Introduction to Neural Codes

Marcello Barbieri

The nervous system is made of three types of neurons: (1) the sensory neurons carry to the brain the signals produced by the sense organs, (2) the motor neurons deliver signals from the brain to the motor organs (muscles and glands), and (3) the intermediate neurons provide a bridge between them. In some cases the sensory neurons are directly connected to the motor neurons, thus forming a reflex arch, a system that provides a quick stimulus-response reaction known as a reflex.
The first nervous systems were probably little more that a collection of reflex arches, and it is likely that the first intermediate neurons came into being as physical extensions of those arches that provided a useful trait-de-union between sensory neurons and motor neurons. Once in existence, however, they could start exploring new possibilities. The behaviour of an animal must take into account a variety of clues from the environment, and to this purpose it is useful that a motor organ receives signals from many sense organs and that a sense organ delivers signals to many motor organs. This requires multi-gated connections between sensory inputs and motor outputs and it was probably the development of such connections that fuelled the evolution of the intermediate brain.
In addition to transmitting electrical signals, however, the intermediate neurons could do something else. They could start processing the signals, and that opened up a whole new world of possibilities. In practice, the processing evolved in two great directions and produced two very different outcomes. One was the appearance of neural networks that formed feedback systems and provided a sort of ‘automatic pilot’ for any physiological function. The other was the generation of feelings and instincts.
The first processing was totally unconscious and was carried out by a component of the primitive brain that can be referred to as the cybernetic brain. The second processing was adopted by another major component of the primitive brain that can be referred to as the instinctive brain. The primitive brain, in short, evolved from a primitive reflex-arch system and was made of two distinct types of neural processing, one completely unconscious and the second based on feelings and instincts (Barbieri, 2011).

Three Modelling Systems

The products of the brain are normally referred to as feelings, instincts, sensations, perceptions, emotions and so on, but in some cases it is convenient to use a term that applies to all of them, and say that they are all brain models. A visual image, for example is a model of the information delivered by the retina, and a feeling of hunger is a model obtained by processing the signals sent by the sense detectors of the digestive apparatus.
The brain can be described in this way as a modelling system, a concept that has been popularized by Thomas Sebeok and has acquired an increasing importance in semiotics (Sebeok and Danesi, 2000). The term was actually coined by Juri Lotman, who described language as the ‘primary modelling system’ of our species (Lotman, 1991), but Sebeok underlined that language evolved from animal systems, and should be regarded as a secondary, if not a tertiary, modelling system. The distinction between primary, secondary and tertiary modelling systems has become a matter of some controversy, so it is important to be clear about it. Here we use those terms to indicate the modelling systems that appeared at three different stages of evolution and were the results of three different types of brain processing.

(1) The First modelling system

This is the system that appeared when the primitive brain managed to produce feelings and instincts. These entities can be divided into two major classes because the sense organs transmit information either about the outside world or about the interior of the body. The first modelling system consists therefore of two types of models, one that represents the environment and one that carries information about the body. Jakob von Uexküll (1909) called these two worlds Umwelt and Innenwelt, names that express very well the idea that every animal lives in two distinct subjective universes. We can say therefore that Innenwelt is the model of the internal body built by the instinctive brain, and that Umwelt is the model of the external world built by the cybernetic brain of an animal. The brain as we know it – the feeling brain – came into being when the primordial brain split into instinctive brain and cybernetic brain, and these gave origin to the first modelling system of all triploblastic animals (vertebrates and invertebrates).

(2) The Second modelling system

Some animals (like snakes) stop chasing a prey when this disappears from sight, whereas others (like mammals) deduce that the prey has temporarily been hidden by an obstacle and continue chasing it. Some can even learn to follow the footsteps of a prey, which reveals a still higher degree of abstraction. This ability to ‘interpret’ the signals from the environment, is based on a new type of neural processing that makes use of memory, learning, the ability to ‘generalize’ and the faculty to ‘jump to conclusions’ (abduction). All together these operations vastly expand the potentialities of the feeling brain and represents a second modelling system, a system that appeared when part of the cybernetic brain became an ‘interpretive brain’.

(3) The Third modelling system

The last major novelty in brain’s history was the origin of language in our species, and that too required a new macroevolution, a new type of neural processing that went far beyond the reach of the interpretive brain because it was capable of coping with mental operations involving symbols. This is why we can say that language truly represents a third modelling system.
There have been, in conclusion, three major transitions in the evolution of the brain and each of them gave origin to a new type of neural processing and to a new modelling system. But how did those major transitions take place?

