I think we are looking at the begining from the end. We are trying to understand emotions from things that contain multiple aspects of emotion generating responses. A song or a painting is made up of both appollian and dyonsian. To understand them as a trigger for emotion you need to break them out into parts. Perhaps an experiment like this. You have an elivator you scan the peoples body tempatures on the elivator and then when the doors open you play diffrent notes and see how the body tempature changes this should be fairly indictive of the types of emotional responses once you have a base you can begin to modfiy the tones pitches stings of notes ect.

the depth of our abaility to divide both aspects is also an important question. of the senses sight is probably linked the most to cognitve reasoning. perhaps this is why it is so easy for us to deferintiate emotional responses from the form of the object vs the texture color. where as a song this seperation becomes more difficult. I have also often wondered if this impacts are abaility to percive correctly how an animal interects with its enviroment for example a dog has a greater sense of smell its "view" of the world is diffrent.

the answer to cold vs warm and how you and your girlfriend painted them has great signifcance to the conversation (largely because it was involantary) but what it tells us i dont know. I think if someone spent alot of time looking at how kids paint and then doing alot of background analysis some useful information could be ascertained. Do abused kids paint mostly cold colors for example