When one first comes across a description of an “observation sentence”, it would seem, prima facie, that such a thing doesn't exist.
How does Quine describe them? In the following way:
“... a sentence is an observation sentence if all verdicts on it depend on present sensory stimulation and on no stored information beyond what goes into understanding the sentence.” (From Quine's 1969 paper 'Epistemology naturalised'.)
Firstly, there are indeed sentences which we utter that do depend on “present sensory stimulation”. They also depend on “stored information”.
For a start, they require the stored information about the meaning of words and our memories (amongst many other things).
However, Quine does qualify his definition by the final clause: “beyond what goes into understanding the sentence”.
That means that the two things I mentioned (i.e., knowledge of the meanings of words and memories of, say, the objects and events in the observation) can be classed as “what goes into understanding the sentence”.
So even though these things are included in a Quinian observation sentence, it's still nonetheless the case that the sentence is about a current observation. (It's about what a person is a present experiencing or observing.) Yes, other things are needed in order to understand the sentence (things which came well before the observation). However, that doesn't stop the sentence - or what we utter - being an observation sentence if the content of that sentence is only about a present observation.
(This isn't that unlike Kant's acknowledgement that even the knowledge of a priori statements requires the experience of what words mean, etc. - and that's obviously, in Kant's terms, a posteriori knowledge.)
“There is a rabbit running before me.”
is about a current observation. Nonetheless, I require previous knowledge in order to make such a statement. Still, the sentence isn't about that past knowledge (or those past memories or experiences). The sentence is about the rabbit. The past knowledge is simply required to make the statement and understand the observation. It's not the subject of the sentence.
Quine himself talks in terms of “stored information”. He writes:
“The very fact that we have learned the language evinces much storing of information, and of information without which we should be in no position to give verdicts on sentences however observational.” (298)
Quine stresses his position on observation sentences by writing:
“... an observation sentence is one on which all speakers of the language give the same verdict when given the same concurrent stimulation.” (298)
All that accepted, there can still be an argument against Quinian observation sentences. Indeed don't Quine's various arguments for holism automatically work against the existence of genuine observation sentences?
Why Observation Sentences?
The naturalistic or scientific bonus of observation sentences – at least according to Quine - is their inter-subjective nature. (Therefore their scientific nature.) In other words, we've moved beyond the first-person pronoun of the aprioristic epistemologist.
As Quine puts it:
“.... the observation sentences are the sentences on which all members of the community will agree under uniform stimulation.” (299)
If inter-subjectivity is stressed, then we must also have moved beyond experiences, sense-data (in the old style) and other personal mental happenings and be focusing instead on the world (or on objects and events in the world). The very fact that there's an element of communal agreement involved when it comes to observation sentences (with a stress on the word “observation”) means that “subjective sensory states” become less relevant or not relevant at all.
Quine put it this way:
“The old tendency to associate observation sentences with a subjective sensory subject matter is rather ironic when we reflect that observation sentences area also meant to be the intersubjective tribunal of scientific hypotheses.” (299)
This alone seems to lead immediately to not only naturalism; but perhaps also to some form of externalism.
Rabbits - rather the experiences of rabbits - are now given priority. (Or at least it seems that way.) Of course the Cartesian/internalist arrived at rabbits via his experience of rabbits. However, the Quinian still arrives at rabbits from communal agreement on the experience of rabbits. So the experiences of rabbits must still be part of the story. It can be said that we've moved to what Roger Scruton calls (if in a different context) a “first person plural” in that it's the grand “we” that determines the nature of rabbits. No single mind can find the truth on rabbits by doing, say, some kind of Cartesian (or Husserlian) reduction.
Clearly we've moved a long way beyond the “Cartesian subject” here. Yet the Cartesian subject ruled epistemology from Descartes (17th century) to the 20th century.