>>>>so I will yield to John Ryan's ...
>>
>>Where is the fun in that? ;-)
>>
>>Here, allow me: sweet is not the opposite of sour. ;-)
>
>yupp, that is the one that was nagging me mostly: if asked to pick the axis/dimension least correlated with sweet, my pick would have been bitter.
>And you are probably thinking directly about the PH scale, where alkaloids are not at the sour end ;-))

Actually the taste space has about five dimensions, with zero or positive coordinates. Not a real vector space, as there's no proper addition - no inverse element to start with. So there's only one extreme on one end - the [0,0,0,0,0], i.e. absolutely bland - and any kind of extreme you may imagine if you push any number of coordinates to its max. Is there a negative sweet? Something that, when mixes with it, neutralizes it? I don't think so.

IOW, I disagree with the question. Or, in yet another words, all offered answers are equally OK, because they are just as independent of the idea they are compared with, just like any two basic tastes are.