Neural networks simulate symbolic thinking

* This text was written by a columnist from TechWorld; Learn more at the end.

Scientists at the University of Texas have discovered a new way in which neural networks can mimic symbolic thinking. In the study, each neuron has a specialized function that can be related to more specific concepts, representing a fascinating discovery, opening the way for the union of deep learning and artificial thinking (AI).

The big difference with this new feature is that the team has removed “backpropagation”, which is the basis of many AI operations. This backpropagation was responsible for unlocking deep neural networks for effective training using descending gradient optimization methods and this, although complex, is not the best way to represent the human brain. This is the big key to this process: the imitation of human thinking; Neural networks are built on the basis of learning.

The project integrates ideas from philosophy of mind, psychology, neuroscience, and artificial intelligence to explore the way the human mind works. Research has shown that ENNs can discover new algorithms, similar to people. Moreover, it represents the potential for a great innovation in neurocognitive theory.

In contrast to previous AI research, the main focus of the University of Texas research has been to understand the brain from a biological perspective, in which neurons are organized in a hierarchical manner to reach complex conclusions. Thus, the forms of organization are myriad and the brain proves once again to be complex as philosophy of mind and neuroscience have always demonstrated.

Fabiano de Abreu Agrila Rodriguez Ph.D., neuroscientist, principal psychoanalyst, biologist, historian, anthropologist, trained in neuropsychology, neurolinguistics, neuroplasticity, artificial intelligence, neuroscience applied to learning, philosophy, journalism and professional training in clinical nutrition. He is the director of the Heraclitus Center for Research and Analysis. Head of the Department of Science and Technology at Logos University International, UniLogos.