I just recently read Richard Dawkin's book "Climbing Mount Improbable." One of his major reasons for writing the book was to counter the common claim that there's something special going on in evolution to explain the development of things like eyes and wings. He explicitly intended to refute the claim that half an eye or half a wing was useless, so evolution couldn't have walked a long narrow trail from no eye or no wing to fully developed eyes or wings.
The argument that he makes is that a useful organ that is 1% better than the previous generation's model often provides more benefits than the previous one did. For instance, if a light detecting patch on the skin of a swimming creature lets it know when to hide, then one that detects edges more often, or gives an indication of the size of the oclusion will provide a better idea about when to hide and when it's not necessary.
Along the way, Dawkins reports that eyes have developed
"independently" 40 different times in the animal kingdom. This is due
to the large differences in the macro structure across animals. There is a fish that has two different lenses in its eyes (one looking forward, one looking down), and they use different designs.
Dawkins discusses the controlling genes that John Clark brought up, and his claim is that this indicates that some very early creature had a light-detecting patch on its skin somewhere. One part of the gene pattern that controlled this development was treated as controlling by all later evolutionary descendents. As various kinds of eyes develop, if there is an easy way to ensure that all the pieces of the apparatus depend on a common switch, then the pieces will find a way to depend on the switch. A gene that tried to develop a cornea when the other genes that know how to build parts of an eye aren't working will quickly be selected out.
Notice that those same genes to develop "abstract eyeness" don't do anything interesting when inserted into a plant. Nothing in the evolutionary history of the plant genese caused any of them to tie themselves to the activation of that gene cluster.
It's not an abstract signal, it's a concrete signal that was useful because all the other genes that built eye machinery were also listening to it.