WEBVTT 00:00:00.000 --> 00:00:02.425 [SQUEAKING] 00:00:02.425 --> 00:00:04.365 [RUSTLING] 00:00:04.365 --> 00:00:06.790 [CLICKING] 00:00:09.533 --> 00:00:11.700 NANCY KANWISHER: So we're talking about navigation-- 00:00:11.700 --> 00:00:13.590 how you know where you are and how 00:00:13.590 --> 00:00:16.680 you can get from here to wherever else you want to go. 00:00:16.680 --> 00:00:19.920 And last time we talked about just the general problems that 00:00:19.920 --> 00:00:22.140 arise in navigation, and we talked 00:00:22.140 --> 00:00:24.690 about the parahippocampal place area 00:00:24.690 --> 00:00:26.970 and other parts of the brain that 00:00:26.970 --> 00:00:28.530 are involved in navigation. 00:00:28.530 --> 00:00:30.130 So today we're going to continue that, 00:00:30.130 --> 00:00:32.850 but we're going to talk more about the actual populations 00:00:32.850 --> 00:00:35.670 of neurons in your head that are involved in doing this. 00:00:35.670 --> 00:00:37.590 And we'll talk about a particular aspect 00:00:37.590 --> 00:00:40.710 of the problem of navigation, which is called reorientation. 00:00:40.710 --> 00:00:43.050 That is what happens when you lose your bearings, 00:00:43.050 --> 00:00:45.120 and you need to figure out where you are again. 00:00:45.120 --> 00:00:50.130 Reset your internal map of where you are. 00:00:50.130 --> 00:00:52.020 And then we'll talk about the idea 00:00:52.020 --> 00:00:53.760 that this whole system for navigation, 00:00:53.760 --> 00:00:57.300 cool as it is and fascinating as navigation itself is, 00:00:57.300 --> 00:01:00.630 is even more interesting because there's increasing evidence 00:01:00.630 --> 00:01:02.670 that we use that same system for lots 00:01:02.670 --> 00:01:05.190 of other aspects of high level cognition 00:01:05.190 --> 00:01:08.010 that have nothing to do with space per se. 00:01:08.010 --> 00:01:10.560 OK that's-- and then we'll have a quiz, a short quiz. 00:01:10.560 --> 00:01:11.550 That's the agenda. 00:01:11.550 --> 00:01:12.130 Here we go. 00:01:12.130 --> 00:01:15.270 So the basic problems of navigation are, one, 00:01:15.270 --> 00:01:16.470 where am I? 00:01:16.470 --> 00:01:20.950 And two, how do I get from here to wherever else I want to go? 00:01:20.950 --> 00:01:23.910 And as I mentioned last time, we can break down each of these 00:01:23.910 --> 00:01:25.920 into a bunch of different components 00:01:25.920 --> 00:01:27.520 and facets of that question. 00:01:27.520 --> 00:01:29.820 So when we want to know where we are, 00:01:29.820 --> 00:01:33.220 that can involve recognizing a familiar location. 00:01:33.220 --> 00:01:35.760 So if you see a photograph or you were plunked down 00:01:35.760 --> 00:01:38.775 spontaneously in an environment someplace you know, 00:01:38.775 --> 00:01:41.038 you would visually recognize it, and that would be 00:01:41.038 --> 00:01:42.330 one way to know where you were. 00:01:42.330 --> 00:01:45.010 Like, this is my living room. 00:01:45.010 --> 00:01:46.793 Even if that location is unfamiliar 00:01:46.793 --> 00:01:48.210 and you're plunked down at random, 00:01:48.210 --> 00:01:50.820 you still have some idea of what kind of a place this is. 00:01:50.820 --> 00:01:52.460 Am I in a natural environment? 00:01:52.460 --> 00:01:53.470 An urban environment? 00:01:53.470 --> 00:01:54.240 Am I inside? 00:01:54.240 --> 00:01:57.520 Am I outside, et cetera? 00:01:57.520 --> 00:01:59.250 And finally, you would have some sense 00:01:59.250 --> 00:02:03.870 of where you are with respect to the immediate bounding 00:02:03.870 --> 00:02:05.730 structures in your immediate environment. 00:02:05.730 --> 00:02:07.950 Like, for example, where you are in this room. 00:02:07.950 --> 00:02:09.570 As I'm talking to you right now, I'm 00:02:09.570 --> 00:02:12.570 aware that there's a wall behind me. 00:02:12.570 --> 00:02:16.740 That kind of immediate spatial location. 00:02:16.740 --> 00:02:18.477 In terms of questions that arise when 00:02:18.477 --> 00:02:21.060 we have to figure out how do we get from here to wherever else 00:02:21.060 --> 00:02:23.940 we want to go. 00:02:23.940 --> 00:02:27.613 If you can directly see or hear your destination, 00:02:27.613 --> 00:02:29.280 then you have the simplest possible kind 00:02:29.280 --> 00:02:31.140 of navigation strategy. 00:02:31.140 --> 00:02:33.150 You just go toward that thing. 00:02:33.150 --> 00:02:36.150 OK, that's called beaconing, and it's like the minimalist case. 00:02:36.150 --> 00:02:39.750 Works great if you can see or hear your destination. 00:02:39.750 --> 00:02:43.020 But when you can't, you need to know, 00:02:43.020 --> 00:02:46.290 where am I in my broader understanding 00:02:46.290 --> 00:02:49.410 of the layout of my environment and where is my goal. 00:02:49.410 --> 00:02:52.168 And for that, you need a mental map of your environment, 00:02:52.168 --> 00:02:53.710 and we'll talk more about that today. 00:02:53.710 --> 00:02:55.740 That's why it's in red. 00:02:55.740 --> 00:02:57.847 You also need to know your current heading 00:02:57.847 --> 00:02:58.680 in that environment. 00:02:58.680 --> 00:03:00.930 It's not enough to know in my map of the world, 00:03:00.930 --> 00:03:02.475 I am here with a dot. 00:03:02.475 --> 00:03:03.960 You needs to know which way you're 00:03:03.960 --> 00:03:06.540 facing in that map of the world in order 00:03:06.540 --> 00:03:09.810 to plan your navigation, and we'll talk about that too. 00:03:09.810 --> 00:03:12.940 We also need to know what routes are possible from here. 00:03:12.940 --> 00:03:15.710 So I may want to go over to Stata and get a cup of coffee. 00:03:15.710 --> 00:03:16.710 But I can't go this way. 00:03:16.710 --> 00:03:19.410 I've got to go around because I can't go through that glass. 00:03:19.410 --> 00:03:23.340 OK, and so finally, this whole magnificent system 00:03:23.340 --> 00:03:25.260 that enables us to process all this stuff 00:03:25.260 --> 00:03:26.970 works pretty impressively. 00:03:26.970 --> 00:03:29.340 But every once in a while, something will go wrong, 00:03:29.340 --> 00:03:32.550 and it will get the wrong signal, and then we're lost. 00:03:32.550 --> 00:03:34.920 And so then we need a way to regain our bearings, 00:03:34.920 --> 00:03:37.140 and we'll talk about that too. 00:03:37.140 --> 00:03:40.650 So last time I talked about a bunch of brain regions 00:03:40.650 --> 00:03:43.020 that are implicated in perceiving scenes 00:03:43.020 --> 00:03:44.430 and in navigation. 00:03:44.430 --> 00:03:46.680 We talked about the parahippocampal place area 00:03:46.680 --> 00:03:50.730 right here and this region over here, formerly known as TOS, 00:03:50.730 --> 00:03:51.780 now known as OPA. 00:03:51.780 --> 00:03:53.280 You don't need to remember all that. 00:03:53.280 --> 00:03:55.290 It's the bit that's out on the lateral surface 00:03:55.290 --> 00:03:57.360 that we can zap because it's out there. 00:03:57.360 --> 00:04:00.090 And both of those regions seem to be involved broadly 00:04:00.090 --> 00:04:04.000 in perceiving the shape of space around you. 00:04:04.000 --> 00:04:06.810 We also talked a bit about retrosplenial cortex, 00:04:06.810 --> 00:04:08.573 that region that's hiding in the sulcus 00:04:08.573 --> 00:04:10.740 here that you can see better when you mathematically 00:04:10.740 --> 00:04:12.840 unfold the sulcus, there it is. 00:04:12.840 --> 00:04:14.970 Responds more to scenes than objects. 00:04:14.970 --> 00:04:17.579 And that region seems to be involved in something 00:04:17.579 --> 00:04:19.890 like getting your bearings-- that is the location 00:04:19.890 --> 00:04:22.980 and orientation of where you are with respect 00:04:22.980 --> 00:04:25.390 to your cognitive map and environment. 00:04:25.390 --> 00:04:27.330 OK, so to make that a little more vivid, 00:04:27.330 --> 00:04:29.880 I gave you one description of a patient before, 00:04:29.880 --> 00:04:31.950 but here's from another study. 00:04:31.950 --> 00:04:36.370 Patients with damage to retrosplenial cortex-- 00:04:36.370 --> 00:04:38.160 so here's from a recent article-- 00:04:38.160 --> 00:04:41.100 in every case, the patient with this damage 00:04:41.100 --> 00:04:44.490 was able to recognize landmarks in their neighborhoods 00:04:44.490 --> 00:04:46.380 and retained a sense of familiarity. 00:04:46.380 --> 00:04:47.370 I know that place. 00:04:47.370 --> 00:04:50.820 That's the coffee shop five blocks from my house. 00:04:50.820 --> 00:04:53.370 But despite that, none of those patients 00:04:53.370 --> 00:04:56.310 were able to find their way in familiar environments, 00:04:56.310 --> 00:04:59.040 and all but one were unable to learn new routes. 00:04:59.040 --> 00:05:03.120 So they can recognize the visual form of a particular place, 00:05:03.120 --> 00:05:04.710 but they don't know how to relate that 00:05:04.710 --> 00:05:06.540 to their cognitive map of the world 00:05:06.540 --> 00:05:11.160 and therefore plan a route from there. 00:05:11.160 --> 00:05:14.310 OK, so the part that I only alluded 00:05:14.310 --> 00:05:17.820 to at the end-- yes, question? 00:05:17.820 --> 00:05:21.180 AUDIENCE: OK, is the retrosplenial cortex 00:05:21.180 --> 00:05:23.727 the home to the cognitive maps, or is it-- 00:05:23.727 --> 00:05:25.060 NANCY KANWISHER: Great question. 00:05:25.060 --> 00:05:26.250 We don't exactly know. 00:05:26.250 --> 00:05:29.760 The typical story is that the home of the cognitive map 00:05:29.760 --> 00:05:32.610 is the hippocampus, which we're about to talk about next 00:05:32.610 --> 00:05:33.960 for reasons I will tell you. 00:05:33.960 --> 00:05:37.718 But all of this is a very active area of research. 00:05:37.718 --> 00:05:39.510 It kills me every time I do these lectures. 00:05:39.510 --> 00:05:41.160 I look at my old notes, and I think 00:05:41.160 --> 00:05:43.070 here are these 10 other awesome studies, 00:05:43.070 --> 00:05:45.570 and then I try to fit them in, and then they just don't fit. 00:05:45.570 --> 00:05:47.153 So actually one question I want to ask 00:05:47.153 --> 00:05:50.040 you guys after this lecture is, should I in future, 00:05:50.040 --> 00:05:52.490 either later in this course or in future courses, 00:05:52.490 --> 00:05:54.990 allocate even more time, or do you guys feel like OK, enough 00:05:54.990 --> 00:05:56.100 already with navigation. 00:05:56.100 --> 00:05:57.808 But I just think it's the coolest system. 00:05:57.808 --> 00:06:00.557 So there's lots of work exactly trying 00:06:00.557 --> 00:06:01.890 to answer that kind of question. 