Animal Vision Symposium - Riccardo Storchi

comparison [Music] I'd like to change the the title of the book to include some of the most recent stuff that we have been doing in the London and um beautiful diversity of ice all this food is in Islamic Mouse but what I hope to give you is some useful insight into a mouse perspective okay uh so as we uh as we all know the visual world is complicated uh and again in spite of this uh complexity uh it is also established that natural scenes share a number of common properties of regularities and one of such regularities for example is the relation between spatial frequency associated with the spatial frequencies right and as humans we tend to uh focus on the on the on the tail of this long-term distribution that we are very focused on the tiny little details of the scene that allows us to for example to manipulate units and this is also reflected in how our visual system are organized so for example primary reasons so the question is what about those personal frequencies right and um and what I want to tell you is that really also low spatial frequencies can provide a very useful information and indeed a whole month of natural scenes is the is the gradient of Life impressively belong with a visual elevation and this is important because it allows us to discriminate between different environments for example if we are in an open field we will see that just above the rise of the life index it increase quite sharply while if we are in a forest and we are shielded from the skyline then this grounded will be much more shallow so here we asked you uh sort of very basic questions because how they do such pattern guide a large respiratory behaviors and secondly which if I do which photo centers are utilized to capture this feature uh and through that we uh we use a a lighter box okay a modified version of the line box this is uh this has been um used for over 40 years in in Pharmacology and and the logic of this type is quite simple so we have a an arena which others the animal can go uh on the right side or on the dark side and my strength of our dark side this is a bit more sophisticated so around these two Chambers we have a much bigger boxes but it acts as a as a diffuser so essentially it dampens now low space high spatial frequency while the low spatial frequencies are retained and we have a bunch of different light sources that provide both visible and utilized and they are a position in different types around the street box and and with this we can recreate the natural Apartments of land intensity a long visual elevation and to measure that we actually use the transmitted which is which is really essentially we have a calculator and we take a bunch of pictures and then these features are sort of rectified and finally average and this other region essentially moves toward the highest patient frequencies but the preserved this vertical gradient of light intensity in the lab and with this now we are ready to go and the the first result that we have paid which was quite uh exciting was that indeed mice do have a preference for specific patterns of division so here you can see we have these two Chambers one is relatively custom like investing all along the elevation and the other one has this typical natural pattern on the ground and we also did another box of tests and one of those was uh with uh with the two Chambers with a relatively constant uh gradient of the elevation but in one chamber you can see the checkerboards so we introduced High explanation frequencies and what we find is that a really nice view like this channel one so the natural one um and instead they don't really seem to care about the high special frequencies for example for this checkered patterns and and I think this is quite interesting also if you think about the the traditional like that box test what the uh what the illumination is more like this responsibility Nation actually explained why uh throughout the years people kept getting like often inconsistent results and what is is that maybe even if you you know if you remove Mouse anxiety maybe still not going to explore the right channel they don't like it maybe they think it's stuck it doesn't look natural right um then we try to deposit this a little bit more so we uh we [Music] we provided you know different uh gradients at different levels of elevation and what we find is that what Minds really seem to care is the is the liking this is so this is what really seemed to drive uh the mouse preference yeah and then this takes us to the the second person so uh which uh photographs are utilized actually is uh this pattern and as we know so so far I've been talking about lighting testing but of course there is also a spectral influence long elevation and this was shown quite nicely from developed by uh Euler lab and they showed that UV light is disproportionately higher above your eyes or compared to Green lines and um and this is also uh reflected in a way for the reception is organized so uh we know that a short wavelength auxes are uh more densely expressed in the in eventually some in the sky and it is also the region where we have bones that can be exclusively expressed this type of oxygen um and a conversely in the innovator we have written again in cells that expressed melanoxin so they are intrinsically sensitive and they are they tend to be more dense so we did we use uh so to this this contributions we use two types of infected animals so the one one group no functional for the transaction and so it relied only on uh and what we find is that when we remove a cold signaling the this reference is still there so the device won't seem to so personally might might be used when they had it but it doesn't seem to be necessary for these for these platforms and conversely when we remove them signal and then um the preference is abolished so just to summarize this information part so those patients uh no special frequency information so in the form of designation guides Mouse exploration and allow the animals and comfort reception is not necessarily well important role right uh and then uh of course now so far we focused on the uh on the properties of the external environment but of course uh as we shown uh repeatedly and very nicely throughout the previous talks uh like our visually those fundamentally determined by our own actions uh and and again it was a beautiful industrated and also very accurately Quantified here in my plan's lab there are a lot of ways in which we can acquire information about our own options uh so understanding how information about our own action is integrated with the flow of visual processing scheme today is one of the fundamental challenges and this problem has also been impressively studied in by using Mouse as a modded system and we have been known now for over addictives that for example in a different stages of the available system can be sometimes dramatically modified and this has been shown a persistently environmental economics but also at earlier stages of visual processing like for example in primary visual patterns and even at the level of written and output line it was shown us yeah one of the the limitations of those studies that I've shown is the fact that most of them were performed in headphix animals so they couldn't really Express the whole range of uh they are natural behaviors