 And welcome, hello everybody, and welcome to another SusageVision talk. As usual, I would like to remind you that our talk series is part of the worldwide neuroinitiative, so I can encourage you to check the upcoming talks from other groups. We'll find all the relevant links in the description below. So today I'm glad to receive Professor Sanké Janssen from Duck University in Durham, North Carolina. Originally trained in mathematics and heart, Sanké Abtein is PhD in biology from the University of North Carolina in 1996, is enjoying different oceanographic institutions as a postdoctoral fellow, and is now a professor in a biology department of Duke University since 2001. I also had that he also was a research associate from the Smithsonian Museum of Natural History. The Sanké studied camouflage, signaling, and non-human visual modalities for the last 27 years, is particularly interested in vision and camouflage in the open ocean, but has also worked on coastal and terrestrial species, magnetoreception, natural ammunition, and human cataracts. His research combines mathematical analysis with biorpherol and morphological studies and in-situ measurement and imaging. His fieldwork currently involves open-research bruises that use cuba diving and deep sea submersibles. In addition to exploring the evolution and diversity of the optical and visual tricks that animals perform, is interested in improving communication between theoretical and experimental scientists, and also between art and science. So with all that, hello Sanké and thanks for accepting our invitation today. Hello everyone. I'm going to share my screen here. All right, good that worked. Yes, good. Hello everyone. This is actually the first time I've done this through YouTube, so yet another way to do this virtually. And what I'd like to talk to you about today is how our own sort of human sensory and cognitive biases influence our studies of visual modalities and using examples of my own research from visual ecology in the open ocean. Now I'm waiting to see if it will actually advance my slide. It didn't. All right, slide you need to advance. I've seen this happen before when it's run through Zoom. There we go. Okay. Yeah, sometimes there's an odd delay to get from the first to second slide. So I've come out of an art background and mathematics and physics, and so I've ended up in a field of biology of visual ecology that's sort of been perfectly suited for me in that I get to look at all these different sort of beautiful animals and think about their visual systems, their camouflage, their signaling, their bioluminescence, the biophotonics of their skin, how they actually produce all these optical effects. And it's really just been sort of a playground for me of all these different things I can work with. And in terms of techniques, we've worked with open ocean scuba diving, open ocean submersible work, remotely operated vehicles in the ocean, various molecular biophysics work, mathematical modeling, microspectrophotometry, a little bit of electrorhetnography, and so on and so on. We like to say that we're a tax on independent lab and a method independent lab. We're very much driven by questions, the questions that drive us and the questions that change over time. What I like to talk to you about today is a very old concept from von Uchsel, known as Umwelt, the idea of that animals live in their own sensory world. And this is something that we think of mostly in the direction of animals having sensory abilities that we do not. For example, animals have electro reception, animals have magneto reception, they have ultraviolet vision, they have infrared vision and so on. But what we sometimes forget is that we have sensory capabilities that other animals don't have, and that we also tend to parse the world in cognitive ways that may not be accurate representations of how animals themselves parse the world, their sensory world. And so because of this, you can end up basically making mistakes in your hypotheses. And some of these are assigning significance to a feature that an animal cannot perceive, assigning significance to a feature that we find striking but that an animal may not, assuming that the lab optical environment is in fact visually relevant and not actually getting the optical environment in the true environment. And finally, assuming that animals subdivide stimuli in the same way that we as scientists do. We as trained, you know, in sort of the Cartesian scientific divide things into orthogonal coordinates sort of way. I realize I'm speaking to the choir here, you know, to people that have thought about these issues a lot. And it's important to me because these are actually some of my own failures. I think it's important for all scientists, especially older or more established scientists, to admit that they fail more often than they succeed. And this is actually real data. I went through and looked at all the different sort of projects that I've initiated over the last 30 years, and all the various ways in which they ended. You can see that a small fraction ended in success, and that a larger number died in various ways, some quite painful. So this story actually this talk I'm giving is sort of a story of three failures. Failures in that the original hypothesis turned out to be completely wrong. But out of this failure, we were able to retrieve some success by realizing a new way of thinking about something that led us to newer hypotheses that were more accurate. And so I guess you could say we have succeeded by failing. The first one is going to be about acuity. The second one is going to be about eye size. And the third one is going to be about polarization vision in the ocean. So to begin with acuity, in general, humans really love color. This is an image from the color festival in India. Color just very much appeals to us as visual scientists. But the actual range of colorability in animals is not so large. This is data collected from a number of different invertebrates and vertebrates looking at the lambda max of different visual pigments. And you can see, for example, in the upper left hand corner that the different visual channels and pass or form birds is actually quite conserved. And that no matter what, you're pretty much always looking at visual pigments over a fairly narrow range, typically what we call the visible and even only a subset of this visible running from about usually 350 or so to about 550. But if you look at acuity, our ability to resolve spatial