“…fluctuations, which always exist in real physical systems, can be used to solve decision-making [action direction] problems.”
“Decision-making—the ability to choose one path out of several options—is generally considered a cognitive ability possessed by biological systems, but not by physical objects. Now in a new study, researchers have shown that any rigid physical (i.e., non-living) object, such as an iron bar, is capable of decision-making by gaining information from its surroundings accompanied by physical fluctuations.”
“Decision-making is typically thought of as something done by intelligent living things and, in modern times, computers. But over the past several years, researchers have demonstrated that physical objects such as a metal bar, liquids, and lasers can also “make decisions” by responding to feedback from their environments. And they have shown that, in some cases, physical objects can potentially make decisions faster and more accurately than what both humans and computers are capable of.”
“In a new study, …lasers (are) capable of decision making and reinforcement learning, which is one of the major components of machine learning.”
“In our demonstration, we utilize the computational power inherent in physical phenomena…The computational power of physical phenomena is based on ‘infinite degrees of freedom,’ and its resulting ‘nonlocality of interactions’ and ‘fluctuations.’ It contains completely new computational principles. Such systems provide huge potential for our future intelligence-oriented society. We call such systems ‘natural Intelligence’ in contrast to artificial intelligence.”
“The researchers demonstrated the laser’s ability by having it solve the multi-armed bandit problem, which is a fundamental task in reinforcement learning. In this problem, the decision-maker plays various slot machines with different winning probabilities, and must find the slot machine with the highest winning probability in order to maximize its total reward. In this game, there is a tradeoff between spending time exploring different slot machines and making a quick decision: exploring may waste time, but if a decision is made too quickly, the best machine may be overlooked.”
“A key to the laser’s ability is combining laser chaos with a decision-making strategy known as “tug of war,” so-called because the decision-maker is constantly being “pulled” toward one slot machine or another, depending on the feedback it receives from its previous play.”
“The most important implication that we wish to claim is that the proposed scheme will provide a new perspective for understanding the information-processing principles of certain lower forms of life…These lower lifeforms exploit their underlying physics without needing any sophisticated neural systems.”
As the researchers explain in their study, the only requirement for a physical object to exhibit an efficient decision-making ability is that the object must be “volume-conserving.” Any rigid object, such as an iron bar, meets this requirement and therefore is subject to a volume conservation law. This means that, when exposed to fluctuations, the object may move slightly to the right or left, but its total volume is always conserved…The researchers explain that the bar’s movements occur due to physical fluctuations.
The researchers also showed that the TOW method implemented by physical objects can solve problems faster than other decision-making algorithms that solve similar problems. The scientists attribute the superior performance to the fact that the new method can update the probabilities on both slot machines even though it plays just one of them. This feature stems from the fact that the system knows the sum of the two reward probabilities in advance, unlike the other decision-making algorithms.
The researchers have already experimentally realized simple versions of a physical object that can make decisions using the TOW method in related work.
“The TOW is suited for physical implementations,” Kim said. “In fact, we have already implemented the TOW in quantum dots, single photons, and atomic switches.”
By showing that decision-making is not limited to biological systems, the new method has potential applications in artificial intelligence… “One example is a device that can make a directional change so as to maximize its light-absorption.” This ability is similar to how a young sunflower turns in the direction of the sun.