How to Innovate: Stop Reading Academic Papers?

One of my professors last semester had this piece of advice for graduate students: if you find an interesting research question, work through how you would attack the problem before doing a literature search. His point was that reading the literature, especially in the problem and research strategy generation phase of a project, stifles your creativity.

A recent NYT article makes a similar point:

This so-called curse of knowledge, a phrase used in a 1989 paper in The Journal of Political Economy, means that once you’ve become an expert in a particular subject, it’s hard to imagine not knowing what you do. Your conversations with others in the field are peppered with catch phrases and jargon that are foreign to the uninitiated. When it’s time to accomplish a task — open a store, build a house, buy new cash registers, sell insurance — those in the know get it done the way it has always been done, stifling innovation as they barrel along the well-worn path.

The rest of the article primarily considers strategies of how to get around the curse of knowledge at the organizational level, whereas my Professor's tactic tries to hit the problem at an individual level. Here are some questions/thoughts I had:

1) Is the curse of knowledge primarily a manifestation of framing, since an analysis presented in a certain way governs the way one thinks of the problem subsequently?

2) Do the benefits of knowledge outweigh the costs? I hear about a lot of projects that stem from "I read this and that paper and saw that I could make the following substantive improvements." I'm guessing that a lot of research is done that way. Such contributions require a fairly deep understanding of the existing body of research and where it is different.

3) Following from (2), I wonder if researchers self-select into the "innovation" (extensive margin) path or the "replicate and extend" (intensive margin) path. Is it possible for a single researcher to do both, or does it make more sense to specialize according to comparative advantage?

4) I'm guessing the innovation path is riskier, though if things pan out, returns will be much higher. If so, to what extent do academic incentives govern whether or not a given researchers chooses to innovate or incrementally add to the literature? For example, in the medical research world (excluded the more basic science oriented work) my guess is that the tacit requirements to publish in quantity drive the marginal researcher towards the "replicate and extend" path. Another example: some argue that academic tenure is justified in that it provides security for researchers with great promise to take on riskier projects.

5) In training graduate students, is their an optimal reading/creating ratio? After all, we learn about approaching problems from how others approach problems. There is really no way around that. But where is the point where we approach negative marginal returns?

6) What is the biological basis of the curse of knowledge? In my first year neuroscience course, one of the professors pointed out that we start out with many more synaptic connections as infants and children than we possess as adults. His take on this was that, with learning and experience, we strengthen the appropriate connections while "weeding out" the others. "This is why as kids we have all these crazy ideas, but as we grow up we have less and less of these," he said. From an evolutionary perspective, I can see how pruning out the crazy ideas makes sense. However, to the extent that evolution is a satisficing process, perhaps the pruning that goes on is a bit much. Are there ways to engage children such that they retain some of this "innovative craziness" later in life? Is there any research on this?