• Ultraviolet Photography
  •  

My Bee Vision Notes, Colour Charts, Paper Links

Insect Vision
17 replies to this topic

#1 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 June 2013 - 01:43

This is not an article, rather it is a list of references, notes and charts which were made as I read about Bee Vision.
Kindly note that my bee vision color charts are pure hogwash! They were fun to attempt, but do not represent authentic bee vision.


Horridge is da Man, for sure !!!!!
Adrian Horridge: Biological sciences, australian national University, canberra, acT, australia
correspondence: adrian horridge 76 Mueller street, Yarralumla, acT 2600, australia
Email horridge@netspeed.com.au

Adrian Horridge
What Does the Honeybee See? And How Do We Know?
https://press.anu.ed...-how-do-we-know
Superb!!
May be downloaded free.



http://adrian-horrid...rs%20on%20bees/

Horridge: How bees distinguish colors
Abstract: Behind each facet of the compound eye, bees have photoreceptors for ultraviolet, green, and blue wavelengths that are excited by sunlight reflected from the surrounding panorama. In experiments that excluded ultraviolet, bees learned to distinguish between black, gray, white, and various colors. To distinguish two targets of differing color, bees detected, learned, and later recognized the strongest preferred inputs, irrespective of which target displayed them.
First preference was the position and measure of blue reflected from white or colored areas.
They also learned the positions and a measure of the green receptor modulation at vertical edges that displayed the strongest green contrast. Modulation is the receptor response to contrast and was summed over the length of a contrasting vertical edge. This also gave them a measure of angular width between outer vertical edges.
Third preference was position and a measure of blue modulation. When they returned for more reward, bees recognized the familiar coincidence of these inputs at that place.
They cared nothing for colors, layout of patterns, or direction of contrast, even at black/white edges.
The mechanism is a new kind of color vision in which a large-field tonic blue input must coincide in time with small-field phasic modulations caused by scanning vertical edges displaying green or blue contrast. This is the kind of system to expect in medium-lowly vision, as found in insects; the next steps are fresh looks at old observations and quantitative models.

Anti-intuitive-bee-vision.pdf
Encycloped2015.pdf
How bees distinguish black from white.pdf
How bees distinguish colors.pdf
How bees distinguish green-and-blue-mod.pdf
How bees distinguish mirror images.pdf
How bees use green-and-blue-modulation.pdf
LostQueensAmerBee J.pdf
Parallel inputs in bee vision of colour.pdf
Tihany2015.pdf
colorvisionAusBeeJ.pdf
green-and-blue-mod_082515.pdf





Lars Chittka Home Page
http://www.sbcs.qmul...arschittka.html

Chittka Publications
http://chittkalab.sb....ac.uk/Pub.html

Bee Sensory and Behavioural Ecology Lab
Chittka Lab
http://chittkalab.sbcs.qmul.ac.uk/

Floral Reflectance Database
http://www.reflectance.co.uk/

Chittka, L., Shmida, A., Troje, N., & Menzel, R. (1994) Ultraviolet as a component of flower reflections, and the colour perception of Hymenoptera. Vision Research, 34, 1489-1508.
Notes:
  • Bee perceptual colour space is trichromatic
    with peaks of UV @ 345nm, blue @440nm, green @535nm.
  • Bee colours are: UV, UV-blue, blue, blue-green, green, UV-green.
  • Bees have two opponent channels.
  • Bees see foliage as uncoloured.
  • Bees do not discriminate brightness, i.e., they don't see black & white.
  • 1063 petals from 573 flower species were studied.
  • 68% of the flowers were UV-absorbing.
  • There are no cyan flowers or UV-reflective black flowers.
  • Green flowers are very rare (i.e., red-absorbtion is very rare in flowers).
  • Most blue flowers have a red component.
  • Orange or brown flowers were not included in the study.
*******************************

From other papers (listed here: http://www.ultraviol...tanical-papers/), we have that:
  • The longwave bee receptor peaks around 540nm
    but has an extended tail which tapers out at about 650nm.
  • Bees can discriminate between green, yellow, orange and red reflecting objects
    regardless of an object's UV reflectivity. Some cues used for this are motion parallax
    and pattern and shape.
  • Leaves moderately reflect green but also some UV and some blue.
  • Leaves appear achromatic to the bee.
In other words, bee colour models should include broadband spectral reflections of objects
and not just monochromatic models.