detail, this can vary enormously, even among animals that we all consider as having relatively good eyes. Here, just looking at a number of different animals plotting eye diameter versus acuity and cycles per degree, we see a range of about four orders of magnitude. And in fact, that many of the animals that we study would be considered legally blind by human standards, meaning that they would have vision worse than what, at least in the US units, we call 2200, which means an acuity one-tenth of what it would normally be to dismiss focus. And even if you look within a smaller group, this is within fish. This is phylogeny here showing acuity as a bar graph in cycles per degree. You can see that there's an enormous range of acuity, even among a relatively smaller group than considering all of animals and all of compound eyes and all these other things. So there are ways to get a sense of what this variation in acuity might mean for what an animal can get out of a scene. It's very important to understand that no matter what technique you use here, you can't actually model what the animal actually sees. All you can really do is provide an estimate of the information remaining in the scene if it's being looked at by a visual system with a lower acuity. This particular method is a Fourier method. You start with an image, you fast Fourier transform it, you multiply that with the modulation transfer function of the imaging system, which is a function of the resolution of the retinal system. And then if you take that product and you do an inverse Fourier transform, you can show what image, what information is remaining in this image. Now, of course, there are things like edge enhancement and a number of other techniques. We call the macros inside the eye and any of these brains that will change the way this actually appears, but it can not actually recover information. And so this gives you an idea of what's available, what's not. And it's also important to remember this is for a static image, a moving image will, of course, provide additional information. So for example here, if you're looking at a deep-sea crab, you know, it looks quite obvious, you know, in its background of its coral, this thing usually lives at about like 500 meters down on deep-sea coral reefs. If you first change the image so that it actually matches the spectral content of the illumination and the visual sensitivity of the eyes looking, you can see that what was, you know, originally quite an obvious animal on the coral is now relatively cryptic. And now if you actually convert this image to what it might look like if it was seen by an animal with a deep-sea visual acuity, which tends to be lower than surface-fish visual acuity, you can see that it's become quite cryptic indeed. So we came to all this sort of in reverse. We were first interested in this project with Eleanor Caves and Tammy Frank in cleaner shrimp. Cleaner shrimp are a group of extremely colorful small crustaceans that have all these beautiful little patterns and spots and so on, and they're known for cleaning. Just like cleaner fish, there's actually cleaner fish in this image. They will set up stations, fish will arrive, they will stop, and then, you know, the shrimp will clean the gills in the mouth and so on. And we're interested in sort of the different sorts of visual communications that were going on. You know, how were the shrimp communicating to the fish? Was there anything about the fish that was being communicated back to the shrimp? And also with all these little color patterns, was there any sort of information being sent back and forth between the shrimp? And we fully expected that the shrimp, you know, being highly colorful animals would, one, have decent color vision and two, that they would have a relatively sharp vision for their size so that they could see these small patterns when they were close to each other and this would signal dominance or sexuality or something of that sort. And we looked at three different species from different locations in the new world and we did two things. The first was actual electro-retinography with chromatic adaptation and to our surprise, we found out that these animals were entirely monochromatic. Even worse than the average crustacean, which is typically dichromatic, these animals had no ability to discern color whatsoever. We worked with Tammy Frank as an extremely good ERG person and tried every single method we could and we were not able to pull out a second visual pigment. So any of the color information these shrimp have is not being passed to each other visually. Then we used an optomotor assay to get a sense of their spatial resolution. Some of you have probably already done this where you put an animal in the middle of an arena with rotating black and white stripes and then you change the width of the stripes to larger and larger and see at which point they can no longer turn to maintain a stable image on their retina. And we did this under several different brightnesses of light and we also looked at what their visual acuity would be predicted by the morphology, the number of facets per degree in their eye and their sensitivity function. And what we found is that their spatial resolution was actually unusually bad even for their eye size. It varied of course by the brightness of the light and so on but typically their minimum resolvable angle was on the order of 10 degrees which is truly terrible. And so our hypothesis that these colors were being used for signaling to each other turned out to be a complete failure. Here we see the animals as we see them you know through the camera you know very obvious striking animals. Here we converted the images to how they would appear you know with the correct visual pigment the fact that they're monochromatic and then here is what they actually look like viewed from only two and a half centimeters away you know with the spatial acuity that they have. Now of course there will be more information when the the animals are actually moving relative to each other but you can see that you know at the very least all the tiny little spots and patterns on the shrimp did not appear to be useful to each other. This led us to think more about visual acuity and how visual acuity and the spatial resolution of patterns might actually play together in an interesting evolutionary way. We thought of a number of different examples that would be fun to think about one being the stability patterns on spider webs it's this