9. Navigation II

Key Takeaways

so we're talking about navigation how you know where you are and how you can get from here to wherever else you want to go and last time we talked about just the general problems that arise in navigation and we talked about the para hippocampal place area and other parts of the brain that are involved in navigation so today we're going to continue that but we're going to talk more about the actual populations of neurons in your head that are involved in doing this and we'll talk about a particular aspect of the problem of navigation which is called reorientation that's what happens when you lose your bearings and you need to figure out where you are again reset your internal map of yours of where you are and then we'll talk about the idea that this whole system for navigation cool as it is and fascinating as navigation itself is is even more interesting because there's increasing evidence that we use that same system for lots of other aspects of high-level cognition that have nothing to do with space per se okay that's the and then we'll have a quiz short quiz that's the agenda here we go so the basic problems of navigation are one where am i and two how do i get from here to wherever else i want to go and as i mentioned last time we can break down each of these into a bunch of different components and facets of that question so when we want to know where we are that can that can involve recognizing a familiar location so if you see a photograph or you were plunked down spontaneously in an environment someplace you know you would visually recognize it and that would be one way to know where you were like this is my living room even if that location is unfamiliar and you're plunked down at random you still have some idea of what kind of a place this is am i in a natural environment an urban environment am i inside am i outside et cetera and finally you had you would have some sense of where you are with respect to the immediate bounding structures in your immediate environment like for example where you are in this room like as i'm talking to you right now i'm aware that there's a wall behind me right that kind of immediate spatial location in terms of questions that arise when we have to figure out how do we get from here to wherever else we want to go if you can directly see or hear your destination then you have the simplest possible kind of navigation strategy you just go toward that thing okay that's called beaconing and it's like the minimalist case works great if you can see or hear your destination but when you can't you need to know where am i in this in my broader understanding of the layout of my environment and where is my goal and for that you need a mental map of your environment and we'll talk more about that today that's why it's in red you also need to know your current heading in that environment it's not enough to know in my map of the world i am here with a dot you need to know which way you're facing in that map of the world in order to plan your navigation we'll talk about that too we also need to know what routes are possible from here so i may want to go over to stata and get a cup of coffee but i can't go this way i got to go around because i can't go through that glass okay and so finally that this whole magnificent system that enables us to process all this stuff works pretty impressively but every once in a while something will go wrong and it will you know get the wrong signal and then we're lost and so then we need a way to regain our bearings and we'll talk about that too um so last time i talked about a bunch of brain regions that are implicated in perceiving scenes and in navigation we talked about the perihipic apple place area right here and this region over here formerly known as tos now known as opa you don't need to remember all that it's the bit that's out on the lateral surface that we can zap because it's out there and both of those regions seem to be involved broadly in perceiving the shape of space around you we also talked a bit about retrosplenial cortex that region that's hiding in the sulcus here that you can see better when you mathematically unfold the sulcus there it is responds more to scenes and objects and that region seems to be involved in something like getting your bearings that is the location and orientation of where you are with respect to your cognitive map and environment okay so to make that a little more vivid i gave you one description of a patient before but here's from another study patients with damage to retrosplenial cortex so here's from a recent article in every case the patient with this damage was able to recognize landmarks in their neighborhoods and retained a sense of familiarity i know that place that's the coffee shop five blocks from my house right but despite that none of those patients were able to find their way in familiar environments and all but one were unable to learn new routes so they can recognize visual the visual form of a particular place but they don't know how to relate that to their cognitive map of the world and their therefore plan a route from there okay okay so um the part that i only alluded to at the end yes question okay is the retros cortex home to the clock on the cops or is it great question we don't exactly know the typical story is that the home of the cognitive map is the hippocampus which we're about to talk talk about next for reasons i will tell you but all of this is kind of this is a very active area of research it kills me every time i do these lectures i look at my old notes and i think here are these 10 other awesome studies and then i try to fit them in and then they just don't fit so actually one question i want to ask you guys after this lecture is should i in future either later in this course or in future courses allocate even more time or do you guys feel like okay enough already with navigation but i just think it's the coolest system so there's lots of work exactly trying to answer that kind of question and i'll give you a current snapshot of the approximate state but all of this is in flux and very much actively investigated okay so cognitive map what do we mean by that just to