Christof Koch – August 25, 2021
The idea that consciousness is widespread is attractive to many for intellectual and, perhaps, also emotional reasons. But can it be tested? Surprisingly, perhaps it can.
Panpsychism is the belief that consciousness is found throughout the universe—not only in people and animals, but also in trees, plants, and bacteria. Panpsychists hold that some aspect of mind is present even in elementary particles. The idea that consciousness is widespread is attractive to many for intellectual and, perhaps, also emotional reasons. But can it be empirically tested? Surprisingly, perhaps it can. That’s because one of the most popular scientific theories of consciousness, integrated information theory (IIT), shares many—though not all—features of panpsychism.
As the American philosopher Thomas Nagel has argued, something is conscious if there is “something that it is like to be” that thing in the state that it is in. A human brain in a state of wakefulness feels like something specific.
IIT specifies a unique number, a system’s integrated information, labeled by the Greek letter φ (pronounced phi). If φ is zero, the system does not feel like anything; indeed, the system does not exist as a whole, as it is fully reducible to its constituent components. The larger φ, the more conscious a system is, and the more irreducible. Given an accurate and complete description of a system, IIT predicts both the quantity and the quality of its experience (if any). IIT predicts that because of the structure of the human brain, people have high values of φ, while animals have smaller (but positive) values and classical digital computers have almost none.
A person’s value of φ is not constant. It increases during early childhood with the development of the self and may decrease with onset of dementia and other cognitive impairments. φ will fluctuate during sleep, growing larger during dreams and smaller in deep, dreamless states.
IIT starts by identifying five true and essential properties of any and every conceivable conscious experience. For example, experiences are definite (exclusion). This means that an experience is not less than it is (experiencing only the sensation of the color blue but not the moving ocean that brought the color to mind), nor is it more than it is (say, experiencing the ocean while also being aware of the canopy of trees behind one’s back). In a second step, IIT derives five associated physical properties that any system—brain, computer, pine tree, sand dune—has to exhibit in order to feel like something. A “mechanism” in IIT is anything that has a causal role in a system; this could be a logical gate in a computer or a neuron in the brain. IIT says that consciousness arises only in systems of mechanisms that have a particular structure. To simplify somewhat, that structure must be maximally integrated—not accurately describable by breaking it into its constituent parts. It must also have cause-and-effect power upon itself, which is to say the current state of a given mechanism must constrain the future states of not only that particular mechanism, but the system as a whole.
Given a precise physical description of a system, the theory provides a way to calculate the φ of that system. The technical details of how this is done are complicated, but the upshot is that one can, in principle, objectively measure the φ of a system so long as one has such a precise description of it. (We can compute the φ of computers because, having built them, we understand them precisely. Computing the φ of a human brain is still an estimate.)
Debating the nature of consciousness might at first sound like an academic exercise, but it has real and important consequences.
Systems can be evaluated at different levels—one could measure the φ of a sugar-cube-size piece of my brain, or of my brain as a whole, or of me and you together. Similarly, one could measure the φ of a silicon atom, of a particular circuit on a microchip, or of an assemblage of microchips that make up a supercomputer. Consciousness, according to the theory, exists for systems for which φ is at a maximum. It exists for all such systems, and only for such systems.
The φ of my brain is bigger than the φ values of any of its parts, however one sets out to subdivide it. So I am conscious. But the φ of me and you together is less than my φ or your φ, so we are not “jointly” conscious. If, however, a future technology could create a dense communication hub between my brain and your brain, then such brain-bridging would create a single mind, distributed across four cortical hemispheres.
Conversely, the φ of a supercomputer is less than the φs of any of the circuits composing it, so a supercomputer—however large and powerful—is not conscious. The theory predicts that even if some deep-learning system could pass the Turing test, it would be a so-called “zombie”—simulating consciousness, but not actually conscious.
Like panpsychism, then, IIT considers consciousness an intrinsic, fundamental property of reality that is graded and most likely widespread in the tree of life, since any system with a non-zero amount of integrated information will feel like something. This does not imply that a bee feels obese or makes weekend plans. But a bee can feel a measure of happiness when returning pollen-laden in the sun to its hive. When a bee dies, it ceases to experience anything. Likewise, given the vast complexity of even a single cell, with millions of proteins interacting, it may feel a teeny-tiny bit like something.
Debating the nature of consciousness might at first sound like an academic exercise, but it has real and important consequences. Most obviously, it matters to how we think about people in vegetative states. Such patients may groan or otherwise move unprovoked but fail to respond to commands to signal in a purposeful manner by moving their eyes or nodding. Are they conscious minds, trapped in their damaged body, able to perceive but unable to respond? Or are they without consciousness?
Evaluating such patients for the presence of consciousness is tricky. IIT proponents have developed a procedure that can test for consciousness in an unresponsive person. First they set up a network of EEG electrodes that can measure electrical activity in the brain. Then they stimulate the brain with a gentle magnetic pulse, and record the echoes of that pulse. They can then calculate a mathematical measure of the complexity of those echoes, called a perturbational complexity index (PCI).
In healthy, conscious individuals—or in people who have brain damage but are clearly conscious—the PCI is always above a particular threshold. On the other hand, 100% of the time, if healthy people are asleep, their PCI is below that threshold (0.31). So it is reasonable to take PCI as a proxy for the presence of a conscious mind. If the PCI of someone in a persistent vegetative state is always measured to be below this threshold, we can with confidence say that this person is not covertly conscious.
This method is being investigated in a number of clinical centers across the US and Europe. Other tests seek to validate the predictions that IIT makes about the location and timing of the footprints of sensory consciousness in the brains of humans, nonhuman primates, and mice.
Unlike panpsychism, the startling claims of IIT can be empirically tested. If they hold up, science may have found a way to cut through a knot that has puzzled philosophers for as long as philosophy has existed.
Christof Koch is the chief scientist of the MindScope program at the Allen Institute for Brain Science in Seattle.
This story was part of our September 2021 issue