big X that you see that makes the spider web more obvious to our eyes and you know at first blush one might think that this might not be a good idea because it makes a filter feeding animal this spider more obvious. But if you look at how this stability of this X pattern looks to the average insect vision it's not visible but it is quite visible to bird vision and so one hypothesis is that they put this extra pattern to avoid accidental destruction of their web by larger animals that of course they would never catch that have sharper vision. Another example are these you know these poison dart frogs that to an insect they remain invisible they're insect prey but to birds that might actually attack them and you know they use warning coloration to warm them off that warning coloration is still visible. An example from the ocean might be the fact that many of the stinging tentacles of jellyfish are actually quite obvious they're either highly pigmented or they're bright white which again makes little sense for a filter feeding animal except when one considers that the animals that they're trying to actually catch have relatively poor acuity and the animals that they don't want to accidentally swim through their web you know their tentacles and destroy them have quite high visual acuity and so we could be looking at acuity specific signals which is something you know we're pursuing further. So in that way we're able to take an original failed hypothesis and turn it into a new way of looking at you know how acuity matters for evolution. The second study is about eye size and depth in oceanic cephalopods and as a collaboration of a number of different people Eric Warrant and Don Eric Nilsson at Lund Michael Vecchione at the Smithsonian and Katie Thomas who was a grad student of mine at the time but is now at the Natural History Museum in London and it came out of a general interest in eye size and the important device size for increasing sensitivity or potentially increasing resolution and cephalopods were particularly interesting because they have an enormous range in body size and even for their range in body size an unusually high range in eye size. Giant squid eyes are the largest eyes on the planet they're about the size of soccer balls which are considerably larger than the eyes for example of whales and so we had a tremendous range of eyes to work with. Katie went to the Smithsonian Museum and studied on the order of about 150 different species of cephalopod got a great number of measurements of body and eye size. Some of these actually had to be lifted off of racks by forklifts because the animals were enormous and she was actually able to get you know measurements from giant squid and so on and so here you see the raw data put together by phylogeny and you're looking at eye size sorting the animals into those that live in bright habitats those that live in dim habitats what we call the mesopelagic portion of the ocean down to about a thousand meters and those that live in completely dark habitats. The first thing you see here is that there's quite a range and that there's a pretty strong phylogenetic signal and then if you were to you know look at this you know on a regression with the size of the bodies you can see that as it's typical the eyes become larger as the bodies become bigger and if you look at it more carefully you see that the size of an eye relative to its body is higher for the animals in the dim habitats those in the ocean between 200 and 1000 meters than those in completely dark habitats and those in bright habitats which is something that's generally seen. One finds you know that animals in bright habitats have relatively small eyes for their size or you could call an average and that is you move to more and more challenging dim environments the eyes become larger and then when you move to completely dark habitats eyes start to shrink again and in some cases you know completely vanish for example in cave animals but the question then was does any of this matter you know how much does this actually affect visual range you know these changes in eye size you know the assumption would be that these eyes are getting larger in these dim deep environments to improve visual range and visual range is important for a number of reasons one you know they could be looking for food they could be looking for each other or they could be looking out for predators and the idea has always been that you know if the eye is larger then that increases visual range and they're able to see a predator from further away and have a better chance to escape. Together with Don Eric Nielsen and Eric Morant we developed a model of visual range underwater that incorporates a number of different factors it incorporates the ambient light it incorporates the visual system its sensitivity the geometry of the eye it also incorporates the murkiness of the water as things move away in the water they become fade away due to the murkiness of the water itself and in the end a whole lot of math can be squished down into a relatively small equation which surprised us and it only depends on a few things depends on the contrast of the target what's being looked at it's width its size the attenuation coefficient of the water which is a number that describes how murky it is the pupil diameter of the viewer the eye sensitivity times the light level available and the integration time of the eye this is pretty much all inside a function known called W which is called a Lambert W function and this is actually what allows this equation to be seen in such impact form the only thing you really need to know about it is that this function rises extremely slowly um with number so in other words as if you go back as everything inside that W parentheses gets larger the number the function changes very little at all so you have to increase eye size by a tremendous amount to really get any gain in visual range is the bottom line um this is just some more math we looked at both this and also um illuminated sources like bioluminescent sources and so what i'm going to show you here are a set of graphs um the x axis is always going to be sighting distance how far you can see a prey item a predator a mate whatever divided by your own body size and so if it's one and you're six feet tall two meters tall you can see something two meters away and this is as a function of depth with the different squid species lined up by their average depth in their habitat um so the first thing you see