The validity of the senses is not an independent axiom; it is a corollary of the fact of consciousness.... If man is conscious of that which is, then his means of awareness are means of awareness, i.e., are valid. One cannot affirm consciousness while denying its primary form, which makes all the others possible. Just as any attack on consciousness negates itself, so does any attack on the senses. If the senses are not valid, neither are any concepts, including the ones used in the attack. [OPAR, ch. 2]
This is a very poor argument. Indeed, most of Rand's "stolen concept" arguments, particularly those relating to epistemology, are very poor. All "attacks" on the senses are ultimately attacks against the view that knowledge refers to something "out there," in "reality." Such attacks cannot be regarded as claims of knowledge; rather, they are radical denials of all knowledge. Although such denials are not true, they are not self-contradictory. But even if they were self-contradictory, it would not help Rand's case. Merely because someone makes a bad argument against x does not prove that x is "necessarily valid." Bad arguments against the validity of the senses cannot be used to establish the validity of the senses!
More to the point, this entire obsession with "validity" which haunts Objectivism is fatuous and tiresome. Imagine if someone declared: Delivery of mail by the post office is necessarily valid. It is an axiom, because any attempt to deny it presupposes it. Now such an assertion would be regarded as palpable absurd. No sensible person would ever say such a thing, because it is a foolish way to think about the issue. What we want to know is not whether delivery of the mail is "valid," but whether it is reliable. The same is true about the senses. It's not validity we seek, but reliability. Using this approach clears up a great many difficulties.
Doesn't the issue of validity solve the issue of reliability? If the senses are "valid," mustn't they ipso facto be reliable as well? But approaching the issue from this vantage point merely creates confusion. It casts the debate between two unpalatable extremes: those who affirm the validity of the senses in all instances on the one side, and those who deny the validity of the sense in all instances on the other. Neither position is all that credible.
The whole issue over the reliability of the senses arose precisely because it was found that some perceptions are not entirely veridical, but are misleading: the bent stick in the water, the railroad lines that appear to be joining in the distance, the circle that appears as an elipse, etc. Claiming that the senses are valid doesn't clear up these issues: it just pushes them further off. These anamolies must be explained; and if they can't be explained in relation to the senses, they must be explained on some other ground. Objectivism achieves the necessary validity of the senses by defining the senses as necessarily valid. But this achieves, at best, a phyric victory. Randall Dipert, in a review of Kelley's Evidence of the Senses, explains the problem with the Objectivist view:
What is the "realist theory of perception" that [Kelley] defends? This is more difficult to say than one would hope, for Kelley oddly is not given to single clear statements of his main positions; he is at his best on the attack. Saying, "Perception is always of existence reality" comes close. So, interestingly, does saying, "Perceptual judgments are never mistaken." This last assertion is of course especially curious, and requires us to turn to Kelley's analyses of "illusions" such as a circle that appears as an ellipse or, still better, a stick half-submerged in water that "appears" bent. In both the case of seeing the stick out of water and then half submerged, I think Kelley wants to say, we perceive the stick. Otherwise, it is not a case of perception at all. Kelley goes on: "The normal look of the stick and the refracted look are simply two different forms in which one can perceive the same external attribute." This external attribute is "the" shape of the stick. Perception now is not just of a stick, but rather of a stick in relation to a background — i.e., whether it is all exposed in air, or half-submerged in water. Perception then is of a relational fact: the stick exposed, or the stick half in water (or, in the case of a circle, of the relational fact formed by the circle and the angle it is being viewed at). Mistaken "perceptual" judgments are then falsely abbreviated judgments about a necessarily veridical percept; "there is no such thing as a nonveridical percept."
My reply to this maneuver is as follows. This is all well and good. You may indeed define for your own purposes, the perceived (the percept) as that which cannot be mistaken. The perception of a stick is "what is common to all appearances of the stick" — at least those in which it is distinguished at all from its background, to avoid (perhaps in an ad hoc way) anything appearing like the stick in suitably bad lighting, etc. But then it is our (abbreviated) perceptual judgment that can be mistaken. Whatever harm — whatever lack of certainty, unreliability, etc., that perceptual relativity formerly injected into your agenda — is now caused by the unreliability of perceptual judgments: how do we know they are correct, reliable, etc. In fact, Kelley comes dangerously close to, if not actually succeeding at, trivializing his entire enterprise. He writes: "Perception should not be defined, then, in terms of a genus that includes hallucinations and the like, as if these were phenomena on a par with perceiving. It should be defined as a type of awareness of external objects, to be contrasted with other types of awareness." But then, when is one certain that one is perceiving an object, and not in another type of awareness?
Kelley's point is, of course, not a new one. It is that perception is a "success" word, like seeing and hearing. One does not say one saw a lake that was not there; one says one appeared to see the lake. But co-opting the word 'perception' for veridical awareness of a certain type (apparently just "when the awareness is a unitary product of physiological causes"), does not give us an interestingly realistic theory of perception. It gives us a theory of perception that is "realist" by definition. The main difficulty for such a tautologous realist is then to decide when he is really perceiving an object, and when he is in one of the other states of awareness. How does he test whether he is perceiving the object? He must determine that his awareness is physiologically caused by an external object. This itself requires perception -- never mind the problem of ascertaining that the object is "external," consider only the problem of determining when one's own awareness is "physiologically caused."
To sum up: Objectivism establishes the "necessary validity" of the senses merely by pushing the issues of perceptual reliability further up the cognitive chain. The senses may be "necessarily valid," but what the senses tell us about the external world is still subject to misinterpretation and error. Sense knowledge, it turns out, is only valid if one's awareness is physiologically caused by external objects. But how does one know whether one's percepts are in fact caused by external objects? How, for that matter, does one know whether one has misinterpreted the "evidence" of the senses and gotten things horribly wrong? The Objectivist "validation" of the senses validates no claim of knowledge. It's another one of Rand's bad arguments in which we find her substituting empty tautological slogans for reasoned, evidence-based discourse.