How bright is "twice as bright"?

That's a very interesting subject. I'm fairly certain that twice as much light output will not be percieved subjectively as twice as bright, if nothing else then because the iris in your eye closes down when the image gets brighter. I would think that this effect will change according to screen size, the bigger the screen the more your iris will be affected by screen brightness relative to room brightness. But in any case, there will be some amount of adaptation going on, so that the subjective difference will be smaller than the actual measured difference.

In practical terms, I would say that a drop from 20 to 14 fL over time, is something that you're not very likely to percieve in reality. As a switch-over, you will definately see the difference, and if you frop from 20 to 14 fL, and then change lamp and go back to 20, you would definately notice it. Side by side it would be very obvious. But I do believe that if you're already on the low side, a 30% drop in light output will be more obvious. I have no scientific data to back this up however, this is just a gut feeling, but to most people the difference between 14 and 20 fL will, on a large screen, be mostly how open your iris is. However, say you drop from 10 to 7 fL, that won't change your iris much as none of them are bright enough to "blind" you enough to really kick in the iris, so the difference will be more noticable. Once again, this is pure speculation to spark the debate, NOT presented as proven fact - this is my theory of this, and I'm very interested in hearing other people's input on this.

However, related to this, dB is a logarithmic scale that is not an absolute number, but only tells you the relation between two numbers. 90 dB in audio is 90 dB more than a specific amount of energy, defined as the threshold of human hearing. a dB scale for light output is probably not needed, but it might be interesting to look at contrast ratio's using a logarithmic scale, as in my opinion our perception of contrast is logarithmic, not linear. That would make 1000:1 30 dB, 10.000:1 40 dB, 100.000:1 50 dB etc. I would say that those numbers more closely represent the difference we percieve for each increase. For instance, the difference between (actual) 10.000:1 on/off and 100.000:1 on/off, isn't really that great, but the number suggests that it's a 10-fold increase. Saying that you go from 40 dB to 50 dB would be a more intuitive number to represent the percieved difference, I would say. But, once again, just throwing a ball up in the air here.

EDIT: Oh, and sorry if i'm clogging up the discussion by throwing in some subjective impressions, I'm just as eager as you to hear if there are any objective studies on this.