is you just see all the different squid species we looked at and the obvious thing that comes out of this is that the visual range plummets with depth and once you get below about 500 meters um it becomes very difficult to see anything at all um we knew that it dropped off but we didn't know it dropped off that severely and we calculated that for a squid um to be able to see what a squid can normally see at the surface to be able to see that at a thousand meters its pupil would have to be one kilometer across um so eye size does let you see further but not by much and so here we've added in sizes for each of the dots that correlates with the actual lens diameter you can see that you know the larger lens diameters do push you know the dots a little bit to the right you know the larger dots are on the right and then this is actually looking at the investment um how much the eye size is larger than you would expect for the given size of squid how much it is above or below a regression line in a plot um but the biggest thing to take away from this is just that eye size does help you in the deep but it helps you far far less um than you would ever expect um this led us to think more about what actually you know can be useful you know in the deep and what we what you're looking at here are the same set of squid looking at targets except now they're looking at targets that are bioluminescent and you can see if you compare back and forth that there's an enormous advantage to having a luminous target even though the amount of light coming from these targets is actually quite low most of this bioluminescence could not be seen in dim room light it's quite dim but just that little bit um all of a sudden allows you to be able to see things that are 10 to 100 meters deep and so now we've been looking at this sort of break that occurs when you go from the top part of the ocean from zero to 500 meters to 500 to 1,000 meters at this 500 meter spot there's a change in what is possible visually you go from a world where you can see just from the light that comes down from the sun to a world where you can really only see bioluminescence and this has led us to think much more carefully about the visual ecology of the animals that are 500 meters and deeper that their camouflage their signaling everything else cannot depend on the light from the sun it has to entirely depend on light from bioluminescence and that changes the visual ecology dramatically and so this has led us to a number of other papers and new avenues of research the last thing I want to talk about is about polarization vision um and in particular whether it's used as a camouflage breaker um this was done with a postdoc at the time Yaqir Gagnon who's now at Lund University Tom Cronin at the University of Maryland and Justin Marshall at the University of Queensland um and this is getting at the whole problem of how difficult predation is in the open ocean how easy it is to be eaten um this picture here shows you that in the open ocean you can be eaten by birds and sharks at the same time um you can be eaten in a three-dimensional way you can be eaten from below you can be eaten from the side you can be eaten from above and there's very little there's nowhere you can hide from I mean there's nothing you can hide behind you can't disappear to a reef you can't disappear behind a tree and so your only real hope in the open ocean to avoid predation if you're not chemically defended or if you're not very fast you know is camouflage and so a primary thing that our lab is interested in is how does camouflage work in the open ocean because in the open ocean you need to actually look like water um which is challenging and there are only four ways that light can deal with matter um it can go through it as in transparency it can bounce off it in a shiny way as in mirrors um it can be produced by matter as in bioluminescence or it can be absorbed selectively and some of it reflected back as in coloration these are the only things that light can do with matter and in most situations the camouflage is entirely based on coloration but in the open ocean where predation is so intense and camouflage is so tricky we see all four of these occurring with great regularity and so it's a great sort of natural laboratory for understanding camouflage tricks and on how animals break camouflage today I'm just going to talk about transparency in mirrors and how that relates to polarization vision as a possible camouflage breaker there are many many um transparent animals in the open ocean um this is a beautiful one it's a transparent octopus called vitro leadenella um but a problem with many transparent animals and here you're seeing three transparent animals looked at through parallel polarizers and then looked at through cross polarizers for those of you who remember polarization microscopy um what this does when you look at something through cost polarizers is it lights up anything that changes the polarization of light by being what we call birefringent anything that either lowers the polarization or changes its direction and so the idea has been for quite a while probably for I don't know the last 30 40 years that because the ocean is polarized ocean light is actually quite polarized when the water is clear and because a number of animals have polarization vision in the ocean such as cephalopods some fish um some crustaceans um that this is used as a camouflage breaker that the animals floating in the water change the polarization of the background light and that this can be seen by the animals with polarization vision and there have been laboratory studies looking at this and so on and we were interested in whether this actually occurred in nature um our original model of how this would work is that light would come down from above it would some of it would be scattered towards the viewer and then the background light with its polarization represented by a red arrow here the polarization would either be shrunk represented by the smaller arrow as it passes through this tina floor or the polarization would actually turn by the tissue but either way the person viewing who looks a lot like Lenin here um would be able to tell this difference and break this transparency camouflage using polarization vision um and we wanted to see if this actually occurred in the real world as opposed to in the laboratory we built two different camera systems the one on the right was extremely expensive um highly complex involved a rotating