When we observe an object for a while, then stop observing, and then 
take up observation again, we don't know what has happened to the object
 between those two periods of observation, so its state at the start of 
the second period of observation is not deducible entirely from ts state
 at the end of the first period of observation. There is an 
epistemological uncertainty involved. 
When
 a quantum object interacts with a piece of apparatus and its state is 
recorded by that apparatus, we can determine what the state of the 
object was at the time of interaction. If the object is allowed to 
continue without interaction with any other object whatsoever for some 
finite duration, then there is no record 
anywhere of what 
happened to the object between that interaction and its subsequent 
interaction, so its state on subsequent interaction is not deducible 
entirely from its state on the prior interaction. However, it seems to 
be the case that this is an 
ontological uncertainty rather than an 
epistemological uncertainty. 
I've come to regard this "superposition of states" for quantum objects as more of a superposition of 
histories.
 In the case of a quantum object passing through a single slit and 
meeting a detector, the superposition of histories is constrained to a 
single slit, so the detector only sees particle-like behaviour. In the 
case of a quantum object encountering double slits and meeting a 
detector, the superposition of histories has to take into account both 
slits, so the detector sees wave-like behaviour. 
The two-slit 
scenario is a highly constrained case of the quantum object "going 
everywhere" (i.e. a case where most of "everywhere" has been narrowed 
down to two slits), this notion of "going everywhere" being what I take 
to be metaphor for "having all possible histories". It may even be the 
case that a non-interacting quantum system can be considered to be 
constituted by all of its "possible histories". This interpretation would also be consistent with the "delayed choice" and the "quantum eraser" variations of the two-slit experiment.
Note that human consciousness 
in the form of a human "observer" is not implicated in these scenarios, but the mystery of 
ontological uncertainty remains.