Knowledge Aided by Observation†

Anscombe seems to think that, even though “the knowledge that a man has of his intentional actions” is not “knowledge by observation”, it can be aided by observation. My aim in this essay is to explain how I think we should understand this thought. I suggest that, in a central class of cases, knowledge of one's intentional action is knowledge whose canonical linguistic expression is an utterance of the form “I am doing something to that G": knowledge in which the subject, at once, knows himself “as self" (and so, not by observation), and knows an outer object “as other” (and so, by observation). To characterise this knowledge either as knowledge by observation, or as knowledge not by observation, is to characterise it in a manner that abstracts away from its fundamental unity.