When you speak of "object" as being "persistent nural patterns", you are speaking from the third person perspective. I believe Korzybski strongly emphasized that "objects" should be approached from the first person perspective.
First person: "I ..."
Second person: "you ..."
Third person "It ..."
The advance in cognitive and brain science has provided a much more advanced and detailed model for the third person perspective, and that can certainly "inform" the first person perspective. Consciousness of abstracting involves an awareness that the objects I see are not the "things" we infer the objcets were abstracted from. The "fact" (current model) that such a direct personal experience, which philosophers might talk about as "introspection", "qualia", etc., can much more easily be talked about in the third person should not distract us from consciousness of our "understanding" that our objects are abstractions that are unique for each person, not what is going on, and reflect our own experiences (mapping).
I would still not consider that "concept" is "reducible" to neurological patterns that instantiate it, as that picks out only one level of abstraction, and we have a multi-level perspective to accomodate. I do like the distinction between "concepts by intuition" and "concepts by postulate", the main difference being that the former is learned, in some sence prior to the words that express it, in an abstracting, implicit, descriptive, extensional, or bottom up manner, and is fuzzy or somewhat ambiguous, while the later is learned by virtue of the words that define it, in an intensional, explicit, defining, or top-down manner, and is unambigous and clear.
Our brains operate with two directions of signaling, incoming data handling known as abstracting, and anticipted modeling, known as predicting, and these, it seems to me correspond to the directions of extensional and intensional, abstraction and prediction. It seems only natural to me that the "binary" classification of concepts fits that neurological pattern. I do believe this would be consistent with George Lakoff's "Philosophy in the Flesh" approach that moves in the direction of an empirically based approach to philosophy.
There's no denying that neural network processes underly every perception and act, every memory, every word, every object we experience, every association we learn, every idea we have. However, I think it somewhat unproductive to be overly "reductionistic" and start to "define" our formulations in terms of such a single level general notion. Sure a "concept" is instantiated uniquely in each person in his or her neural processes. But this does not help us "explain" in formulations what we traditionally have meant. That requires formulations that provide some bracketing as to what formulations are reasonable to communicate such a notion in terms of time-binding learning capability. One might just as well say that a "dog", or any other non-proper noun, is a pattern of neural activity. It is "trivially" true in our current model. but it is at a level of abstraction that is not going to communicate an ability to use the term consistent with the time-binding record - that is to provide the ability to start where our parents left off - by such an extremely reductionistic approach.
If we are not to provide such a formulation for the term 'concept' then, of course, as "general semantics" has proclaimed, we should drop the use of the term in favor of "abstraction" and "definition" respectively, and talk about both as "formulations".
So, the challenge is to come up with a formulation that has not only a description of the general class of neurological process that we model as underlying the formulation, but also a verbal level description that can differentiate what we mean (collectively) by the term 'concept' from what we mean by other terms.
Empirically, we can examine the time-binding record for the use of the term, and as lexicographers do, abstract a concise (concept by intuition) "dictionary definition" to act as a standard for teaching the young how to join the time-binding vis-a-vis "concept".
Newly initiated time-binders acquire most "concepts" first by a "dictionary" definition or another formulation provided by a fellow, more experienced, time-binder. After a few such exposures, and through hearing and reading the term in more contexts, they gain an ability to perform their own abstraction into their own "concept by intuition", thus producing an ability to "initiate" more time-binder by providing their own formulation. And the cycle goes on and on with quite a bit of "corrective feedback" that makes some "concepts" much more precise while allowing others to remain ambiguous and fuzzy.
Because "concept" (along with idea, thought, etc.) has been conceived of in the Platonic tradition and in Frege's sense as something apart from the physical world, something in a "mental" substance or existence, it has been outside of empirical investigation independent of language. We can only investigate empirically how the term has been used in language contexts (- until now. But our brain imaging techniques are not sufficient to find detailed specifics during the use of individual words.)
We need a "general semantics" "definition" or "formulation" for the word 'concept' that is not limited to one level of abstraction, or, failing to come up with one, stop using the term.