00:06:01.890 --> 00:06:04.650 And I'll give you a current snapshot 00:06:04.650 --> 00:06:06.690 of the approximate state, but all of this 00:06:06.690 --> 00:06:12.060 is in flux and very much actively investigated. 00:06:12.060 --> 00:06:16.230 OK, so cognitive map, what do we mean by that? 00:06:16.230 --> 00:06:19.890 Just to remind you of this classic study from the 1940s 00:06:19.890 --> 00:06:23.248 in rats, where the rat, when they learned this route 00:06:23.248 --> 00:06:25.290 and then went up here and found their gold block, 00:06:25.290 --> 00:06:27.248 the rat immediately comes out and goes straight 00:06:27.248 --> 00:06:28.050 toward the goal. 00:06:28.050 --> 00:06:30.750 Telling you they've learned something much more interesting 00:06:30.750 --> 00:06:33.630 than the series of left and right turns to get to the goal. 00:06:33.630 --> 00:06:37.170 They must have done something much more like actually learned 00:06:37.170 --> 00:06:40.165 the layout of space and the relative position of that goal 00:06:40.165 --> 00:06:41.790 so they could come up with a new vector 00:06:41.790 --> 00:06:44.730 to get there when the original route was blocked. 00:06:44.730 --> 00:06:48.240 OK, and you guys can do this too. 00:06:48.240 --> 00:06:51.930 When your route is blocked, you come up with a novel route. 00:06:51.930 --> 00:06:55.830 And you do that by having some knowledge of your environment, 00:06:55.830 --> 00:06:58.590 something tantamount to that in your head, 00:06:58.590 --> 00:07:00.240 some version of that. 00:07:00.240 --> 00:07:03.030 And further, you know where you are in that map. 00:07:03.030 --> 00:07:06.540 Like right now, you know where you are. 00:07:06.540 --> 00:07:08.010 Now here's the cool thing-- 00:07:08.010 --> 00:07:10.440 specific neurons in your hippocampus 00:07:10.440 --> 00:07:16.560 right now are firing telling you that you are right there. 00:07:16.560 --> 00:07:19.540 So these neurons are called place cells, 00:07:19.540 --> 00:07:20.850 and this is what they do. 00:07:20.850 --> 00:07:23.580 OK, so I'll be a place cell, or rather what I'll do 00:07:23.580 --> 00:07:27.000 is I will act out the activity of a place cell 00:07:27.000 --> 00:07:29.610 by a series of clicks I will make as I walk around. 00:07:29.610 --> 00:07:32.740 So imagine there's an electrode in my hippocampus, 00:07:32.740 --> 00:07:35.400 and you are hearing the activity of a single neuron 00:07:35.400 --> 00:07:37.660 in my hippocampus as I walk around. 00:07:37.660 --> 00:07:39.150 And here's what it would do. 00:07:39.150 --> 00:07:40.560 You'd hear background firing. 00:07:40.560 --> 00:07:43.920 So it's going to go click, click, click, click, click. 00:07:43.920 --> 00:07:45.392 Noisy background firing. 00:07:45.392 --> 00:07:47.100 Click, click, click, click, click, click, 00:07:47.100 --> 00:07:47.900 click, click, click, click. 00:07:47.900 --> 00:07:49.300 Click, click, click, click, click, click, click, 00:07:49.300 --> 00:07:50.700 click, click, click, click, click, click, click. 00:07:50.700 --> 00:07:52.300 Click, click, click, click, click, click, click, click, 00:07:52.300 --> 00:07:53.700 click, click, click, click, click, click, click. 00:07:53.700 --> 00:07:55.300 Click, click, click, click, click, click, click, click. 00:07:55.300 --> 00:07:57.600 Click, click, click, click, click, click, click, click. 00:07:57.600 --> 00:07:58.540 OK, I'm not going to go down there. 00:07:58.540 --> 00:08:00.140 Click, click, click, click, click, click, click, click. 00:08:00.140 --> 00:08:01.540 Click, click, click, click, click, click, click. 00:08:01.540 --> 00:08:03.140 Click, click, click, click, click, click, click, click. 00:08:03.140 --> 00:08:04.540 Click, click, click, click, click, click, click. 00:08:04.540 --> 00:08:06.140 Click, click, click, click, click, click, click, click, 00:08:06.140 --> 00:08:08.140 click, click, click, click, click, click, click. 00:08:08.140 --> 00:08:10.650 So that's one neuron that fires only when I'm 00:08:10.650 --> 00:08:12.780 right over there, that place. 00:08:12.780 --> 00:08:14.940 It's not where I'm facing over there. 00:08:14.940 --> 00:08:17.040 It's not what I'm looking at particularly. 00:08:17.040 --> 00:08:19.740 It's when I'm right there. 00:08:19.740 --> 00:08:22.180 OK, that's a place cell. 00:08:22.180 --> 00:08:25.500 And so there's lots of place cells in your hippocampus 00:08:25.500 --> 00:08:27.690 that do that, and they do it for different locations 00:08:27.690 --> 00:08:29.790 in your environment. 00:08:29.790 --> 00:08:31.950 And all of this was first worked out, of course, 00:08:31.950 --> 00:08:34.320 in rodents, who were running around 00:08:34.320 --> 00:08:36.809 who had electrodes in their hippocampus, 00:08:36.809 --> 00:08:39.659 but where those electrodes were connected with a loose tether 00:08:39.659 --> 00:08:42.240 so that the rodent could move around in their environment 00:08:42.240 --> 00:08:45.210 while recording from individual neurons in the hippocampus. 00:08:45.210 --> 00:08:46.930 OK, so that's the setup. 00:08:46.930 --> 00:08:49.680 And so I'm going to show you a movie of an aerial view 00:08:49.680 --> 00:08:52.650 of a rodent moving around-- 00:08:52.650 --> 00:08:54.570 a rat moving around in its environment. 00:08:54.570 --> 00:08:56.670 Can you see the little rat there? 00:08:56.670 --> 00:08:59.310 And what's happening is this video 00:08:59.310 --> 00:09:02.220 is tracing out the rat's path with the light gray. 00:09:02.220 --> 00:09:04.590 And every time that-- and it's recording from one 00:09:04.590 --> 00:09:08.550 neuron-- every time that neuron fires, it makes a red dot. 00:09:08.550 --> 00:09:10.440 So this is obviously sped up. 00:09:10.440 --> 00:09:13.200 But as the rat moves around in his environment, 00:09:13.200 --> 00:09:15.930 you see an accumulation like more firing 00:09:15.930 --> 00:09:17.550 when the rat is right there. 00:09:17.550 --> 00:09:19.515 It's not which direction the rat is going 00:09:19.515 --> 00:09:20.640 through when he goes there. 00:09:20.640 --> 00:09:22.170 Just basically whenever he passes 00:09:22.170 --> 00:09:24.870 through that in any direction, neurons 00:09:24.870 --> 00:09:27.460 fire more than anywhere else. 00:09:27.460 --> 00:09:29.550 And then if we take that and blur it, 00:09:29.550 --> 00:09:33.270 as scientists like to do to make nice idealized pictures, 00:09:33.270 --> 00:09:36.570 that is the place cell for that neuron. 00:09:36.570 --> 00:09:40.770 That is the place in space that that animal has 00:09:40.770 --> 00:09:44.200 to be to make that neuron fire. 00:09:44.200 --> 00:09:45.550 Yeah, question. 00:09:45.550 --> 00:09:47.580 AUDIENCE: Is it one dimension? 00:09:47.580 --> 00:09:51.540 I mean, can multiple places be mapped to the same neuron? 00:09:51.540 --> 00:09:53.040 NANCY KANWISHER: That's complicated. 00:09:53.040 --> 00:09:56.190 In an immediate environment like this, generally not. 00:09:56.190 --> 00:09:58.710 OK, I'll show you some examples in a moment. 00:09:58.710 --> 00:10:00.660 It's more complicated if you follow 00:10:00.660 --> 00:10:03.170 that cell when the animal moves to a new location. 00:10:03.170 --> 00:10:05.670 So let me say a few more things, and then if it's not clear, 00:10:05.670 --> 00:10:06.503 I'll take questions. 00:10:06.503 --> 00:10:09.090 Oops, we're going to see that again. 00:10:09.090 --> 00:10:12.810 OK, right so in answer to Sasha's question, here 00:10:12.810 --> 00:10:15.480 are a bunch of place cells from a rodent exploring 00:10:15.480 --> 00:10:17.110 the same environment. 00:10:17.110 --> 00:10:19.860 So you might say, well, there's a hotspot here 00:10:19.860 --> 00:10:21.580 and a little sub one there. 00:10:21.580 --> 00:10:23.850 But in general, most of these cells 00:10:23.850 --> 00:10:27.660 respond with a hotspot in a particular single location 00:10:27.660 --> 00:10:30.240 in this particular environment. 00:10:30.240 --> 00:10:32.640 OK, did you have a different question about that? 00:10:32.640 --> 00:10:35.340 AUDIENCE: Yes, so that all depends 00:10:35.340 --> 00:10:39.000 on the rat being conscious of the fact that it's 00:10:39.000 --> 00:10:39.610 in that place? 00:10:39.610 --> 00:10:40.110 NANCY KANWISHER: Uh-huh. 00:10:40.110 --> 00:10:41.070 Uh-huh. 00:10:41.070 --> 00:10:45.390 If the rat was anesthetized or if he was blindfolded 00:10:45.390 --> 00:10:47.520 and you passively moved him around in that space 00:10:47.520 --> 00:10:50.190 and he had no idea-- no way to tell where he was, 00:10:50.190 --> 00:10:51.370 that wouldn't work. 00:10:51.370 --> 00:10:53.670 However, if the rat knows the environment 00:10:53.670 --> 00:10:57.270 and then you do this in a darkened room where 00:10:57.270 --> 00:11:00.240 he's actively locomoting around, these things 00:11:00.240 --> 00:11:01.950 will still work pretty well because rats 00:11:01.950 --> 00:11:04.800 are very good at keeping track of where they are, even 00:11:04.800 --> 00:11:07.110 without visual cues if they know the environment. 00:11:07.110 --> 00:11:09.210 They'll have other cues like tactile cues, 00:11:09.210 --> 00:11:12.520 and they will know how far they went in each direction. 00:11:12.520 --> 00:11:14.970 Remember I talked briefly about the Tunisian ants 00:11:14.970 --> 00:11:16.890 doing dead reckoning. 00:11:16.890 --> 00:11:19.608 Keeping track of their vector and speed at each moment 00:11:19.608 --> 00:11:21.900 and integrating the whole thing to know where they are. 00:11:21.900 --> 00:11:23.108 That's called dead reckoning. 00:11:23.108 --> 00:11:24.495 Rats are pretty good at that too. 00:11:24.495 --> 00:11:25.620 Another question over here? 00:11:25.620 --> 00:11:26.645 Yeah. 00:11:26.645 --> 00:11:28.020 AUDIENCE: For the place cells, do 00:11:28.020 --> 00:11:31.535 they have a map as well [INAUDIBLE].. 00:11:31.535 --> 00:11:32.910 NANCY KANWISHER: We'll get there. 00:11:32.910 --> 00:11:33.600 Great question. 00:11:33.600 --> 00:11:34.720 We'll get there. 00:11:34.720 --> 00:11:35.970 I'll just give you the answer. 00:11:35.970 --> 00:11:36.990 No, they don't. 00:11:36.990 --> 00:11:39.150 It's too bad. 00:11:39.150 --> 00:11:39.990 They could have. 00:11:39.990 --> 00:11:41.460 They could have been all organized, 00:11:41.460 --> 00:11:42.750 but it's actually a little complicated. 00:11:42.750 --> 00:11:44.010 How would you organize them? 00:11:44.010 --> 00:11:47.830 What if you learned more stuff off of the edge of space? 00:11:47.830 --> 00:11:50.310 What if you had a whole other piece of your hippocampus? 