We have grammars in our heads. That's why we can produce and understand unlimited numbers of sentences; why people who suffer damage to their heads often lose some or all of their linguistic abilities; why PET scans show increased blood flow in particular bits of our brains when we carry out linguistic tasks under experimental conditions. The list is almost endless. But are the grammars (or I-languages) in our respective heads “psychologically real” as well as neurophysiologically real? The question has seemed unnecessarily vexed to many linguists and it has certainly attracted a mass of debate. Why should there be a problem?

Many psychologists and philosophers are happy with the idea that we have something in our heads which accounts for these phenomena. What some balk at is the complexity and opacity of the linguist's account of what we have in our heads. It is unexceptionable to suggest that there is a rule specifying that verbs precede their objects in English, because we can see immediately what the effect of contradicting that rule is: silly sentences like John onions eats instead of the correct John eats onions. It is not so obvious that the correct analysis of John was too clever to catch should contain three empty categories of the kind we saw in the previous chapter. You may by now be convinced that the evidence for such empty categories is pretty good, but postulating three of them in a six-word sentence still strikes many as excessive, just as the physicist's claim that the universe has ten (or eleven or twenty-six) dimensions seems unnecessarily baroque.

One response is to temporize and deny that such complexity is really necessary. Some linguists may like to talk about this plethora of empty categories, for instance, but they are not psychologically real: they are not represented in our heads as nouns and verbs are. In Chomsky's view, this reaction is either patronizing (when the linguists grow up they'll see the error of their ways) or uncomprehending, or both.