remind you of this classic study from the 1940s in rats where the rat when they learned this route and then went up here and found their gold block the rat immediately comes out and goes straight toward the goal telling you they've learned something much more interesting than the series of left and right turns to get to the goal they must have done something much more like actually learned the layout of space and the relative position of that goal so they could come up with a new vector to get there when the original route was blocked okay and you guys can do this too right when your root is blocked you come up with a novel root right uh and you do that by having some knowledge of your environment something tantamount to that in your head some version of that and further you know where you are in that map like right now you know where you are okay now here's the cool thing specific neurons in your hippocampus right now are firing telling you that you are right there okay so these neurons are called place cells and this is what they do okay so i'll be a place cell or what rather what i'll do is i will act out the activity of a place cell by a series of clicks i will make as i walk around so imagine there's an electrode in my hippocampus and you are hearing the activity of a single neuron in my hippocampus as i walk around and here's what it would do you'd hear background firing so it's going to go click click click click click noisy background finding firing click click click click click click click click click click click click click click click click click click click click click click click okay i'm not going to go down there click click click click click click look click click click click click click click click click click click click click click click so that's one neuron that fires only when i'm right over there that place it's not where i'm facing over there it's not what i'm looking at particularly it's when i'm right there okay that's a place cell and so there's lots of place cells in your hippocampus that do that and they do it for different locations in your environment and all of this was first worked out of course in rodents who were running around who had electrodes in their hippocampus but where those electrodes were connected with a loose tether so that the rodent could move around in their environment while recording from individual neurons in the campus okay so that's the setup and so i'm going to show you a movie of an aerial view of a rodent moving around a rat moving around in its environment can you see the little rat there um and what's happening is this um video is tracing out the rat's path with the light gray and every time that it's recording from one neuron every time that neuron fires it makes a red dot and so this is obviously sped up but as the rat moves around in his environment you see an accumulation like more firing when the rat is right there it's not which direction the rat is going through when he goes there just basically whenever he passes through that in any direction neurons fire more than anywhere else and then if we take that and blur it as scientists like to do to make nice idealized pictures that is the place cell for that neuron that is the place in space that that animal has to be to make that neuron fire yeah question is it one dimension i mean can multiple cases be mapped to save your home um that's complicated in an immediate environment like this generally not okay i'll show you some examples in a moment it's more complicated if you follow that cell when the animal moves to new locations so let me say a few more things and then i'll then if it's not clear i'll take questions oops i'm going to see it again okay right so so in answer to josh's question here are a bunch of place cells from a rodent exploring the same environment so you might say well there's like a hot spot here and a little sub one there but in general most of these cells respond with a hot spot in a particular single location in this particular environment okay did you have a different question about that yes so that all depends on the wrath um the unconscious of the fact whether it's in that place if the rat was anesthetized or if he was you know blindfolded and you passively moved him around in that space and he had no idea no way to tell where he was that wouldn't work however if the rat knows the environment and then you do this in a darkened room where he's actively locomoting around these things will still work pretty well because rats are very good at keeping track of where they are even without visual cues if they know the environment they'll have other cues like tactile cues and like they will know how far they went in each direction remember i talked briefly about the tunisian ants doing dead reckoning right keeping track of their vector and speed at each moment and integrating the whole thing to know where they are that's called dead reckoning rats are pretty good at that too another question over here yeah um the police officers do they have like uh look at math as well like how like we'll get there great question we'll get there i'll just give you the answer no they don't it's too bad um they could have they could have been all organized but it's actually a little complicated how would you organize them then what if you like learned more stuff off of the edge of space well like i had a whole other piece of your hippocampus i would be inconvenienced so maybe that's why it doesn't work whereas with visual space you know you're your retina topic information always stays the same we don't have to suddenly add a whole new part of retinotopic space thereby screwing up our retina topic maps in the brain i'm just making that up as a possible reason i don't know if that's why yeah what's sorry behind you david who what's tell me your name justin yeah right hi um so i was wondering like if you're in a smaller space comparatively or like a bigger space right like more areas of these specific uh place cells that like what they're mapping to um will they also scale up right now it's a great question i don't know the answer my guess is they'll scale according to the space right like so if if i had if that if my place my