Abstract
This talk concerns the highly vexing issue of how a confirmation theory ought to handle post-hoc monsters, that is, post-hocly constructed or modified hypotheses like Velikovsky's theory or the various Ptolemaic systems of astronomy. One approach to this issue is to demonise post-hocness itself, arguing that no hypothesis earns support from evidence that has been used in its construction or modification. The problem with this approach is that there are plenty of examples of fruitful and even true hypotheses that were constructed or modified using post-hoc methods. Another approach is to attempt to segregate the monstrous from the non-monstrous post-hoc hypotheses and to argue that only the latter earn support from accommodated evidence. The problem with this approach is that there are compelling logical reasons why post-hoc monsters must earn some support from accommodated evidence. I will put forth a more subtle approach which I call the 'Frankenstein' theory of confirmation. According to this approach, post-hoc monsters earn confirmation from accommodated evidence but the confirmation earned does not spread uniformly throughout the content of such hypotheses.