crystal in the front go back and forth 90 degrees every other frame and to actually give you you know really wonderful polarization information the one on the left was two $500 cameras with extremely cheap polarizing filters on top of each lens and we just pushed the buttons at the same time which one do you think worked it was the cheap one um the expensive one was never really what we wanted it to be but the cheap one actually worked very well so one lesson you know is always use the cheap equipment in your research because it'll probably be more reliable um and this will give us images like this um you're looking at the red the green and the blue channel um don't worry about the top row in this case the M row is what it looks like um you know just sort of under natural light and then the P is the percent polarization and what you see is that the polarization image isn't in any way more obvious than the um the normal unpolarized image and this is of course a very crude way of looking at this which we're going to get back to in a moment because this is not exactly address how a visual system might do this but what we it did allow us and we looked at many many different species in many different orientations and different kinds of water we never saw that beautiful birefringence that I showed you in that first picture between the cross polarizers we never saw these animals become more obvious to polarization vision than to normal vision which sort of broke our hypothesis um and if there's one lesson from this is that it's always dangerous to fall in love as a visual ecologist with your pretty pictures um examples of this are people who study fluorescence um fluorescence can look really beautiful when you have a lot of excitation light and then a really strong filter to block all of the light from coming back except the emission wavelength then you know the rocks on this upper left hand corner can look really stunning but the rocks on the right are being illuminated by the same emission I mean excitation spectrum but you don't see any real polars and any real strong fluorescence at all because you haven't filtered out everything else and so it can be dangerous to you know look at something under an extremely artificial lighting system and then try to convert it to what's going on in the real world another example is the beautiful iridescence from the comorows of tina floors which you see in the lower left these sort of rainbow effects of the body and people thought you know what is the visual purpose of this but this only works under highly directed illumination like for example from a flash flash of a camera if you were to look at them underwater normally unless you're very close to the surface the lighting is so diffuse in the ocean that you don't see any of this iridescence at all so the most important thing I think in all visual ecology is to at some point be inside the optical environment of your animal to get a better sense of what optical effects truly matter and which ones don't um so we wanted to know you know why we weren't seeing you know this birefringence in these images and so we could unfortunately not directly measure the light field in all directions this turns out to be very difficult but because the ocean is optically very simple we were able to model it um and this is done by looking at what are called the inherent optical properties which are sort of a description of how murky the water is and then there are instruments that measure this which we had and then you use software that will take you know this measurement about how murky the water is and grind it through a whole bunch of math and then tell you exactly how much light there is in all directions at all wavelengths depending on how much light there is above the ocean where the sun is the amount of waves and so on um there's a side reaction where a lot of your money goes away because the software costs ten thousand dollars but they do give you a t-shirt so you get something out of it um and what we found in these models is that the downward light the downward radiance compared to the horizontal radius was much much brighter so much brighter than we didn't even expect and if you were to look at the ratio of this downward light to this horizontal light it's on the order of a hundred for most wavelengths and then gets all the way up to about five thousand um so in this waters that we're dealing with even down to significant depths down to a hundred meters the light coming down is much much brighter than the light you see from the side so our original image actually needs to look like this it needs to look like a lot of downward radiance just a little bit of horizontal radiance and what we see from the animal is actually the scattered downward radiance even though only a few percent of the light that goes down is scattered horizontally to the viewer that ends up overwhelming the system um and so we that downward radius is not polarized at all and so you don't see any birefringes and so this initial failure of trying to see in nature what people had seen in the lab for the last 30 years allowed us to get a better understanding of the structure of light in the ocean so for the very last thing i wanted to talk about polarization vision and how it might work to break camouflage in silvery animals which has also been a hypothesis for a while many many open ocean animals are quite silvery and when we see them on land of course they look very obvious but it turns out that if you hold a mirror vertically in the open ocean it's quite a good piece of camouflage so how does this work so imagine you have an oceanic predator in this case it's my wife and she's looking at a fish this sort of cross section here and even though the fish's body is curved there are these mirrors represented by the short black lines that are all vertical the underwater light field is often fairly symmetrical about the vertical axis and so that when she is looking at the fish she sees light behind her head that is of the same radiance same brightness as the light as if she looked directly through and so you literally are playing a trick with mirrors when you have a mirror in the open ocean it's more complicated than this because the underwater light field isn't as symmetric as we would like but this is how the basic principle works this was unearthed a long time ago in the 60s by Eric Denton at the Plymouth marine lab and he found that indeed the mirrors seen here in a you're looking at a cross section of a