Thanks all for contributing to this interesting debate.
I am definitely no expert but I wonder if the perceived colour changes in different environments have less to do with brightness and more to do with different colour temperatures. Eg we might perceive a bird as 'glowing' in morning or evening light which probably has as much to do with the warmer colours rather than the brightness of the light. However, perception of colour could also have to do with the amount of reflectance (is that a word?) in feathers, which would change depending on the direction and brightness of the light. A classic example would be hummingbirds in the US, where the colour can be quite drab in some lights and vivid in others. (NB I had written this sentence before I saw Ian's most recent post so it might have been overtaken by that.)
I suppose the difficulty with what I could call the 'perceptive' approach is that there is no absolute - it becomes about what the photographer wants to depict (perhaps based on some vague 'memory of the scene'). This has the potential to lead to a high variation in colour, and also the use of colour for dramatic effect (as seems to be routinely happen with landscapes these days). If we think of bird photography from a 'creative' perspective perhaps this doesn't matter - the image impact, drama, mood, message might be the main consideration. However if we think of bird photography more from a 'scientific' perspective (ie a record of the species) we might give more credence to 'accuracy', however determined. Overall, I tend to prefer a more subdued look, but accept that sometimes it doesn't make for a particularly interesting or dramatic photo.
Part of the challenge the 'non-expert' photographer faces is the different approaches taken by different post processing software, in their default approaches. I did a quick check of three different software programs on my computer using a particularly difficult type of photo, the male Australian King Parrot. This photo depicted the bird sitting on a greyish brown branch, with dark green foliage as the background, so there was not really a decent target for setting the white balance using the eye-dropper. I used 'neutral' in DPP as the baseline (with things like sharpening turned off, and had the white balance set to 'daylight' (5200K) in all cases purely to enable consistency across products. I compared this to using RawTherapee 5.8 (using their 'neutral' settings) and DxO Photolab 3.2 using the 'No Correction' preset which sets all sliders to 0. Obviously this isn't a scientific test in any way. However in each product the rendering was visibly different. DxO rendered the bird a warmer red; while there is obviously quite a bit of variation between the RGB values pixel by pixel, it had reds about 250-ish, but greens typically about 20 points higher than blues. DPP had about the same value in the red, but with the greens and blues with relatively equal values - a bit more green than blue. The bird appeared, to my eye, a cooler red, with a bit more pink on the screen. RT measures values as a percentage rather than on the 0-256 scale, but again the green and blue values were reasonably similar (close to 50%) with a little more blue than green. The actual appearance on the screen was different from either of the others, being much greyer. I would have liked to compare the RGB value for a specific pixel but unfortunately DxO does not appear to allow me to pick a particular pixel on the x/y matrix. My screen is a Dell, calibrated with Spyder 5 Pro. I have an old version of Lightroom, but I didn't bother looking at that. If nothing else, this illustrates the difficulty the typical photographer will have is simply operating 'by eye' on the screen using different software products in isolation. I find it very difficult to decide how, exactly, to set the red for these birds, and it can change perceptively in different lights.
Perhaps, at least for Canon, one option might be to use DPP neutral as a visual guide to help set the starting point for other software, given that it has been shown to be reasonably accurate.
Simon