One of These is Not Like the Other
Artificial Neural Networks vs. Biological Neural Networks
One of the biggest misconceptions about artificial neural networks is that they have anything to do with actual (‘actual’ could be a massive overstatement) neural networks in the brain. Other than sharing the words “neural” and “networks” they do not.
Artificial neural networks are real, actually existing things
They can be manipulated, changed, intentionally made larger or smaller, slower or faster, etc. through programming and/or hardware changes.
Neural networks in the brain are a hypothetical (possibly theoretical) structural/functional component that may not actually exist
Or, if they do exist that may not be relevant to how the human brain (and other parts of the nervous system, depending on which version of the hypothesis/theory you ascribe to) actually works. They cannot be changed (intentionally at least) or if they can, through, for instance, surgical techniques or drugs, the outcomes of those surgeries or drug regimens cannot be entirely predictable/knowable. This is unlike the case of an artificial neural network in which the outcome of any programming/hardware changes can be entirely predictable.
Artificial neural networks are composed non living materials
To vastly oversimplify let’s call them silicon and wires.
Neural networks in the brain are (if they exist and are relevant structural/functional components) are composed of/emerge from living materials
Biological cells call neurons.
Rarely are two things which share the same name so different, and yet they are conflated as if one is interchangeable with the other. This is absolutely not the case and cannot ever be the case. Logically, cannot be the case, for the concepts exist in different logical categories. To suggest that they are the same is to commit a category error.
Can we say the two are analogous? Use the one to help illuminate or clarify some aspect of the other in a unique or interesting way? Sure, we are free to do that. At least there is no logical fallacy inherent in doing that. However, I would suggest that in this case it is particularly risky, as it implies a certain level of similarity between the two that is not justified. It also implies, though does not require, both things be real, actually existing things, which in this instance may not be the case for one of the two (biological neural networks).