*******************************

In the Bee Colour charts below the flower colours and the mappings to the bee colours are taken from the above referenced paper - except for orange flowers which I added. Basic flower colours in the paper are blue, purple/violet, magenta/pink, white, cream, yellow and red. Orange, brown and green flowers are not considered.

I note that Chittka uses a colour hexagon of dark magenta, blue and green to represent a bee colour space.
Also note that the UV reflectivity in these charts is "all or nothing". This does not model reality well because we quite certainly have some flowers which are somewhere in between, neither UV-dark or UV-bright.

I have not added green foliage to these charts.

Cream coloured flowers, mentioned in the paper, are an unsaturated yellow.
Bees do not discriminate brightness, but I have used 75% grey for the achromatic case.

The bee vision colours produced by stimulation of their G and/or B receptors only (no UV stimulation) are easy to model with our human green, cyan and blue colours.

For a flower coloured pure red having no UV reflection and thus no stimulation to any bee receptors, I've assigned a light grey colour. I suppose in a sense this flower colour (+R-G-B-UV) is a "bee-black" (-G-B-UV). But if bees have no brightness discrimination, then grey makes more sense in a model. Note, however, (and mentioned somewhere in one of the papers) that there prolly are not any "pure red" flowers in nature. Most all flowers we consider "red" also reflect some blue.

It is when a flower colour reflects UV that we run into problems because we must assign some colour to the bee-colour "UV". As with red, there are prolly no "pure bee-UV coloured" flowers in nature, but we must include this in a model anyway.

CHART 1:
For the first chart the bee UV-reflective colours are modeled with a 120º spread of colours from purple to orange (longwave to shortwavew). This mimics the 120º spread between blue & green for the bee UV-absorbing colours in such a way that UV-reflective and UV-absorbing colours do not overlap in any way.
In this chart the bee-colour B+UV, or bee-blue, is assigned purple (128,0,255), the bee-colour G+UV, or bee-green, is assigned cerise (255,0,128) and the bee-colour UV, or bee-UV is assigned orange (255,128,0).
Why do I not like this chart? Because for "bee white" - when all 3 bee receptors are stimulated equally - we wind up with a magenta representation. That is so very counter-intuitive.
Attached Image: BeeChart120Deg.jpg

CHART 2:
For the second chart I decided to emulate "bee-white" with white which makes a little more sense to me. This permits yellow (255,255,0) to be assigned for UV-reflective yellow flowers, orange (255,128,0) to UV-reflective orange flowers and red to UV-reflective red flowers. Again we have the problem of using a model containing a colour which bees cannot see, but as a model it holds together better that the preceding chart, imho. At a glance any UV floral photograph coloured with this model would show which parts are UV reflective if you see colours having some amount of red in them.
Attached Image: BeeChartBeeWhite.jpg

CHART 3
This chart is a result of my thinking that UV-reflective flower colours should be modeled with very light colours. I simply lightened the preceding UV-reflective bee colours to make this chart. The snag here is that if bees do not discriminate brightness, then light/dark colours in a model are not quite right. Oh well, a model is simply a model and not the real thing.
Attached Image: BeeChartReflective.jpg


*************

RED FLOWERS

23. Chittka, L. & Waser, N.M. (1997). Why red flowers are not invisible for bees. Israel Journal of Plant Sciences, 45: 169-183 http://chittkalab.sb...r_J_Pla_Sci.PDF

Boundary between orange & red between 593nm and 625nm with a mean of 611nm.
So red starts beyond 611nm.

Bee visible spectrum reaches up to red at about 650nm,
but bees discriminate single wavelengths only in the range from near UV 350nm to green 550nm.
So when adjusted for equal brightness (Abney's law),
monochromatic lights from green to red are indistinguishable for bees.

Most flowers have simple broadband reflectance functions.
Reflectance curves of many flowers are essentially step functions. (41% in experiment.)

