It seems to me there's a long debate regarding the Neural Network and whether or not disabling it or keeping it enabled is the right way to tune the fueling on the newer Mopars.
I for one love the challenge and disable the network every chance I get to (Not much point in disabling it on stock cars).
Researching into the Neural Network for more details, the conclusion I've come to is that this is a type of "Artificial Intelligence" so to say, that puts many factors (IAT, VCT, ECT, VE, EQ Ratio, Airflow, Baro, Humidity, etc, etc) into consideration before outputting a signal.
I've created a couple images to explain my reason behind disabling the network. These images are a representation, and an idea of the strategy used for both operations. There are many many factors that go here, but for a 1+1=2 equation, that is probably the best I could come up.
Neural Network.jpgDisabled Neural Network.jpg
As you can see from the Enabled Neural Network, there is too much going on, so when a lot of people say they tune their VE tables but don't see a change, the image helps to show why you probably wouldn't get the end result you wanted. The ECU is calculating every sensor used in the Neural Network process before actually executing it's end result. The trick people use here to get fueling inline while keeping this network enabled is to scale the fuel injectors, which I've seen done 10 years ago on the Ford platform, but nothing since until now.
With the network disabled, you get to take control of the fueling process to get your desired AFR. This may take longer to tune then usual since you will have to re-do your VE tables and fight your adaptive trims to get fueling inline. I get -/+ 3% STFTs on cams, strokers, bigger injectors, superchargers with LTFTs @ 0%! (I guess to me disabling it is the right way to go.)