John Danaher at "Philosophical Disquisitions" has a nice post discussing what are variously called physiological, physical, perverted faculty, or (in this case) impeded function arguments against gay sex. There are a few minor problems with the discussion. For instance, it turns out to be remarkably difficult to find any natural law theorists who put much weight on this argument. A small handful have, but it really is a small handful: they would be virtually insignificant if they didn't have a few very clever philosophical analyses to their names. (I haven't read this one in particular, but the most famous seems to be Richard Connell's "A Defense of Humanae Vitae," in Laval Philosophique et Theologique vol 26, no. 1 (February 1970).) When it is used by others, it is used almost exclusively to establish a much weaker conclusion than Danaher's (4); in particular, it is used to argue for something like the claim that gay sex is wrong unless there is excellent reason to do it. That is, it is used to claim that when it comes to sexuality there is a moral presumption in favor of sex acts in which there is a possibility of procreation and a moral presumption against sex acts in which there is no such possibility. Other arguments are usually brought in to try to strengthen this to any sort of absolute prohibition. So we should be rather careful about suggesting that natural law theorists usually put much weight on the impeded function argument. It is, to be sure, a popular argument, but it is popular in the way that popular utilitarian arguments are popular: whatever flaws there might be, any such use of utilitarian arguments for or against something can't be treated as generally representative of utilitarian arguments on that subject; they are simply (at best, assuming no gross distortion, which is always a danger with popularizations of arguments) the most accessible fragments of the more accessible versions of utilitarianism. So here.
In any case, even if we take the pure form of the argument, we have to be careful about counterexamples. Take, for instance, the example used of blindfolding and sight. Your eyes have as their natural function letting in the light, and closing your eyes or blindfolding impedes that. It's not so clear, however, that this sort of counterexample works; first, because eyes are part of a sensory system that is in actual life fairly integrated, and whose function is (one would suppose) to gather information about the environment. This overarching function will sometimes be impeded by blindfolding but sometimes won't; and, moreover, it isn't clear that this, rather than sight as such, is the function of the eyes, with letting in the light simply being the particular means of fulfilling that function. At least, it needs to be argued rather than left to unstated assumptions. Likewise, just as your eyelids close to protect your eye, closing your eyes to protect your eyes actually furthers the functions of sight by protecting them from what is dangerous to them. And it is certainly the case that there are plenty of circumstances in which one will think there is really, and perhaps morally, something wrong with a person who goes around blindfolded or with eyes closed: driving, for instance, or walking across the street, or the like. One could, of course, account for these in other ways, but that these other ways are superior would, again, need to be argued rather than assumed. Similar problems arise with many other alleged counterexamples -- showing that they are actually counterexamples is much trickier than Danaher makes it sound.
A problem related to both of these issues is that Danher's (5) is unnecessarily ambiguous between "An action is not always immoral if it impedes the natural function of a bodily organ" and "No action is immoral merely because it impedes the natural function of a bodily organ". Establishing the latter would be a decisive refutation, but it is tricky to establish, and structurally cannot be established by counterexamples alone; the former can be established by well-developed counterexamples. But if we go the former route, it clearly leaves open the possibility of a probabilistic version of the impeded function argument, and a probabilistic version of the argument is difficult to refute by counterexamples alone. The problem with counterexamples in this sort of context is that they depend crucially on precise issues of interpretation and application, and they get their force entirely from the supposition that the proponent of the argument assumes it to be demonstrative rather than probabilistic, presumptive, defeasible under precisely determinable conditions, or any of the other ways arguments can be taken.
And a final worry is the oddity of Danaher's proposed alternative, (13). The impeded faculty argument is not that all biological functions should be fulfilled as much as possible, but simply the more moderate view that they should not be deliberately foiled in those acts in which they are, in fact, biologically involved. But (13) makes a worrisome confusion between biological functions and "basic goods", understood in a consequentialist way, which is, frankly, a bizarre kind of consequentialism. It would imply, for instance, that it is always morally acceptable to destroy or suppress parts of your brain geared toward and necessary for human sympathy if under the circumstances doing so would make it easier to get more of (all three of) sex, food, and sleep.
But while there is some looseness in the precision and nuances that aren't fully addressed, the basic argument Danaher is presenting is an important and essentially correct one. In genuinely traditional natural law the relevant principles of natural law are species-level and therefore, while they establish important priorities, if they are left unsupplemented do not establish anything definitive about individual actions. In genuinely traditional natural law, of course, they are not thought to be unsupplemented; but this gets into rather difficult and sometimes murky territory. But that is as it should be. Sex impinges on so many facets of life that it seems a bit foolish to assume that there are any definitive arguments to exceptionless conclusions on the subject that can be set down in short syllogisms. And one of the things the counterexample approach, and Danaher's presentation of it, does do well is raise over and over again the worry that things are much more complicated than they might initially have seemed to the proponent of the impeded function argument.