Insects as Art Lovers: Bees for Van Gogh

Installation artist Julian Walker and insect scientist Lars Chittka collaborated on an experiment to see if bees were art critics, distinguishing between yellows, flowers and artists.

A series of flower species as seen in the visible light (left) and in the ultraviolet (right) which bees but not humans can perceive. Photos: David Pye and Lars Chittka.

The visual aesthetics of animals, and the ways they perceive the world, often differ fundamentally from those of humans. A biologist's view is that these differences have, at least in part, evolutionary roots. In an attempt to provoke thinking about the subjectiveness of visual appearance, and its biological relevance, a biologist and an installation artist got together to launch a Sci-Art project in which bees were confronted with a series of paintings highly appreciated in Western society, such as Van Gogh’s Sunflowers.

Two bumblebee workers attempting to imbibe nectar from a copy of one of Vincent Van Gogh‘s Sunflowers (1888).

A "Colour and Design" symposium of the Linnean Society in 2003 was dominated by physicists and artists. Yet, colour is neither purely physics nor a domain of the arts: it is, to a large extent, biology. The coloured world we see is not the real or the physical world - instead, the colours we perceive are filtered through the specific sense organs that we have acquired in evolutionary history. Colour vision systems differ widely between different animal species, and the reason is that different aspects of the coloured world are biologically relevant for different species. Our goal was to raise appreciation of this fact in an audience not specifically trained in the biology of vision, and to use live bees’ attraction to a series of paintings as a vehicle towards this goal.

The insight that flowers (and their colours) have not been created solely to please us humans dates back to the 18th century. The idea that flowers are in fact sex organs, designed to attract the services of pollinators, is commonly attributed to Sprengel, who entitled his 1793 book The uncovered mystery of nature. Flower colours are clearly important signals to bees, since flowers provide bees with nectar and pollen. But how do insect pollinators see colours? In 1874, Lord Rayleigh pointed out that “The assumed attractiveness of bright colours to insects would appear to involve the supposition that the colour vision of insects is approximately the same as our own. Surely this is a good deal to take for granted”. Lord Rayleigh was right: in 1924, Kuhn discovered that bees see ultraviolet light, and in subsequent decades a wealth of information has been collected on how bees process colour information. Bees (including the familiar bumblebees and honeybees) have three colour receptor types, with maximum sensitivities in the ultraviolet (UV), blue, and green. Brightness, a parameter so fundamental to our own visual experience, has a relatively minor role in bee colour discrimination. But bees use a single colour receptor, the green receptor, for detection of flowers from a longer distance.

Old world primates, including humans, have three colour receptor types (typically called blue, green and red receptors). The light sensitive pigment of human photoreceptors has some sensitivity to UV light, but such radiation never reaches the retina because it is absorbed by the lens. Claude Monet, an avid painter of flowers, had the lens removed from his right eye in 1923 due to a cataract, and would therefore have been able to see UV patterns of flowers.

It is thought that the mammalian ancestors of primates had only blue and green receptors, and that the red receptor is an adaptation to frugivory (eating fruit). Flowers do not play a major role in the diet of humans and other primates; the biological significance of human attraction to flowers is discussed later. It is clear from the above, however, that there will be differences both in perception and in meaning for human and bee observers of floral colours – and that perceptual differences have evolved alongside the biological significance of the objects in question.

The authors of the current article, a biologist and an installation artist, were drawn to each other’s work by the fact that bees and people obviously are both drawn to flowers, and that one of the most obvious ways that humans express this in Western culture is by creating and appreciating paintings of flowers. By presenting such paintings to bees, we hope to address people with an interest in colour (but not necessarily a training in the biology of colour vision). We hope to stimulate thinking about the fundamental philosophical issue of whether perception reflects reality, about the nature of the image as object, and about the biological meaning of colour for different receivers.

What are bees’ favourite paintings?
We chose reproductions of two paintings that contained flowers, and two that didn’t. These were: Vincent Van Gogh’s Sunflowers, Paul Gauguin’s A Vase of Flowers, Patrick Caulfield’s Pottery and Fernand Léger’s Still Life with Beer Mug. Bumblebee nest boxes were connected to a flight arena. These bees had never seen natural flowers prior to or during the experiments. We placed the paintings onto the floor of the flight arena and bees’ responses to objects in the paintings were filmed using a digital video camera.
Van Gogh’s Sunflowers proved to be the most popular: of 146 approach flights by bees to the painting, 99 were to flowers. Bees mostly approached the high contrast margins of flowers, or the contrast between periphery and centre. Interestingly, 17 approaches were to the blue-on-yellow Vincent signature.
15 landings were recorded in total, of which 13 were on flowers. Caulfield’s Pottery came second in terms of approach flights (138) but only 4 landings were observed. Gauguin’s A Vase of Flowers attracted only 81 approaches, of which 25 were to blue flowers. Two landings occurred on the blue flowers in the upper right, 9 were distributed over other flowers of the painting. On Léger’s Still Life with Beer Mug, a light blue square was frequented most strongly (24 out of 117 total approaches). In summary, the fraction of approach flights that terminated in landing was substantially higher in the paintings with flowers (11%) than it was in the paintings without flowers (4%). Thus, there is evidence that the flower paintings have captured the essence of floral features from the viewpoint of a bee, and that these features are recognised by bees that have never been exposed to flowers before.

