Perceptual-Recognitional Abilities, Barn Facades and the problem of Individuating Environments

One line of thought in the literature on perceptual knowledge makes important use of the idea of perceptual-recognitional abilities. Such an ability would be, for example, the ability to recognize by looking whether or not it is an azalea that sits on a hill in front of you. (If you couldn't tell an azalea from a daffodill, or if the hill is salted with silk-made fraud azaleas, then you wouldn't have such an ability).

In his forthcoming paper "Perceptual Recognitional Abilities and Perceptual Knowledge", Alan Millar advances the idea that the exercising of perceptual-recogntional abilities is required for perceptual knowledge. Given that, even to the most seasoned horticulturalist, a hill of fake (and perceptually indistinguishable from real) azaleas would create a situation in which our horticultiralist no longer has the ability to recognize an azalea, Millar points out that 'whether' one has such an ability will depend, in part, on facts about the environment within which one is inquiring.

I confess that I don't have a clear idea how ‘environments’ are supposed to be individuated. Suppose, for example, that we take on one hand Goldman’s valley salted with barn facades and determine that that valley constitutes an environment. Suppose, further, that you are lost while driving in the general area. You come to a fork in the road; one road leads down Goldman’s barn façade valley, the other road leads down a valley of all genuine barns. You flip a coin and happen to take the road that leads down the road with all genuine barns. It is unclear to me whether this road constitutes a ‘good’ environment simply by virtue of the fact that there are no barn facades in the valley to potentially deceive you. Or, should we say that it is part of a ‘bad’ environment (i.e. one in which you do not have barn-recognizing perceptual recognitional abilities) by virtue of the fact that you so easily could have driven down Goldman lane. Alan Millar, in conversation, has suggested to me that—at least on his view—it will be inevitable that there will be some indeterminate cases in which the issue of whether one has exercised perceptual-recognitional abilities in a certain environment will not be clear given that the delineation of the environment relative to which one would have such abilities is obfuscated.

Some quesitons for discussion:

1. Is the project of indexing perceptual-recognitional abilities to environments undermined by the fact that some cases will be indeterminate?
2. Is the project doomed to concede that, within its framework, we can never 'know' that we know? It seems to me that this would have to be conceded because, even in cases in which the environments are such that, if we were to learn all the facts about them, they would be 'clear cut' cases, it's not the case that during the time in which we make perceptual judgments we have all the facts about the particular environments in which we make these judgments. Put another way, to know that we know, we must know that we are in an environment apposite to our exercising our perceptual-recognitional abilities--a tough antecedent order, indeed! Thoughts/comments welcome.