Abstract The relationship between face recognition and visual word recognition/reading has received increasing attention lately. A core question is whether face and word recognition rely on cognitive and cerebral processes that are largely independent, or rather processes that are distributed and highly shared. This question has been investigated using experimental, neuropsychological, and neuroimaging methods in both healthy and clinical groups. Finding comparable tests of face and word processing is not as straightforward as might be expected. Typically, a test of object processing is also included, as a control, which makes designing experiments all the more challenging. Three main strategies have been used to overcome this problem, each of which has limitations: 1) Compare performances on typical tests of the three stimulus types (e.g., a Face Memory Test, an Object recognition test, and a test of reading out loud); 2) Compare typical effects found in normal participants or clinical groups with the different categories of stimuli (e.g., the face inversion effect and the word length effect) 3) Test the three stimulus categories in the same experimental setup (e.g., delayed matching tasks). None of these methods, however, has provided measurements that enable direct comparison of performances across categories. We propose a simple framework for classifying tests of face, object, and word recognition according to the level of perceptual processing required to perform each test. Using this framework to classify tests and experiments aiming to compare processing across these categories, it becomes apparent that core differences in characteristics (visual and semantic) between the stimuli make the problem of designing comparable tests an insoluble conundrum. By analyzing the experimental paradigms, we might be contributing to our understanding of the differences between word and face processing. The suggested framework may help researchers in creating the least inappropriate experimental designs to test their hypotheses. Meeting abstract presented at VSS 2017