00:11:50.310 --> 00:11:52.200 It would be inconvenient, so maybe that's 00:11:52.200 --> 00:11:53.070 why it doesn't work. 00:11:53.070 --> 00:11:56.460 Whereas with visual space, your retinotopic information 00:11:56.460 --> 00:11:57.730 always stays the same. 00:11:57.730 --> 00:12:00.570 We don't have to suddenly add a whole new part 00:12:00.570 --> 00:12:02.580 of retinotopic space, thereby screwing up 00:12:02.580 --> 00:12:04.220 our retinotopic maps in the brain. 00:12:04.220 --> 00:12:06.225 I'm just making that up as a possible reason. 00:12:06.225 --> 00:12:07.350 I don't know if that's why. 00:12:07.350 --> 00:12:10.560 Yeah, sorry, behind you David, tell me your name. 00:12:10.560 --> 00:12:11.400 AUDIENCE: Justice. 00:12:11.400 --> 00:12:12.780 NANCY KANWISHER: Yeah, right, hi. 00:12:12.780 --> 00:12:16.560 AUDIENCE: So I was wondering if you're in a smaller space 00:12:16.560 --> 00:12:20.910 comparatively or a bigger space, will 00:12:20.910 --> 00:12:26.580 areas of these specific place cells, what they're mapping to, 00:12:26.580 --> 00:12:28.182 will they also scale up right now? 00:12:28.182 --> 00:12:29.890 NANCY KANWISHER: That's a great question. 00:12:29.890 --> 00:12:30.890 I don't know the answer. 00:12:30.890 --> 00:12:33.720 My guess is they'll scale according to the space. 00:12:33.720 --> 00:12:39.330 So if my fake place cell fields that I just acted out 00:12:39.330 --> 00:12:42.245 over there is maybe five feet across, if I was then confined 00:12:42.245 --> 00:12:43.620 to a little space, you'd probably 00:12:43.620 --> 00:12:46.770 have smaller ones for that space, but I don't know. 00:12:46.770 --> 00:12:48.630 Let me say a little bit more about this. 00:12:48.630 --> 00:12:51.090 So just to cash this out, the place field 00:12:51.090 --> 00:12:53.250 is the location in space the animal 00:12:53.250 --> 00:12:56.610 has to be to make that hippocampal cell fire. 00:12:56.610 --> 00:13:01.080 OK, so let's distinguish that from a receptive 00:13:01.080 --> 00:13:03.510 field in visual cortex, which is a similar idea 00:13:03.510 --> 00:13:04.620 but a different one. 00:13:04.620 --> 00:13:06.690 A receptive field and visual cortex 00:13:06.690 --> 00:13:08.670 is the location in the visual field 00:13:08.670 --> 00:13:12.607 where a stimulus has to be to make a visual neuron fire. 00:13:12.607 --> 00:13:14.190 Not where the animal itself has to be, 00:13:14.190 --> 00:13:15.960 where the stimulus has to be. 00:13:15.960 --> 00:13:17.950 So keep those ideas separate. 00:13:17.950 --> 00:13:20.490 They're related but different. 00:13:20.490 --> 00:13:24.240 OK, so what about we and rodents tend to go around mostly 00:13:24.240 --> 00:13:26.250 on a 2D plane. 00:13:26.250 --> 00:13:28.920 That is we have buildings and trees and stuff. 00:13:28.920 --> 00:13:30.640 We sometimes go up in the z-axis, 00:13:30.640 --> 00:13:33.330 but mostly we live in a 2D plane, 00:13:33.330 --> 00:13:35.320 but that's not true of all animals. 00:13:35.320 --> 00:13:37.650 So recall the bat that I mentioned last time. 00:13:37.650 --> 00:13:42.210 These amazing flyers and navigators who fly in 3D 00:13:42.210 --> 00:13:44.400 and complicated trajectories and yet 00:13:44.400 --> 00:13:47.250 have amazing abilities to keep track 00:13:47.250 --> 00:13:51.750 of where they are over 30 to 50 miles that they fly at night 00:13:51.750 --> 00:13:54.600 and even as they change their orientation. 00:13:54.600 --> 00:13:57.960 Well, it turns out that in the hippocampus of bats, 00:13:57.960 --> 00:14:03.880 there's a bunch of work where people have put remote-- 00:14:03.880 --> 00:14:06.300 what do you call these things-- recording devices 00:14:06.300 --> 00:14:09.900 on bats, where you can remotely record neural activity 00:14:09.900 --> 00:14:13.530 in the hippocampus as the bat flies around. 00:14:13.530 --> 00:14:16.290 And it turns out that bats have place cells too, 00:14:16.290 --> 00:14:19.670 and their place cells, as they also 00:14:19.670 --> 00:14:22.170 can do this in a lab environment where they're flying around 00:14:22.170 --> 00:14:25.050 and you keep track of their location with cameras. 00:14:25.050 --> 00:14:28.200 So you know exactly where they are in 3D space. 00:14:28.200 --> 00:14:30.840 And it turns out that place cells in bats 00:14:30.840 --> 00:14:32.730 are three dimensional because bats live 00:14:32.730 --> 00:14:34.470 in a three dimensional world. 00:14:34.470 --> 00:14:38.880 So whereas rodent-- these would be a bunch of schematized place 00:14:38.880 --> 00:14:43.200 cells for different hippocampal cells in a rodent, 00:14:43.200 --> 00:14:48.150 these are different place cells for different hippocampal cells 00:14:48.150 --> 00:14:49.320 in a bat. 00:14:49.320 --> 00:14:49.950 Make sense? 00:14:49.950 --> 00:14:50.730 Bats need this. 00:14:50.730 --> 00:14:52.470 They need to know-- not making sense? 00:14:52.470 --> 00:14:54.990 OK, so the bat is moving around in three dimensions. 00:14:54.990 --> 00:14:57.690 Its place field isn't just like the one I did there. 00:14:57.690 --> 00:14:59.760 I can't act this out because I can't fly, 00:14:59.760 --> 00:15:03.150 but that place cell might fire over in that location. 00:15:03.150 --> 00:15:06.360 But then if the bat flew directly above it, it wouldn't. 00:15:06.360 --> 00:15:07.945 So it's got three dimensions. 00:15:07.945 --> 00:15:08.445 OK. 00:15:12.320 --> 00:15:15.590 OK, so I said before that I had one of those, 00:15:15.590 --> 00:15:16.610 and I acted it out. 00:15:16.610 --> 00:15:18.770 But what's the evidence for that? 00:15:18.770 --> 00:15:23.925 The evidence in humans came way after the evidence in rodents. 00:15:23.925 --> 00:15:25.550 Because as you can imagine, it's harder 00:15:25.550 --> 00:15:28.040 to arrange to record from individual neurons 00:15:28.040 --> 00:15:30.110 in human hippocampus. 00:15:30.110 --> 00:15:33.290 Nonetheless, as I've mentioned a few times 00:15:33.290 --> 00:15:35.330 there are occasional opportunities 00:15:35.330 --> 00:15:37.730 where a neurosurgeon has stuck an electrode 00:15:37.730 --> 00:15:41.030 in an interesting part of the brain for clinical reasons, 00:15:41.030 --> 00:15:42.830 and the patient and the neurosurgeon 00:15:42.830 --> 00:15:46.718 are nice enough to let scientists collect data. 00:15:46.718 --> 00:15:49.010 So I'm going to show you a really gross bloody picture. 00:15:49.010 --> 00:15:50.990 If that's going to bother you, just look away. 00:15:50.990 --> 00:15:55.250 OK, so this is neurosurgery. 00:15:55.250 --> 00:15:58.880 You take the skull off, you take the dura off. 00:15:58.880 --> 00:16:01.310 That's the direct surface of the brain. 00:16:01.310 --> 00:16:03.140 The neurosurgeons stick electrodes 00:16:03.140 --> 00:16:04.290 right on top of there. 00:16:04.290 --> 00:16:07.280 And in this case, they put them deep inside the brain. 00:16:07.280 --> 00:16:08.640 OK, the gross pictures are gone. 00:16:08.640 --> 00:16:13.350 We have just a nice clean X-ray here. 00:16:13.350 --> 00:16:17.510 So in these cases, this is a patient 00:16:17.510 --> 00:16:20.090 who's got an electrode sticking straight 00:16:20.090 --> 00:16:22.490 into the brain from the surface straight 00:16:22.490 --> 00:16:25.010 down n to the hippocampus. 00:16:25.010 --> 00:16:29.660 OK, kind of horrifying, but sometimes clinically 00:16:29.660 --> 00:16:30.680 called for. 00:16:30.680 --> 00:16:32.970 Seizures very often start in the hippocampus, 00:16:32.970 --> 00:16:37.670 so this is a commonplace for clinicians to put electrodes. 00:16:37.670 --> 00:16:40.760 And so what would you do if you had a patient who 00:16:40.760 --> 00:16:43.400 was willing to do your short experiment while hanging out 00:16:43.400 --> 00:16:46.010 in the hospital waiting to have a seizure with electrodes 00:16:46.010 --> 00:16:47.150 in their hippocampus? 00:16:47.150 --> 00:16:51.320 Well, you'd have them play a little game 00:16:51.320 --> 00:16:54.650 in a virtual space in some kind of-- you don't even need VR. 00:16:54.650 --> 00:16:57.555 You can use a pretty cheesy little video game, 00:16:57.555 --> 00:16:59.180 and I'm sure this one was quite cheesy. 00:16:59.180 --> 00:17:01.670 This study was done back in 2003. 00:17:01.670 --> 00:17:05.720 So they had patients navigate through a space-- this 00:17:05.720 --> 00:17:07.430 is an aerial view of the space. 00:17:07.430 --> 00:17:08.810 The patients didn't see that. 00:17:08.810 --> 00:17:12.440 They saw this front view, and they navigated around 00:17:12.440 --> 00:17:14.460 with the joystick in that space. 00:17:14.460 --> 00:17:19.053 And there were three visually recognizable locations 00:17:19.053 --> 00:17:20.720 in that space, and they had to do things 00:17:20.720 --> 00:17:23.089 to go from one location to another. 00:17:23.089 --> 00:17:25.130 OK, details don't really matter. 00:17:25.130 --> 00:17:28.850 So all the while, Ekstrom and colleagues 00:17:28.850 --> 00:17:31.100 are recording from individual neurons 00:17:31.100 --> 00:17:33.740 in this patient's hippocampus. 00:17:33.740 --> 00:17:37.050 OK, so here's an example of a place cell. 00:17:37.050 --> 00:17:39.890 So this is a diagram of the space I just showed you, 00:17:39.890 --> 00:17:43.730 with those three recognizable locations and other locations 00:17:43.730 --> 00:17:46.130 that the patient could virtually navigate through 00:17:46.130 --> 00:17:47.090 with the joystick. 00:17:47.090 --> 00:17:49.280 The red lines are the patient's trajectory 00:17:49.280 --> 00:17:51.470 as they moved around in that space. 00:17:51.470 --> 00:17:53.930 And the colors within each square 00:17:53.930 --> 00:17:59.150 are the average firing rate when the patient navigated 00:17:59.150 --> 00:18:01.230 through that location. 00:18:01.230 --> 00:18:04.850 And so this is the place field of that individual cell 00:18:04.850 --> 00:18:07.310 in this patient's brain as they went through this space. 00:18:07.310 --> 00:18:10.190 Because the firing rate there was around five hertz 00:18:10.190 --> 00:18:12.710 compared to three hertz for some other locations 00:18:12.710 --> 00:18:15.050 and mostly lower than that. 00:18:15.050 --> 00:18:16.200 OK, does that make sense? 00:18:16.200 --> 00:18:18.080 So just like the rodent experiment, 00:18:18.080 --> 00:18:21.950 but it's a person with a joystick looking at this space 00:18:21.950 --> 00:18:24.410 as they go through this virtual environment, 00:18:24.410 --> 00:18:27.480 and we're mapping out their place fields like that. 00:18:27.480 --> 00:18:30.080 OK, so that shows that humans have 00:18:30.080 --> 00:18:31.790 place fields in their hippocampus 00:18:31.790 --> 00:18:33.560 just as rodents and bats do. 00:18:33.560 --> 00:18:34.370 Yeah? 00:18:34.370 --> 00:18:38.863 AUDIENCE: Well, this is independent of landmarks? 