Suppose we are asked, ``How did Junior fly from Glasgow to London?'' and want to respond with facts about taking a taxi to the airport, presenting his ticket at the check-in counter, going to the gate, getting on the airplane, taking his assigned seat, etc. We can add this additional narrative with its intermediate situations, and we can throw in reading the book if we like. There is no reason to discard occurs(does(Junior, fly(Glasgow,London)),S0). We merely have a redundant way of reaching the same conclusion. However, we would like a sentence relating the more detailed narrative to the less detailed narrative, i.e. of asserting that one realizes the other. For this we will at least need narratives as objects, and this has not yet been introduced.
Note that the relation elaborates(N2,N1), when we get around to introducing it, will not be like the relation between a subroutine call and the subroutine. N1 will not in any sense be the definition of N2. N2 could be realized in a number of ways, only one of which corresponds to N1.
The elaboration involved in telling about Junior reading a book is of a different kind from that involved in telling about his taking a taxi to the airport, because reading the book is a parallel operation rather than a means of accomplishing part of the travel. Reading should be simpler to treat. In fact it may be more like Daddy stacking blocks in New York.
Suppose, however, that we want to treat reading the book as a simple sequential situation calculus account using the function result(a,s). We will need to encapsulate the reading narrative in some way. The obvious way to do it is by using a context in which we do the reasoning about reading, e.g. what has to be read first in order to understand the subsequent chapters of the book, etc. It is not obvious what to call this context, but for now let's give it an arbitrary name c21. By the way, using the context theory of [] requires that the sentences of the whole narrative be true in an outer context--call it c0.
What should be the language of this context c21, and what should the initial sentences p such that ist(c21,p)? c21 should also have some nonmonotonic rules specific to reasoning within it. For example, it may assume some kinds of normality of the reader, e.g. that he knows the language of his reading material. This assumption will be realized by applying some defaults to facts in the common sense database about reading.
There are two possible approaches to forming c21, i.e. to asserting what is true in it. The first approach is to derive these facts from the situation in some way. The second approach is to state them by fiat. (As Russell put it, the advantages of the axiomatic method are the same as the advantages of theft over honest toil.) We take the latter approach with the consequence that we will later have more work to do when we want to lift conclusions from c21 to an outer context. c21 should be adapted to precisely the reasoning that has to be done about Junior's reading. Thus if we have the formalism in good shape, nothing about the fact that Junior is travelling by airplane should need to affect c21.