Sometimes it is suggested that there cannot be such things as sense data, because nothing can have the properties our perceptual contents allegedly do have. In particular, it is sometimes suggested that the contents of our perceptions are vague or indeterminate in a way nothing could actually be. If you find this line of thought compelling, consider the following two step argument:
1. Nothing is indeterminate.
2. Our perceptual states are not indeterminate.
or this one:
1. Our perceptual states are indeterminate.
2. Something is indeterminate.
These arguments seem obviously valid, so if it is the case that nothing is indeterminate, then neither are our perceptual states. Conversely, if our perceptual states are indeterminate, then something in the world is indeterminate. Perhaps it will be said that this confuses the features of a representation with its content. The representational state is not itself indeterminate; it is rather that it is not determinate what these states, themselves determinate, represent. But if representation is supposed to be some sort of relation between a thing and its content, then this relation can no more be indeterminate than its relata. If there is no such thing as the representation relation, however, then nothing can be true of it, not even that 'it' is indeterminate. And this can't simply be a way of speaking about the relata or their monadic properties, for according to the advocate of determinacy neither these nor any sum of them can be indeterminate, and it cannot be true that it is indeterminate whether p if there is no relevant indeterminate state of affairs to make it true (at least for those who accept a kind of "truthmaker principle", as I do). If all the relevant things and their properties are determinate (including the propositions!), how can it be indeterminate whether some proposition is true of them? Some may say, as Lewis did, that indeterminacy or vagueness is simply semantic indecision on our part; there are our representations, their candidate represntata, and our simple failure to make up our mind about exactly which entities are the representata of a given representation. But the relevant entities, along with our decisions, are surely determinate. Take the term "old person". A Lewisian will say that we have not made a decision about precisely which people to cover by this term. But the question here is whether the term "old person" has an extension or not. If it does, then it must be indeterminate precisely which persons are in the extension of the term, because we have made no decision regarding the borderline cases; but if so, then according to this theory the borderline cases will neither determinately have the property of being in the extension of the term nor determinately lack it. (How does this differ from saying that a given object neither determinately has nor determinately lacks the property of being red? It seems there is no such thing as a purely semantic account of vagueness, for any such account will involve ontic vagueness in the semantic relation itself). An advocate of determinacy cannot admit this, so they will, if they are consistent, say that the term "old person", as we use it, has no extension. But those who advocate the indeterminacy of perception cannot take this stance; if, for example, they want to account for the phenomenon of our seeing the red and green segments of a bar without seeing where they meet, when the border falls on our blind-spot, they cannot locate this indeterminacy in the perception itself. By hypothesis it can no more be indeterminate than the term "old person", nor can they locate it, by the above argument, in the relation of our perceptual state to its contents. And it cannot be that they have no contents, for in that case we would not have perceptual states. Epistemicism will not work here, for if the contents of perception are determinate then our perceptual states should be as well (Though of course an epistemicist could escape this by adhering to first-person skepticism: there really is a border in our perception of the red and green segments, we just can't detect it. Of course, if this response is true, everything involved in perception is determinate, and the epistemicist cannot use the argument from indeterminacy against sense data). If a perceptual state is indeterminate, and it is identical to a brain state, then by an admittedly bizarre application of Leibniz's Law that brain state must be indeterminate. The same is true of causal roles, functional roles, and adverbial states.
Sense data, therefore, are on no more a shaky ground than any other entities invoked to account for perception: If perception is indeterminate, then something in the world, whether it be a sense datum, brain state, adverbial state, causal role, functional role, or representation relation, is indeterminate. If nothing in the world is indeterminate, then perception cannot be either, including the entities involved in it. There are good arguments against sense data, but this is not one of them, for what is a problem for all theories can't be blamed on any one in particular.
 Pointed out by V.S. Ramachandran in "Filling in Gaps in Perception: Part I". Current Directions in Psychological Science. 2:199-205 (1992)