00:18:38.863 --> 00:18:41.030 NANCY KANWISHER: That's a very complicated question. 00:18:41.030 --> 00:18:43.220 This patient had access to landmarks. 00:18:43.220 --> 00:18:45.330 They are seeing as they go through. 00:18:45.330 --> 00:18:47.810 So one could ask, for example, if you did it 00:18:47.810 --> 00:18:50.600 with your eyes closed and you had to go by dead reckoning 00:18:50.600 --> 00:18:52.280 remembering the left and right turns 00:18:52.280 --> 00:18:53.900 you had in a familiar environment, 00:18:53.900 --> 00:18:56.390 how well could these things go, they would go for at least 00:18:56.390 --> 00:18:57.380 a while. 00:18:57.380 --> 00:18:59.330 They'd probably go for longer in rodents 00:18:59.330 --> 00:19:01.910 because rodents are more accustomed to navigating 00:19:01.910 --> 00:19:02.450 in the dark. 00:19:02.450 --> 00:19:06.590 And they rely less on visual cues and more on other cues. 00:19:06.590 --> 00:19:11.390 But yeah, place cells aren't just visually responsive. 00:19:11.390 --> 00:19:14.330 So if we had, for example, if we set up 00:19:14.330 --> 00:19:18.680 a distinctive sound source in this corner 00:19:18.680 --> 00:19:21.350 of the room and a different-- like say somebody 00:19:21.350 --> 00:19:26.210 was singing quietly over here, and we tied a dog over there 00:19:26.210 --> 00:19:27.800 who was barking. 00:19:27.800 --> 00:19:29.930 And you walked around in this room with your eyes 00:19:29.930 --> 00:19:33.892 closed, you'd have a good way to keep track of your bearings 00:19:33.892 --> 00:19:36.350 as you moved around because you'd know that the singing was 00:19:36.350 --> 00:19:38.808 coming from here and the dog barking was coming from there. 00:19:38.808 --> 00:19:40.212 You wouldn't be seeing anything. 00:19:40.212 --> 00:19:42.170 Your eyes would be closed, but your place cells 00:19:42.170 --> 00:19:43.600 would work pretty well. 00:19:43.600 --> 00:19:45.620 OK, so whenever you have some basis 00:19:45.620 --> 00:19:48.350 for knowing where you are, no matter what modality 00:19:48.350 --> 00:19:49.640 is telling you that-- 00:19:49.640 --> 00:19:52.070 and usually it's many modalities-- 00:19:52.070 --> 00:19:53.300 those place cells will go. 00:19:55.910 --> 00:19:57.785 OK, so humans have these things too. 00:19:57.785 --> 00:20:00.035 So you can think of the place cell as the kind of "you 00:20:00.035 --> 00:20:03.810 are here" system that is the whole set of place cells. 00:20:03.810 --> 00:20:05.480 Any one place will only tell you are you 00:20:05.480 --> 00:20:07.560 in this particular location or not. 00:20:07.560 --> 00:20:09.080 But you have a whole array of them, 00:20:09.080 --> 00:20:12.320 then collectively, that whole representation 00:20:12.320 --> 00:20:14.300 across all of those neurons can tell you 00:20:14.300 --> 00:20:17.630 where you are in your familiar environment. 00:20:17.630 --> 00:20:22.590 OK, but if you want to not just know where you are 00:20:22.590 --> 00:20:25.910 but you want to go somewhere else, like there, 00:20:25.910 --> 00:20:28.430 you also need to know your current heading 00:20:28.430 --> 00:20:31.760 as we discussed last time. 00:20:31.760 --> 00:20:34.580 So it turns out that there is a whole other batch 00:20:34.580 --> 00:20:38.030 of cells that tell you what way you're heading. 00:20:38.030 --> 00:20:40.340 OK, these are called head direction cells, 00:20:40.340 --> 00:20:42.980 also first studied in rodents. 00:20:42.980 --> 00:20:45.770 And each head direction cell responds 00:20:45.770 --> 00:20:49.550 when that rodent is heading in a particular direction, 00:20:49.550 --> 00:20:51.470 not in another direction. 00:20:51.470 --> 00:20:56.540 OK, so for example, if we're mapping along 00:20:56.540 --> 00:20:58.362 the x-axis different heading directions. 00:20:58.362 --> 00:21:00.320 So the rodent is facing in different directions 00:21:00.320 --> 00:21:01.460 in his environment. 00:21:01.460 --> 00:21:04.250 You map up the whole 360 degrees, 00:21:04.250 --> 00:21:06.170 this would be the response of one cell 00:21:06.170 --> 00:21:08.780 as that rodent moves around. 00:21:08.780 --> 00:21:11.400 This one would be tuned to this particular direction. 00:21:11.400 --> 00:21:13.858 It would fire only when the rodent was facing this way, 00:21:13.858 --> 00:21:15.650 not when it was facing this way or this way 00:21:15.650 --> 00:21:18.410 or this way or this way. 00:21:18.410 --> 00:21:21.260 So does everybody get how where you are in space 00:21:21.260 --> 00:21:25.520 is different-- that's not a very good way to show this. 00:21:25.520 --> 00:21:28.460 Where you are in space is different from where 00:21:28.460 --> 00:21:31.430 you're aimed and headed in that location. 00:21:31.430 --> 00:21:35.750 OK, two orthogonal axes of relevant to your location. 00:21:35.750 --> 00:21:36.755 Yeah? 00:21:36.755 --> 00:21:38.130 AUDIENCE: So this isn't the angle 00:21:38.130 --> 00:21:39.980 of the head in respect to the body, right? 00:21:39.980 --> 00:21:42.210 It's the entire-- 00:21:42.210 --> 00:21:44.120 NANCY KANWISHER: I think I meant to look that 00:21:44.120 --> 00:21:46.880 up again because this question always arises. 00:21:46.880 --> 00:21:48.855 I think that there's some muck about that 00:21:48.855 --> 00:21:50.480 in the literature, which is why I never 00:21:50.480 --> 00:21:51.890 remember a clear answer. 00:21:51.890 --> 00:21:54.200 Usually in a rodent, especially the same. 00:21:56.930 --> 00:21:59.390 Because rodents can turn their heads a little bit, 00:21:59.390 --> 00:22:02.390 but mostly they're going to keep it aimed the way 00:22:02.390 --> 00:22:04.100 they're moving. 00:22:04.100 --> 00:22:04.850 So I don't know. 00:22:04.850 --> 00:22:06.710 This is a long, complicated excuse 00:22:06.710 --> 00:22:09.350 that I forget what the answer to that is. 00:22:09.350 --> 00:22:11.150 But send me an email, and I'll look it up. 00:22:11.150 --> 00:22:12.620 I meant to before this lecture. 00:22:12.620 --> 00:22:15.800 I just ran out of time. 00:22:15.800 --> 00:22:18.350 OK, most of the time, they'll be the same. 00:22:18.350 --> 00:22:21.260 Actually, I'm pretty sure it's which way your body is facing. 00:22:21.260 --> 00:22:22.890 Because if I turn like this-- 00:22:22.890 --> 00:22:24.890 well, anyway, I'm not going that way. 00:22:24.890 --> 00:22:25.610 Yeah. 00:22:25.610 --> 00:22:30.943 AUDIENCE: Have you found at least 360 cells for each angle? 00:22:30.943 --> 00:22:33.110 NANCY KANWISHER: You mean, are there cells for each? 00:22:33.110 --> 00:22:37.880 Yes, yes, they pretty much evenly tile the 360 degrees 00:22:37.880 --> 00:22:39.110 around the animal. 00:22:39.110 --> 00:22:42.290 Yeah, so collectively, that whole set of cells, 00:22:42.290 --> 00:22:44.690 just as a collective set of place cells, 00:22:44.690 --> 00:22:46.940 is sufficient to tell the animal where it is. 00:22:46.940 --> 00:22:49.010 A collective set of head direction cells 00:22:49.010 --> 00:22:54.042 is sufficient to tell the animal which way it's oriented. 00:22:54.042 --> 00:22:56.000 OK, I think we just said-- all these things are 00:22:56.000 --> 00:22:57.997 in a structure called the-- well, first found 00:22:57.997 --> 00:22:59.580 in the structure called the subiculum, 00:22:59.580 --> 00:23:00.955 which is part of the hippocampus. 00:23:00.955 --> 00:23:02.390 But since then, they've been found 00:23:02.390 --> 00:23:03.598 in lots of different regions. 00:23:03.598 --> 00:23:06.090 You don't need to remember that. 00:23:06.090 --> 00:23:08.420 So they get input from lots of different information. 00:23:08.420 --> 00:23:12.200 There's many different ways to know which way we're oriented. 00:23:12.200 --> 00:23:15.910 For example, too bad we don't have a rotating chair. 00:23:15.910 --> 00:23:18.410 If we did, I would have done the following ridiculous thing. 00:23:18.410 --> 00:23:19.610 I would have sat one of you in it, 00:23:19.610 --> 00:23:21.080 and told you to close your eyes. 00:23:21.080 --> 00:23:23.030 And I would suddenly turn it. 00:23:23.030 --> 00:23:25.160 And the person in the chair would notice that. 00:23:25.160 --> 00:23:27.230 That's your vestibular system that 00:23:27.230 --> 00:23:28.880 tells you if your body is being turned, 00:23:28.880 --> 00:23:31.280 even if you yourself don't decide to turn it. 00:23:31.280 --> 00:23:32.930 It will tell you if you get turned. 00:23:32.930 --> 00:23:36.470 That's another cue that provides input to the head direction 00:23:36.470 --> 00:23:38.390 cells, just as visual information does 00:23:38.390 --> 00:23:40.490 and potentially auditory information and lots 00:23:40.490 --> 00:23:42.543 of other kinds of information. 00:23:42.543 --> 00:23:44.210 So many different sources of information 00:23:44.210 --> 00:23:46.130 feed in to inform these head direction 00:23:46.130 --> 00:23:47.885 cells about the orientation of the animal. 00:23:50.900 --> 00:23:54.530 All right, so you can think of this as the brain's compass 00:23:54.530 --> 00:23:57.980 telling the organism what way they're facing. 00:23:57.980 --> 00:24:01.760 And lots of organisms have versions of this. 00:24:01.760 --> 00:24:03.890 In the fly, there's an amazing structure 00:24:03.890 --> 00:24:05.930 that was discovered just a couple of years 00:24:05.930 --> 00:24:08.570 ago, where there's a whole layout 00:24:08.570 --> 00:24:10.067 of this little neural structure-- 00:24:10.067 --> 00:24:11.150 I forget what it's called. 00:24:11.150 --> 00:24:13.610 But actually spatially in that structure, 00:24:13.610 --> 00:24:17.120 there's a little array of direction cells. 00:24:17.120 --> 00:24:21.710 So actually you can see it in a little spatial map of direction 00:24:21.710 --> 00:24:24.230 in that little structure in the fly. 00:24:24.230 --> 00:24:26.190 In humans and primates and rodents, 00:24:26.190 --> 00:24:29.720 it's not organized spatially like a literal map 00:24:29.720 --> 00:24:31.460 of direction. 00:24:31.460 --> 00:24:36.470 OK, so now we have where you are and which way you're facing. 00:24:36.470 --> 00:24:38.570 One pool of cells, place cells for where you are. 00:24:38.570 --> 00:24:42.230 Another pool of cells for which way you're heading. 00:24:42.230 --> 00:24:45.050 But those are just-- we're just getting going here. 00:24:45.050 --> 00:24:48.830 The coolest navigation-related cells 00:24:48.830 --> 00:24:51.120 are grid cells and entorhinal cortex. 