fish the bleak um and the little short little red lines are the mirrors you can see that even though the fish's body is curved the mirrors are vertical the whole way around and that you're actually these things have evolved this very fine tuned photonic structure for using mirror as a piece of camouflage in the open ocean these mirrors are made of bands of guanine crystals that act as a whole series of different quarter wave stacks for different wavelengths and because of this they in theory at least should change the polarization of light when you shine light off a structure like that it should change the polarization and the idea again has been for about 30 or 40 years that this should allow animals with polarization vision to see them but the question is does this actually occur in the real world you know does a predator with you know any sort of polarization vision you know looking at this animal now able to break this camouflage again we have more cameras including on the left and even more expensive camera um made by um victor grove at um washington university um and that it gave us these images which show in these four columns the intensity the degree of polarization of the image the angle polarization of the image and something we call profile which i want to talk about in a moment you can see immediately the problem with all these polarization things is that you're false coloring an image you can't directly compare for example the linear polarization image to the intensity image because you've just picked a false color map and you could pick a false color map that would make this look very obvious or make it not look very obvious so we needed a way to compare polarization contrast to intensity contrast in a physiologically relevant way um fortunately this had been worked out by martin howe and just in marshal only a year or two before this um and they looked at you know basically a polarization sensitive system um in this case it's set up you know with an invert system um with a channel of micro villi going one direction a channel of micro villi going the other direction a comparison between different neurons and so on and eventually you end up with a way of modeling this quite well and then using that to determine polarization thresholds so that you can actually compare brightness vision to polarization vision in a real way this is just to show you you know there's some math involved um the thing that they didn't look at though was how does the polarization change as it moves to the water if we want to understand how far away this animal can be seen due to polarization vision we have to also accept that the polarization may change as it moves through the water and this is where the normal way of our separating polarization into two orthogonal axes of degree of polarization and angle of polarization don't work because it turns out these two things talk to each other here is just a graph to demonstrate this it's showing how the degree of polarization of a target as you move away from it in um attenuation lengths just think of them as arbitrary units depends on the angle of that polarization relative to the angle of the background and so you can't just measure how polarization attenuates through water you need to also know the angles of it which ends up making a real mess of it um and it's a way of sort of saying that our cognitive bias towards how we like to divide the world may not be how the animals actually divide the world may not be a good way to do this instead you have to look at a different formulation that turns out to be a lot closer to how um a visual system might do it which is known as a stokes vector i'm not going to belabor the math of it but it allows you to come up with unambiguous ways of looking at how polarization attenuates underwater in a way that makes a lot more sense with how we know that polarization visual systems work and so in the end you can end it with something like this here r the row r is just a normal radiance image like normal black and white vision and then p is the polarization image of just one particular fish and the first thing you can see is that this polarization image is not nearly as obvious as the radiance image and these have all been corrected to match um thresholds so this you know they should match you know to the eye if they had the same brightness and this led us to be suspicious that perhaps our hypothesis that polarization vision broke silvery camouflage was false we then looked at it um you know actually you know calculated all the numbers and we found that this was indeed true um this figure is sort of a a map of increasing desperation um the only thing you really need to know from it is that if the data is in the sort of orange triangle that means that normal vision works better than polarization vision for seeing these animals at a distance and that if it was in the data is in the purple upper triangle that polarization vision is better and our hypothesis was true um and you can see that you know for all these different graphs it's always in the lower orange triangle meeting our hypothesis was very false the reason there are so many different graphs are that we once we found that we didn't get an answer we started to think are there any conditions under which we would get a positive answer or polarization vision actually helped the row no row lines are no rotation and rotation meaning if we allowed the animal to rotate its head so that it got the strongest polarization signal from the fish would that be enough to allow polarization vision to be helpful and it wasn't the three rows are increasing sensitivity of the polarization sensor all the way up at the bottom to pretty unreasonably strong polarization sensitivities and the left two versus the right two um columns are the ones on the right only are for water in which the degree of polarization was at least 25 um to basically say well what if we only use the water that is the clearest and is the highest background polarization does that help in the end none of it helped um it turns out that polarization vision did not allow these fish to see each other at farther away and this led us actually to what we call the five stages of scientific grief because we have written an entire um proposal and gotten it funded on basically showing that this was true um and it was not true and just and I you know at the same time first went through denial um you know we must have made a mistake there's something wrong we kept sending our