I saw this on a thread on /lit/ and was thinking. Can neural networks be used as an argument for the plausibility of materialistic consciousness?

Neural networks can seemingly hold some kind of "knowledge", "inner representation", whatever you want to call it, that allows them to make "decisions" (classifications, for instance), while only using combinations of very simple components.
We are only able to really fully interpret, in human terms, the input and output layers. Even though we could know all the variables of the hidden layers (the activation in each neuron and all the weights), we don't understand exactly how it's holding the "knowledge" to make "decisions" in a human understandable form.

What this shows us, though, is that without using anything but simple fully material components, it's possible to create representations of things that can be used to allow for decision making.

Before anyone gets angry about comparing Neural Networks and brains, that's not the point. The point is, could we then imagine that neurons might be able to hold representations and allow us to make decisions while purely using material/physical components? Does this reasoning make sense?