Connectionism, a model of neuronal functioning

Understanding how the brain works is one of the greatest challenges facing psychology. Hence the existence of diverse approaches and perspectives. In fact, after the appearance of cognitive psychology and the Turing machine there was a revolution in this field. From this moment on, the brain began to be seen as an information processor.

The first theory created to explain the functioning of the brain was the computational metaphor, but soon it began to have errors. Taking this situation into account, cognitive psychologists, with the intention of seeking new explanations, created a theory known as connectionism.

However, before explaining what connectionism is, it is important to understand cognitive psychology’s view of the brain. In this way, we will understand the implications and failures of the computational metaphor. For this reason, we will review the main aspects of this branch of psychology in the following section.

Cognitive psychology and the computational metaphor

Cognitive psychology understands the human brain as an information processor.. This means that it is a system that is capable of encoding data from its environment, modifying it and extracting new information from it. Furthermore, this new data is incorporated into the system in a continuum of inputs and outputs.

The computational metaphor explains that the brain is like a computer. Through a series of programmed algorithms, it transforms the inputs of information in a series of outputs. This may at first seem to make sense, since we can study some human behaviors that fit this model. Now, if we explore a little further, we begin to detect flaws in this perspective.

The most relevant errors are the speed with which we process information, the flexibility with which we act and the imprecision of our responses. If our brain had programmed algorithms we would have different types of responses: slower due to all the processing steps to be carried out, more rigid and much more precise than they are. In short, we would be like computers, and At first glance we see that this is not the case.

Although we can make attempts to adapt this theory to new evidence, changing the rigidity of the programmed algorithms for others that are more flexible and capable of learning, we would still continue to identify flaws in the computational metaphor. And this is where it comes in connectionism, a current that is simpler than the previous one, and that explains the brain’s information processing in a more satisfactory way.

What is connectionism?

Connectionism leaves behind computational algorithms and explains that Information is processed through activation propagation patterns. But what are these patterns? In simpler language, this means that when an input of information enters your brain, neurons begin to activate in a specific pattern, which will produce a specific output. This will form networks between neurons that will process information quickly and without the need for pre-programmed algorithms.

To understand this let’s take a simple example. Imagine that a person tells you to define what a dog is. When the word reaches your ear, automatically the set of neurons associated with it will be activated in your brain. The activation of this group of cells will spread to others with which it is connected, such as those related to words mammal, bark either hair. And this will activate a pattern in which these characteristics are included, which will lead you to define a dog as ‘a mammal with fur that barks’.

Properties of connectionist systems

According to this perspective, for these systems to function as the human brain seems to behave, they have to meet certain conditions. The basic properties that must follow are the following:

Activation spread. This means that neurons, when activated, influence those with which they are connected. This can occur by facilitating its activation or inhibiting it. In the previous example, the neurons of dog facilitate the mammal, but they inhibit those of reptile.Neural learning. Learning and experience affect the connections between neurons. Thus, if we see many dogs that have hair, the connections between the neurons related to both concepts will be strengthened. This would be the way in which the neural networks that help us with processing would be created.Parallel processing. Obviously this is not a serial process, the neurons are not activated one after another. The activation spreads in parallel between all neurons. And it is not necessary to process one activation pattern after another, multiple can occur at the same time. Thanks to this we are able to interpret a large amount of data at once, although there is a limit to our capacity.Neural networks. The system would be a large network of neurons grouped together, through inhibition and activation mechanisms. Within these networks there would also be the inputs of information and outputs behavioral. These groupings would represent the structured information that the brain has, and the activation patterns would be the way in which the processing of said information occurs.

Conclusions

This way of interpreting neuronal functioning not only seems very interesting, but also The studies around it seem fruitful. Nowadays, computer simulations of connectionist systems on memory and language have been created, which closely resemble human behavior. However, we cannot yet say that this is the exact way the brain works.

Furthermore, this model has not only helped contribute to the study of psychology in all its fields. Also We find multiple applications of these connectionist systems in computing. Above all, the theory has represented a great advance in studies about artificial intelligence.

To conclude, it is important to understand that The complexity of connectionism is much greater than that raised in this article. Here we can find a simplified version of what it really is, useful only as an approximation. If your curiosity about the topic has been aroused, do not hesitate to continue researching this theory and its implications.

You might be interested…