> For instance, there would be much more effort made to check to see if various artistic works actually communicated things accurately by (say) studying whether

> For instance, there would be much more effort made to check to see if various artistic works actually communicated things accurately by (say) studying whether viewers who had an experience were surprised by aspects of the experience despite having previously consumed the relevant art. This is a fascinating idea. Imagine Facebook injecting various works of art into the timeline of 1 billion people (since the effect size will almost certainly be minuscule, you're going to need a large N). Then, as these people report various life experiences (ended relationship, etc), their happiness/despair/resilience scores (which are probably already being computed) could be assessed and the relationship to the art, if any, could be ascertained. (Caveat being that conducting a trial in the fashion I described would be pragmatic, but at least arguably unethical.)