From a Lacanian (or any less physicalist idealist) point of view, ideas are the only place knowledge can arise.
Whether you are imagining the phone of the future, or guessing where your cat is hiding, you are imagining possible scenarios, and the two things are not really different except in their degree of 'futurity'. You will or will not succeed at making that phone or finding the cat. But the ideas about the cat will be tested more immediately and more completely than those about the phone, which will have to be combined with other imaginary criteria over the course of months or years before they are validated or abandoned.
If an idea is validated by experience, it becomes an 'observation' and if it associates itself with existing observations it becomes a 'symbol', but the only thing that can be represented in thought is an idea arising from the imagination.
This is a psychological elaboration of the older notions behind idealism as an epistemology. We have to build a model of the world out of imaginary projections validated by experience instead of being able to build it out of actual experiences. There is no actual experience except for a mere sense impression, and we know that our picture of the world does not consist primarily of raw sense impressions. It is made up of objects, people, actions, interpretations etc. Those things are representations that we have to have imagined and tested out as models. They capture sense impressions that validate or challenge them, but they are whole objects with identities in a way not directly determined by those impressions.
I like this model as an approach to perception, science and representation because it has a built-in dynamic.
In this model, there are three 'realms': the Ideal, the Symbolic and the Real, from which we seek different kinds of knowledge. The Ideal realm determines what is logically possible, the Real realm determines what is actually effective, the Symbolic realm determines whether what is contextually relevant.
Of course we live communally in environments that lead to our sharing a lot of experiences, so our collections of facts are similar, and we bind those similar facts together into cultural symbols which shape our future imagination. So we converge, under pressure from reality on sets of symbols that we reference in language. But since the detailed models are constructed internally, we do not share exact copies of these symbols and meanings.