Can a single molecule possess intelligence? The answer depends on what one means by intelligence. One tends to associate intelligence with what a human brain can do: perception, memory, thinking, problem solving, learning, innovation, creativity, etc. A personal digital computer can do some of these tasks and, in performing certain types of tasks, appears to surpass the human brain. But in terms of the more sophisticated aspects of intelligence such as pattern recognition in an ambiguous situation and creativity, a personal computer is no match even to a modest human brain.

A human can do much more than "walk and chew gum" at the same time. While a human being is listening, talking and thinking, the body parts are quietly performing the diverse tasks of providing oxygen and nutrients to individual cells and managing wastes generated in the process of nutrient utilization, fighting invading microorganisms and repairing damaged parts, all without the conscious intervention of a human being. In the computerese language, what the human body possesses is the capability of massively parallel distributed information processing.

In contrast to humans, a personal computer has "one-track" mind, capable of doing only one thing at a time. However, computers have such astonishing raw power and speed in terms of number crunching and chore management that they gave us the illusion of being capable of doing many things simultaneously and serving many users at the same time (time-sharing).

In reality, this cannot be farther from the truth. In fact, most of the machine parts are idling most of the time. Imagine the situation if governmental decisions at all levels required the personal approval of the President, one at a time. In view of this bureaucratic obstacle, known as the "von Neumann bottleneck," remedy has been sought in terms of parallel distributed processing. But conventional parallel processing is like setting up several agencies to handle similar and related tasks: they don't communicate well with each other and certainly not on a minute-by-minute basis. Thus, the need to explore radically different computer architectures so as to improve the overall intelligence of a computer is especially acute.

Another important determinant in pursuing improved computer performance is miniaturization. The increasing degree of miniaturization of the individual components (integrated circuits, or simply IC) results in increasing capabilities of each device component, increasing speed of operation of these devices, decreasing consumption of energy, decreasing size and weight of the finished product, and, last but not least, a decrease in price. It may seem that continuing miniaturization could make possible the implementation of better computer architectures, but physicists and engineers see the dead end as the physical limits imposed by quantum and thermal fluctuation phenomena are rapidly approached and the operation of these miniature devices becomes less and less reliable.

The recognition of these ultimate limits has inspired computer scientists to seek inspiration from biology. This is because a living organism operates with functional elements which are of molecular dimensions (about one thousandth of the size of a transistor) and which actually exploit quantum and thermal fluctuation phenomena. The hope of breaking the barrier of miniaturization seems to lie in the utilization of organic and biological materials, and the exploitation of their chemistry, and in the utilization of radically different computer architectures. This line of thinking has ushered in a new science and technology: molecular electronics, which is sometimes also referred to as nanotechnology.

The new science and technology calls for research and development in three related areas: novel materials, novel fabrication technology, and novel computer architectures. These three aspects actually go hand in hand; the progress in one area depends on that made in the other two areas. Novel computer architectures require materials with extraordinary properties to implement, and new techniques to assemble the required "hardware." The entire field of molecular electronics thus requires joint, integrated efforts of scientists and engineers of different backgrounds; it is truly an interdisciplinary and multidisciplinary endeavor.

As we shall see, the development of intelligent materials is fundamentally important not just for the goal of further device miniaturization but also for the evolution of the ultimate machine intelligence - the kind of intelligence that allows for learning and innovation, and allows for decision-making in a fuzzy situation when many conflicting requirements coexist.

Until recently, biomaterials have not been seriously considered for device construction because they were perceived as too fragile and not durable enough. A number of years ago, Nikolai Vsevolodov and his colleagues [1] in the Institute of Biological Physics, Pushchino, Russia, excited the biomedical research community by producing the first imaging device and microfilm made primarily of biological materials and entirely organic materials (named the "Biochrom" film). The key substance in this device is bacteriorhodopsin.

More recently, Robert Birge's group at Syracuse University [2] has devoted considerable efforts to developing a high-speed optical random access memory based on bacteriorhodopsin. In the remaining portion of this article, we shall examine some salient features of bacteriorhodopsin to illustrate the concept of intelligent materials. With the advent of genetic engineering, the intelligence of a biomolecule originally acquired through evolution can be further improved by breeding it in the laboratory in a much shorter time. Thus, molecular engineering will fast become one of the key technologies for the implementation of molecular electronics.

