Gary Marcus

If we don't know the answer to (2), how can we be sure about the answer to (1)?

We know enough of the basics of how individual neurons turn inputs to outputs to create an abstraction of that process. That is the solid statement that supports (1). (2) is dependent on how neurons interact & what further complexities are needed to capture all the brain does.

It may be fruitful & fashionable to say we dont understand anything about the brain & so cant model it. But that's simply not how modeling works. Its iterative and you go with your best guess until contradictory evidence provides you with a better one.

But the problem Grace is that the advocates of ANNs do not take your position: They do not want to iterate to a better (biologically) model, but instead double down on this particular abstraction at this particular point in time.

With all due respect, I don't know who these practitioners you speak of are. Grace's position is the mainstream, in my opinion.

Think this might also just be a diff of positions due to diff end goals? Those who might view AI as way to understand our brain & intel & leverage it have Graces position. Versus those who see AI as a blanket term wrt to data science+ML & focused on solns to specific problems.

Certainly there are people who use & design ANNs w/o any interest in interating on models of the brain. And that's fine. But that doesnt mean their artificial neurons arent abstractions of real neurons (unless they really take them far off in some other direction)

And people who do want to iterate and make better brain models are on solid enough ground when they start with ANNs