00:24:51.120 --> 00:24:54.900 OK, so this is a slice of the brain like this, 00:24:54.900 --> 00:24:58.700 showing the hippocampus is folded up thing right here. 00:24:58.700 --> 00:25:03.590 And entorhinal cortex is just right next door. 00:25:03.590 --> 00:25:08.990 OK, so an entorhinal cortex, these things 00:25:08.990 --> 00:25:15.200 were discovered around a dozen years ago, maybe 15 years ago. 00:25:15.200 --> 00:25:17.900 And I'm going to show you a video of a rodent moving around 00:25:17.900 --> 00:25:21.510 his environment mapping out activity, like we saw before. 00:25:21.510 --> 00:25:23.480 But now we're in entorhinal cortex, 00:25:23.480 --> 00:25:25.710 and this neuron is going to be a grid cell, 00:25:25.710 --> 00:25:28.190 and you'll see why as it moves around in its space. 00:25:31.340 --> 00:25:31.840 Maybe. 00:25:35.430 --> 00:25:36.470 Come on. 00:25:36.470 --> 00:25:37.790 Here we go. 00:25:37.790 --> 00:25:38.930 OK, so there's a rodent. 00:25:38.930 --> 00:25:39.740 He's moving around. 00:25:39.740 --> 00:25:41.840 That's the tether taking the neural activity. 00:25:41.840 --> 00:25:44.990 The white dots are every time this one neuron fires, 00:25:44.990 --> 00:25:47.420 we're following one neuron this whole time. 00:25:47.420 --> 00:25:50.240 And the rodent is moving around, sped up video 00:25:50.240 --> 00:25:51.650 so you can see this happening. 00:25:51.650 --> 00:25:54.300 And at first, it looks like completely random. 00:25:54.300 --> 00:25:58.190 But as a rodent keeps migrating around in his space there, 00:25:58.190 --> 00:26:00.860 you start to see that they're like blobs in there. 00:26:00.860 --> 00:26:02.220 It's not totally random. 00:26:02.220 --> 00:26:04.790 They're particular blobs that are clustered. 00:26:04.790 --> 00:26:11.030 And oh my god, those blobs are organized in a hexagonal grid. 00:26:11.030 --> 00:26:13.940 It's a hexagon. 00:26:13.940 --> 00:26:15.200 Isn't that awesome? 00:26:15.200 --> 00:26:18.380 That's a grid cell. 00:26:18.380 --> 00:26:20.550 And whoops, here we go. 00:26:20.550 --> 00:26:22.860 We don't need to see it again. 00:26:22.860 --> 00:26:25.070 So this is a picture of what you just saw, 00:26:25.070 --> 00:26:29.630 the trajectory of the animal and the hot spots in that array. 00:26:29.630 --> 00:26:31.940 And here's a smoothed mathy version 00:26:31.940 --> 00:26:35.060 of where the firing is significant in that space, both 00:26:35.060 --> 00:26:38.210 showing you hexagonal grid cells. 00:26:38.210 --> 00:26:43.230 OK, so this is at first glance a very weird thing. 00:26:43.230 --> 00:26:48.740 Why would it help to have essentially a place 00:26:48.740 --> 00:26:50.690 field that has multiple different places that 00:26:50.690 --> 00:26:51.775 make it fire? 00:26:51.775 --> 00:26:53.900 OK, and actually somebody else before whether place 00:26:53.900 --> 00:26:55.040 cells have two hot spots. 00:26:55.040 --> 00:26:57.830 Place cells generally have one, but grid cells, as you see, 00:26:57.830 --> 00:26:59.990 have many organized in this grid. 00:27:03.170 --> 00:27:08.570 So the kind of circuitry and math of this whole system 00:27:08.570 --> 00:27:10.860 is mind blowing and super exciting, 00:27:10.860 --> 00:27:13.880 and the talk that I mentioned yesterday was on this topic. 00:27:13.880 --> 00:27:15.380 And many people are working on this, 00:27:15.380 --> 00:27:18.200 and they're working out like really deep interesting math 00:27:18.200 --> 00:27:19.940 about how you can take these cells, 00:27:19.940 --> 00:27:23.390 how they're arranged spatially in the brain at multiple scales 00:27:23.390 --> 00:27:26.990 and how you can use them to do path integration 00:27:26.990 --> 00:27:29.120 and keep track of how far an animal has 00:27:29.120 --> 00:27:30.255 gone along its trajectory. 00:27:30.255 --> 00:27:31.880 It's a little bit much for this course, 00:27:31.880 --> 00:27:34.400 but I'll just say the current thinking is 00:27:34.400 --> 00:27:37.310 what these cells enable us to do is 00:27:37.310 --> 00:27:40.100 to keep track of how far we've gone in each direction, 00:27:40.100 --> 00:27:42.110 and that's really crucial in navigation. 00:27:42.110 --> 00:27:45.960 We need to know where we are, not just by the landmarks 00:27:45.960 --> 00:27:46.460 we see. 00:27:46.460 --> 00:27:48.980 We need to know how far we've gone in a given direction 00:27:48.980 --> 00:27:50.780 and the thought is that that's the function 00:27:50.780 --> 00:27:55.100 that these grid cells primarily serve in navigation. 00:27:55.100 --> 00:27:58.220 And so that's especially important for dead reckoning, 00:27:58.220 --> 00:28:00.620 like integrating where you've gone 00:28:00.620 --> 00:28:02.790 according to your trajectories. 00:28:02.790 --> 00:28:05.760 OK, so you also need head direction cells at each point. 00:28:05.760 --> 00:28:07.790 So you can of the head direction cells 00:28:07.790 --> 00:28:11.360 as telling you the orientation of your vector 00:28:11.360 --> 00:28:13.010 and the grid cells of telling you 00:28:13.010 --> 00:28:15.362 the magnitude of the vector of how far you went. 00:28:15.362 --> 00:28:17.070 And then you take a whole bunch of those, 00:28:17.070 --> 00:28:19.280 and you integrate them, and you know where you've 00:28:19.280 --> 00:28:21.140 gone from your starting point. 00:28:21.140 --> 00:28:24.680 And lots of animals do all that math in their head. 00:28:24.680 --> 00:28:27.180 It's pretty complicated integrals, 00:28:27.180 --> 00:28:28.310 but they all do that. 00:28:30.860 --> 00:28:33.290 OK, so this is super awesome work. 00:28:33.290 --> 00:28:36.350 And fittingly, the 2014 Nobel Prize 00:28:36.350 --> 00:28:42.740 was awarded to the Mosers, a then husband-wife team, who 00:28:42.740 --> 00:28:45.590 discovered the grid cells and also 00:28:45.590 --> 00:28:50.210 to John O'Keefe, who discovered place cells decades earlier. 00:28:50.210 --> 00:28:54.890 And it's a super exciting line of work and continuing 00:28:54.890 --> 00:28:56.990 to be very exciting one. 00:28:56.990 --> 00:29:01.910 OK, so so far we've talked about place cells in the hippocampus, 00:29:01.910 --> 00:29:05.420 direction cells in the subiculum and lots of other places, 00:29:05.420 --> 00:29:08.940 and entorhinal grid cells and entorhinal cortex. 00:29:08.940 --> 00:29:10.790 And this is just a schematic diagram 00:29:10.790 --> 00:29:12.050 of where those locations are. 00:29:12.050 --> 00:29:14.810 The anatomy is complicated, and you don't need to know it. 00:29:14.810 --> 00:29:18.620 Know they're all in the hippocampus and its neighboring 00:29:18.620 --> 00:29:19.580 structures. 00:29:19.580 --> 00:29:23.280 That's good enough for here. 00:29:23.280 --> 00:29:26.415 Well, OK, know that the grid cells are in entorhinal cortex 00:29:26.415 --> 00:29:28.040 and the place cells are in hippocampus. 00:29:28.040 --> 00:29:29.030 That's worth knowing. 00:29:29.030 --> 00:29:33.350 OK, direction cells are kind of all over. 00:29:33.350 --> 00:29:38.400 OK, so that's cool, but there's one more cool kind of cell-- 00:29:38.400 --> 00:29:39.650 actually there's several more. 00:29:39.650 --> 00:29:42.680 The new one I never heard of was reported in this job talk 00:29:42.680 --> 00:29:44.240 yesterday, but we won't go there. 00:29:44.240 --> 00:29:46.430 We'll try to keep it simple. 00:29:46.430 --> 00:29:51.200 Another well-established one is called a border cell. 00:29:51.200 --> 00:29:55.460 So these are the place fields of three different neurons 00:29:55.460 --> 00:29:59.300 from an animal moving around in this space. 00:29:59.300 --> 00:30:02.300 OK, so you see how these are very interesting kind 00:30:02.300 --> 00:30:03.320 of place fields. 00:30:03.320 --> 00:30:05.090 They're not just a nice round blob. 00:30:05.090 --> 00:30:07.070 They stretch around a whole order 00:30:07.070 --> 00:30:09.390 of the animal's environment. 00:30:09.390 --> 00:30:12.530 OK, so does that make you think of anything? 00:30:12.530 --> 00:30:14.510 Does that ring any bells with other stuff 00:30:14.510 --> 00:30:15.680 we've talked about in here? 00:30:19.300 --> 00:30:21.170 I think we've talked a bunch about how 00:30:21.170 --> 00:30:24.290 the parahippocampal place area cares about the shape of space 00:30:24.290 --> 00:30:25.455 around you. 00:30:25.455 --> 00:30:27.080 Well, you might think that you'd really 00:30:27.080 --> 00:30:30.110 want to have awareness of where you are with respect 00:30:30.110 --> 00:30:32.570 to navigational barriers. 00:30:32.570 --> 00:30:35.600 It turns out border cells respond not just to walls. 00:30:35.600 --> 00:30:37.490 If you put a rodent in an environment 00:30:37.490 --> 00:30:39.380 where there's a cliff they can't go off, 00:30:39.380 --> 00:30:42.950 the border cells also respond to the edge of that cliff. 00:30:42.950 --> 00:30:45.770 OK, so any navigational barrier basically 00:30:45.770 --> 00:30:47.870 telling you where you are with respect 00:30:47.870 --> 00:30:49.025 to navigational barriers. 00:30:51.530 --> 00:30:54.410 OK, all right. 00:30:54.410 --> 00:30:55.340 Blah, blah, blah. 00:30:55.340 --> 00:31:00.620 OK, so as I mentioned in the last lecture 00:31:00.620 --> 00:31:02.930 when we talked about the parahippocampal place area, 00:31:02.930 --> 00:31:04.850 the shape of space around you has 00:31:04.850 --> 00:31:09.020 this kind of privileged role in many aspects of navigation. 00:31:09.020 --> 00:31:12.470 OK, so now we're going to talk about this problem 00:31:12.470 --> 00:31:16.850 of reorienting or regaining your sense of direction 00:31:16.850 --> 00:31:19.430 once you've been disoriented. 00:31:19.430 --> 00:31:22.870 And so, again, I mentioned this before. 00:31:22.870 --> 00:31:25.370 But just to give you in give you the intuition of what we're 00:31:25.370 --> 00:31:29.090 talking about here, you come up from the subway in Manhattan 00:31:29.090 --> 00:31:32.360 or any other environment that's rectilinear 00:31:32.360 --> 00:31:35.880 that you know and you know which stop you're coming at up at. 00:31:35.880 --> 00:31:38.960 So you kind of know where you are, but you come out 00:31:38.960 --> 00:31:40.670 and you don't know which way to head. 00:31:40.670 --> 00:31:43.460 You don't know which way is which. 00:31:43.460 --> 00:31:47.027 So that's a modern version of a classic problem that animals 00:31:47.027 --> 00:31:48.110 face in their environment. 