This seems fairly obvious to me; self control involves doing something extra. For instance, take the Stroop test (where a color word is shown in another color, e.g. "red" in blue ink). When asked to read the word, you don't have to do anything extra (assuming you are an experienced reader); you kind of automatically just "know" what the word is, it is your strongest identification with the written text. Identifying the color isn't any harder than reading the word, but you have to override your automatic propensity to say what's written.<p>Thinking of it in programming terms, you're not just doing "sayColor()" instead of "sayText()"; you <i>always</i> do "sayText()", but you also have to do "blockSayText()" and "sayColor()" in parallel.