Anscombe introduces the notion of "non-observational knowledge" by taking the knowledge one usually has of the position of his limbs as an example. According to her definition two requirements need to be met when we speak of "observing something": first

we can speak of separately describable sensations (call it the SD condition); second

having such sensations is in some sense our criterion for saying something (call it the CS condition). The "sensations of position"-so called by Anscombe-play a central role in understanding the knowledge we usually have of our bodily position as

non-observational

. But

do we really have such sensations? If yes

how can we tell whether they are "separately describable" or not? In what sense can having a given sensation be or be not our criterion of saying something? And

why (and how) are the above question to be related to our understanding the knowledge that we usually have of our bodily position as non-observational knowledge? By clarifying the ambiguity in the possible meanings of "sensation of X"

I'll try to defend my understanding of Anscombe's answers to the above questions

namely

that we do have sensations of position

and usually they are neither separately describable nor our criterion of saying something-in the sen

se that we usually don't identify our bodily position by identifying the sensations of position

and that