There seems to be a common intuition that parts of a system can't understand the system they are in without stepping outside of it. This is mostly applied to ideological and political issues ("Ideology is everywhere, so you can't step outside it and thus can never fully understand it"), but I've seen it applied to artificial intelligence as well ("A computer program can't fully understand itself" is treated as self-evident by some). Is there something to this intuition, or is it just rhetoric? I can't think of any obviously necessary reason why a part of a system shouldn't be able to perceive or understand the system as a whole.

I reply here knowing full-well I am out of my depth - I know there will be others who will probe this question far more adequately than I. But this disclaimer is itself a type of question imbedded within your query: what is "the whole" of any system? As a philosopher I know the limits of my knowledge by coming up against them time and again; as much as I love Hegel, there is no system of consciousness about which consummate knowing is possible. So let me suggest that, in the abstract, as you have framed it, I do not find there to be a compelling answer without greater specificity of what system one wishes to analyze. Indeed, the word "understand" or literally, "to stand under" suggests a relation other than from within. One can make too much of the insider/outsider standpoint as necessary for critiques aimed at the whole, but there seems to be a compelling case for some epistemic advantage from which "the whole" appears more clearly. Examples abound, such as how African Americans can shed light on...