fake place cell field that i just acted out over there is maybe five feet across if i was then confined to a little space you'd probably have smaller ones for that space but i don't know let me say a little bit more about this so so just to catch this out the place field is the location in space the animal has to be to make that hippocampal cell fire okay so let's distinguish that from a receptive field in visual cortex which is a similar idea but a different one a receptive field in visual cortex is the location in the visual field where a stimulus has to be to make a visual neuron fire not where the animal itself has to be where the stimulus has to be okay so keep those ideas separate they're related but different okay so what about you know we and rodents tend to go around mostly on a 2d plane that is we have buildings and trees and stuff we sometimes go up in the z-axis but most we live in a 2d plane but that's not true of all animals so recall the bat that i mentioned last time these amazing flyers and navigators who fly in 3d and complicated trajectories and yet have amazing abilities to keep track of where they are over 30 to 50 miles that they fly at night and even as they change their orientation well it turns out that in the hippocampus of bats there's a bunch of work where people have put um remote what do you call these things recording devices on bats where you can remotely record neural activity in the hippocampus as the bat flies around and turns out that bats have play cells too and their place cells as they they all also can do this in a lab environment where they're flying around and you keep track of their location uh with cameras so you know exactly where they are in 3d space and it turns out that place cells in bats are three-dimensional because bats live in a three-dimensional world so whereas rodent these would be a bunch of kind of schematized place cells for uh of different hippocampal cells in a rodent these are different place cells for different hippocampal cells and a bat makes sense bats need this they need to know not making sense okay so the bat is moving around in three dimensions its place field isn't just like the one i did there i can't act this out because i can't fly but that play cell might fire over in that location but then if the bat flew directly above it it wouldn't so it's got three dimensions okay um okay so i said before that i had one of those and i acted it out but what's the evidence for that that evidence in humans came way after the evidence in rodents um because as you can imagine it's harder to arrange to record from individual neurons in human hippocampuses nonetheless as i've mentioned a few times there are occasional opportunities where a neurosurgeon has stuck an electrode in an interesting part of the brain for clinical reasons and the patient and the neurosurgeon are nice enough to let scientists collect data so i'm going to show you a really gross bloody picture if that's going to bother you just look away okay so um this is neurosurgery you take the skull off you take the dura off that's the direct surface of the brain the neurosurgeons stick electrodes right on top of there and in this case they put them deep inside the brain okay the gross pictures are gone we have just a nice clean uh x-ray here um so in these cases these are um this is a patient who's got an electrode sticking straight into the brain from the surface straight down into the hippocampus okay kind of horrifying but sometimes clinically called for seizures very often start in hippocampus so this is a common place for clinicians to put electrodes um and so uh what would you do if you had a patient who was willing to do your short experiment while hanging out in the hospital waiting to have a seizure with electrodes in their hippocampus well duh you'd have them play a little um a little game in a virtual space and some kind of you don't even need vr you can use a pretty cheesy little video game and i'm sure this one was quite cheesy this study was done back in 2003 so they had patients navigate through a space this is an aerial view of the space the patients didn't see that they saw this kind of front view and they kind of navigated around with the joystick in that space and there were three kind of visually recognizable kind of locations in that space and they had to do things to go from one location to another okay details don't really matter so all the while um ekstrom and colleagues are recording from individual neurons in this patient's hippocampus okay so here's an example of a place cell so this is a diagram of the space i just showed you right with those three recognizable locations and other locations that the patient could virtually navigate through with the joystick the red lines are the patient's trajectory as they moved around in that space and the colors within each square are the average firing rate when the patient navigated through that location and so this is the place field of that individual cell in this patient's brain as they went through this space because the firing rate there was around five hertz compared to you know three hertz for some other locations and mostly lower than that okay does that make sense so we're just just like the rodent experiment but it's a person with a joystick looking at this space as they go through this virtual environment and we're mapping out their place fields like that okay so that shows that humans have placed fields in their hippocampus just as rodents and bats do yeah all this is like independent of landmarks um that's a very complicated question this patient had access to landmarks they are seeing as they go through so one could ask for example if you did it with your eyes closed and you had to go by dead reckoning remembering the left and right turns you had in a familiar environment how well could these things go they would go for at least a while they'd probably go for longer in rodents because rodents are more accustomed to navigating in the dark and they rely less on visual cues and more on other cues but yeah these are not place cells aren't just visually responsive right so if we had like for example if we set up a a um a distinctive sound source