Interesting exercise. As I moved the tiles around, my perception of a given tile changed depending on it's location in the line. Had a bit of fun just playing with that.
And since we see with our brain, many other factors can influence the outcome at any given time as well and renders it somewhat subjective.
For the morbidly curious - Age 58 male, calibrated '06 iMac, score 7(trouble with blue/green). At this time.
It's well known that the perception of a specific color will change depending on its proximity to an adjoining color. This effect is also very obvious with shades of black and white. A pale grey square next to a deeply black square may appear as white. But the same pale grey square next to an even paler grey square will appear grey, and the paler grey will then seem white.
Checking on the RGB values of the squares in the first row in this test, I find a few pairs of values that are probably too close to differentiate, no matter how well-calibrated your monitor or how good your eyesight, but I could be wrong .
They are: (1) 164, 114, 92, and 164, 114, 88, (2) 163, 121, 81, and 163, 119, 82, (3) 162, 117, 84, and 163, 115, 86.
The 0-255 system of 8 bits per channel creates a maximum range of 16.7 million colors. There's no way any person could visually distinguish between all those different shades (outside of numerical description). Most images, even high resolution images processed in the ProPhoto color space in 16 bit mode (capable of several billion different shades), in practice contain only a few thousand, or fewer, perhaps only a few hundred, visually distinguishably different shades, so I believe, but you understand I haven't personally counted all the different shades that may exist in any one of my photographs.