I've said it before (here and here) - one of my biggest criticisms of Postrel's much heralded Substance of Style is that it does a good job of demonstrating that the aesthetic imperative exists, but does NOT provide a good explanation for why.

This article by Dennis Dutton presents my favorite angle for providing the answer -- the realm of evolutionary psychology:

The most famous example of sexual selection is the peacock’s tail. This huge display, far from enhancing survival in the wild, makes peacocks more prone to predation. The tails are heavy, requiring much energy to grow and to drag around. This seems to be nature’s point: simply being able to manage with a tail like that functions as an advertisement to peahens: “Look at what a strong, healthy, fit peacock I am.”

...The human tendency to create amusements, to elaborate and decorate everywhere in life, is therefore a result of mate choices, accounting for the evolution of dancing, body decoration, clothing, jewellery, hair styling, architecture, furniture, gardens, artefact design, images from cave paintings to calendars, creative uses of language, popular entertainments from religious pageants to TV soaps, and music of all kinds. Artistic expression in general, like vocabulary creation and verbal display, has its origins according to sexual selection in its utility as a fitness indicator: “Applied to human art, this suggests that beauty equals difficulty and high cost. We find attractive those things that could have been produced only by people with attractive, high-fitness qualities such as health, energy, endurance, hand-eye coordination, fine motor control, intelligence, creativity, access to rare materials, the ability to learn difficult skills, and lots of free time”

...Works of art, especially of fine art, therefore engage the higher faculties, and the pleasures they afford are of a different order than sexual or gustatory sensations of pleasure.

This is not a distinction many evolutionary psychologists have fully appreciated. For example, Randy Thornhill, agreeing with Donald Symons, says that “Pleasure, like all experiences, is the product of brain mechanisms, and brain mechanisms are the products of evolution...by selection” (Thornhill 1998). They leave no room here for any distinctions between pleasures directly implicated in the satisfaction of desires and the contemplative pleasure historically identified as aesthetic and artistic.

A key point that Dutton makes is echo'ed in Pinker's Blank Slate - people of all shades often fail to appreciate the crucial distinction between proximate and ultimate causality in many contexts.   The evolutionary psychologists that Dutton critiques above seem overly fixated on looking for proximate causality - A therefore B - to explain pleasure.   But this overlooks so many complex aesthetic experiences - for example, why is a movie like Lord of the Rings so much fun despite not possessing any glaring sexual overtones? 

Here, we must acknowledge the contribution that a theory of ultimate causality provides.   For example, the human preference & recognition of the lush green landscapes in LoR's "good" worlds arguably have an evolutionary basis.   As individuals, we aren't necessarily consciously aware of why we like the green worlds relative to the parched, volcanicly blackened worlds of the Orcs (or for that matter the rotted-meat, corpse-like appearances of the Orcs themselves).   However, a subconcious preference for this environment would have induced early man to pursue it which in turn would have contributed to a greater food supply and finally better sexual 'wealth' for that individual.