I have been reading about abortion recently and came across a ‘thought experiment’ used by Judis Jarvis Thomson about an expanding baby. The scenario is that you're in your house when your baby starts expanding rapidly. You realise that you have no chance of getting out and the only way to survive is to pop and kill the baby. The idea is that this is an analogy for mothers who will die if an abortion is not performed i.e. is it ok to kill in this form of self-defence? These thought experiments are designed to provoke a moral attitude which can then be applied to discover your true feelings on a particular issue. My instant reaction was that yes, it was ok to pop the baby in order to survive and therefore I believe abortion is ok if it saves the life of the mother. However, imagine that the baby is now an analogy not for abortion but for a virus like AIDS, by the same thought experiment it could be argued that saying yes would justify killing everyone who had AIDS in order to save everyone else in...

I agree with Richard that Thomson's analogy doesn't apply to the AIDS victims, as opposed to the virus. I wanted to add something about the reliability of thought experiments in general, though. Philosophers like Thomson (and Kamm) employ imaginary examples in a quasi scientific manner. The example, or rather a consideration of the example, is like an experiment. Our intuitive reactions to the examples are the results of the experiment--the data. We are then supposed to construct moral theories that fit the data. The problem is that our moral intuitions are influenced by all kinds of factors, including ones that those same moral intuitions tell us are morally irrelevant. A pretty good examination of this problem is contained in Peter Unger's excellent (but slightly annoyingly written) book Living High and Letting Die . This doesn't mean that we should abandon imaginary examples altogether. They can serve as a pretty good consistency check on a position, for example. But we certainly shouldn't...