in this corner of the room and a different you know like say somebody was singing quietly over here and we um tied a dog over there who was barking right and you you walked around in this room with your eyes closed you'd have a pretty you'd have a good way to keep track of your bearings as you moved around because you'd know that the singing was coming from here and the dog barking was coming from there you wouldn't be seeing anything your eyes would be closed but your place cells would work pretty well okay so whenever you have some basis for knowing where you are no matter what modality is telling you that usually it's many modalities those place cells will go okay okay so humans have these things too so you can think of the place cell as the kind of you are here system right that is the whole set of place cells any one place will only tell you are you in this particular location or not but you have a whole array of them then collectively that that whole representation across all of those neurons can tell you where you are in your familiar environment okay but if you want to not just know where you are but you want to go somewhere else like there you also need to know your current heading as we discussed last time okay so it turns out that there's a whole other batch of cells that tell you what way you're heading okay these are called head direction cells also first studied in rodents and each head direction cell responds when that rodent is heading in a particular direction not in another direction okay so for example you have you know if we're mapping along the x-axis different heading direction so the rodent is facing in different directions in his environment you map up the whole 360 degrees this would be the response of one cell as that rodent moves around this one will be tuned to this particular direction it would fire only when the rodent was facing this way not when it was facing this way or this way or this way or this way okay so everybody get how where you are in space is different that's not a very good way to show this where you are in space is different from where you're aimed and headed in that location okay two orthogonal axes of relevant to your location yeah so this isn't the angle of the head respect the body right it's the entire um i think i meant to look that up again because this question always arises um i think that there's some muck about that in the literature which is why i never remember a clear answer usually in a rodent especially they're the same um and whether you know because you know runs can turn their heads a little bit but you know mostly they're gonna keep it aimed the way they're moving um so i don't know this is a long complicated excuse that i forget what the answer to that is um but send me an email and i'll look it up i meant to before this lecture just ran out of time um okay most of the time they'll be the same i actually i'm i'm pretty sure it's which way your body's facing because if i turn like this well anyway i'm not going that way yeah and have you found at least 360 cents for each sort of angle i mean are there cells for each yes yes they they pretty much evenly tile the 360 degrees around the animal yeah so collectively that whole set of cells just as a collective set of place cells is sufficient to tell the animal where it is a collective set of head direction cells is sufficient to tell the animal which way it's oriented okay okay i think we just said all these things are in a structure called the well first found in a structure called the subiculum which is part of the hippocampus but since then they've been found in lots of different regions you don't need to remember that um so they get input from lots of different information there's many different ways to know which way we're oriented for example too bad we don't have a rotating chair if we did i would have done the following ridiculous thing i would have sat one of you in it and told you to close your eyes and i would suddenly turn it and the person in the chair would notice that right that's your vestibular system that tells you if your body's being turned even if you yourself don't decide to turn it it will tell you if you get turned that's another cue that provides input to the head direction cells just as visual information does and potentially auditory information and lots of other kinds of information so many different sources of information feed in to inform these head direction cells about the orientation of the animal okay all right so you can think of this as the brain's compass right telling the organism what way they're facing okay and lots of organisms have versions of this in the fly there's an amazing structure that was discovered just a couple years ago where the there's a whole kind of layout of this little neural structure i forget what it's called but actually spatially in that structure there's a little array of direction cells so actually you could see it in a little spatial map of direction in that little structure in the fly okay in humans and primates and rodents it's not organized spatially like a literal map of direction okay so now we have where you are and which way you're facing okay you know one pool of cells place cells for where you are another pool of cells for which way you're heading but those are just we're just getting going here the coolest navigation related cells are grid cells in entering cortex okay so this is a slice of the brain like this showing the hippocampus this folded up thing right here an interrental cortex is just just right next door okay so in internal cortex these things were discovered a little around a dozen years ago maybe 15 years ago and i'm going to show you a video of a rodent moving around his environment mapping out activity like we saw before but now we're in entering cortex and this neuron is going to be a grid cell and you'll see why as it moves around in its space maybe come on here we go okay so there's the rodent he's moving around that's the tether taking the neural activity the white dots are every time this one neuron fires we're following one neuron this whole time and the rodent is moving around sped up video so you can see this happening and at