What the Cuttlefish Sees

Shrimp in Natural and Polarized light
A Shrimp - As We See it vs. How They See It. Courtesy Dr Shelby Temple, University of Bristol

Dr. Shelby Temple  at the University of Bristol and colleagues are doing some fascinating work with the visual systems of cuttlefish, animals that – despite their name – are most closely related to squid and octopi.  The cuttlefish are color-blind, so at first the second, rainbow-colored image might be a bit puzzling.  According to Temple, the cuttlefish are sensitive to differences in angle of polarization as small as a degree.  Explaining polarization is a bit tricky, because quite frankly you are blind to it.    Light can have an orientation (although it doesn’t necessarily need to have one) in addition to the other factors of light that your eyes actually are sensitive to, intensity and color (and if you are blind or color-blind, I apologize for my insensitivity).    In order to demonstrate this, purchase some polarized sun glasses or take some glasses home from your next 3D movie.  Look at a polarized light source (your monitor or TV, if they are LCD, are the closest), slowly cock your head, and note that the light appears to get brighter and darker.  You could measure the angle of polarization by noting at which angle your head is cocked when the light is at its brightest or, conversely, at its darkest.  If you had multiple sources at different orientations (take one of your TVs and rotate it on its side) you wouldn’t be able to turn your head so that everything was bright simultaneously.  And turning your head to take multiple measurements of the same scene is a waste of time and energy.  The cuttlefish doesn’t need to turn its head.  At every point in its visual field, it can measure this angle automatically.  To attempt to represent what it might see using this super-power, we have the rainbow shrimp.  Each color represents a slightly different angle of polarization.  However…

Shelby et al are careful to note that this is not a literal representation of what cuttlefish sees, but is intended to represent the large amount of polarization data accessible to the cuttlefish eye (polarization angles can be measured from 0 to 180 degrees, and the colored wedge represents the small variation in angle actually present within the image).   The image has mostly saturated colors (presumably the few dark patches are areas where there was no measurable polarization).  And quite frankly, it is awfully confusing.  The shrimp and the background on the left-hand side are both suffused with lots of oranges and greens.  The rock and the right-hand background are both blue and purple.  It’s hard to tell what’s what.  If I was hungry for a shrimp dinner, I might accidentally attack the background to the upper left of the actual shrimp.   Several different data sets have been thrown out in the process, data sets which I suspect the cuttlefish is still using.  The first is the intensity of the light, both the polarized and non-polarized components.  We’re quite lucky in this regard.  This data is not in the polarized image, but does exist in the natural, “human vision” image.  To imagine what the intensity looks like, just convert that first image to black and white.  I decided to use this as the Lightness channel in a new image (some of you may be familiar with the more traditional method of storing color data in Red, Green, and Blue channels – Hue, Saturation, and Lightness are an alternative method of storing this data).

The second data set I referred to was not completely available for me to work with.  This would be the relative amount of polarization.  Some light is not polarized and no matter how much you cock your head wearing your cool polarized sunglasses, it always stays the same brightness.  This I would represent as unsaturated, i.e. some value of gray (anywhere from black to white).  Completely polarized light (when cocking your head, there’s some point at which it becomes completely black) would be represented as a completely saturated color, the precise hue of which is determined by the angle.  Light that is partially polarized would be only partially saturated. I found some black spots in the Lightness channel of the original polarized image.  I interpreted the larger dark spots to be regions where the camera was unable to record any polarization whatsoever.  Sprinkled throughout were smaller black and white spots which seemed to follow the contours of certain colors, especially the light blue.  These are simply due to the particular choice of colors to represent the angles, and probably have no connection to the relative amount of polarization in each pixel.  I filtered these out leaving only the larger (or longer) dark areas, and used this as the Saturation channel in a new image.   Unfortunately this method tends to represent partially polarized light just as saturated as completely polarized light, but I had no alternative.  The Hue channel was left unchanged.

Here is my interpretation of what the cuttlefish sees:

Shrimp in "natural" polarized light
Barak's version. Original courtesy Dr Shelby Temple, University of Bristol


Against a dark background, the shrimp is going to be quite visible even to creatures such as us.  I’m fairly certain this was photographed in an aquarium.  Perhaps out in the open ocean against a brighter background, the shrimp would blend in.  Shelby and team are taking their “polarized light” camera to Lizard Island, and I anticipate many fascinating images from their work.

One Reply to “What the Cuttlefish Sees”

Leave a Reply

Your email address will not be published. Required fields are marked *