2009 : WHAT WILL CHANGE EVERYTHING?

[ print ]

Cognitive Scientist, UC, Irvine; Author, Visual Intelligence
THE LAPTOP QUANTUM COMPUTER

Everything will change with the advent of the laptop quantum computer (QC). The transition from PCs to QCs will not merely continue the doubling of computing power, in accord with Moore's Law. It will induce a paradigm shift, both in the power of computing (at least for certain problems) and in the conceptual frameworks we use to understand computation, intelligence, neuroscience, social interactions, and sensory perception.

Today's PCs depend, of course, on quantum mechanics for their proper operation. But their computations do not exploit two computational resources unique to quantum theory: superposition and entanglement. To call them computational resources is already a major conceptual shift. Until recently, superposition and entanglement have been regarded primarily as mathematically well-defined by psychologically incomprehensible oddities of the quantum world—fodder for interminable and apparently unfruitful philosophical debate. But they turn out to be more than idle curiosities. They are bona fide computational resources that can solve certain problems that are intractable with classical computers. The best known example is Peter Shor's quantum algorithm which can, in principle, break encryptions that are impenetrable to classical algorithms.

The issue is the "in principle" part. Quantum theory is well established and quantum computation, although a relatively young discipline, has an impressive array of algorithms that can in principle run circles around classical algorithms on several important problems. But what about in practice? Not yet, and not by a long shot. There are formidable materials-science problems that must be solved—such as instantiating quantum bits (qubits) and quantum gates, and avoiding an unwanted noise called decoherence—before the promise of quantum computation can be fulfilled by tangible quantum computers. Many experts bet the problems can't adequately be solved. I think this bet is premature. We will have laptop QCs, and they will transform our world.

When laptop QCs become commonplace, they will naturally lead us to rethink the notion of intelligence. At present, intelligence is modeled by computations, sometimes simple and sometimes complex, that allow a system to learn, often by interacting with its environment, how to plan, reason, generalize and act to achieve goals. The computations might be serial or parallel, but they have heretofore been taken to be classical.

One hallmark of a classical computation is that it can be traced, i.e., one can in principle observe the states of all the variables at each step of the computation. This is helpful for debugging. But one hallmark of quantum computations is that they cannot in general be traced. Once the qubits have been initialized and the computation started, you cannot observe intermediate stages of the computation without destroying it. You aren't allowed to peak at a quantum computation while it is in progress.

The full horsepower of a quantum computation is only unleashed when, so to speak, you don't look. This is jarring. It clashes with our classical way of thinking about computation. It also clashes with our classical notion of intelligence. In the quantum realm, intelligence happens when you don't look. Insist on looking, and you destroy this intelligence. We will be forced to reconsider what we mean by intelligence in light of quantum computation. In the process we might find new conceptual tools for understanding those creative insights that seem to come from the blue, i.e., whose origin and development can't seem to be traced.

Laptop QCs will make us rethink neuroscience. A few decades ago we peered inside brains and saw complex telephone switch boards. Now we peer inside brains and see complex classical computations, both serial and parallel. What will see once we have thoroughly absorbed the mind set of quantum computation? Some say we will still find only classical computations, because the brain and its neurons are too massive for quantum effects to survive. But evolution by natural selection leads to surprising adaptations, and there might in fact be selective pressures toward quantum computations.

One case in point arises in a classic problem of social interaction: the prisoner's dilemma. In one version of this dilemma, someone yells "FIre!" in a crowded theater. Each person in the crowd has a choice. They can cooperate with everyone else, by exiting in turn in an orderly fashion. Or they can defect, and bolt for the exit. Everyone cooperating would be best for the whole crowd; it is a so-called Pareto optimal solution. But defecting is best for each individual; it is a so-called Nash equilibrium.

What happens is that everyone defects, and the crowd as a whole suffers. But this problem of the prisoner's dilemma, viz., that the Nash equilibrium is not Pareto optimal, is an artifact of the classical computational approach to the dilemma. There are quantum strategies, involving superpositions of cooperation and defection, for which the Nash equilibrium is Pareto optimal. In other words, the prisoner's dilemma can be resolved, and the crowd as a whole needn't suffer if quantum strategies are available. If the prisoner's dilemma is played out in an evolutionary context, there are quantum strategies that drive all classical strategies to extinction. This is suggestive. Could there be selective pressures that built quantum strategies into our nervous systems, and into our social interactions? Do such strategies provide an alternative way to rethink the notion of altruism, perhaps as a superposition of cooperation and defection?

Laptop QCs will alter our view of sensory perception. Superposition seems to be telling us that our sensory representations, which carve the world into discrete objects with properties such as position and momentum, simply are an inadequate description of reality: No definite position or momentum can be ascribed to, say, an electron when it is not being observed. Entanglement seems to be telling us that the very act of carving the world into discrete objects is an inadequate description of reality: Two electrons, billions of light years apart in our sensory representations, are in fact intimately and instantly linked as a single entity.

When superposition and entanglement cease to be abstract curiosities, and become computational resources indispensable to the function of our laptops, they will transform our understanding of perception and of the relation between perception and reality.