As for judging colors, now with computers and the color picker tool anyone can tell what a color "should" look like just based on simple math and experience. Finding neutral is easy by looking at the numbers in RGB or Lab. As long as your RGB are equal it's neutral. As long as your a and b channels in Lab are 0 the color is neutral.
Skin tones are a bit different. You have to know that in Lab color space, skin is always more yellow than blue (a is a positive number) and always more magenta than green (b is a positive number). You also have to know that skin, except for babies, is always more yellow than magenta (b > a) but only a little bit. In CMYK color space for skin you should have no black, little cyan, about 2-3x more magenta and a bit more yellow than magenta.
If you look at skin color in your original photo the cheek on the girl is 80,11,29 in Lab color space which indicates that it is way more yellow than magenta (11 vs. 29). You can also look in the boy's shirt (90,-1,16) and see that the magenta is probably right but the yellow is way off (-1 is almost 0, 16 is way yellow!). In CMYK you have 9,27,48,0 on the girl's cheek. That means your cyan is fine, your magenta is 3x the cyan so it's still correct but your yellow is way too high with 48%.
So you know that the image is way too yellow all around. What you need to do is remove all that yellow. The easiest to do is in Lab color space by adjusting the b curve. You can also try to adjust it in RGB by adjusting the B channel. In CMYK you'd adjust the Y channel. There are many ways to skin this cat.
But if I had to guess, I think the image was taken under nasty fluorescent lights and it was not white balanced correctly in raw. If you have the raw file you'd want to white balance correct in your raw converter and set neutral off the boy's shirt, as that would correct your color in linear space. It's a lot harder to color correct in gamma corrected color space such as sRGB.
Oh and I most certainly hope and assume that you are using a color corrected work flow and you are looking at these pictures on a calibrated and profiled monitor. Otherwise there is no way for us to tell what we're looking at is the same thing that you're looking at. Even with profiled and calibrated monitors there will be differences due to gamut variance in monitors but at least it'll be as close to a level playing field as possible. You can't do color correction by eye without having a properly adjusted viewing environment. This is where color correction by numbers still helps though.
And to close with an interesting link just to show how adaptive and deceiving our eyes are take a look at how quickly and easily we adapt to chromatic differences:
http://www.sciencebuddies.org/scienc...Beh_p008.shtml