The JPEG converted by ViewNX should be the same as the camera JPEG, both are software processed. You have a Nikon so the camera JPEG is the result of Nikon's EXPEED image processor. ViewNX, also from Nikon, should produce the same JPEG.
Neither are trustworthy relative to your question.
You're asking an intriguing and uncommon question. It rarely occurs to most photographers because most photographers haven't had the experience of using different RAW file converters and getting very different results. It's far more typical that a photographer will invest in something like Light Room from Adobe, start using it, and never question that LR is pre-processing their RAW files -- spin applied up front.
It's necessary that the software pre-process your RAW files. There's no other way to get an image up on your display. What you've discovered is that different software from different commercial and non-commercial suppliers do that pre-processing with variation. Leading you to then very rightly conclude that what they're doing may be interfering with your desire for feedback on what you did behind the camera.
This came up in a post last week. You should read through it:
http://www.thephotoforum.com/forum/beyond-basics/260217-question-interpreting-histogram-lr3.html
What you're experiencing is unsettling -- me too, but it's not disabling. The variation I see between RAW file processors irritates me and I wish the industry would adopt an appropriate standard. In the meantime those differences aren't huge and there is some real hard physical activity going on that the software engineers can't cover up. If you peg a photo receptor on your camera's sensor, nothing is going to recover data that's not there. If you wanted to test different RAW file processors the way to proceed would be to hand each of them a bracket set of files beginning with major overexposure and see when in the bracket set they manage to successfully start recovering highlight detail. This would be tied to a physical reality and I suspect the various programs would perform with reasonable consistency. I'm going to do that one of these day.
Joe