Red flowers of 3 types: UV peak, blue peak, red peak.
  • reflectance peak below 400nm in the UV: bee sees bee-UV. Papaver rhoeas.
  • substantial reflectance in blue: bee sees bee-blue. Dianthus carthusianorum.
  • low reflectance over 300-600nm range & reflectance everything above 600nm. Ipomopsis aggregata. bee sees ?
UV absorbing red flowers: Sarcotheca celebica, Mimulus cardinalis, Zinnia elegans, Salvia splendens(bumblebee), Lobelia cardinalis(bumblebee&solitaryBee), Nonea pulla.
Also beetle pollinated Anemona coronaria, Ranunculus asiaticus, Tulipa agensis.

There are several other examples of bee-visited red flowers whose UV reflectance still needs to be examined. Carpenter bees (xylocopa californica) rob the tubular "hummingbird" flowers of chuparosa, Justicia californica in warm deserts of the USA and honeybees secondarily use the resulting holes to obtain nectar. (N. Waser, unpublished). The same species of carpenter bee robs (but also pollinates) the red tubular flowers of ocotillo, Fouquieria splendens and these flowers also attract other solitary bees and honeybees along with many other insects and hummingbirds (Waser
1979).

Red flowers visited by bees: Pedicularis densiflora, Penstemon centranthifolius, Corydalis cava, Peraxilla colensoi.

Red flowers tend to be rare in areas that lack pollinating birds including central Europe.

*******************

CONICAL CELLS & SPARKLE

Glover, B.J. & Martin C. (1998) Heredity 80, 778-784
They argue that conical cell sparkling sheen bedazzles and attracts bees and perform an experiment about this.

But(??) Dafni, Lehrer & Kevan Biol.Rev. 72, 239-282 say that angular resolution of a bee's eye is far too coarse for it to see sparkle.

But bees easily learn to distinguish petals of different textures with their antennae or feet.
Menzel, R. in Neurobiology of Comparative Cognition, 1990.

BEES COLOUR PERCEPTION IS PLASTIC
Reser, Witharanage, Rosa & Dyer (2012) Honeybees (Apis mellifera) Learn Color Discriminations via Differential Conditioning Independent of Long Wavelength (Green) Photoreceptor Modulation LINK

BEE VISION: Each facet "sees" differently
Wakakuwa, Kurasawa, Giurfa & Arikawa (2005) Spectral heterogeneity of honeybee ommatidia LINK

INTERPRETING UV "COLORS"
Kevan, Peter G. (1979) Vegetation and Floral Colors Revealed by Ultraviolet Light: Interpretational Difficulties for Functional Significance. Am. Journal of Botany. 66(6): 749-751
Andrea G. Blum
Often found hanging out with flowers & bees.

#2 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 2,735 posts
  • Location: Sørumsand, Norway

Posted 02 June 2013 - 08:31

After having a quick look at the FRED data base, it strikes me that were the scientists building this also UV photographers, conclusions would be different. We could be polite and saying that a scientists only measuring a single aspect of an object will have difficulty grasping the whole as it were. The spectrometer is a poor substitute for a lens even though both use reflected light as a source of information to process.

However, let us build a data base of UV signatures comprising at least the same number of species and then try to see if we can come up with generalisations of our own. We already have documented that low UV reflectivity does not entail the flower is dark in UV light (thanks to the conical cell structure, an aspect of flowers not existing in the FRED data base).

#3 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 June 2013 - 15:44

I note that the Chittka Lab researchers do seem to be aware that singling out any one aspect of the UV mystery is, as they say, "improper". Please see this link to a list of reference papers in another post:
http://www.ultraviol...tanical-papers/

I have extracted from a paper listed there - Chittka & Kevan (2005) Flower Color as Advertisement - some of their observations:
  • Singling out UV reflections is improper in studies of pollination.
  • Ignoring the nature of the colours of the backgrounds against which flowers are presented is improper.
  • Floral colours should be considered with:
    • Animal's colour vision system
    • Flower's background colour (.....we have been talking about this...)
    • Flower/background contrast
    • Visual, olfactory or other cues (....again, iridescence as cue....)
    • Presence & exposure of rewards
    • Other co-flowering species
They have also written about conical cells.
And have noted that bees can see iridescence and do indeed use it for detecting targets.

Ideally, our database and their database will both be useful to researchers.
I hope so, anyway. :)
Andrea G. Blum
Often found hanging out with flowers & bees.

