Admittedly the "filling-in" could fill in something into a non-metric representation. But I believe that most people who have thought about "filling in" do not consider this possibility. Certainly when the concept is used in the literature on brightness perception, some kind of cortical map having metric properties is assumed to be completed by a filling-in process (e.g. Grossberg & Mingolla, 1985).
Hebb's (1949) theory has similarities to this. Interestingly, Hochberg (1984) notes that a similar view was also expressed in Helmholtz's classic theory of perception, where what is perceived are expected contingencies, not what is on the retina or some iconic derivative of it.
The notion of selective attention could be invoked here instead of 'interrogation' or 'wondering about'. However the idea of attention sometimes carries with it the idea of sensitizing or activating a metric-preserving (or at least retinotopically organized) cortical map. I want to avoid this notion, since it is a tempting step to take to then imagine that what we have the impression of "seeing" is what is activated on such a map. While retinotopic cortical maps undoubtedly exist, what we "see" is not what is on these maps, but rather what schematic (semantic) representations we construct on the basis of what the maps signal.
It is amusing to reflect on the Wittgensteinian nature of this idea: In the famous proposition 7 of the Tractatus Logico-Philosophicus Wittgenstein (1961) said: "What we cannot speak about we must pass over in silence" ("Worüber mann nicht sprechen kann, darüber muss mann schweigen"). This seems to correspond to our point (b), above, for which we could say here something like: (b) "What we cannot see must remain hidden from us!" (this sounds more impressive in german: "Was mann nicht erkennen kann, das bleibt uns verborgen!"). The equivalent of (a) would be: "We cannot see what we don't look at! (or perhaps better: we only see what we notice)" ("Wir sehen nicht was wir nicht anschauen."). For memory we might say: (b) "We recollect only what we can" ("Das an dem mann sich nicht erinnern kann, das bleibt vergessen") and (a) "We recollect only what we try to !" ("Was mann sich nicht in Erinnerung ruft, das bleibt vergessen"). All these statements are only seemingly tautological!
This analogy was suggested by Donald MacKay in a popular lecture I attended in about 1965.
In fact, the skin of the body covers a very large surface. We generally pay little attention to the contact our bodies make with our clothes, chairs, the floor, etc., because we are not using our whole bodies to probe the outside world. Possibly blind persons make more use of their bodies as "seeing" organs. This may be related to the notion of "Whole-Body-Seer" referred to by the author of a recent account of what it is like to become blind (Hull, 1991).
Is this a perceptual version of sollipsism? Perception is an illusion created by the desire to look...!
 This may be part of the reason why under stabilized vision we see in a fragmentary way. Of course there may be additional mechanisms of habituation -- but then these may be present precisely because seeing only requires changes.
It is unclear exactly what constitutes a "greater" perturbation. I doubt that any purely visual definition of salience would work: I suspect that even a (normally) perfectly obvious flash occurring during a fixation might not be noticed if the observer is being very attentive to some aspect of a display (as in playing a difficult video game).