While visiting the Fitzwilliam Museum in Cambridge a few years ago, I happened upon the French Impressionist Claude Monet's Rocks at Port Coton, The Lion Rock.
Painted during his 1886 visit to the French island of Belle-Île, the work depicts the famous Lion Rock, sitting in the water below a cliff whose top perfectly aligns with the distant horizon.
I didn't like it. But I also had a good idea as to why I didn't like it.
In 1999, influential neuroscientist Professor V. S. Ramachandran and American philosopher Professor William Hirstein proposed several principles they believed underpinned people's aesthetic preferences in images.
One principle was that people prefer generic viewpoints, where the spatial relationship between objects is largely similar despite small changes in the viewer's position.
For example, in the cartoon on the left in Figure 1 below, the foreground tree would appear somewhere between the peak and valley of the background hills even if the observer moved a little to the left or the right.
So, the general composition of the work is maintained despite small alterations in our viewpoint.
But if we look at the image on the right in Figure 1, the alignment between the foreground tree and the valley between the background hills would occur only from a single, specific viewpoint - so if the image were created with the observer moved a little to the left or the right, this alignment would be broken.
In general, generic viewpoints are more common, making the spatial coincidences arising from specific viewpoints relatively unlikely.
It is argued that these kind of spatial coincidences are seen by the brain as suspicious given their unlikely nature and, as a result, are unappealing.
In the case of foreground and background objects, it's possible that the parts of the brain processing this visual information need to exert additional effort to ensure that a spatial coincidence is just that - a chance finding - rather than indicating the background and foreground are somehow physically connected.
In the case of the Monet image, the coincidence between the clifftop and the horizon likely makes my brain need a few extra moments to decide that the distant cliff isn't physically continuous with the horizon, but merely appears so from this particular vantage point.
Over the 20 years since Ramachandran and Hirstein proposed that generic views are more aesthetically appealing, their work has been highly cited. Indeed, their paper is in Ramachandran's top five most cited works, which is impressive given he is a particularly highly cited neuroscientist.
Despite this, experimental support for their idea is limited.
Given this - and prompted by my encounter with Monet's Lion Rock - a team of students from the Department of Optometry and Vision Sciences and I set out to test Ramachandran and Hirstein's proposal.
We had people view a range of image pairs containing both specific and generic viewpoints between foreground and background objects and we asked them to choose which image they liked more.
Our results found strong experimental support for the idea that people find generic viewpoints more appealing.
This includes when people are presented with a Monet image manipulated to give a more generic view (Figure 2). This would have been the result if Monet were (substantially) shorter - or, more realistically, if he had found a vantage point with a different elevation.
Of course, there is more to a beautiful work of art than simply the spatial relationship between foreground and background objects. Various factors influence what we find appealing and this is reflected in the substantial variation in responses between our experimental observers.
While our manipulations improve the overall appeal of Monet's Lion Rock relative to the original, most people still find this manipulated work less appealing than many other works painted during his Belle-Île visit (Figure 3).
So really, the influence of the specific viewpoint in The Lion Rock is small compared to the sum of other factors influencing aesthetic appeal.
Indeed, Ramachandran and Hirstein outline a number of visual factors that they think influence what we find appealing - along with evolutionary rationales for why this might be so.
Using careful experiments to quantify how important these factors are might then allow us to better appreciate what non-visual factors - like cultural factors - affect what we find appealing.
Experiments like this might also help identify artworks whose appeal is in stark contrast to what we might expect, purely based on visual processing.
This could help expose a whole new area to explore to more fully appreciate what makes an artwork aesthetically appealing.
It's possible that the lack of unusual spatial alignments in generic viewpoints help our brains to easily separate foreground from background objects, and those visual cues that make processing the world a bit easier are also the things we find appealing.
While our research doesn't provide a 'formula' for creating good art, it does provide an insight into how certain aspects of what we find appealing are closely linked to fundamental aspects of how the eye and brain process visual information.
While it is unlikely that Monet's The Lion Rock would've been more appealing if he had been a couple of centimetres shorter, it might well have been more appealing had he moved further down the hill.
Better still, though, appears to be his decision to turn his easel towards completely different sections of the coastline.
