Empiricism and the Epistemic Status of Imaging Technologies

This starting point for this project was the question of how to understand the epistemic status of mathematized imaging technologies such as positron emission tomography (PET) and confocal microscopy. These sorts of instruments play an increasingly important role in virtually all areas of biology and medicine. Some of these technologies have been widely celebrated as having revolutionized various fields of studies while others have been the target of substantial criticism. Thus, it is essential that we be able to assess these sorts of technologies as methods of producing evidence. They differ from one another in many respects, but one feature they all have in common is the use of multiple layers of statistical and mathematical processing that are essential to data production. This feature alone means that they do not fit neatly into any standard empiricist account of evidence. Yet this failure to be accommodated by philosophical accounts of good evidence does not indicate a general inadequacy on their part since, by many measures, they very often produce very high quality evidence. In order to understand how they can do so, we must look more closely at old philosophical questions concerning the role of experience and observation in acquiring knowledge about the external world. Doing so leads us to a new, grounded version of empiricism.After distinguishing between a weaker and a stronger, anthropomorphic version of empiricism, I argue that most contemporary accounts of observation are what I call benchmark strategies that, implicitly or explicitly, rely on the stronger version according to which human sense experience holds a place of unique privilege. They attempt to extend the bounds of observation - and the epistemic privilege accorded to it - by establishing some type of relevant similarity to the benchmark of human perception. These accounts fail because they are unable to establish an epistemically motivated account of what relevant similarity consists of. The last best chance for any benchmark approach, and, indeed, for anthropomorphic empiricism, is to supplement a benchmark strategy with a grounding strategy. Toward this end, I examine the Grounded Benchmark Criterion which defines relevant similarity to human perception to be defined in terms of the reliability-making features of human perception. This account, too, must fail due to our inability to specify these features given the current state of understanding of the human visual system. However, this failure reveals that it is reliability alone that is epistemically relevant, not any other sort of similarity to human perception.Current accounts of reliability suffer from a number of difficulties, so I develop a novel account of reliability that is based on the concept of granularity. My account of reliability in terms of a granularity match both provides the means to refine the weaker version of empiricism and allows us to establish when and why imaging technologies are reliable. Finally, I use this account of granularity is examining the importance of the fact that the output of imaging technologies usually is images