58�Թ�

�������ٰ�������:��

In this paper I present a problem for a certain very traditional conception of epistemology.   According to this traditional conception, the epistemic status of one’s doxastic state – belief, disbelief, agnosticism – is determined by factors that are limited to (i) one’s current evidence/corpus of beliefs and/or (ii) one’s currently available cognitive processes (together with their reliability score).  I argue that any epistemology that embraces this conception faces what I call the “road sign problem.”   This problem arises when we recognize the possibility that what a road sign enjoins us to do – for example, (in the US) refrain from turning right on red – is something regarding which we might have all sorts of relevant evidence.  The difficulty is that even if we have overwhelming evidence that p (e.g. that there is no car coming now), acting on the assumption that p would not excuse us if, having violated the road sign’s injunction, we were to cause an accident.  This is so even if the assumption that p is overwhelmingly supported by our total evidence (and by any information we might acquire by relying at the moment on our available cognitive processes).  I hope to use this example to provide yet more evidence that sometimes the epistemic evaluation of a person’s doxastic state appeals to things that we properly expect them to have done (whether or not they did them) – that is, by appeal to things that go beyond (i) and (ii).