postdoc here back you know to show to find the mistake then Justin in particular got very angry because he had always thought this was true and then we went back to more bargaining how can we look at this in a way that does make polarization vision you know matter um and then we just sort of resigned ourselves to it after a short period of depression we accepted it and then we send it off to publication and watch the reviewers go through the same five steps as they again would not believe it but the bottom line is these animals you know based on the actual data we had weren't more visible and that this polarization vision camouflage breaking out hypothesis for the particular animals at least turned out not to be true um we did a little more exploration as to why um we found that the degree of polarization of the background um was quite close to the degree of polarization of the fish that's shown in the left graph here you can see that the degree of linear polarization of the background the x-axis that data is just a little bit below the one to one relationship and if you looked at on the right graph the difference in the angle of polarization of the light coming off the fish versus the angle of polarization of the background you can see again that it was mostly centered around zero with not many data points far from zero so these animals even though we thought the light would change polarization as it bounced off the fish it actually did not or at least not enough to really make a difference i'm going to skip this modeling part here um and just end with one statement that this doesn't mean the polarization vision is useless i mean it's obviously there for a purpose but that it's apparently not useful for seeing things as far away as possible a nice analogy of this is color vision underwater um when we see something close up for example this sea turtle in the upper left we see all the color and the color makes it more obvious and the color makes it perhaps more identifiable to species or to sex or so on but as the turtle swims away in the lower right um the color information goes completely away and all we're left with is brightness information so it seems that as animals move away underwater um the last information remaining is always just the brightness information the animal is always just a little brighter a little darker than the background but with roughly the same hue and with no real polarization information but that doesn't mean that closer up polarization can't serve a purpose and with that i'd like to thank you for listening and i'll be happy to take any questions thank you thank you that was uh really interesting before we sorry remove this stupid mask before we move to question i'd like to invite uh everybody on on the audience to join us on zoom right now we've shared a link for this uh zoom room so if you want to come and join us to ask your question or further discuss the topic you're most welcome to do so so please join us and turn off your youtube video sound um i mean i had a couple of questions but you answer most of them during during your talk um when you were discussing uh for interest from uh daniel or so you asked how do you measure equity um yeah so yeah so this is one of these things where you know any model is only good as the information that goes into it so you know there are multiple ways to measure acuity um you know you can measure acuity behaviorally you know with an optomotor assay which is you know what we had done with the shrimp um you know you can you know it's the classic one where you spin black and white stripes and then make watch the stripes get smaller and smaller and see at which point they can no longer do it the danger of any assay of that sort is you can only tell if an animal won't do the activity not whether it can't which is of course a problem with all thresholds um another way people have measured acuity is by you know using morphology um you know looking at you know if you look at the spacing of photoreceptors in a camera eye or if you know you look at um you know the interometrial angles and the um the acceptance angles of compound eyes you know you can get that information but again you don't know how close to this theoretical maximum the animal can go those are two of the the biggies so there's a whole set of um sort of subversions of you know these different ways of getting acuity um acuity of course can vary across a retina as we know with our own fovea um so you know we could have extremely high acuity in one part of our eye lower acuity in another and there's a whole complicated interplay between the information you can receive from a static info a static image versus a moving image you know there are these wonderful demonstrations of a bunch of lines where it doesn't look like anything's there and then when one set of that lines moves across the other you see that those set of lines represents um sort of a bird shape and so you do get information additionally from moving and actually we've been converting this particular software to also work with movie inputs to give a better sense um so these sort of Fourier transform methods like I said they can't tell you the actual perception of an animal which is almost fundamentally impossible um but they can tell you the information available so for example like it's kind of like looking at the resolution limit of a microscope you can see detail below the resolution of a microscope but it's false detail you know you can do certain sharpening things in that sort but you can't recover detail unless of course you use some of the weirder new super resolution microscopy techniques and so they're really measures of information availability than they are a true representation of the perception of the animal that you did uh I guess that also answered uh the question asked by Damien Joubert Joubert sorry um we have Michael Doe with us today so um Michael Doe wonderful talk thank you are animals specialized which is a camouflage conspecific for example do the particular pigments present within otherwise transparent bodies storage I mean can they see the camouflage counter species the domain question can you say that again animals with uh those special animals can they see the camouflage conspecific the own species oh can an animal see the camouflage of its own concept that's a really good question um you know that sort of leads to the it's related to the whole private channel concept you know can animals signal to each other in ways that other animals