Essentially higher frequencies are represented using fewer number of shades and many of them will go to zero. Not only due to human cortex but also that there are not much information at high frequency region most images.
True, although there will be more in images with a lot of fine, sharp detail, which is part of why those images will tend to look particularly bad when heavily compressed. An image of small text, for example, will have a ton of ringing artifacts around the edges, which gives it that “deep fried” look.