first it looks like completely random but as the rodent keeps migrating around in his space there you start to see that they're like blobs in there it's not totally random they're particular blobs that are clustered and oh my god those blobs are organized in a hexagonal grid it's a hexagon isn't that awesome that's a grid cell and oops there we go we don't need to see it again so this is a picture of what you just saw the trajectory of the animal and the hot spots uh in that array and here's a kind of smooth mathy version of where the firing is significant in that space both showing you hexagonal grid cells okay so this is at first glance a very weird thing why would it help to have a um a kind of plate essentially a place field that has multiple different places that make it fire okay and actually somebody asked before where their place cells have two hot spots place cells generally have one but grid cells as you see have many organized in this grid okay um so the the kind of circuitry and math of this whole system is mind-blowing and super exciting and the talk that i mentioned yesterday uh was on this topic and many people are working on this and they're working out like really deep interesting math about how you can take these cells how they remain arranged spatially in the brain at multiple scales and how you can use them to do path integration and keep track of how far you an animal has gone along this trajectory it's a little bit much for this course but i'll just say the current thinking is what these what these cells enable us to do is to keep track of how far we've gone in each direction and that's really crucial in navigation we need to know uh where we are not just by the landmarks we see we need to know how far we've gone in a given direction and the thought is that that's the function that these grid cells primarily serve in navigation um and so that's especially important for dead reckoning like integrating you know where you've gone according to your trajectories okay so you also need head direction cells at each point so you can think of the head direction cells is telling you the orientation of your vector and the grid cells of telling you the magnitude of the vector of how far you went and you take a whole bunch of those and you integrate them and you know where you've gone from your starting point and lots of animals do all that math in their head like it's pretty complicated integrals right but they all do that um okay so this is super awesome work and uh fittingly the 2014 nobel prize was awarded to um the the mosers a husband then husband wife team who discovered the grid cells uh and for and also to john o'keefe who discovered place cells uh decades earlier um and it's um it's a super exciting line of work and continuing continuing to be very exciting one okay so so far we've talked about place cells in the hippocampus direction cells in the subiculum and lots of other places and end to rhino grid cells in enter rhino cortex and this is just a schematic diagram of where those locations are the anatomy is complicated and you don't need to know it no they're all sort of in the hippocampus and it's neighboring structures that's good enough for for here well okay know that the grid cells are in internal cortex and the place cells are in hippocampus that's worth knowing okay direction cells are kind of all over um okay so that's cool but there's one more cool kind of cell actually there's several more the new one a new one i never heard of was reported in this job talk yesterday but we won't go there we'll try to keep it simple uh another well established one is called a border cell okay so this is the these are the place fields of three different neurons from an animal moving around in this space okay so you see how these are very interesting kind of place fields they're not just like a nice round blob they stretch around a whole border of the animal's environment okay so does that make you think of anything does that ring any bells with other stuff we've talked about in here right think we've talked a bunch about how the parahippocampal place area cares about the shape of space around you right well you might think that you'd really want to have awareness of where you are with respect to navigational barriers it turns out border cells respond not just to walls if you put a rodent in an environment where there's a cliff they can't go off the border cells also respond to the edge of that cliff okay so any navigational barrier basically telling you where you are with respect to navigational barriers okay okay all right and blah blah blah okay so as i mentioned in the last lecture when we talked about the para hippocampal place area the shape of space around you has this kind of privileged role in many aspects of navigation okay so now we're going to talk about this problem of reorienting or regaining your um your sense of direction once you've been disoriented okay and so um again i mentioned this before but just to give you in the give you the intuition of what we're talking about here you come up from the subway in manhattan or any other environment that you that's rectilinear that you know and you know which stop you're coming at up at so you kind of know where you are but you come out and you don't know which way to head right you don't know which way is which all right so that's a modern version of a classic problem that animals face in their environment they may know where they are but that doesn't tell them which way they're facing so just to be really concrete about this so here's an aerial view of a person you're standing here you have a cognitive map in your mind and your place cells are telling you your location in that map okay so you know where you are in that map but to you and you're looking down a street so you know that you're oriented with respect to some external axis like this but you don't know how your mental map should be aligned with that street are you facing like this facing north in manhattan or are you facing south right so that's the problem of reorientation is figuring out your particular orientation not just your location but which way you're facing in a known environment okay and we've all