I think you misuderstand my point... There is no way to observe anything but behavior. My observations about the Turing Test is not that it is bad because it observes behavior, but that it is an avoidance of classifying any of the aspects of the characteristic it is supposed to be testing for. The result being that you therefore will not actually determine whether or not you found the unnamed thing you are looking for.
The second paragraph that you refer to I am trying to communicate the reasons that I think it important that we recognize intelligence as a large collection of characteristics, rather than trying to reduce it to definitions so oversimplified they become useless.
To quote Einstein: "everything should be made as simple as possible, but not simpler".
My article is trying to show that both extremes have been ineffectual.