The neural code

Feelings and sensations are produced in the brain from signals that come from the sense organs. Mechanical stimuli, for example, are detected at the surface of the body by pressure-receptors, and are transformed into tactile sensations in the brain. Rats have mechano-receptors on the tip of their whiskers while we have them on the tip of our fingers, and there is no doubt that we and rats explore the world in different ways, but do we use different transformation mechanisms? The experimental evidence is that we don’t. The physiological processes that transform mechanical stimuli into tactile sensations seem to be the same in all animals (Nicolelis and Ribeiro, 2006).
What is most important is that this appears to be true for all sense organs. Countless experiments on animal brains suggest that the transformations of sense stimuli in neural sensations take place according to universal mechanisms, and there does seem to be a rational explanation for that. The sense organs arise from the basic histological tissues of the body, and these tissues (epithelial, connective, muscular and nervous tissues) are the same in all triploblastic animals. All signals that are sent to the brain, in other words, come from organs produced by a limited number of universal tissues, and that does make it plausible that they represent a limited number of universal inputs. But do we also have a limited number of universal outputs?
The neural correlates of the sense organs (sensations) can be recognized by the actions that they produce, and there is ample evidence that all triploblastic animals have the same basic instincts. They all have the imperative to survive and to reproduce. They all seem to experience hunger and thirst, fear and aggression, and they are all capable of reacting to stimuli such as light, sound and smells. The neural correlates of the basic histological tissues, in short, are associated with the basic animal instincts and these appear to be virtually the same in all triploblastic animals.
What we observe, in conclusion, is a universal set of basic histological tissues on one side, a universal set of basic animal instincts on the other side, and a set of neural transformation processes in between. The most parsimonious explanation is that the neural processes in between are also a universal set of operations. And since there is no necessary physical link between sense organs and feelings, we can conclude that the bridge between them can only be the result of a virtually universal neural code. All of which suggests that there has been a universal neural code at the origin of mind just as there has been as a universal genetic code at the origin of life.

The evolution of the brain

The origin of mind was the origin of the primary modelling system that allows animals to build a model of their body (an Innenwelt) and a model of their environment (an Umwelt). That was a true macroevolution because it brought absolute novelties into existence, and the key event that made it possible was the origin of a virtually universal neural code.
The second and the third modelling systems were also the results of macroevolutionary events, and this suggests that they were based on new types of neural codes, because only codes can bring absolute novelties into existence. On this point, however, it is necessary to take into account the full complexity of the systems.
The second modelling system allows animals to interpret the signals from the environment, but a process of interpretation cannot be reduced to a single code because it requires a variety of faculties such as memory, learning, the ability to generalize (categorization) and the ability to ‘jump to conclusions’ (abduction). These faculties can be studied with models obtained from artificial neural networks (Hopfield, 1982; Rumelhart and McClelland, 1986; Holland 1992) because it has been shown that these networks are capable not only of learning and memory but also of more complex operations like categorization and abduction (Kohonen, 1984). These faculties are based on many codes in artificial systems and this strongly suggests that the same is true in living systems. It is likely therefore that the origin of interpretation was not based on a single code but on a complex system of neural codes, and the same is probably true also for the origin of language.
The evolution of the brain, in conclusion, took place with a whole stream of many neural codes, and was characterized by three major transitions that gave origin to three distinct modeling systems.


Barbieri, M. (2011). Origin and Evolution of the Brain. Biosemiotics 4(3), 369-399.
Holland J.A. (1992). Adaptation in Natural and Artificial Systems. MIT Press, Cambridge MA  
Hopfield J.J. (1982). Neural networks and physical systems with emergent collective computational abilities. Proceedings of the National Academy of Sciences, USA, 79, 2554-58.  
Kohonen T. (1984) Self-Organization and Associative Memory. Springer, New York.  
Lotman J. (1991) Universe of the Mind: A Semiotic Theory of Culture. Indiana University Press, Bloomington.  
Nicolelis M. and Ribeiro S. (2006) Seeking the Neural Code. Scientific American, 295, 70-77.
Rumelhart D.E. and McClelland J.L. (1986). Parallel Distributed Processing: Explorations in the Microstructure of Cognition. MIT Press, Cambridge MA.  
Sebeok T.A. and Danesi M. (2000) The Forms of Meaning: Modeling Systems Theory and Semiotic Analysis. Mouton de Gruyter, Berlin.  
Uexküll J. von (1909) Umwelt und Innenwelt der Tiere. Julius Springer, Berlin.