#4 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 2,735 posts
  • Location: Sørumsand, Norway

Posted 02 June 2013 - 16:06

The data presented in the FREd base is not given in a proper botanical context. Thus, the frequent abuse of botanical terms is strange given this activity is conducted in a university setting. Often measurements are cited as taken from a "whole flower", probably meaning it is the average of a petal? They use the term "leaf" for petals and did not recognise the capitulum of Asteraceae as being different from a solitary flower all of which I find pretty confusing. And so on. I noted also they did not present Pilosella and true Hieracium as distinct genera, the measurements overlapped, and they also named a lot of Hieracium microspecies which is at variance with the lax attitude towards botany elsewhere. Taraxacum "officinale" (probably refering to a microspecies in sect. Ruderalia) is cited as having a calyx which is plainly wrong (it has an involucre wih two whorls of phyllaries) and I find it strange the refererence is to measurments of its "upper side". That would be the inside of the phyllaries which normally are not exposed at all. Or do they refer to the outer whorl of bract-like phyllaries in the involucre? As the human colour is claimed to be yellow, and neither interpretation of their term resolves to anything yellow, I am none the wiser. Probably will have to write the research lab for an explanation?

Well, enough of that for the time being. We have to continue our own quest for amassing UV appearances and later, when the data set has reached a critical mass, try to correlate our observations with existing measurements.

#5 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 June 2013 - 17:45

Good points, all.
Thank you for posting your concerns so that we may learn about them eventually.

Here is the link for the FReD methodology: http://chittkalab.sb...LOS1%202010.pdf
I have not read this myself, but will try to get to it soon.
Andrea G. Blum
Often found hanging out with flowers & bees.

#6 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 2,735 posts
  • Location: Sørumsand, Norway

Posted 02 June 2013 - 17:54

What I found most disturbing in the write-up of FRED methodology is that spectrometric measurements are said to be averaged over entire flowers, or petals. We know in many cases a flower is divided into regions with different UV properties and frankly just averaging away such details is food for thought in regard to later data analysis.

#7 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 June 2013 - 18:27

I have strong misgivings about such spectrometric averaging over the entire flower if/when it is done for flowers which exhibit variations in UV reflectivity. It is good that you have noted this so that it can be taken into account by FReD users.
Andrea G. Blum
Often found hanging out with flowers & bees.

#8 Nico

    Nico Chalwatzis

  • Members(+)
  • 121 posts
  • Location: Bensheim, Germany

Posted 02 June 2013 - 19:19

Andrea & Bjørn,
A very interesting discussion. Thanks also Andrea for posting the colour-charts. Do you have a way to generate these bee-colours in a (composite) image?
If I get it right you put UV in the red-channel ...? I’ve also been experimenting with this one example, below. The last one seems close to your colour scheme. - Is that right?
Best,
Nico


Ranunculus ficaria:

visible light:
Attached Image: NCH_P1070923_130414.JPG
image reference: NCH_P1070923_130414

UV-photo:
Attached Image: NCH_P1070924_130414.JPG
image reference: NCH_P1070924_130414

simulated bee-colours: RGB -> G, B, UV
Attached Image: NCH_P1070924_RGB_UV_b_130414.JPG
image reference: NCH_P1070924_RGB_UV_b_130414

simulated bee-colours: RGB -> UV, G, B
Attached Image: NCH_P1070924_RGB_UV_b_rev_130501.JPG
image reference:NCH_P1070924_RGB_UV_b_rev_130501

#9 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 2,735 posts
  • Location: Sørumsand, Norway

Posted 02 June 2013 - 19:19

Andrea: Show me a flower that is entirely uniform in regard to UV - I haven't seen that so far.

Nico: I think one should look into the remapping software made by B Lincoln for making multispectral or multichannel colour composites.

On a general note: you can map in principle a three-stimulus colour response to any 3D colour space you like. The colours are equally false. The scheme proposed by Chittka uses the human colour equivalent so is not a complementary (primary colour additive or subtractive) system. Whilst the labels confer the underlying components, so has a typological nicety to it, the colour scheme itself makes comparison between species complicated.