00:31:48.110 --> 00:31:50.540 They may know where they are, but that doesn't tell them 00:31:50.540 --> 00:31:52.220 which way they're facing. 00:31:52.220 --> 00:31:54.480 So just to be really concrete about this, 00:31:54.480 --> 00:31:56.060 so here's an aerial view of a person. 00:31:56.060 --> 00:31:57.980 You're standing here. 00:31:57.980 --> 00:32:01.910 You have a cognitive map in your mind, 00:32:01.910 --> 00:32:06.060 and your place cells are telling you your location in that map. 00:32:06.060 --> 00:32:10.190 OK, so you know where you are in that map. 00:32:10.190 --> 00:32:11.630 And you're looking down a street, 00:32:11.630 --> 00:32:13.580 so you know that you're oriented with respect 00:32:13.580 --> 00:32:17.450 to some external axis like this. 00:32:17.450 --> 00:32:20.720 But you don't know how your mental maps should 00:32:20.720 --> 00:32:22.530 be aligned with that street. 00:32:22.530 --> 00:32:28.250 Are you facing like this, facing north in Manhattan, 00:32:28.250 --> 00:32:31.160 or are you facing south? 00:32:31.160 --> 00:32:34.650 All right, so that's the problem of reorientation 00:32:34.650 --> 00:32:37.560 is figuring out your particular orientation, not just 00:32:37.560 --> 00:32:39.180 your location but which way you're 00:32:39.180 --> 00:32:42.270 facing in a known environment. 00:32:42.270 --> 00:32:44.610 And we've all faced some version of this presumably 00:32:44.610 --> 00:32:46.798 at some point, and it's annoying. 00:32:46.798 --> 00:32:48.090 It takes a while to figure out. 00:32:48.090 --> 00:32:50.340 And then I don't know if anybody's had this experience. 00:32:50.340 --> 00:32:51.870 I've had it only in Manhattan because that's 00:32:51.870 --> 00:32:53.398 where this arises for me, but I'm 00:32:53.398 --> 00:32:54.690 sure there are other locations. 00:32:54.690 --> 00:32:57.273 Where you come up and you think you're going one way, and then 00:32:57.273 --> 00:33:02.012 all of a sudden, it's like your whole mental map goes kaboom. 00:33:02.012 --> 00:33:03.720 How many people have had that experience? 00:33:03.720 --> 00:33:05.130 It's very sudden and punctate. 00:33:05.130 --> 00:33:07.950 Yeah, it turns out that when that happens, 00:33:07.950 --> 00:33:11.040 all of your neurons flip together in unison. 00:33:11.040 --> 00:33:12.390 Like they're all in cahoots. 00:33:12.390 --> 00:33:14.100 They have one version of this. 00:33:14.100 --> 00:33:15.690 When you have that experience, it's 00:33:15.690 --> 00:33:17.273 because they're all flipping together, 00:33:17.273 --> 00:33:19.450 and I'll show you some data on that in a second. 00:33:19.450 --> 00:33:21.690 OK, all right. 00:33:21.690 --> 00:33:24.420 So there's a very evolutionary old system 00:33:24.420 --> 00:33:26.550 for solving just this problem. 00:33:26.550 --> 00:33:28.778 And it's a wonderful little piece of the literature 00:33:28.778 --> 00:33:30.570 that I'm going to spend a couple of minutes 00:33:30.570 --> 00:33:34.050 on because it's so classic and so cool. 00:33:34.050 --> 00:33:37.680 And this started with work by Randy Gallistel in the 1980s. 00:33:37.680 --> 00:33:40.860 And so what he did was he studied this problem 00:33:40.860 --> 00:33:44.640 of reorientation-- that is figuring out your orientation 00:33:44.640 --> 00:33:47.700 in a known environment once you've been disoriented. 00:33:47.700 --> 00:33:51.360 It's a very particular aspect of the problem of navigation. 00:33:51.360 --> 00:33:54.270 So he put rats in a rectangular environment, 00:33:54.270 --> 00:33:56.820 and he had them explore the environment. 00:33:56.820 --> 00:34:00.357 And then he hid some rat-relevant thing, 00:34:00.357 --> 00:34:02.190 like a little piece of food, say a chocolate 00:34:02.190 --> 00:34:03.600 chip in that corner. 00:34:03.600 --> 00:34:06.190 OK, rat sees that happen, rat is interested. 00:34:06.190 --> 00:34:08.010 Take rat out of box before they get 00:34:08.010 --> 00:34:12.100 to go take the chocolate chip, and then you disorient the rat. 00:34:12.100 --> 00:34:14.350 You don't grab them by the tail and swing them around, 00:34:14.350 --> 00:34:16.034 but you do some slower version. 00:34:16.034 --> 00:34:17.159 You want to make them sick. 00:34:17.159 --> 00:34:18.850 You do some slower version of that 00:34:18.850 --> 00:34:21.600 so they've lost track of which way they're facing. 00:34:21.600 --> 00:34:24.870 OK, now you put them in a new box-- 00:34:24.870 --> 00:34:26.969 new box because you don't want the smell 00:34:26.969 --> 00:34:32.670 to still be there, new box, and you see which way the rat goes. 00:34:32.670 --> 00:34:38.253 And you find that the rat goes 50-50 to those two corners. 00:34:38.253 --> 00:34:39.920 What does that mean the rat has encoded? 00:34:43.380 --> 00:34:46.139 He doesn't go randomly to any corner. 00:34:46.139 --> 00:34:46.980 He goes to corners-- 00:34:46.980 --> 00:34:48.105 he knew it was in a corner. 00:34:48.105 --> 00:34:49.800 He doesn't go randomly to any corner. 00:34:49.800 --> 00:34:51.360 Yeah, Ben? 00:34:51.360 --> 00:34:52.270 Jack, I'm sorry. 00:34:52.270 --> 00:34:54.270 AUDIENCE: Or you can turn it around to the left. 00:34:54.270 --> 00:34:55.395 NANCY KANWISHER: Say again. 00:34:55.395 --> 00:34:58.130 AUDIENCE: Like it's specifically in one of these directions. 00:34:58.130 --> 00:34:59.402 So the left [INAUDIBLE]. 00:34:59.402 --> 00:35:01.860 NANCY KANWISHER: You've got to say a little more than that. 00:35:01.860 --> 00:35:02.940 What's to the left? 00:35:02.940 --> 00:35:05.790 What's different about those two corners than the other two? 00:35:05.790 --> 00:35:06.885 Yeah, Isabel? 00:35:06.885 --> 00:35:09.510 AUDIENCE: Well, if he's looking at the shape of the room-- i.e. 00:35:09.510 --> 00:35:11.427 these two longer walls and two shorter walls-- 00:35:11.427 --> 00:35:14.220 he recognizes that the space [INAUDIBLE] 00:35:14.220 --> 00:35:18.278 he has to go to what looks like the right [INAUDIBLE].. 00:35:18.278 --> 00:35:19.320 NANCY KANWISHER: Exactly. 00:35:19.320 --> 00:35:23.070 He has to have encoded the axis-- 00:35:23.070 --> 00:35:27.900 the fact that the room is longer on one axis than another. 00:35:27.900 --> 00:35:31.860 And he's essentially encoded that chocolate chip was 00:35:31.860 --> 00:35:34.050 on the right side of the long wall 00:35:34.050 --> 00:35:36.818 or the left side of the short wall, and both of those corners 00:35:36.818 --> 00:35:37.860 are consistent with that. 00:35:37.860 --> 00:35:40.350 That's why he goes 50-50 to them. 00:35:40.350 --> 00:35:42.658 He can't go 100% of the time to the right corner 00:35:42.658 --> 00:35:44.700 because he has no information that would tell him 00:35:44.700 --> 00:35:46.020 that in this experiment. 00:35:46.020 --> 00:35:47.070 OK, everybody clear? 00:35:47.070 --> 00:35:50.370 So it tells you he learned where the thing is with respect 00:35:50.370 --> 00:35:54.120 to the shape of the room and its particular aspect ratio. 00:35:54.120 --> 00:35:58.140 OK, so now the plot thickens, and now 00:35:58.140 --> 00:35:59.920 they repeat the experiment. 00:35:59.920 --> 00:36:03.420 But this time, they make some very rat-salient asymmetry 00:36:03.420 --> 00:36:04.230 over here. 00:36:04.230 --> 00:36:07.530 You make a color and a texture, and you 00:36:07.530 --> 00:36:12.850 make other things to make this wall very saliently different. 00:36:12.850 --> 00:36:15.520 So you would think the rat, motivated to find the chocolate 00:36:15.520 --> 00:36:18.550 chip, would now go 100% to that corner 00:36:18.550 --> 00:36:20.800 when we put them in the new box with the same landmark 00:36:20.800 --> 00:36:22.570 cue over there. 00:36:22.570 --> 00:36:27.670 But no, the rat goes 50-50 to the same two corners. 00:36:27.670 --> 00:36:30.837 And in control experiments, many control conditions, 00:36:30.837 --> 00:36:32.920 you can show-- and I'll show you one in a moment-- 00:36:32.920 --> 00:36:34.960 the rat absolutely knows about this wall. 00:36:34.960 --> 00:36:37.900 He's encoded the presence of that asymmetric wall, 00:36:37.900 --> 00:36:40.090 so he has the information that should 00:36:40.090 --> 00:36:42.070 enable him to break the symmetry, 00:36:42.070 --> 00:36:45.330 but he doesn't use it. 00:36:45.330 --> 00:36:45.920 That's weird. 00:36:45.920 --> 00:36:47.420 You should be surprised. 00:36:47.420 --> 00:36:49.957 OK, everybody get why that's weird? 00:36:49.957 --> 00:36:52.040 He could have solved this one perfectly this time. 00:36:52.040 --> 00:36:53.180 He has the information. 00:36:53.180 --> 00:36:56.120 He's not using that information. 00:36:56.120 --> 00:37:00.260 OK, so that's weird. 00:37:00.260 --> 00:37:02.540 But then Liz Spelke and her colleagues 00:37:02.540 --> 00:37:05.060 came along 10 years later and said, 00:37:05.060 --> 00:37:07.580 let's try this with infants. 00:37:07.580 --> 00:37:09.200 And so they did the infant version, 00:37:09.200 --> 00:37:13.460 where you put the infant in a room with a symmetrical-- 00:37:13.460 --> 00:37:16.670 in a rectangular room, and you hide the doors so the infant 00:37:16.670 --> 00:37:20.015 doesn't have any cues other than the shape of the room. 00:37:22.910 --> 00:37:26.990 18 to 24-month-old infants, and then you 00:37:26.990 --> 00:37:30.020 hide a toy in a corner, and you see what the infant does. 00:37:32.235 --> 00:37:33.860 Actually what you do with the infant is 00:37:33.860 --> 00:37:37.130 you make this wall really salient in all kinds of ways. 00:37:37.130 --> 00:37:39.680 In one case, it was red velvet, and they first 00:37:39.680 --> 00:37:42.830 showed the-- and these are, I guess, toddlers. 00:37:42.830 --> 00:37:45.470 They first show them that when you knock on the red wall, 00:37:45.470 --> 00:37:46.700 music happens. 00:37:46.700 --> 00:37:49.670 Totally cool, riveting for a little kid. 00:37:49.670 --> 00:37:50.580 They totally get it. 00:37:50.580 --> 00:37:52.550 They know all about the music wall. 00:37:52.550 --> 00:37:54.320 Very salient to them. 00:37:54.320 --> 00:37:56.960 Nonetheless, you put them in this experiment, 00:37:56.960 --> 00:37:58.700 and they behave just like rodents. 00:37:58.700 --> 00:38:01.100 They go 50-50 to the two corners. 00:38:01.100 --> 00:38:03.920 Even though they notice the red music wall, 00:38:03.920 --> 00:38:06.470 and it could have solved the problem for them perfectly. 00:38:06.470 --> 00:38:07.880 And they were motivated, but they 00:38:07.880 --> 00:38:09.650 didn't use the information. 00:38:09.650 --> 00:38:12.410 Everybody get why that's kind of interesting 00:38:12.410 --> 00:38:15.560 and kind of surprising? 