The main findings in this book are as follows.
People enjoy experiencing negative emotions when looking at art because they know they are not about real events. Just by knowing something is art, allows people to simultaneously have a negative and a positive emotion.
Arnheim is more right than Goodman because formal properties of images cannot be arbitrarily learned/associated with emotion, but are pre-trained by experience to be so linked. This accounts for a lot of cross-cultural agreement in the connotations arising from abstract forms. There are pre-trained isomorphisms between physical and mental life, between physical and mental states that are exploited by abstract art allowing visual metaphors such as black rage, fiery courage, to make sense.
In agreement with Greyson Perry's book "Playing to the Gallery", the quality of a piece of art is related to its familiarity, although Perry prefers to say its linked to the social status rather than the familiarity of the art per se, which I think makes sense.
Apparently a neural network has been trained to distinguish children's drawings from abstract expressionist art, although I am suspicious of overfitting and need to further investigate this.
Hans Eysenck's experiment on Visual Aesthetic Sensitivity is rejected as only a test of regularity of form, rather than of quality. Winner alludes to but provides no evidence for polysemy (many layers of meaning), complexity, non-predictability, and non-sentimentality as objective criteria for aesthetic judgement.
I was hoping for more Arnheimian details about the objective properties of artistic experience that humans valued, but there was not much of that, I suppose thats really quite challenging and hard to make valid.