In this video from about 1:15, it is stated that if you have an RGB value of (0.5,0.5,0.5) it is only 22% as bright as (1,1,1) rather than the expected 50% as bright.
Does this mean that RGB is adjusted for a logarithmic scale so that we perceive the (0.5,0.5,0.5) to be half as bright, or have a misunderstood the video?
Despite searching around a lot I can not find any other source of information that discusses this.
Two different effects are causing the observation mentioned in the video.
On one side, the vast majority of screens have a non linear response: if the RGB value is half as much, the emitted light intensity is not half as much. This behavior originally comes from cathode ray tube (CRT) displays, which produced different light intensities by varying the voltage used to generate the electron beam. From the article:
CRTs have a pronounced triode characteristic, which results in significant gamma (a nonlinear relationship in an electron gun between applied video voltage and beam intensity).
As CRT became ubiquitous, systems assumed this behavior to be present and took it into account. Nowadays CRT have mostly been replaced by LCD technology, which emulate this behavior for compatibility.
Gamma is usually considered to be around $2.2$, which means the light intensity follows a function $I=x^{2.2}$. This is where the 22% mentioned in the video comes from: ($0.5^{2.2} \approx 0.218$).
On the other side, the human vision is not linear either. According to Wikipedia, a typical sunset on a clear day is about 400 lux, and clear sky daylight is about 10000 to 25000 lux. Yet the latter doesn't feel over 25 times as bright as the former.
In the end, the non linear response of screens is a good thing, because it gives more precision where it is needed. Note that it is not logarithmic though, but closer to a square function. If you want to read more on the topic, you can look for articles on "gamma correction".
