Data processing in observation

Although it has become extremely common for scientists to make use of sophisticated instruments in combination with computational treatments of data in their observational practices, only the former has caused a debate among philosophers. Whether it is appropriate to make use of such instruments as telescopes, microscopes, PET scans, MRI devices and so forth is a question that has received various answers, depending on how the concept of observation is to be understood. The classical empiricist view, that sees unaided perception as the only way that human beings can really learn about facts of the world, has notably been followed by van Fraassen [1] while many other philosophers have defended the idea that instruments can legitimately be used in observation, so long as the processes involved in data acquisition meet some requirements. Hence, Maxwell [2], Shapere [3], Hacking [4] and others have opposed not only to the observational/theoretical distinction that shapes the empiricist view, most strikingly in its logical form, but also to the observable/unobservable distinction. Indeed, to defend that instruments play a role in observation is to acknowledge that what is observable at a given time is a function of our technology and therefore, that no fixed distinction can be established in the long term regarding what is observable and what is not. Among the defenders of a concept of observation that is not exclusively associated to human perception, it is unfortunate that none has tried to propose a systematic account of the different ways that the data acquired with an instrument are processed before they are actually used to observe some entity or phenomenon. Their analysis