cannot detect um we wrote a long review about this arguing that maybe there aren't as many private channels as people originally thought um but that's always been an idea and the the idea that you know can animals see each other's camouflage sort of falls in that private channel idea um you know all camouflages are you know unless they're truly perfect um can be broken by different modalities you know for example an animal with higher acuity than another may be better at breaking certain forms of pattern camouflage um animals with color vision can break a can can break a camouflage that only works for animals with you know let's say only monochromatic vision and so on um but I don't think it's been studied a great deal um whether on average well camouflaged animals are particularly good at finding each other visually you know within their own species um because of course you know there are other modalities as well and behaviors also change so you know if an animal approaches a conspecific um the conspecific was camouflaged may release that camouflage for example a flounder may pop come up off the ground or in particular a cephalopod might switch from camouflage form to any of its number of different um sexual signaling forms um so it would be a little bit tricky to do but it's a really interesting question um sorry I'm reading the question as I discovering it uh so we have another one from Michael Doe is there an inverse relationship between camouflage and the degree of social interaction across species I guess there's a shrimp in her fish you showed earlier oh god has anybody ever looked at that um I would I would doubt it I mean I mean of course you know anytime you think about it's something true across all species and is there a general pattern it's like you know be scary to ever guess but there's certainly many counter examples I mean there are many I mean you know they're extremely social schools of fish that are all quite silvery and quite well camouflaged um but it does lead to funny questions so I mean that um octopus I showed you know is extraordinarily well camouflaged the transparent octopus um but the only way that it can reproduce is to actively copulate it has to find another completely transparent octopus and mate and we know they do um because we have movies of them actually mating and it's not a common animal at all it's quite a rare octopus and of course the ocean's very large um and the depth ranges over which this animal operates is also quite large um so somehow they managed to figure it out and it's tricky to you know think about sometimes you would think well maybe they're using another modality um but for something like a transparent octopus I mean maybe it would be a chemical signal but in the plagic world chemical signals because they're everybody's moving in the same bulk current they're left with just simple diffusion which is slow um and of course you know things like that aren't using sonar or things that sort so the whole question of how animals balance camouflage and social interaction I mean it's it's a long old one I mean John Endler was of course very interested in this and how guppies you know balance being able to signal with each other but also not be too obvious you know to predators um and we've thought about it in a practical way you know can you design lures for fish that don't catch sea turtles you know what are the ways in which sea turtle vision differs from swordfish vision so that you can catch one and have another be different you know can you play with temporal characteristics or spatial characteristics um but yeah it's a long-standing question whether there's an overall relationship across all species I would doubt it just because biology is always so messy yeah complicated topic um just to let people know audience know I will shut down the stream after this last question so if you want to join us please do so in the meantime the link is still in the chat you won't be accessible after that um I have one last question from Tom Baden did you look at any possible wavelength dependence in your trade-off between polarisation and brightness um let's see we did look at we we had the different channels um so you know we looked at you know in this case the camera systems you know we were looking at a red a blue and a green channel so we didn't have you know fine spectral information like with hyperspectral imager we didn't see anything um now it's the thing of the the tricky part of that though is there's almost no red channel so you know we're working in a blue ocean at some depth and so there's almost no red information at all and so that you know channel is quite noisy and so all that we're really left doing is comparing the blue and the green channel um but no we don't have any hyperspectral information to look at that question it'd be interesting to look at um in general when we've looked at some of these things like transparency and mirroring and so on look at wavelength dependence we haven't found anything particularly striking the one place fascinatingly enough where we found a strong dependence we've done work on ultra black fish um fish that have extremely low reflectances um you know you know far below one percent like on the order of 0.05 percent and we found that even though the reflectance across the visible spectrum is extremely low it's far lower right at the wavelengths that are used by bioluminescence at about 480 nanometers um you know the values there are one third to one quarter what they are at the blue and red edge of the spectrum um so those animals even though you think it wouldn't matter anymore because they're so extraordinarily dark to begin with are still have a very strong wavelength dependence of their camouflage um but you know some of the very first work I did which was on transparent animals looking at wavelength dependence there there was never anything interesting there which surprised me I always assumed that there would be a stronger um you know that all these traits would become more prominent as one moved to the portion of the spectrum where most of the light occurred which is in the ocean about at 480 nanometers but it turned out not to be true unfortunately all right thank you for that um last chance for audience to join us on zoom um thanks in case it was very interesting don't like me and so I'd like to thank everybody that was there today and I'll see you next week for another the sex vision yeah all right well thank you all right take care and we are