so what we did here was to try to understand how pastoral movements are can enter the business system specifically the most visual economies during more natural and real money the exploration and so to do that we did uh recordings while we were performing 3D construction of the mouse and then we use this 3D data to capture some of the to capture a number of distinct behavioral variables that allow us to measure different postures or movements of the animal initially perform all these experiments in the dark where you can kind of see that the selling pattern seem to correlate either with a specific postures or you know the animal or with another state of movements so and then we wanted to uh quantify this more systematically so we uh we used a little bit of machine learning so uh we we build a model potentially predict uh neural siding rates based on a specific behavioral variables and then we measure the correlation between predicted and measuring firearmed and that was our measure of cutting a sizable fraction of neots even in primary visual problems are at least to some extent According to some behavioral variables in about right and sometimes the the the size of this happening was not it was relatively modest and I think part of it is because um of the neurons don't have very high firing rights so so they didn't have the dynamic range of encoding more continuous variables but this effect becomes more apparent when we put together more and more units and we use this population failure rate to actually particular boxes yes so so the first many results that we obtain is that we're happening between postures and movements in the primary visual elements can actually be explained by a few variables okay so the important message here is that some variables are more relevant than others so for example here we counted how many units were coupled to different behavioral variables okay so for example if we look at this live pride and the body difference they were only coupled to a very few neurons instead when we look at this and postures that captured the change in positive when the animal is happening now then those were a strong believer typically the second thing with the second question that we asked was how many couplings can be expressed by individual units how many different ways uh single units can happen to a positive movements and to do that we performed in a relatively simple clustering analysis maybe represent each unit as a as a 2d histogram where the referee rate is that polluted as a function of the two starscreen so either is up and down phosphorus or this other level of movements and what we find is essentially two groups of nodes that we sort of called uh limits look at neurons and to fire most when the animal means is moving a lot and when it is looking up and look down nodes okay to so yeah a lot when the animal is moving but that's to be looking down and instead they are almost silent something now okay so how would the results have been showing you uh when obtaining complete darkness so we know visual uh then we repeated all these experiments on the black line yeah and uh well I don't have to go through all the results again but essentially we were all uh it's quantitative equipment is that the Scotland is actually stronger on the right lines um and the important thing to to show though is that this type of behavior theory was preserved on the bright like so here I show you the the cross correlation between firing rate and particular behavioral variables measure either in darkness or on the bright light and as you can see the cross correlation is very very similar in these team conditions and this is true across most of our data set so to summarize is the second part so assuming that neurons specific visual model behaviors via this [Music] I'm going to show you and she worked with me to perform the experiments for the second part that actually and that's it learning [Music] predict the fire and pay then the features that you use for this Behavior are am I understood correctly that's like the sheet from the points yeah that's the distance in the beginning I'm not sure if I understood it because uh behavioral viables are yes yes yes so uh some of them essentially there were two type of variables so one what type was measuring postures so we um what we do what we did was to um essentially do must we have this 3D reconstruction of Denmark we superimposed for the courses in the data set and we run and essentially brings for component analysis to capture the direction of you know hand and body movements and that was quite deformative the results were made a lot of sense so when the physical component is kind of body sort of arch learning ahead Arch along the the main axis left and right and so on the other uh so so this was just the uh so some of these components are some of these variables are essentially the green star components and then we added some more uh sort of handcrafted posters like Emmanuel in the angle as a function I think ground and so on but then we um then we also measure the the temporary of some of the environments so those were the movements and at the top of it we had The Locomotion in a session by measuring how you know how fast the body scent of the animal is moving on the on the plane and then we and and an additional measure when we made when we sort of uh calculate the distance between all the three points and one thing and what is so I can think of two reasons why these neurons might increase their firing rate so typically the behaviors and they're not mutually exclusive the neuron could be doing both and one is to tell other neurons that this stuff is going on but the other is to actually it's symptomatically it's changing it's threshold of the Chinese sensitivism if you have any data indicating whether that's happening at the moment you only showed correlations yes yeah no I I totally agree you could it could just be that some form of game modulation another example is or you could actually be something more sophisticated like and as I said there are no mutualism it might actually be exciting because it could be yes these neurons are sensing particular huge these two particular cues which are very important of that particular sort of thing yes I think that's one of the that's one of the ideas so in a way like you can think about it game modulation on a global skill so I mean I'm more alert so I have a game evolution of course but you can think about the more sophisticated you know start mentioned by depending on the specific tasks those nails that are relevant for particular tasks but I think it's very exciting to be able to break down both the behavior and the neural responses so yes the first picture what the noise like your left mice are experiencing in their cage and if it's not like what they prefer where does that means come from a woman in their homepage yeah no that's a good question no idea I guess it's very different so natural conditions it might actually be different it depends um but thank you okay yeah um which inputs into the LDN using articles in these neurons are up and down if it's coming from other corporate so to think something some other areas currency and I think we thank you you want to be the answer we are looking into this relationships in the winter Parks but as you said thank you okay so thank you very much foreign