"…the form that any account must take which invokes…”? What is the antecedent of "which"?

Question

For while it may not show that a reductive mechanistic account is impossible, a proof that we are inescapably embodied agents to ourselves does show the form that any account must take which invokes our own self-understanding.
_{Philosophical Arguments by Charles Taylor}

I had a discussion with my teacher today, we have a difference of opinion on what the word "which" refers to. On my reading, I think it refers to "the form", but on the reading of that one who discuss with we, it refers to "any account". I don't really know what is the reason that make my reading like this, it's just my feel of reading make me to think in this way. But my teacher did give me a persuasive reason for his reading, the reason that the word "which" refers to account is it's more closer to the word "any account" than "the form".

For all of these, I am here to ask the question, that is, my title. In addition, I would like to ask one more extra question, is the word "which" grammatically necessarily refers to "any account"?

Answer 1

1. For while it may not show that a reductive mechanistic account is impossible,
2. a proof that we are inescapably embodied agents to ourselves
3. does show the form that any account must take
4. which invokes our own self-understanding.

Part 1 is a statement about what this proof we're talking about does not show.

Part 2 describes what the proof proves.

Let's combine those two parts into "This proof".

Part 3 tells us what "this proof" does - it shows (or demonstrates) the form of an account or a statement or exposition of reasons, causes, or motives.

Part 4 something we've already referred to invokes or appeals to or cites as authority, our own self-understanding

Simplified paraphrase:

This proof (of this assertion), (while it doesn't show that a certain type of account is impossible), demonstrates the form (or structure) of an account which invokes our own self-understanding.

I believe it is the account that invokes our own self-understanding, because it is closer to the clause than the form of that account and because an account is more likely to invoke something as a reason for something else. The form (or shape or nature) of an account doesn't itself appeal to our self-understanding as support for for a proof.

It is not grammatically necessary that the "which" clause refer to "account". It is possible that it refers to "form" and not "account", but it's hard to say without reading and understanding more of the context.

Formal logic (that doesn't involve electronic buffers and gates) isn't really in my wheelhouse though.