faced some version of this presumably at some point and it's annoying it takes a while to figure out and then i don't know if anybody's had this experience i've had it only in manhattan because that's where this arises for me but i'm sure there are other locations where you come up and you think you're going one way and then all of a sudden like your whole like it's like your whole mental map goes kaboom so how many people have had that experience like it's very sudden and punk date yeah turns out that when that happens all of your neurons flip together in unison like they're all in cahoots they have one version of this when you have that experience it's because they're all flipping together and i'll show you some data on that in a second okay all right so there's a very evolutionary old system for solving just this problem and it's a wonderful little kind of piece of the literature that i'm going to spend a couple minutes on because it's so classic and so cool and this started with work by randy gallistail in the 1980s and so what he did was he studied this problem of reorientation that is figuring out your orientation in a known environment once you've been disoriented okay it's a very particular aspect of the problem of navigation so he put rats in a rectangular environment and he had them explore the environment and then he hid some rat relevant thing like a little piece of food say a chocolate chip in that corner okay rat sees that happen rad is interested take rat out of box before they get to go take the the chocolate chip and then you disorient the rat you don't grab them by the tail and swing them around but you do some slower version you don't want to make them sick you do some slower version of that so they've lost track of which way they're facing okay now you put them in a new box new box because you don't want the smell to still be there new box and you see which way the rat goes and you find that the rat goes 50 50 to those two corners what does that mean the rat has encoded he doesn't go randomly to any corner he goes to corners you know it was in a corner he doesn't go randomly to any corner yeah i'm jack i'm sorry like it's specifically in like one of these directions you've got to say a little more than that what's to the left what's different about those two corners and the other two yeah isabel well he's looking at the shape of the room there's two longer walls and two shoulder walls he recognizes that he's basically involved he has to go to what looks like the right complaint exactly he has to have encoded the axis um the the orient the fact that the the room has a is longer on one axis than another and he's essentially encoded that chocolate chip was on the right side of the long wall or the left side of the short wall and both of those corners are consistent with that that's why he goes 50 50 to them he can't go 100 of the time to the right corner because he has no information that would tell him that in this experiment okay everybody clear so it tells you he learned where the thing is with respect to the shape of the room and its particular aspect ratio okay so now the plot thickens and now they repeat the experiment but this time they make some very rat salient asymmetry over here you make a color and a texture and you make other things to make this wall very saliently different okay so you would think the rat motivated to find the chocolate chip would now go 100 to that corner when we put them in the new box with the same landmark queue over there but no the rat goes 50 50 to the same two corners and in control experiments many controlled conditions you can show and i'll show you one in a moment the rat absolutely knows about this wall he's encoded the presence of that asymmetric wall so he has the information that should enable him to break the symmetry but he doesn't use it that's weird you should be surprised okay everybody get why that's weird he could have solved this one perfectly this time he has the information he's not using that information okay so that's weird but then liz spelke and her colleagues came along 10 years later and said let's try this with infants and so they did the infant version where you put the infant in a room with a symmetrical with a um in a rectangular room and you hide the doors so the infant doesn't have any uh cues other than the shape of the room um um 18 to 24 month old infants um and then uh you hide a toy in a corner and you see what the infant does and uh oh actually what you do with the infant is you make this wall really salient in all kinds of ways you in one case that was red velvet and they first showed the and these aren't like these are i guess toddlers right they first show them that when you knock on the red wall music happens totally cool riveting for a little kid they totally get it they know all about the music wall very salient to them nonetheless you put them in this experiment and they behave just like rodents they go 50 50 to the two corners even though they notice the red music wall and it could have solved the problem for them perfectly and they were motivated they didn't use the information everybody get why that's kind of interesting and kind of surprising okay now you might say okay rodents infants they're dummies uh we wouldn't do that us smart adult humans would we but oh yes you would under certain circumstances if we tied up your language system and there's lots of ways of doing that one way is called shadowing so it's kind of like simultaneous translation but you don't translate try this sometime i do this occasionally when i'm bored in my car just because it's amusingly difficult turn on the radio listen to somebody talking and just repeat everything they say after they say it right i'm not even translating it's still demanding right so you have to you have to be listening and producing right okay a little running thing okay so that's called verbal shadowing and it's an established way to really like tie up your language system and kind of take it offline so you can't really use it when you do this experiment on human adults if they're verbally shadowing and their language system is tied up they behave just like rodents and infants that is they