is the gray matter of the brain’s surface—the famous cortex. The cerebral cortex is a laminated layer of intricately interacting regions of nervous tissue, about the size of a 14-inch pizza. Two highly wrinkled hemispheres, with their hundreds of millions of connections (white matter), are packed into the skull. All available evidence suggests that the neocortex generates sensations.
We can pinpoint the location of consciousness even more precisely. Consider experiments where different stimuli are presented to each eye—say, Donald Trump to the left eye and Hillary Clinton to the right. In your mind, a strange superposition of Trump and Clinton appears. For a few seconds, you see Trump, then he disappears and Clinton appears, then she disappears and Trump returns. This endless dance, called binocular rivalry, occurs because the brain receives ambiguous information and cannot decide who it’s dealing with—Trump or Clinton.
If you undergo this experiment inside an MRI scanner, which records your brain activity, researchers will find that several cortical areas are active. Collectively, these are called the “posterior hot zone”—the parietal, occipital, and temporal regions at the back of the cortex, which play a key role in tracking visual images. Interestingly, the primary visual cortex, which receives and transmits information from the eyes, does not signal what the subject sees. A similar division of labor seems to apply to the cortex for hearing and touch: the primary auditory and somatosensory cortices do not directly contribute to the content of auditory or somatosensory experiences. Consciousness perceives images—including those of Trump and Clinton—thanks to activity not in the primary, but in the posterior cortex, specifically its hot zone.
Two clinical sources provide a clearer picture of these causal relationships: electrical stimulation of cortical tissue and studies of patients who have lost parts of the cortex due to injury or disease. For example, before removing a brain tumor or epileptic focus, neurosurgeons map the functions of nearby cortical tissue by directly stimulating it with electrodes. Stimulating the posterior hot zone produces a wide variety of experiences: flashes of light, geometric shapes, distorted perception of faces, auditory and visual hallucinations, feelings of triviality or unreality, urges to move a limb, and so on. Stimulating the front of the cortex is a different story: it generally does not directly cause any experiences.
Another important source is neurological patients from the first half of the 20th century. Surgeons sometimes had to remove the entire belt of the prefrontal cortex to excise tumors or reduce epileptic seizures. Remarkably, the behavior of these patients was unremarkable. Loss of part of the frontal lobe did have negative effects: patients lost control over inappropriate emotions and actions, had motor problems, and showed involuntary repetition of actions and words. However, after surgery, their personality and IQ often improved, they lived for many years, and nothing suggested that radical removal of frontal tissue significantly affected their consciousness. In contrast, removing even small parts of the posterior hot zone led to the loss of entire classes of experiences: patients could not recognize faces, track movement, or perceive colors and space.
So, it appears that visual, auditory, and other images of our experiences are generated by regions of the posterior cortex. As far as we can tell, almost all experiences arise there. Why doesn’t most of the prefrontal cortex have the same direct relationship to subjective content as the posterior hot zone? Unfortunately, we don’t know. But that’s okay! Recent discoveries in neurobiology seem to have brought us closer to answering this question, and it’s thrilling.
Measuring Consciousness
There is a strong demand in medicine for a device that can clearly detect the presence or absence of consciousness in people who are in a weakened or completely incapacitated state. For example, anesthesia is used during surgery so that the patient doesn’t move or feel pain, maintains stable blood pressure, and doesn’t have unpleasant memories afterward. Unfortunately, this isn’t always achieved: every year, hundreds of patients remain partially conscious under anesthesia.
There are also patients with severe brain injuries from accidents, infections, or acute poisoning. They may live for years unable to speak or respond to verbal questions. Doctors often struggle to determine whether such patients have experiences. Imagine an astronaut in space who hears mission control trying to contact him but can’t respond because his radio only receives. The astronaut is lost to the world. Similarly, a patient whose damaged brain prevents communication is in the ultimate form of solitary confinement.
In the early 2000s, Giulio Tononi of the University of Wisconsin–Madison and Marcello Massimini, now at the University of Milan, Italy, first used the “zap and zip” technology to determine the presence or absence of consciousness. Here’s how it works: an induction coil is placed on the scalp and “zaps” the brain with a powerful magnetic pulse, inducing a brief electric current in nearby brain neurons. This current excites and suppresses partner cells, spreading through neighboring areas. Before fading, the waves of electric current travel throughout the cortex. Sensors on the head record this as an EEG, producing a movie showing how electrical impulses arise in different brain regions over time.
The resulting recordings don’t show a stereotypical pattern, but they’re not completely random either. The key point: the more predictable the spread of electrical waves, the more likely consciousness is absent. Using a data compression algorithm (like those used to archive computer files—“zip”), researchers quantified this relationship. They learned to assess the complexity of the brain’s response. In volunteers who were awake, the PCI—perturbational complexity index—ranged from 0.31 to 0.70, but dropped below 0.31 during deep sleep or anesthesia. Massimini and Tononi tested “zap and zip” on 48 awake patients with brain injuries who could still respond to questions. In all 48 cases, the PCI matched behavioral signs of consciousness.
