Before getting involved in the arguments for the theory dependence of observation, it is worth first thinking why it is attractive that theory and observation should be distinct. That issue sets the stage for assessing the significance of any particular view of the inter-penetration of theory and observation. The worry is, surely, that without a separation, observation – which lies at the heart of the last three hundred years of science – cannot be an impartial test of theory.
The lecture, readings and lecture notes outlined 4 potential arguments for the theory dependence of observation and one argment against (Fodor's).
1: Duhem’s argument is suggestive but does not provide a clear argument that observation reports have to be – rather than merely typically are – theory-laden. Duhem rather undermines his own case by talking of two factors involved in experiment: 'In the first place it consists in the observation of certain facts… In the second place, it consists in the interpretation of those facts…' That suggests that there can, afterall, be a distinction between observation and theory. (That said, it is worth thinking what the former aspect might be given Duhem's helpfully rich description. It seems always already quite theoretical withy its talk of batteries etc)
2: Hanson’s argument is that there is more to seeing than a retinal image. The duck-rabbit undermines any such thought because what strikes the retina is shared between seeing it as a duck and as a rabbit. Hanson takes it that whilst the ability to switch between one ‘aspect’ and another may be uncommon, all seeing is seeing-as: all seeing is conceptually structured.
We need to be a bit careful here. My cat Sootica can see the television: she steers round it / jumps on it. But she cannot see it as a TV or see that there is a TV. Still, what we report/say when asked what we see is conceptually structured. So conceptual structure in what we report is important for science.
But it is one thing to say that observation is conceptually structured, it is another to say it is structured by whatever theory is in play. Perhaps we could limit the concepts in play to very low-level observational concepts?
3: Churchland attempts to provide a stronger argument. First, the sensations or qualia we experience have no role in fixing the meaning of observation reports. Even if aliens had the same experiences we have when we see colour but when they detected heat, we should not link their utterances to colours but to heat: their theory of what in the world they are detecting. Second, we could connect our ‘heat sensations’ to a different theory. If so we would directly talk of ‘caloric’ eg. So in both cases what matters to observation reports is, he argues, the theory held not the observational 'sensations' or qualia in play.
Churchland’s argument is that what fixes the meaning of observation reports isn’t observational ‘sensations’ but the theory we have of the properties in the world we think we’re detecting. So there is no split between observation and theory.
4: Hesse has another argument. Part of our scientific picture is a distinction between what we think we can see directly and what we have to infer. For example, in normal light, we can tell by just looking, what colours things are. Colour judgements are, in those normal conditions, observational rather than a matter of theoretical inference. But - and this is the complication - what is normal lighting is itself a theoretical notion. So the distinction of observation versus theory (in this example: of colours) itself depends on a background of theory (normal light). But the point of the theory-observation split is to provide a neutral test of theories which this theory-relative distinction cannot provide. (Imagine that we discover that lighting on earth has changed gradually over the last couple of years. If so, then all our direct colour judgements over the last couple of years will have been mistaken. Real colour judgement would become a matter of theoretical inference again.) This is in Hesse, M. (1980). Revolutions and Reconstructions in the Philosophy of Science. Brighton: Harvester.
5: Fodor provides an argument against the view that theory and observation interpenetrate. Or rather, he has an argument that whilst observation is dependent on some theory, it is only a limited more or less hard wired theory. Thus changes in our broader theoretical views of the world will not change the hard wired sub-set of theory which infects observation. Thus Fodor disagrees with Hanson and Churchland by saying that they assume meaning is holistic. They think that the meaning of 'observation' claims depends on the broader theory one holds. Thus changes in any of one's theoretical belief can change what one sees. But isn’t it significant, Fodor says, that even though we know that the Muller Lyer lines are the same length, we cannot help seeing them as distinct? So Fodor thinks that, as a matter of fact, our perceptual systems are fairly insulated from the rest of our beliefs. Thus the theory that affects them is fairly fixed and low level. Hence that provides a distinction between theory and observation of sorts. (Click here for a discussion of sentence parsing, Pooh and Piglet.)
But is it enough?
Thus we have two questions:
Can Fodor really save a distinction between theory and observation? (One way to think about this is to ask: would Fodor's account satisfy the logical empiricists? Would they be reassured to know that although our observations depend on some theory, the theory is more or less fixed and hard wired?)
If he cannot, should we worry that theory faces no impartial observational test?