On Reasons in Support of Spontaneous Beliefs

This is another example of a universally quantified statement that seems counter-instanced in broadly Chisholmian epistemologists.

On p. 216 Duncan writes:

"No one seriously holds that one needs reasons to support one's spontaneous belief [such as that one is cold]..."

But evidentialists think that all beliefs need reasons in support of them, whether doxastic or non-doxastic. So suppose someone believes I'm cold. Now either that belief is caused by an experience of being cold or inferred from some other supporting belief such as that I'm standing in the rain in a tee-shirt and I've got goose bumps or some such thing. If it does, then that's all well and good, but if not isn't it *clear* that something's wrong. Wouldn't the person be irrational at some level if there was *nothing* that supported the belief?