I get the author's beef with the idea that discrete brain areas peform discrete tasks - if there's one thing my neuroscience MSc taught me it's that the brain processes things in parallel rather than series, which is what most of those theories rest on - but I think part of the issue is that we try to impose existing words, like "memory", onto the brain. One of the more productive lines of thinking I used when studying was trying to throw out all of those definitions and try to work from first principles about the kind of processing various areas were doing. Of course it's impossibly complex to get understand it in totality (which is probably partially what drives specialising in particular brain areas), but discarding the limits imposed by language really helped me think more flexibly about what it might be doing.
Also, a small nit about the way it was presented was using plasticity (i.e. areas area able to perform different kinds of processing if the inputs change, e.g. blindfold + braille) as evidence of non-specialisation. That's doesn't follow. Just because neurons can be flexible doesn't mean they're not performing a discrete task, although I agree with the general conclusion.