What do bees aesthetic preferences tell us?
In more standard, controlled laboratory measurements of visual pattern attractiveness for flower-naive bees, we had previously found that bees will prefer blue and especially violet over other colours, which is what was also found in this study. The evolutionary explanation is that flowers with these colours offer high nectar rewards in nature. Hence, 'favourite colours’ (in bees) have biological significance; we assume that selection has favoured individuals which prefer colours associated with nutritional desirability.
How do human observers react to presentations of bees visiting flowers in paintings? When our study was first published online in 2005, press reaction was more interpretative than we had expected, the research being described as “important” (Discovery Channel), “cutting edge” (Art Monthly), and even meriting a cartoon in the Sydney Morning Herald. The study was covered on BBC and ABC television news, Science, New Scientist, The Times, and multiple radio stations). We were excited by the readiness with which reviewers proposed that the bees’ response to the picture proved its value, as if the biological ‘rightness’ of the image confirmed its aesthetic value. Such willingness to delegate aesthetic judgement to bees raised the possibility that our responses could be linked to a wider biological visual reaction based on survival and the assessment of resources.

Some viewers commented on the absurdity and surrealism of seeing live bees in an out-of-place context (paintings), yet in another sense the bees do seem to belong (since the paintings contain flowers). In some cases there was an assumption that the fact that the bees were attracted to the centres of the flowers in Van Gogh’s painting indicated that the artist had ‘unwittingly’ captured some essence of the flower, which rendered the painted flower attractive to bees. Elsewhere, scepticism that the bees were attracted merely by the choice and distribution of colours was mixed with some concern at the possibility of applying biological determinism to visual art. Some artists, however, also felt that bees were mistaken, or were indeed ‘invading’ the painting, whereas biologists felt that the intimate signal-receiver relationship between flower and bees had been strangely thwarted. Inherent in all these interpretations is the implication that flowers in paintings are not really meant for bees. They are created by humans for human observers. This raises an interesting question: why is it obvious that flowers rendered by painters should be different from those which have evolved to attract bees?

Indeed, for thousands of years, humans have reshaped flowers to their liking, either through horticulture or through pictorial representation. Flowers play a major role in most cultures, and the flower trade is a global multibillion dollar enterprise. For example, the Netherlands alone exported cut flowers for more than 2 billion dollars in 1992. Could there be an evolutionary explanation for human attraction to flowers, and the fact that humans obviously prefer different floral features than those which selection has acted on to address bees? In our evolutionary history, paying close attention to flowers might have conferred strong selective advantages. Even if flowers may play only a minor role as food for primates, they can be indicators of resource availability: they might correlate with the presence of water, and indicate future availability of fruits, nuts and honey, and they can be used to identify plants for medicinal purposes. Is human aesthetic appreciation of flowers in part based on a primordial interpretation of a landscape with flowers as one that could support human foraging?

If flowers carry different information for humans than for bees, then human horticultural selection and pictorial representation is expected to emphasise the traits that indicate relevant resources for humans. One floral feature that has been clearly exaggerated by humans is flower size, and the number of floral petals and sepals. It remains to be determined whether these floral traits are indicative of future fruit set or water availability. What about flower colour? It is clear that human colour selection on flowers would have excluded the ultraviolet, but even within the human visual range, qualitative inspection of any flower store indicates that flower colours have been strongly altered to match human preference. Blue flowers seem under-represented, whereas red and orange colours are common, despite being rare in natural flowers in European temperate habitats. Curiously, however, these are the typical colours of primate-eaten fruits. Could human flower colour preference be a result of our primordial lifestyle as frugivorous mammals, a lifestyle which has shaped the way we see colours? Clearly, a SciArt project such as this one cannot provide scientific answers to these questions. For that, we will have to employ conventional scientific practice.

This is a slightly abridged version of Insects as art lovers: Bees for Van Gogh originally commissioned by and published in Antennae: the Journal of Nature in Visual Culture, Issue 3 Vol 2, 2007 (

Lars Chittka is a Professor of Sensory and Behavioural Ecology at Queen Mary College, University of London. He works on the visual perception, communication and psychology of insects.
Julian Walker is an installation artist trained at the London Guildhall University and Central St Martins School of Art.

Support independent writing on the visual arts. Subscribe or donate here.