#10 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 June 2013 - 20:41

Bjørn, there are some flowers that are "almost uniform" when observed at some distance by the camera.
See the recent Biscuitroots I posted. Geez are they ever bland in UV !!
For example: http://www.ultraviol...af-biscuitroot/

But, yes, if one were to get closer and photograph a portion of such a flower,
then no doubt variation in UV-reflectivity will appear.
I think you would have to define some kind of frequency metric about how rapidly the UV-reflectivity is changing.
Then using that you could decide upon a reasonable definition of "uniform".
Perhaps "uniform except upon a set of measure zero." (...sorry, a little mathematical jest there....)
**************************

Nico, no, I don't think my bee colour "flight of fancy" fits into an RGB channel model which, naturally,
works best only for human eyes.

Added Later: Well, maybe RGB channels could work. Your RGB-> UV,G,B works well enough for the R. ficaria.
Would it work for other flowers? Don't know. I'll leave it to you to experiment. :)

The flower_colour to bee_colour_name mapping is correct as per the cited papers.
And the bee blue, bee blue-green and bee green - which I mapped to human blue, human cyan and human green -
seems reasonable as a false colour mapping because bees do perceive some version of those colours.

But my interpretation of the bee UV, UV-blue and UV-green is my own.

My bee UV-blue mapping to a human violet/purple (or whatever it is called)
seemed reasonable because violet/purple naturally occurs under blue.
The only difficulty is that R+B = magenta in human vision.
Ugh!! I did not want to suggest red by using magenta there.
So I picked an r+B colour to simulate human violet and thus to simulate bee UV-blue.
I think the colour started out as (128,0,255) before jpg compression and web colours messed it up.

Similarly, my bee UV-green mapping to human yellow seemed reasonable
because yellow naturally occurs above green.
I'm showing R+G yellow (255,255,0). But if I reasoned as I did for the bee UV-blue,
I should have chosen r+G (128,255,0) which would be more greenish-yellow. :D

The mapping for bee UV (i.e., UV-reflective red flowers) is the one which gives me fits.
I chose that dark magenta simply to finish the charts. I don't like it.

So for your Ranunculus ficaria, the UV-reflective human-yellow outer portion of the petals
would map to yellow or greenish yellow.
And the UV-absorbing human-yellow center would map to green.
You'd probably have to paint it in in Photoshop. :)
[Added Later: I didn't see your last yellow & green R. ficaria until after I had written this!! See note above about RGB->UV,G,B mapping.]

(Interestingly, I note that the Precision U filter would record R. ficaria as having bright yellow tips and a dark greenish center after white balancing.)
Andrea G. Blum
Often found hanging out with flowers & bees.

#11 Alaun

    Member

  • Members(+)
  • 311 posts

Posted 01 January 2014 - 16:22

A very nice discussion on the visual perception of bee's vision, on the historical scientific work as on actual scientific publications, is given by Adrian Horrige:

"WHAT DOES THE HONEYBEE SEE ? and HOW DO WE KNOW? A CRITIQUE OF SCIENTIFIC REASON"
published by 2009 ANU (Australian National University) E Press

Link to E Press:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3292233/

Link to the publication there:
http://epress.anu.edu.au/wp-content/uploads/2011/02/whole_book5.pdf


If you go to page 141 there and take a look at Fig.6.8 e (indicating the response on the different colors comparing bees and humans), it comes to mind, that just a one to one transposing of the "three bee colours" to the "three human colours" might not be close to the bees' perception of colours at all.
Werner

#12 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 02 January 2014 - 15:13

That is a good overview of the research on Bee Vision. Thank you for the link.

Indeed, we can never know how another creature perceives its world. :)
Our simple models of Bee Vision do give us some insights and make it obvious (at least I think so)
that the way the bee sees a flower is different from the way our UV-pass filtered cameras see a flower.
So that is interesting and useful.
Bee vision models are also educational for the general public and school children -
as long as it is made clear that a model is not the actual perception. :rolleyes:
Andrea G. Blum
Often found hanging out with flowers & bees.

#13 Bill De Jager

    Member

  • Members
  • 265 posts
  • Location: Northern California, USA

Posted 09 January 2014 - 02:59

Meanwhile, other pollinators should not be forgotten as they are very important in many parts of the world. Here's an insect that sees red, green, and UV as primary colors: http://www.ncbi.nlm....pubmed/22526111 . I wonder how much of the variety we see in the UV behavior of various flowers has to do with what the target pollinator(s) are? I would think quite a bit.

