What is Hebbian Theory?

The Hebbian theory was first introduced in 1949 by a neuroscientist from Nova Scotia named Donald Hebb. The theory describes the synaptic plasticity mechanism, in which the repeated stimulation of the postsynaptic cells by their presynaptic counterparts results in increased synaptic efficacy. This theory is also known as the cell assembly theory, Hebb’s postulate, and Hebb’s rule.

It is frequently evoked when explaining some associative learning techniques, where activating cells simultaneously results in significant improvements in synaptic strength. This is known as Hebbian learning.

The Hebbian theory forms the foundation for the development of modern-day artificial neural networks. It centers on the idea that algorithms (machine rules) can be designed to update the weights (strengths) of artificial neural connections.

Artificial neuron networks are computational tools inspired by the intricate network of neurons in biological nervous systems. They consist of artificial neuron arrays joined together with different connection strengths, which evolve based on specific learning algorithms.

A major reason why the Hebbian theory is so appealing to engineers is the idea that they can get different results from complex artificial neural networks simply by changing their neural associations and weights. The principle offers a method of determining how these weights can be altered.

If two model neurons activate concurrently, then the weight between them increases. On the other hand, if they activate separately, then the weight between them decreases. The nodes that exhibit strong weightings are often either both negative or both positive simultaneously. Those with opposite polarity tend to have strong negative weights.

Examples of real-world applications of artificial neuron networks based on the Hebbian theory include:

• Artificial intelligence and machine learning
• Banking fraud prevention
• Relevant search results based on your online activities
• Voice-to-text technology for smart personal assistants like Siri, Alexa, and Google Assistant
• Social media facial-recognition, DeepText, and emoji auto-suggests