use the shape of the space but they don't use salient landmarks that could help them solve it perfectly they go 50 50 to the two corners they become rats and infants we become rats and infants okay so liz spelke has spun a whole fascinating big theoretical story about how what this really means well let me just say a little bit more about about this first before i do her whole big story okay yeah so the idea is so first of all why would it make sense for rodents at least let's just consider the rats to use only the shape of space to reorient themselves when they're disoriented at first glance that seems really crazy but if you think about rodents in natural environments the idea is that like actually in natural environments features change snow comes and goes plants come and go odors change all those kinds of features of the environment can change but the shape of the environment like that there's a slope like this and a barrier here and a cliff there those are more stable features of the environment so it actually makes evolutionary sense for disoriented rodents at least to use the shape of space more than the features the colors and textures and odors of a space as landmarks to reorient themselves does that make sense and so the idea is that rodents have through evolution kind of evolved this system for reorienting themselves when they lose their bearings that relies only on the shape of space so restrictively that even if another cue becomes relevant and important they don't use it okay the further idea is that we have some version of this system in our heads as well and as smart adult humans we learn all kinds of other strategies to get beyond this we're not trapped with only being able to use this one system to solve it we can use other systems possibly language to help us say things to ourselves like it's on the left side of the short wall right that's what spelkey thinks is there's some version in your head of it's on the left side of the short wall and that's why adults can do this when their language system isn't tied up i don't think that's exactly right but it's a beautiful story and there's some evidence for it okay anyway part of the reason i go through this whole thing well one i think these experiments are cool but it's also been the basis of a kind of core idea in cognitive science and that idea is called informational encapsulation so think about just lots of syllables for a pretty simple idea you have this system for reorientation and it is designed to use the shape of space around you as the cue that you use to reorient yourself when you're disoriented that system is kind of hardwired to do just that and if some other part of your brain has information that could solve the problem like the presence of a relevant feature that could you could use you don't have that's this your reorientation system doesn't have access to that information it's informationally encapsulated it only has access to the particular inputs that are kind of hardwired into it okay and so um 20 years ago a lot of people kind of went wild with us and said that all the you know brain regions that i've talked about and cognitive systems that we're considering this course are informationally encapsulated it's kind of an extreme idea that goes far beyond functional specificity to say like the inputs are extremely restricted to each region and that's probably not true but there's some limitations on the information that each of these processors we're considering in this course has access to and this is kind of the kind of classic evidence behavioral evidence that some of those systems have very restricted inputs does that make sense idea of informational encapsulation not is it you know absolute truth about the brain but as an idea that is interesting to consider individually for each of the systems we study there's been pushback about the extremeness of this claim that infants and rodents only use the shape of space there are circumstances where you can get them to use other information but it's definitely true that the shape of space is the dominant cue for reorienting in in rodents and infants all right so when you're lost as i've mentioned there's two questions you need to answer where are you and which way you're oriented right this last stuff we were talking about is about the which way you're oriented question okay and i just showed you some evidence for this general finding that the geometric cues the shape of space uh are the dominant cubes you use to reorient yourself to to get your heading back when you're disoriented okay um but do we really know that those cues are different for for place recognition and for heading direction right so i've sort of said here are two different parts of the problem but do they function differently do we really use different cues do we use the shape of space more for heading direction and maybe other cues for place recognition for knowing where we are okay so i'm going to show you a very elegant behavioral experiment in mice that does this all all at once in one experiment so this is josh julian a former lab tech in my lab i get no credit for this whatsoever i'm proud even though i shouldn't be proud he was just an endogenously smart guy who went on and did an awesome experiment after he left my lab and went off to grad school and here's his awesome experiment okay so he said let's get mice to do both of these tasks they have to know where they are and which way they're oriented okay we're gonna do the same disorientation thing take them out turn them around till they're disoriented but these mice have to learn two different environments okay one environment has the vertical stripes on the short wall on one of the short walls the other environment has horizontal stripes on the short wall okay so you do the same experiment you bait one corner and you see where the rodent goes does he go to the two opposite corners exactly the same experiment but he has to remember which room is to solve the problem he has to know whether he's he has to discover rediscover the vertical stripes or the horizontal stripes and act accordingly because when he's in the vertical context the um the thing gets hit he does this over repeated