00:38:15.560 --> 00:38:20.120 OK, now you might say, OK, rodents, infants, 00:38:20.120 --> 00:38:22.040 they're dummies. 00:38:22.040 --> 00:38:25.250 We wouldn't do that, us smart adult humans. 00:38:25.250 --> 00:38:27.080 Would we? 00:38:27.080 --> 00:38:30.830 But oh yes you would under certain circumstances. 00:38:30.830 --> 00:38:33.230 If we tied up your language system-- 00:38:33.230 --> 00:38:35.010 and there's lots of ways of doing that. 00:38:35.010 --> 00:38:37.340 One way is called shadowing. 00:38:37.340 --> 00:38:39.610 So it's like simultaneous translation 00:38:39.610 --> 00:38:40.610 but you don't translate. 00:38:40.610 --> 00:38:41.442 Try this sometime. 00:38:41.442 --> 00:38:43.400 I do this occasionally when I'm bored in my car 00:38:43.400 --> 00:38:46.460 just because it's amusingly difficult. Turn on the radio, 00:38:46.460 --> 00:38:48.740 listen to somebody talking, and just repeat everything 00:38:48.740 --> 00:38:52.618 they say after they say it. 00:38:52.618 --> 00:38:53.660 I'm not even translating. 00:38:53.660 --> 00:38:54.680 It's still demanding. 00:38:54.680 --> 00:38:57.270 So you have to be listening and producing. 00:38:57.270 --> 00:38:58.250 OK, running thing. 00:38:58.250 --> 00:39:00.350 OK, so that's called verbal shadowing, 00:39:00.350 --> 00:39:02.870 and it's an established way to really tie up 00:39:02.870 --> 00:39:05.510 your language system and take it offline so you can't really 00:39:05.510 --> 00:39:06.740 use it. 00:39:06.740 --> 00:39:10.520 When you do this experiment on human adults, 00:39:10.520 --> 00:39:13.250 if they're verbally shadowing and their language system 00:39:13.250 --> 00:39:16.580 is tied up, they behave just like rodents and infants. 00:39:16.580 --> 00:39:19.940 That is they use the shape of the space, 00:39:19.940 --> 00:39:22.160 but they don't use salient landmarks that could 00:39:22.160 --> 00:39:23.660 help them solve it perfectly. 00:39:23.660 --> 00:39:26.990 They go 50-50 to the two corners. 00:39:26.990 --> 00:39:29.980 They become rats and infants. 00:39:29.980 --> 00:39:32.350 We become rats and infants. 00:39:32.350 --> 00:39:36.640 OK, so Liz Spelke has spun a whole fascinating big 00:39:36.640 --> 00:39:40.577 theoretical story about what this really means. 00:39:40.577 --> 00:39:42.910 Well, let me just say a little bit more about this first 00:39:42.910 --> 00:39:44.470 before I do her whole big story. 00:39:44.470 --> 00:39:47.360 OK, yeah so the idea is-- so first of all, 00:39:47.360 --> 00:39:49.990 why would it make sense for rodents at least-- 00:39:49.990 --> 00:39:51.280 let's just consider the rats-- 00:39:51.280 --> 00:39:54.550 to use only the shape of space to reorient themselves 00:39:54.550 --> 00:39:55.870 when they're disoriented? 00:39:55.870 --> 00:39:58.270 At first glance, that seems really crazy. 00:39:58.270 --> 00:40:01.660 But if you think about rodents in natural environments, 00:40:01.660 --> 00:40:05.290 the idea is that actually in natural environments, 00:40:05.290 --> 00:40:06.610 features change. 00:40:06.610 --> 00:40:07.960 Snow comes and goes. 00:40:07.960 --> 00:40:09.250 Plants come and goes. 00:40:09.250 --> 00:40:11.650 Odors change. 00:40:11.650 --> 00:40:15.190 All those kinds of features of the environment can change, 00:40:15.190 --> 00:40:17.500 but the shape of the environment, 00:40:17.500 --> 00:40:20.110 like that there's a slope like this and a barrier here 00:40:20.110 --> 00:40:22.930 and a cliff there, those are more stable features 00:40:22.930 --> 00:40:24.260 of the environment. 00:40:24.260 --> 00:40:26.470 So it actually makes evolutionary sense 00:40:26.470 --> 00:40:31.360 for disoriented rodents at least to use the shape of space more 00:40:31.360 --> 00:40:34.180 than the features-- the colors and textures and odors 00:40:34.180 --> 00:40:35.260 of a space-- 00:40:35.260 --> 00:40:37.330 as landmarks to reorient themselves. 00:40:37.330 --> 00:40:38.720 Does that makes sense? 00:40:38.720 --> 00:40:44.500 And so the idea is that rodents have through evolution evolved 00:40:44.500 --> 00:40:46.450 this system for reorienting themselves 00:40:46.450 --> 00:40:49.090 when they lose their bearings that relies only 00:40:49.090 --> 00:40:53.050 on the shape of space so restrictively that even 00:40:53.050 --> 00:40:55.420 if another cube becomes relevant and important, 00:40:55.420 --> 00:40:58.480 they don't use it. 00:40:58.480 --> 00:41:00.040 And the further idea is that we have 00:41:00.040 --> 00:41:03.280 some version of this system in our heads as well. 00:41:03.280 --> 00:41:05.260 And as smart adult humans, we learn 00:41:05.260 --> 00:41:07.930 all kinds of other strategies to get beyond this. 00:41:07.930 --> 00:41:09.550 We're not trapped with only being 00:41:09.550 --> 00:41:11.290 able to use this one system to solve it. 00:41:11.290 --> 00:41:13.180 We can use other systems-- 00:41:13.180 --> 00:41:16.000 possibly language to help us say things to ourselves, 00:41:16.000 --> 00:41:19.000 like it's on the left side of the short wall. 00:41:19.000 --> 00:41:20.110 That's what Spelke thinks. 00:41:20.110 --> 00:41:21.693 There is some version in your head of, 00:41:21.693 --> 00:41:23.440 it's on the left side of the short wall, 00:41:23.440 --> 00:41:26.410 and that's why adults can do this when their language 00:41:26.410 --> 00:41:27.700 system isn't tied up. 00:41:27.700 --> 00:41:29.200 I don't think that's exactly right, 00:41:29.200 --> 00:41:32.230 but it's a beautiful story, and there's some evidence for it. 00:41:32.230 --> 00:41:36.670 OK, anyway, part of the reason I go through this whole thing-- 00:41:36.670 --> 00:41:38.650 well, one, I think these experiments are cool, 00:41:38.650 --> 00:41:42.460 but it's also been the basis of a core idea 00:41:42.460 --> 00:41:44.650 in cognitive science, and that idea 00:41:44.650 --> 00:41:47.030 is called informational encapsulation. 00:41:47.030 --> 00:41:49.030 So think about, it's just lots of syllables 00:41:49.030 --> 00:41:50.740 for a pretty simple idea. 00:41:50.740 --> 00:41:53.110 That you have this system for reorientation, 00:41:53.110 --> 00:41:55.180 and it is designed to use the shape of space 00:41:55.180 --> 00:41:58.270 around you as the cue that you use to reorient yourself 00:41:58.270 --> 00:41:59.650 when you're disoriented. 00:41:59.650 --> 00:42:03.220 That system is hardwired to do just that. 00:42:03.220 --> 00:42:05.140 And if some other part of your brain 00:42:05.140 --> 00:42:07.360 has information that could solve the problem, 00:42:07.360 --> 00:42:11.810 like the presence of a relevant feature that you could use, 00:42:11.810 --> 00:42:12.850 you don't have-- 00:42:12.850 --> 00:42:14.830 your re-orientation system doesn't have access 00:42:14.830 --> 00:42:15.700 to that information. 00:42:15.700 --> 00:42:17.950 It's informationally encapsulated. 00:42:17.950 --> 00:42:20.680 It only has access to the particular inputs 00:42:20.680 --> 00:42:23.830 that are hardwired into it. 00:42:23.830 --> 00:42:28.240 And so 20 years ago, a lot of people went wild with this 00:42:28.240 --> 00:42:31.660 and said that all the brain regions that I've talked about 00:42:31.660 --> 00:42:34.480 and cognitive systems that we're considering in this course 00:42:34.480 --> 00:42:36.280 are informationally encapsulated. 00:42:36.280 --> 00:42:38.320 It's kind of an extreme idea that 00:42:38.320 --> 00:42:40.270 goes far beyond functional specificity 00:42:40.270 --> 00:42:44.290 to say the inputs are extremely restricted to each region, 00:42:44.290 --> 00:42:46.180 and that's probably not true. 00:42:46.180 --> 00:42:49.330 But there's some limitations on the information 00:42:49.330 --> 00:42:51.460 that each of these processors were considering 00:42:51.460 --> 00:42:53.110 in this course has access to. 00:42:53.110 --> 00:42:57.220 And this is the classic evidence, behavioral evidence 00:42:57.220 --> 00:42:59.827 that some of those systems have very restricted inputs. 00:42:59.827 --> 00:43:02.410 Does that make sense, the idea of informational encapsulation? 00:43:02.410 --> 00:43:05.090 Not as an absolute truth about the brain, 00:43:05.090 --> 00:43:08.560 but as an idea that is interesting to consider 00:43:08.560 --> 00:43:11.870 individually for each of the systems we study. 00:43:11.870 --> 00:43:15.020 There's been pushback about the extremeness of this claim 00:43:15.020 --> 00:43:18.265 that infants and rodents only use the shape of space. 00:43:18.265 --> 00:43:19.640 There are circumstances where you 00:43:19.640 --> 00:43:21.680 can get them to use other information, 00:43:21.680 --> 00:43:23.900 but it's definitely true that the shape of space 00:43:23.900 --> 00:43:26.330 is the dominant cue for reorienting 00:43:26.330 --> 00:43:30.590 in rodents and in infants. 00:43:30.590 --> 00:43:34.660 All right, so when you're lost, as I've mentioned, 00:43:34.660 --> 00:43:36.410 there's two questions you need to answer-- 00:43:36.410 --> 00:43:38.738 where are you and which way you're oriented. 00:43:38.738 --> 00:43:40.280 This last stuff we were talking about 00:43:40.280 --> 00:43:42.770 is about which way you're oriented question. 00:43:42.770 --> 00:43:44.840 OK, and I just showed you some evidence 00:43:44.840 --> 00:43:47.570 for this general finding that the geometric cues, 00:43:47.570 --> 00:43:50.870 the shape of space are the dominant cues 00:43:50.870 --> 00:43:54.290 you use to reorient yourself, to get your heading back 00:43:54.290 --> 00:43:57.320 when you're disoriented. 00:43:57.320 --> 00:44:00.590 But do we really know that those cues 00:44:00.590 --> 00:44:03.530 are different for place recognition 00:44:03.530 --> 00:44:05.330 and for heading direction? 00:44:05.330 --> 00:44:07.790 So I've said, here are two different parts of the problem. 00:44:07.790 --> 00:44:10.070 But do they function differently? 00:44:10.070 --> 00:44:11.930 Do we really use different cues? 00:44:11.930 --> 00:44:14.120 Do we use the shape of space more 00:44:14.120 --> 00:44:17.530 for heading direction and maybe other cues for place 00:44:17.530 --> 00:44:19.790 recognition, for knowing where we are? 00:44:19.790 --> 00:44:21.920 OK, so I'm going to show you a very elegant 00:44:21.920 --> 00:44:24.650 behavioral experiment in mice that does this 00:44:24.650 --> 00:44:27.090 all at once in one experiment. 00:44:27.090 --> 00:44:31.100 So this is Josh Julian, a former lab tech in my lab. 00:44:31.100 --> 00:44:32.758 I get no credit for this whatsoever. 00:44:32.758 --> 00:44:34.550 I'm proud even though I shouldn't be proud. 