The car analogy isn't that bad, though it's a bit incomplete. Imagine you want to study ways to optimize traffic in a city. You can use a car in a garage, since you have all the information necessary inside it (and you do, technically). Or you can use a model of the city with many toy cars inside it. Which will help you most? It's about the level of abstraction you choose to use. And it's important not only because some models are more efficient then others, but because some models have inherent biases. In this case, considering consciousness trapped inside the skull makes it a lot harder to see things which would be obvious when defining it as interaction between brains and environment.
That the environment is an integral part of consciousness.
I think he is saying not just that the environment "affects" consciousness, but is part of what we mean when we use the word "consciousness." In other words, the idea that you can "see" consciousness in an MRI scan is fundamentally in error. You need to also examine the environment and context in which that scan was taken, in order to make any progress in understanding what consciousness is.
Despite claims to the contrary, it really seems to me like the philosopher in this article is marking territory for a turf war.
The analogy with car was really horrible, not to mention the point of it is unclear.
It is understandable however that many people would find the idea of consciousness as product of neural interaction uncomfortable. It strips human of certain degree of divinity. I haven't read the book, but from the article this seems to be author's major concern.