Classical Theism, Philosophy, and Religion Forum / Connectionism

**Posted by iwpoe** 12/30/2015 3:40 pm

So, I'm dealing with a guy who wants to defend the claim that the mind is software on neural hardware by way of appeal to neural network programming. I know that he's running a number of things together, but my education with respect to computer related theories of mind was merely sufficient to let me understand the rudimentary disagreement between computationalism and connectionalism. The objections of Searl and Dreyfus, which I am familiar with, are mainly directed against computationalism. I have no idea what to say to somebody talking about neural networks and mind other than to say that they in principle are not accessing universals, to which he will reply some handwaving about Bayesian association of data, or something like that.

I'm not sure what a substantial reply would amount to. Neural networks do seem to be able to recognize shapes and faces and play jeopardy and any number of other things, and I'm not sure what to say about it in replied to that.

**Posted by z10** 12/31/2015 3:00 am

Neural networks only recognise shapes insofar as we recognise them recognising said shapes. Could they recognise the shape without an intentional subject around to interpret the results?

**Posted by iwpoe** 12/31/2015 1:29 pm

I mean, I think they can group certain shapes automatically, but I'm not sure how much bite your objection has. The point will be made that no "concept" or "universal" was necessary to perform the task, which undercuts that series of objections to a computer centered theory of mind.

Now, yes, the computer can't reflect on its own task, but I'm not sure that's an in principal objection rather than one against present technology.

**Posted by Mysterious Brony** 12/31/2015 11:47 pm

@iwpoe
What about the James Ross argument that Dr. Feser uses most of the time? Does that argument have any force against such view?

**Posted by z10** 1/01/2016 12:10 am

iwpoe wrote:
I mean, I think they can group certain shapes automatically, but I'm not sure how much bite your objection has. The point will be made that no "concept" or "universal" was necessary to perform the task, which undercuts that series of objections to a computer centered theory of mind.

Now, yes, the computer can't reflect on its own task, but I'm not sure that's an in principal objection rather than one against present technology.

I think I would be tempted to deny that any task has been performed at all without an intentional subject to recognise a task. Does air perform the task of being lighter than water? I'm not sure I follow what distinguishes the neural network from this example if no subjects are around. Perhaps I'm missing the argument here!