The epistemological basis of [the logical-factual dichotomy] is the view that a concept consists only of its definition. According to the dichotomy, it is logically impermissible to contradict the definition of a concept; what one asserts by this means is "logically" impossible. But to contradict any of the non-defining characteristics of a concept's referents, is regarded as logically permissible; what one asserts in such a case is merely "empirically" impossible.
Thus, a "married bachelor" contradicts the definition of "bachelor" and hence is regarded as "logically" impossible. But a "bachelor who can fly to the moon by means of flapping his arms" is regarded a "logically" possible, because the definition of "bachelor" ("an unmarried man") does not specify his means of locomotion. [IOTE, 115]
Implicit in this criticism is the view that concepts include all the characteristics of a concept's referent. In practical terms, that means all proposotions about a concept, including theories, would presumably be included in the concept. For Objectivism, a concept is not a symbolic meaning used to represent something outside itself; it is, rather, a container which includes everything known (or potentially knowable) about the concept's referent. As Peikoff puts it, "the concept 'man' ... includes all the characteristics of the 'man.'" 
I've criticized this view in another post. Here I wish to focus on the Objectivist conflation of logical identity with predication. At the core of the Objectivist epistemology is a confusion between identity and understanding. Human knowledge is fundamentally representational and symbolic. If these symbols are to be used to convey truths about reality, a certain logical consistency in the use of the symbol becomes necessary. It would be confusing to define bachelor as an "unmarried man," and then turn around and apply the identical symbol to women or married men. However, the fact that a symbol has a specific meaning does not in itself constitute knowledge. It is possible to have symbols that have little or no reference to reality, such as phlogiston, griffen, kryptonite, centaur, etc. Whether a specific symbol actually represents something real is not discovered exclusively by logic, but mainly through observation. Hence the origin of the so-called logical-factual dichotomy. This dichotomy does not so much separate logic and fact as it distinguishes them, while more clearly defining their extent. Logic is applicable to our symbols. However, it provides no necessary guarantee of truth. There are an infinite number of possible logical syllogisms; but only a finite number of those will be true of empirical reality. In his criticism of logical-factual dichotomy, Peikoff seems to be oblivious of the fact that logical validity does not guarantee factual truth.
While identifying a symbol with its referent in reality involves some level of knowledge, that level is rather superficial. It doesn't require much knowledge to identify and name entities in reality. Understanding things, processes, and events in reality takes considerably more knowledge and acumen. A person may be able to identify an elephant in the sense he knows what to call it when he sees it; but he may know nothing about elephants. Identity does not guarantee understanding. Rand had no trouble identifying men and distinguishing them from other creatures. But many of her chief notions about men were exaggerated and flawed. Identity is, cognitively speaking, relatively easy. Understanding tends to be more difficult; sometimes much more difficult.
In an earlier post, I criticized the Objectivist view of concepts for its quasi-Platonic implications. I also noted the confusion of identity and understanding. To conclude this post, I wish to offer a few more elaborations of this criticism, tying it to earlier post criticizing the Objectivist view of identity . In that post, I argued that there were several types of identity: logical identity (Rand's A is A), the identity of the symbol with something real (A is), and the identity of characteristics (A is B). Now the tendency of Objectivism is to conflate these types of identity under a single concept. As Peikoff puts it, "All truths are the product of a logical identification of the facts of experience." Knowledge, for an Objectivist, is identification all the way through. You identify existing entitites and you identify their characteristics and that's how you go about creating the concept. And all these identifications are basically of the relatively simple, unproblematic, A is A, logical type of identity.
Since knowledge is symbolic through and through, identity is not a very apt way of describing the ideal of knowledge. The symbol is really nothing like its object, nor is knowledge a mirror. When Rand wrote of "unit economy" and cognitive efficiency," she seemed at least to imply, if not explicitly avow, that knowledge is not identical with its object. But in other parts of her epistemology, she forgets all about her unit economy and reverts to a crude literalism, where the tautology A is A becomes the catch phrase for all types identity, in defiance of logic and good sense. But logical identity is not identical with either the identification of objects or the characteristics of objects. Knowing what a symbol is and what it stands for in reality is not knowledge in the deeper sense of the word. It's merely knowledge of symbols and/or knowledge of naming conventions. Nor is logical identity fully compatible with predication (i.e., knowledge of characteristics). As I noted in my post about identity, while blood is blood and red is red, and while red can be predicated of blood, red is not identical to blood nor is blood identical to red. A is A is not logically compatible with A is B. These are different forms of identity and it will not do to conflate them. Concepts are meanings; and in that sense, one could say that a concept is its definition. Predications are not further elaborations of a concept; on the contrary, they are merely assertions about a concept, stated in terms of propositions and theories. Although this distinction is not mandatory, it does clear up many of bogs and swamps that infest the Objectivist epistemology and opens the way for a proposition-centric, theory-centric approach to philosophy that is more in keeping with the sort of empirical responsbility that should be the ideal of human cognition. Let's concentrate on facts and theories, not words and symbols. Is that too much to ask for?