#14 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 09 January 2014 - 17:33

Thank you for the link.

Indeed, pollination is a fascinating and complex topic.

Some botanical scientists think the pollinators and flowers co-evolved. Others think the flowers evolved to attract pollinators.

Flowers have visible signatures, UV signatures, scent lures and nectar lures. The presence of conical cells on flower lip landing platforms plays a physical role too.

And yes, we should not forget non-bee pollinators. There are beetles, bugs, wasps, flies, night-flying moths and more. As well as birds - some (all??) of which also have some UV vision. Pollination may also occur via wind and rainwater - or even simple gravity in "self-pollinators". I think certain rodents are also considered pollinators for some species of desert flowers. I frequently find mobs of beetles in cactus flowers.

Did you know that flowers pollinated by night-flying moths are usually white? This must make them more visible in very low light or moonlight.

We are lucky to have discovered such interesting subjects & topics via our UV photography explorations.
Andrea G. Blum
Often found hanging out with flowers & bees.

#15 DaveO

    Aussie Bunyip

  • Members+G
  • 735 posts
  • Location: Maldon, Victoria, Australia

Posted 10 January 2014 - 09:02

In one of those serendipitous moments, I recently posted images of Nicotiana suaveolens whose flowers open at night, and yes, it is white in visible light and blue in UV.
Cheers,
Dave

Edited by DaveO, 10 January 2014 - 09:02.


#16 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 24 October 2015 - 17:48

The latest but not last incarnation of mapping between Flower Colours and Bee Vision.
I no longer like +R+UV to be Red, so the next chart will reflect a change.
I also need to include Brown as a flower colour.
Also, here I used grey as the Achromatic tone rather than black/white. This was because of reading that Bees do not detect brightness quite like we would think. But models have to make sense also to the human viewer, so I might go back to black/white

Attached Image: BeeColorChartsV23 copy.jpg
Andrea G. Blum
Often found hanging out with flowers & bees.

#17 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 05 May 2018 - 18:55

Free (1970) found that bees preferred symmetrical radial patterns, then bilateral symmetry and then irregular patterns, and also that bees landed at the edge of a plain target, but on a spot at the centre of a circle. Bumblebees prefer to land on flowers that are symmetrical.





In his earliest experiments, von Frisch found that flower-like radial or concentric patterns of the same size were easily distinguished, but that triangles, squares, discs and ellipses were not (Figures 1.2 and 1.4), and a chequered pattern of squares was not distinguished from one of triangles.




Later, Hertz found three classes of patterns—stars, circles and irregular blobs (Figure 1.4)—that were discriminated from each other irrespective of the length of edge, location or orientation.




It was also noticed that when flying bees landed on bilaterally symmetrical flowers, they lined up with the direction of the axis (Jones and Buchmann 1974).

(They preferred radial patterns of any sort and a vertical axis of bilateral symmetry, but avoided concentric circles.)



Horridge, Summary of the Model

Input Features: processed separately and not reassembled.
If Areas are not related to Edges, then there is no Shape.
  • Areas
  • Edges
Feature Detectors:
  • Intensity - respond to areas' size, colour and brightness. Requires steady signal. The photoreceptors themselves are the detectors.
  • Modulation/Orientation - respond to passing edges.
    Detect contrast at edges but are insensitive to changes in brightness.
    • Pure modulation detector - heterochromatic modulation but not edge orientation
    • Edge orientation detector - green sensitive and colorblind
    • Sequential modulation detectors - detect direction of motion of a contrast across the eye
Feature detector responses are summed by type and position in each local region of the eye
to form Cues (which destroy a local pattern).

Bee Cues:
  • modulation (in a local region of the eye)(of an electrical response caused by change in light intensity)
  • area
  • position of center
  • black spot
  • colour
  • radial edges
  • bilateral symmetry
  • average orientation
  • tangential or circular edges
  • heterchromatic modulation
  • green-sensitive edge modulation
  • absence of a Cue is itself a Cue.
The coincidence of different cues in a local region (of the eye) form a Landmark Label.
Bees learn Landmark Labels to identify a place and find the reward.


Horridge, Chapter 6, Processing and Color Vision

Few animals have been tested for colour vision.

