Mexico City /
Understanding how the human brain works is one of the most important scientific projects of our time.
Today we know that the ability to think resides in the neocortex, where a structure of neurons is observed in the form of columns that are repeated identically throughout the thin sheet of gray matter. The cerebral cortex is where thoughts, judgments and decision making occur. It is that thin layer just a few millimeters thick that developed as new and distinct tissue in the most recent stage of mammalian evolution; It is also what inspired the invention of artificial neural networks in the eighties that is today recognized with the 2024 Nobel Prize in Physics.
In the cerebral cortex the uniform structure of cortical columns appears to form modular units that repeat identically. The vertical cylinders may well constitute elementary units and their uniform multiplicity has led specialists to think that there is a single algorithm that works to produce all the brain’s tasks. We have around 150 thousand cortical columns, all identical, although different regions of the cortex perform apparently different functions.
Many people have worked for a long time developing models of the brain. Their results have been contrasted with reality and although neural networks implemented in computers ended up being a useful tool for analysis and pattern recognition activities, they also represent that: a model that emulates the way in which information is stored in the brain, how simple logical tasks are performed and how decisions are made based on algorithms and machine learning.
This is also the relevance of the work that Geoffrey Hinton and John Hopfield carried out at the beginning of the eighties, although for now the applications of artificial intelligence are the ones that cause the most sensation. Its impact on all areas of human endeavor is already a fact.
The idea of modeling neurons goes back further in time. Were Warren McCulloch and Walter Pittsin 1943, the first to conceptualize neurons as “linear threshold units.” This is perhaps the antecedent of great developments, both in the biological knowledge of the brain and in the implementation of artificial networks, which could simulate the nervous system.
However, already at that time, the research approaches were divided, which developed in two directions: on the one hand, the study of the biological processes of the brain, and, on the other, applications for artificial intelligence.
This year’s Nobel Prize is awarded for the second aspect and is certainly an important gesture to promote discussion and reflection around artificial intelligence that is on the horizon and that promises great advances and predicts great dangers.
Geoffrey Hinton is not a physicist, but it is not the first time that a “layman”, not initiated in the field of physical sciences, has received the Nobel Prize in Physics. The engineers received it Guglielmo Marconi in 1909 for the development of wireless communication, Nils Gustaf Dalén in 1912 for the invention of a valve for his gas lamps, and Simon van der Meer in 1984 for the development of accelerator technology that allowed the construction of instruments that have been fundamental in the discovery of much of what is known in that area. Many others have received it for inventing or developing techniques or measuring instruments that would end up having an impact on the way we see nature. This year’s Nobel Prize can be placed very well in that category.
For his part, John Hopfield was always linked to materials physics. He is known in that union for his closeness to Philip Andersonlegendary condensed matter physicist, Nobel Prize winner and profound thinker in the philosophy of science.
Anderson went so far as to say that Hopfield was his hidden collaborator in the famous work “Anderson’s impurity model” in which an atom is studied contaminating the sea of delocalized electrons that represent a metal. The model was of unusual importance in understanding how these atoms affect the properties of materials. It was one of the most important works of the great physicist and without a doubt one of those that brought him closer to the Nobel Prize.
Hinton was inspired by the “spin glass” work that Anderson also developed. They are disordered magnetic systems where the spins do not align leading to unusual magnetic properties. Hinton seems to have begun to imagine artificial neural networks based on the way those systems are modeled in nature.
Next December, Geoffrey Hinton and John Hopfield will receive the highest honor awarded in science for more than a century. The work underlying the honors is one of the most popular. Artificial Intelligence has reached our WhatsApp and more and more people ask questions again and again about the most diverse topics. Artificial Intelligence based on neural networks is the great technological advance of the moment.
AQ
