The idea that observation is theory dependent is central to many debates in the philosophy of science. This concept is a reaction to the idea that disputes about the way the world is can be easily and simply resolved by simply ‘looking at the facts’, or performing some sort of experiment or observation. The problem with this answer is that there is no ‘neutral’ vantage point from which such facts can be gathered or interpreted. Rather, empirical evidence is always interpreted within the context of one’s preexisting ideas, conceptions, and expectations, which can often have a dramatic effect on how observations are understood or what they are taken to mean.
Historians of science have given many examples of instances where proponents of rival theories have interpreted the same empirical evidence in very different ways, in accordance with their theoretical commitments. An interesting illustrative case can be found in a popular drawing called the ‘duck-rabbit’, a sketch which can be interpreted as either a drawing of a duck or of a rabbit, depending on the ‘theory’ one applies in interpreting the pattern of lines. While in this particular case both interpretations are equally ‘correct’, in many cases scientific and philosophical disputes, however, it is often unclear whether one, both, or none of the differing interpretations of the relevant facts are correct.
Another problem with observation is that there are always far too many empirical facts for us to consider all of them. One must have some way of selecting which facts are ‘relevant’ and which are not, an activity which naturally requires the use of some theory, a theory which in turn may be widely disputed. For example, scientists do not spent their time counting the number of blades of grass on every lawn, even though this would result in the collection of more facts, because we have reason to think that this fact has no real importance or significance. If, however, we believed that the number of blades of grass on a field was a form of communication from an extraterrestrial race, or a sign from some divine being, then our attitude towards the significance and meaning of the very same facts would doubtless be very different. This dispute could not simply be resolved by ‘looking at the facts’, because which facts we regard as relevant would depend upon which theory we accepted.
How to resolve such problems has been the subject of considerable philosophical attention, and remains an ongoing problem for any attempt to provide a comprehensive philosophical underpinning for scientific inquiry.
Facts are often theory-dependent: a readable introduction to the idea for a philosophy course
Theory dependence of observation: useful and detailed discussion of the problem
Underdetermination of scientific theory: an overview and discussion of various formulations of and responses to the problem from the Stanford Encyclopedia of Philosophy
The theory-ladeness of observation: more advanced discussion of the problem