Even if we understood our own colour vision, it would not be a model for insects.

Third-class colour vision: discriminates at least one wavelength from all shades of grey or separates two colours irrespective of intensity.

Second-class colour vision: discriminates different colours from one another irrespective of saturation and intensity.

First-class colour vision: recognises a colour even when the wavelength mixture in the illumination changes, irrespective of saturation or intensity, as in humans and bees, but confuses some mixtures of wavelengths that combine to make the same colour, just as green for humans can be made from various mixtures of blue and yellow.

Insects also have receptor-specific vision, in which the outputs of different colour-coded receptors are used separately for different responses.

All insects tested detect motion via the receptors with a peak in the green

Receptors with a peak in the ultraviolet occur in the eyes of all insects—male and female—and there is sufficient ultraviolet in daylight for it to be a useful colour.




Red is little used by insects except by dragonflies, some wasps and butterflies with red markings.




In all insects studied, the usual responses in motion perception, edge detection and several behavioural responses, except colour discrimination itself, all turn out to be colourblind, with inputs only from green receptors.





[For the bee] dorsal light response, escape to the light and the sky compass depend only on UV receptors;
landing, scanning, landmarks and behaviour that depends on motion, only on the green receptors;
dim light vision is colourblind, with ultraviolet, blue and green simply summed together;
and finally all three receptor types collaborate in antagonistic interactions in first-class colour vision.




Bees have three types of large retinula cells with broad spectral peaks in the green, blue and ultraviolet, with uneven distribution over the eye. In the honeybee, unlike other insects, colour vision has been studied in sufficient detail to show that bees discriminate colour from all shades of grey in their selection of food targets and at the hive entrance. Certain mixtures of colours interact together to produce ‘bee white’, which is close to human green. Colours are discriminated irrespective of the amount of bee white (called saturation) in the mixture. Data on trained bees strongly suggest that they detect the relative positions of at least two neighbouring patches of different colours.




Over the region of the spectrum where the spectral sensitivities of the receptors overlap (Figure 6.7b), the honeybee can discriminate differences of about 20nm in wavelength irrespective of intensity (Figure 6.7c). In dim light, the spectral types are apparently added together, so increasing the sensitivity, but bees are poor at discrimination of brightness differences, which suggests that the colour discrimination system has only opponent neurons. Neurons with opponent wavelengths in both centre and surround, as in primates, have not been found in insects.




Electrophysiology of the bee’s optic lobe (Yang et al. 2004) revealed at least 10 colour types of neuron without antagonistic responses and eight types with antagonistic responses—for example, excited by blue and inhibited by green and ultraviolet.




In the bee, as in humans, colour discrimination is recalibrated according to the colour of the illuminating light, as though the weighting functions of the three receptors are modified so that known objects such as clouds or leaves are detected as expected.





A useful model of colour vision that necessarily fits a system of receptor types with broadly overlapping spectral sensitivities is a colour triangle for three receptor types (Figure 6.8) or a tetrahedron for four types. QV.




In the bee, Yang et al. (2004) recently recorded non- opponent and also opponent cells, some broadband and others fed from one type of receptor. Combinations such as (UV + B – G –), (UV – B + G +) and (UV – B + G –) were recorded—in all, 50 types—but no spatial opponency. In dim light, the opponency disappears at the same level as colour vision is lost.
[UV, cyan, blue]
Andrea G. Blum
Often found hanging out with flowers & bees.

#18 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,998 posts
  • Location: USA

Posted 17 May 2018 - 19:48

Bees have an entirely new kind of natural visual system.
They do not see colours as we do.

Bees locate and measure amounts of blue in areas
and, separately,
quantities of green contrast at edges, and the angle between.
They do not identify colours except by these features.
(blue amount and green contrast)

To bees, white is an intense blue and black is zero blue with maximum edge contrast.

The worker bee has three colour types in each ommatidium of the compound eye (behind each facet);
  • one cell sensitive to blue,
  • six to green,
  • and one to ultraviolet (UV) (Figure 3, left).
The UV cells indicate for the bee the direction of the sky for escape and level flying,
and have never been shown to be essential for foraging.

UV is not a color to the bee. It is a signal.
Andrea G. Blum
Often found hanging out with flowers & bees.