Several yours ago I posted a blog on the subject of Nikon's "new" TTL-BL flash. I had tried to convince someone who posted a blog of how it works [based on the fact that he helped design the chip before he retired] because it didn't seem to behave like the description in the manual. He didn't want to hear what I was saying - human nature, resistance to change etc. so I did a series of tests that proved it no longer only works for back-lit situations, but rather is superior to TTL flash. After all, since it has access to lens distance info and the meter reading it should be. The fact was that during the introduction of the D200 Nikon had decided to use the technology more and programmed the camera system to lie to TTL-BL when the background was totally underexposed, telling it that the background was perfectly exposed, so it would also try to perfectly expose the subject. Previously the program had simply tried to make a subject as well lit as the background on a bright day. This meant that if the background was underexposed then the program would make the subject look the same - underexposed.
A few years later I did the same tests again to prove something to someone and found that the results differed slightly - but in exactly the same way that matrix metering behaviour had changed over the year - this lead to the conclusion that TTL-BL flash is linked to the matrix metering program of your camera.
Basically TTL-BL takes ambient exposure into account as per this graph.
Here are the images from the TTL-BL vs matrix metering tests. TTL-BL in my D90, D40 and D5100 all behaved sightly differently with changes to the outer focus point, in exactly the same way that matrix metering changed between the cameras.
I recently ready a very true comment on a blog discussing equipment something along the lines of "Try to buy a camera body and flash from the same time period, they will work better together."