"Recognize text" is Acrobat X speak for OCR. The text characters you "see" are just pieces of the scanned image you used OCR on.
The OCR output (unless Clearscan is used) is a hidden "layer" (note - *not* a "Layer" of an optional content group).
The OCR output provides fonts in rendering mode 3 (no stroke, no fill - so hidden/invisible).
With Acrobat X's "Edit text" tool you can try selecting a piece of this, going into Properties (right click for context menu) and change font color.
If this "takes" you can see what is happening to the OCR output characters as you select-delete, backspace, etc.
Why hidden/invisible OCR output? It is there to facilitate search/find and not there to replace the image.
Use Clearscan to replace recognized character images.