Bacteriorhodopsin is a purple-colored pigment occurring in natural abundance in the cell membrane of Halobacterium halobium [3]. The pigment derives its name from the resemblance of its chemical structure and properties to those of the eye pigment rhodopsin (the key material component in the "reusable microfilm" in our eye). Curiously enough, the natural function of this pigment is to convert sunlight into electrical and chemical energy for the benefit of the bacteria. In other words, these bacteria utilize an eye pigment for photosynthesis.

It is tempting to describe bacteriorhodopsin as a naturally occurring solar cell, but that would be an oversimplification as well as a misleading notion. This is because the operation of bacteriorhodopsin requires that the photopigment be incorporated in a membrane: it cannot work alone. Its operating principle is analogous to that of a hydroelectric power generator. The immediate source of energy in hydroelectric power generation is the energy stored as the difference in the water level across a dam; the difference in the water level is of course created by the solar energy through evaporation of water and subsequent rainfalls upstream to the dam.

In the case of the purple membrane, the membrane is the "dam" and bacteriorhodopsin is the "pump" that utilizes solar energy to move protons across the membrane, resulting in a difference in proton levels (a proton is the nucleus of a hydrogen atom). This difference in proton levels can then be utilized to turn on the ATP synthetase (the "turbine"), which synthesizes the universal energy currency in the world of living cells known as ATP (adenosine triphosphate).

The basic mechanism just mentioned was discovered almost two decades ago by Walther Stoeckenius (University of California, San Francisco) and Dieter Oesterhelt (Max Planck Institute, Munich) [4]. Obviously, the basic research on bacteriorhodopsin is very important in explaining the fundamental principle of photosynthesis and solar energy conversion.

My own interest in this subject stems from the discovery of a fast light-induced electric signal in the bacteriorhodopsin membrane by Mauricio Montal (University of California, San Diego) [5] and his colleagues. This electric signal bears a striking resemblance to a bioelectric signal known as the early receptor potential, which was discovered by Kenneth Brown (University of California, San Francisco) and Motohiko Murakami (Keio University School of Medicine, Tokyo) in monkey eyes in 1964 [6].

To make a long story short, we discovered that these electric signals can be generated from two types of light-induced charge separation. One type of charge separation is completely confined to the inside of the membrane when the chromophore (the light-absorbing part of the molecule, which is the attached vitamin A aldehyde) is excited by light. This process generates an electric signal which appears almost instantly after the photon is absorbed (i.e., the signal appears only with the delay of a few trillionths of a second at most). This signal is called the R1 component of the early receptor potential and the B1 component of the similar signal of bacterial origin.

The second type of charge separation occurs at the membrane surface. This process occurs when the membrane-bound pigment binds a proton from the adjacent intracellular water phase and when it releases a proton to the extracellular water environment. The latter process at the two membrane surfaces generates two electric signal components which we named B2 and B2', respectively. Our task was to distinguish between the two types of charge separation experimentally and to resolve (separate) the two components B2 and B2', of similar origins.

The experimental strategy is readily suggested by visualizing the bacteriorhodopsin molecule as the continental US, which borders on two oceans (two aqueous solutions on either side of the membrane). Imagine that protons are being imported, transported across the continent by highway, and exported from the opposite coast. Clearly, it takes a longer time to move the merchandise across the continent than to load it or to unload it at the docks. Therefore, warehouses are needed at both coasts to store the merchandise temporarily. Thus, in the short term, the loading and unloading activities at the two coasts can be considered independent of each other.

For example, a longshoremen's strike at one coast will not affect the loading/unloading activities at the other coast for a short period until the stock is running low or until the warehouse is overfilled. In contrast, the news about the strike gets transmitted across the continent via electronic media almost instantly.

Another related analogy suggests an experimental approach to resolve the B2 from the B2' component. That is, the weather forecast for Korea seldom apply to India. This simple conclusion is translated into a concept of local reaction conditions for a membrane-bound protein such as bacteriorhodopsin. The proton transfer reactions at each exposed surface of bacteriorhodopsin depend on the conditions of the adjacent water phase but not directly on those of the opposite water phase, at least on a short time scale.

Thus, if the proton concentration in the intracellular water phase is changed, the B2 component should be affected but not the B2' component. The situation will be similar with the role of B2 and B2' reversed, if the proton concentration in the extracellular water phase is changed. The B1 component, being generated in a region which is not exposed to water, is insensitive to the proton concentration in the two water phases on either side of the membrane. This was indeed the case when the suggested experiment was performed.

