This is another way of expressing the "Theory-ladenness of observation", which is one of the central points of post-modernism. So a large quantity has been written, but it is often expressed from various viewpoints that are hard to follow. It is often more clearly obvious in experimental results in the modern psychology of perception.
You may want to start with early forms, then skip the middle part full of pretentiously thick prose, and leave philosophy for psychology. Two of the simplest expressions of this principle within philosophy are commonly expressed here: 1) in the arguments of late Wittgenstein and Quine and 2) in the models of science due to Popper and Kuhn.
It is directly related to Wittgenstein's notion of the complementarity of usage and meaning through the 'game' that coordinates them. Meaning predicts how others will understand us. Usage is all we can see. We harmonize the two via experimentation with what forms of expression win or lose other's attention and agreement.
Some viewpoints in semiotics propose that this is the underlying model for all knowledge. Since everything we know is somehow expressed or represented -- to ourselves in memory, if not to others in language -- 'usage' and 'meaning' do not have to be restricted to the notion of language. The model can apply to experience in general.
Good science meeting Popper's demarcation criterion reiterates this form. Another way of expressing his notion of falsifiability is that we cannot learn anything scientifically without making predictions and being aware of how those predictions may or may not fail. This is often because it is very hard to see failures that we can conceivably predict, so the notion of failure is itself also a prediction, and it is possible to totally miss the point and to defend presumptions that are too deeply ingrained.
Relevant psychological data include that we begin to react to most things before we are aware of them. So prediction precedes attention. And that memories are quite mutable, so attention is selective -- by one dominant theory, it only every contains confirmed predictions, and all the rest is created ad hoc. But I will not elaborate these, because you can get better information in the SEs on psychology, cognition and the other social sciences.