trials the the the food gets hid um on the um uh hey let me get this right long wall on the left yeah right um yeah so when the long wall is on the left of the rodent okay that corner the long wall is on the left everybody oriented whereas when he's in the in the blue context the reward here the long wall is on the right okay so you have to learn those two different environments and that the the relevant shape cues are opposite in each okay everybody got that okay now what you find is that the rodent can learn that just fine okay so this shows that when you put the rodent in the vertical context in a room like this um they go more to these two corners than those two corners whereas when you put him in a um in a horizontal context with horizontal stripes he goes more to those two corners than those two corners that tells you the rodent has used the orientation of the stripes to figure out which room he's in and hence which two corners are the right ones everybody got that but here's the amazing thing even though in this experiment the very same animals in the very same trials are using those stripes to figure out which room they're in they don't use those stripes at all to break the symmetry and to go only to the great corner which they could do but don't okay so once you've trained the rodents on these two things that the reward is here uh in um the ver the vertical context and they're in the horizontal context you disorient them you put them back in you find that when you have vertical stripes they go to these two corners i'm just repeating the data when there are horizontal stripes they go to those two corners okay they've learned that but why do they go to those two corners they learn the damn stripes they use them to know which room they're in but they don't use them to break the asymmetry and decide which is the correct corner okay so this is like a microcosm of everything i've been saying so far all in one experiment right the rodents are noticing those feature cues using them to figure out which room they're in where are they but failing to use those features the orientation of the stripes to figure out which of the two corners is the correct one they're not even encoding food is near stripes like duh that should have been easy right so this is a beautiful i mean this is even more evidence for informational encapsulation of this system because it shows us on the very same trial they used the stripe information to know which room they were they failed to use it to figure out their orientation in that room this sort of making sense i realized kind of subtle it's sort of simple and subtle at the same time um yeah so now the themselves that you showed us on the very first maps what are they doing here yeah exactly great question let's look at that that's what we're doing next it's a great question what are the damn place cells doing here great question okay let's say a little bit more and then we'll think about what the play cells are doing okay so let me just like restate cash out the findings here the mice are using the features to figure out which place they're in are they in this one or that one but they are failing to use those features to figure out which is the correct corner they're still 50 50 for the two corners even though logically they have that information and they could use it and they should use it they don't okay so that means the mice are using features in this case orientation for place recognition but not for um regaining their orientation within that place i've just i'm just repeating what i said before is that making sense okay so now david's question what are the place cells doing here great question let's look it's mice so we can do that or k nathat all can do that and josh julian my amazing former lab tech so again i get no credit whatsoever okay so what do they do they allow the mice to forage for crumbs in a box like this okay they disorient the mouse before each trial take him out turn him around so he doesn't know which way he's facing put him in the box okay and they find that place cells have a particular location in that box not surprising that's what place cells do so here are two different trials two different cells that were mapped out in a rodent doing this this cell responds always in that corner another cell responds only in that corner okay these are just place cells like we described before doing what place cells do okay but now sometimes those place cells are off by 180 degrees even though the stripes should resolve the ambiguity okay so those same cells on other trials respond to the opposite corner so the place cells are doing just what the rodent is doing the place cells are confused am i facing you know am i am i phase am i oriented like this or am i oriented like that the place cells don't know and the rodent doesn't know and the coolest thing about this experiment is that these things are linked on the trials where the rodent goes to the wrong corner the place cells are also in the wrong corner okay they systematically determine which way the animal will go okay so um oh and also as i mentioned before like all those cells are in cahoots they're all in sync going the same way so when one of the cells rotates to the opposite corner all the other ones rotate to the opposite corner so it's as though somehow on trial to trial the rodent thinks he's oriented in one way he's actually 50 50 which way he's oriented he's not using the feature cues and his behavior according to where he looks for the food exactly follows that way he's oriented and so do all of his place cells okay that's that whole system goes together that tells you that those place cells are relevant behaviorally they are the they are the system that either directly determines or is tightly linked to the system that determines which way the animal thinks he's facing okay i realize this is a little bit complicated um does it make sense to you that you know as we've been talking about with reorientation even though the animal should know from this stripe the difference between that corner and this corner he doesn't know behaviorally he's looking for food right there and yet he goes 50 50. weird and stupid right place cells do t