But why do the pH changes affect the two signal components? There are two possibilities: one is expected from the law of mass action and the other is the pH-induced change of the binding constant of the surface proton binding groups.

Getting back to the analogy of proton import/export across a continent, the proton import at the east coast will be affected by the availability of the merchandise from the supplier (higher proton concentration there means greater availability). Likewise, the proton export at the west coast will be affected by the demand of the merchandise from the trade partner (lower proton concentration in water means greater demand).

The other possibility, given unfavorable conditions in supply and demand, is to change the trading policy. Thus, if the supply of protons to the east coast is diminishing, the trade policy would call for increasing proton grabbing power there. Likewise, if the demand of protons at the west coast is diminishing, the trade negotiation could force the partner politically or persuade the partner diplomatically to buy them even if there is no real demand.

The experimental result showed the preponderance of the second possibility. Decreased intracellular proton concentration actually increases the proton uptake by bacteriorhodopsin, and increased extracellular proton concentration increases the proton release there, contrary to the expectation of the law of mass action. This finding demonstrates the intelligence of bacteriorhodopsin because this peculiar feature has a survival value.

Theoretically speaking, the proton pump would work even without the pH-dependent changes of the proton binding constant. However, like any mechanical pump, the efficiency of the pump can be diminished as a result of its own pumping action. The hydrostatic pressure built up by pumping water into a reservoir will render the pump less and less efficient. Likewise, proton pumping by the purple membrane will increase the extracellular proton concentration but will reduce the intracellular concentration. These resulting proton concentration changes will inhibit proton binding at the intracellular surface and inhibit proton release at the extracellular surface, and, thus, diminishes the pumping efficiency.

If, however, the resulting pH changes also causes the protein to readjust its proton binding constants, the difficulty can be partially relieved. It is as if a water pump could increase its horsepower when it encountered the back pressure built up in the reservoir. In this way, bacteriorhodopsin adjusts its functionality to meet the demands of environmental changes. This behavior is unique to macromolecules and is summed up in the concept of intelligent materials. According to the Minister of State for Science and Technology of Japan [7], intelligent materials are "substances/materials with the ability to respond to environmental conditions intelligently and manifest their functions."

The concept of intelligent materials was initially proposed to promote the idea of designing/synthesizing materials with a microstructure so that both sensors and actuators are embedded throughout [8]. For example, in the construction of airplane wings, the purpose is to allow the material to sense the changing loads or the condition of external stress as a result of damages, so as to adjust its mechanical characteristics in order to compensate for the changes.

Several scientific workshops on this topic have been held in the past few years. As the concept evolved through those workshops, the consensus of the definition of the concept began to diverge, and the concept became less well-defined. This trend is apparently the consequence of increased research activities in this areas. As new and different facets of intelligence in materials are being revealed, the definition must then be modified and generalized to accommodate the new findings.

Proteins are particularly suitable to be exploited as intelligent materials, as they have already acquired significant degrees of intelligence through evolution. Proteins could be further modified by genetic engineering to custom-tailor their functional properties to suit the intended technological applications. The use of a bacteriorhodopsin mutant as a reversible holographic medium (Paper by Norbert Hampp [9] in this Symposium) and the use of chemically modified bacteriorhodopsin for construction of "Biochrom" films (Papers by N. N. Vsevolodov and by A. B. Druzhko in this Symposium) are existing successful examples.

Richard Needleman at the Department of Biochemistry, Wayne State University, recently developed a new expression system for the purpose of engineering mutants of bacteriorhodopsin [10]. Preliminary results showed that by changing the genetic code of one out of 248 amino acid units, the pigment changes its electrical characteristics and photonic characteristics dramatically. This raises the hope of breeding mutants with the right type of intelligence for the intended design of the molecular device.

The development of intelligent materials is in keeping with the goal of miniaturization at the nanometer scale (one nanometer = one billionth of a meter) (nanotechnology). For example, by allowing sensor/processor/actuator capabilities to be packaged into a single molecule or a supramolecular cluster, avenues are open in the design of integrated information processing systems with massively parallel distributed processing capabilities. Thus, the progress made in the research of intelligent materials will pave the road towards the development of novel information processing systems so as to overcome the much-dreaded "von Neumann bottleneck" that characterizes conventional computers.