They then tested the technology on 81 patients, some showing minimal signs of consciousness, others in a vegetative state. For the first group, consciousness was correctly detected in 36 out of 38 cases, with only two false negatives. Of 43 vegetative patients (all unresponsive), absence of consciousness was diagnosed in 34. The brains of the remaining nine responded as if consciousness were present. Perhaps these patients are among those who are conscious but unable to communicate, even with loved ones.
Researchers are now working to improve and standardize “zap and zip” for neurological patients and to extend its use to psychiatric and pediatric patients. Sooner or later, scientists will figure out which neural mechanisms generate experiences, which will greatly impact medicine and likely ease the suffering of many whose loved ones have severe brain injuries. However, fundamental questions remain. Why these neurons and not others? Why this frequency and not another? Where is the answer to how and why a highly organized piece of active matter gives rise to consciousness? This is truly an eternal mystery! After all, the brain, like any other organ, obeys the laws of physics—the same as the heart or liver. What’s the difference? What feature of the biophysics of brain tissue allows it to turn gray goo into the magnificent surround sound and vibrant color that make up our daily experiences?
Ultimately, we need a solid scientific theory of consciousness that can answer under what conditions a physical system—whether a complex network of neurons or silicon transistors—generates experiences. This theory should also answer other questions. Why do experiences have different qualities? Why does a clear blue sky feel so different from the screech of an out-of-tune violin? Do these differences in sensation serve a function, and if so, what is it? With a solid theory of consciousness, we could reliably predict when the systems we create will be conscious. Until then, discussions of machine consciousness are based only on intuition, which, as history shows, is not a reliable guide.
Leading Theories of Consciousness
Today, two main theories of consciousness are most prominent, and both are hotly debated. One is the Global Neuronal Workspace (GNW) theory, developed by psychologist Bernard J. Baars and neuroscientists Stanislas Dehaene and Jean-Pierre Changeux. This theory arose from the observation that when you act consciously, many areas of your brain have access to information about that action, but when you act unconsciously, information is localized within the specific sensorimotor system involved. For example, you type quickly and automatically, and if asked how you do it, you can’t explain—the information bypasses your consciousness, residing in neural circuits connecting your eyes to your fingers’ movements.
Toward a Fundamental Theory
GNW theory claims that consciousness arises from a specific type of information processing—the same kind used in early artificial intelligence systems, where all specialized programs had access to a shared, limited information store. Any data appearing on this “board” immediately became available to modules responsible for working memory, language, planning, or other higher functions. According to GNW, consciousness arises when sensory information is globally broadcast to many cognitive systems, which process it for speaking, remembering, recalling, or acting.
Since space on the board is limited, we can only be conscious of a small amount of information at any one time. Scientists believe the network of neurons transmitting these messages is in the frontal and parietal lobes. When this sparse data is broadcast in the network and globally accessible, the information is conscious—it enters the subject’s field of awareness. Although today’s machines haven’t reached this level of cognitive complexity, it’s only a matter of time. GNW theory predicts that future computers will be conscious.
The Integrated Information Theory (IIT), developed by Tononi and colleagues (including myself), is based on different observations: the experiences themselves. Every experience has certain essential properties. It is intrinsic, existing only for the subject as its “owner”; it is structured (a yellow cab brakes to avoid hitting a brown dog crossing the street); and it is specific—different from any other experience (just as a movie frame differs from all others). It is also unified and concrete. When you sit on a park bench on a warm sunny day watching children play, the different fragments of that experience—like the gentle breeze in your hair and the joy from your child’s laughter—cannot be separated without turning it into a different experience.
Tononi postulates that any complex mechanism with interconnected elements whose structure encodes a set of causal relationships possesses these properties and, therefore, a certain level of consciousness. Such a mechanism feels the world as external to itself. But if, as with the cerebellum, the mechanism lacks integration and complexity, it is not conscious. According to IIT, consciousness is an intrinsic causal power associated with such complex mechanisms, like the human brain.
IIT also introduces a measure of the complexity of the underlying structure—an irreducible number Φ (pronounced “phi”), which quantifies consciousness. If Φ is zero, the system does not feel itself. As Φ increases, the system’s intrinsic causal power—and thus its consciousness—increases. The human brain, with its vast and highly specific connectivity, has a very high Φ, indicating a high level of consciousness. IIT explains many neurobiological facts—for example, why the cerebellum is not directly involved in consciousness and why consciousness can be measured using “zap and zip” (which, albeit roughly, estimates Φ).
Among other things, IIT predicts that no matter how carefully a digital computer simulates the human brain, it will not be conscious, even if it learns to speak exactly like a person. Just as a simulation of a black hole’s gravity cannot warp real space-time around the computer running the simulation, programming consciousness cannot create a conscious computer. Consciousness cannot be computed; it must be built into the system’s structure.
Two challenges lie ahead. One is to further pinpoint the neural localization of consciousness using ever more advanced tools to observe and study the vast coalitions of highly diverse neurons in the brain. The Byzantine complexity of the central nervous system suggests this will take decades. The other challenge is to verify or falsify the two leading theories of consciousness—or perhaps to create a new, better theory from fragments of both, one that could provide a scientific solution to the main mystery of our existence: how a three-pound organ with the consistency of tofu gives rise to the feeling of life.