00:44:34.550 --> 00:44:36.410 He was just an endogenously smart guy 00:44:36.410 --> 00:44:39.710 who went on and did an awesome experiment after he left my lab 00:44:39.710 --> 00:44:43.250 and went off to grad school, and here's his awesome experiment. 00:44:43.250 --> 00:44:47.120 OK, so he said, let's get mice to do both of these tasks. 00:44:47.120 --> 00:44:50.593 They have to know where they are and which way they're oriented. 00:44:50.593 --> 00:44:52.760 OK, we're going to do the same disorientation thing. 00:44:52.760 --> 00:44:55.520 Take them out, turn them around till they're disoriented. 00:44:55.520 --> 00:44:58.680 But these mice have to learn two different environments. 00:44:58.680 --> 00:45:02.000 OK, one environment has the vertical stripes 00:45:02.000 --> 00:45:04.500 on the short wall, on one of the short walls. 00:45:04.500 --> 00:45:07.070 The other environment has horizontal stripes 00:45:07.070 --> 00:45:09.800 on the short wall. 00:45:09.800 --> 00:45:11.360 So you do the same experiment. 00:45:11.360 --> 00:45:14.450 You bait one corner, and you see where the rodent goes does. 00:45:14.450 --> 00:45:18.710 Does he go to the two opposite corners? 00:45:18.710 --> 00:45:21.920 Exactly the same experiment, but he has to remember which room 00:45:21.920 --> 00:45:23.720 is-- to solve the problem, he has to know-- 00:45:23.720 --> 00:45:26.750 he has to discover, rediscover the vertical stripes 00:45:26.750 --> 00:45:29.330 or the horizontal stripes and act accordingly. 00:45:29.330 --> 00:45:31.610 Because when he's in the vertical context, 00:45:31.610 --> 00:45:34.020 the thing gets hid. 00:45:34.020 --> 00:45:35.510 He does this over repeated trials. 00:45:35.510 --> 00:45:40.698 The food gets hid on the-- 00:45:40.698 --> 00:45:41.990 hang on, let me get this right. 00:45:41.990 --> 00:45:43.660 Long wall on the left. 00:45:43.660 --> 00:45:44.160 Do, do, do. 00:45:44.160 --> 00:45:46.610 Yeah, right. 00:45:46.610 --> 00:45:51.860 Yeah, so when the long wall is on the left of the rodent. 00:45:51.860 --> 00:45:53.850 OK, that corner, the long wall is on the left. 00:45:53.850 --> 00:45:54.860 Everybody oriented? 00:45:54.860 --> 00:45:59.750 Whereas when he's in the blue context, the reward here, 00:45:59.750 --> 00:46:01.928 the long wall is on the right. 00:46:01.928 --> 00:46:04.220 OK, so he has to learn those two different environments 00:46:04.220 --> 00:46:08.120 and that the relevant shape cues are opposite in each. 00:46:08.120 --> 00:46:09.740 OK, everybody got that? 00:46:09.740 --> 00:46:13.610 OK, now what you find is that the rodent 00:46:13.610 --> 00:46:15.020 can learn that just fine. 00:46:15.020 --> 00:46:17.840 OK, so this shows that when you put the rodent 00:46:17.840 --> 00:46:21.290 in the vertical context in a room like this, 00:46:21.290 --> 00:46:26.330 they go more to these two corners than those two corners. 00:46:26.330 --> 00:46:29.840 Whereas when you put him in a horizontal context 00:46:29.840 --> 00:46:32.420 with horizontal stripes, he goes more to those two corners 00:46:32.420 --> 00:46:34.400 than those two corners. 00:46:34.400 --> 00:46:36.860 That tells you the rodent has used 00:46:36.860 --> 00:46:39.530 the orientation of the stripes to figure out 00:46:39.530 --> 00:46:41.750 which room he's in and hence, which 00:46:41.750 --> 00:46:44.690 two corners are the right ones. 00:46:44.690 --> 00:46:46.700 Everybody got that? 00:46:46.700 --> 00:46:48.110 But here's the amazing thing-- 00:46:48.110 --> 00:46:51.020 even though in this experiment, the very same animals 00:46:51.020 --> 00:46:55.160 in the very same trials, are using those stripes to figure 00:46:55.160 --> 00:46:59.030 out which room they're in, they don't use those stripes at all 00:46:59.030 --> 00:47:02.370 to break the symmetry and to go only to the correct corner, 00:47:02.370 --> 00:47:06.380 which they could do but don't. 00:47:06.380 --> 00:47:10.970 So once you've trained the rodents on these two things 00:47:10.970 --> 00:47:15.660 that the reward is here in the vertical context 00:47:15.660 --> 00:47:18.680 and they're in the horizontal context, you disorient them, 00:47:18.680 --> 00:47:19.745 you put them back in. 00:47:19.745 --> 00:47:21.620 You find that when you have vertical stripes, 00:47:21.620 --> 00:47:22.950 they go to these two corners-- 00:47:22.950 --> 00:47:24.283 I'm just repeating the data. 00:47:24.283 --> 00:47:25.700 When there are horizontal stripes, 00:47:25.700 --> 00:47:26.908 they go to those two corners. 00:47:26.908 --> 00:47:28.490 OK, they've learned that. 00:47:28.490 --> 00:47:30.710 But why do they go to those two corners? 00:47:30.710 --> 00:47:32.330 They learned the damn stripes. 00:47:32.330 --> 00:47:35.130 They used them to know which room they're in, 00:47:35.130 --> 00:47:37.530 but they don't use them to break the asymmetry 00:47:37.530 --> 00:47:39.155 and decide which is the correct corner. 00:47:42.260 --> 00:47:44.240 OK, so this is like a microcosm of everything 00:47:44.240 --> 00:47:46.970 I've been saying so far all in one experiment. 00:47:46.970 --> 00:47:49.640 The rodents are noticing those feature cues, 00:47:49.640 --> 00:47:52.100 using them to figure out which room they're in, 00:47:52.100 --> 00:47:54.920 where are they, but failing to use 00:47:54.920 --> 00:47:57.890 those features, the orientation of the stripes, 00:47:57.890 --> 00:48:01.340 to figure out which of the two corners is the correct one. 00:48:01.340 --> 00:48:04.430 They're not even encoding food is near stripes. 00:48:04.430 --> 00:48:08.220 Like, duh, that should have been easy. 00:48:08.220 --> 00:48:10.318 All right, so this is a beautiful-- 00:48:10.318 --> 00:48:12.860 I mean, this is more evidence for informational encapsulation 00:48:12.860 --> 00:48:13.610 of this system. 00:48:13.610 --> 00:48:15.703 Because it shows us on the very same trial, 00:48:15.703 --> 00:48:17.870 they used the stripe information to know which room. 00:48:17.870 --> 00:48:20.330 They failed to use it to figure out 00:48:20.330 --> 00:48:21.800 their orientation in that room. 00:48:25.172 --> 00:48:26.380 Is this sort of making sense? 00:48:26.380 --> 00:48:27.630 I realize it's kind of subtle. 00:48:27.630 --> 00:48:31.660 It's sort of simple and subtle at the same time. 00:48:31.660 --> 00:48:32.465 Yeah? 00:48:32.465 --> 00:48:34.390 AUDIENCE: So with cells that you showed us 00:48:34.390 --> 00:48:37.030 in the very first maps-- 00:48:37.030 --> 00:48:38.780 NANCY KANWISHER: What are they doing here? 00:48:38.780 --> 00:48:39.550 AUDIENCE: Yeah, exactly. 00:48:39.550 --> 00:48:40.883 NANCY KANWISHER: Great question. 00:48:40.883 --> 00:48:41.675 Let's look at that. 00:48:41.675 --> 00:48:42.883 That's what we're doing next. 00:48:42.883 --> 00:48:43.810 It's a great question. 00:48:43.810 --> 00:48:46.180 What are the damn place cells doing here? 00:48:46.180 --> 00:48:47.972 Great question. 00:48:47.972 --> 00:48:49.930 OK, let's say a little bit more, and then we'll 00:48:49.930 --> 00:48:51.790 think about what the place cells are doing. 00:48:51.790 --> 00:48:56.080 OK, so let me just restate, cash out the findings here. 00:48:56.080 --> 00:48:58.930 The mice are using the features to figure out 00:48:58.930 --> 00:48:59.930 which place they're in. 00:48:59.930 --> 00:49:02.170 Are they in this one or that one? 00:49:02.170 --> 00:49:06.280 But they are failing to use those features to figure out 00:49:06.280 --> 00:49:07.510 which is a correct corner. 00:49:07.510 --> 00:49:09.670 They're still 50-50 for the two corners, 00:49:09.670 --> 00:49:12.280 even though logically they have that information, 00:49:12.280 --> 00:49:16.780 and they could use it, and they should use it, they don't. 00:49:16.780 --> 00:49:19.210 So that means the mice are using features-- 00:49:19.210 --> 00:49:22.300 in this case, orientation-- for place recognition, 00:49:22.300 --> 00:49:27.947 but not for regaining their orientation within that place. 00:49:27.947 --> 00:49:29.530 I'm just repeating what I said before. 00:49:29.530 --> 00:49:31.000 Is that making sense? 00:49:31.000 --> 00:49:35.410 OK, so now David's question, what are the place cells 00:49:35.410 --> 00:49:36.190 doing here? 00:49:36.190 --> 00:49:37.210 Great question. 00:49:37.210 --> 00:49:37.960 Let's look. 00:49:37.960 --> 00:49:40.990 It's mice, so we can do that or Keinath et al. 00:49:40.990 --> 00:49:44.050 can do that and Josh Julian, my amazing former lab tech. 00:49:44.050 --> 00:49:47.750 So again, I get no credit whatsoever. 00:49:47.750 --> 00:49:50.110 So what do they do? 00:49:50.110 --> 00:49:54.790 They allow the mice to forage for crumbs in a box like this. 00:49:54.790 --> 00:49:59.280 OK, they disorient the mouse before each trial. 00:49:59.280 --> 00:50:01.460 Take them out, turn them around so he doesn't 00:50:01.460 --> 00:50:02.690 know which way he's facing. 00:50:02.690 --> 00:50:03.530 Put them in the box. 00:50:06.350 --> 00:50:10.040 And they find that place cells have a particular location 00:50:10.040 --> 00:50:11.000 in that box. 00:50:11.000 --> 00:50:11.660 Not surprising. 00:50:11.660 --> 00:50:14.220 That's what place cells do. 00:50:14.220 --> 00:50:16.460 So here are two different trials-- 00:50:16.460 --> 00:50:18.260 two different cells that were mapped out 00:50:18.260 --> 00:50:19.505 in a rodent doing this. 00:50:19.505 --> 00:50:23.360 This cell responds always in that corner. 00:50:23.360 --> 00:50:26.367 Another cell responds only in that corner. 00:50:26.367 --> 00:50:27.950 OK, these are just place cells like we 00:50:27.950 --> 00:50:30.035 described before doing what place cells do. 00:50:33.880 --> 00:50:40.830 But now, sometimes those place cells are off by 180 degrees, 00:50:40.830 --> 00:50:44.530 even though the stripes should resolve the ambiguity. 00:50:44.530 --> 00:50:49.620 OK, so those same cells on other trials 00:50:49.620 --> 00:50:51.030 respond to the opposite corner. 00:50:54.670 --> 00:50:58.960 So the place cells are doing just what the rodent is doing. 00:50:58.960 --> 00:51:00.760 The place cells are confused. 00:51:00.760 --> 00:51:03.770 Am I facing-- 00:51:03.770 --> 00:51:06.460 I am oriented like this, or am I oriented like that? 00:51:06.460 --> 00:51:10.660 The place cells don't know, and the rodent doesn't know. 00:51:10.660 --> 00:51:12.880 And the coolest thing about this experiment 00:51:12.880 --> 00:51:14.650 is that these things are linked. 00:51:14.650 --> 00:51:17.530 On the trials where the rodent goes to the wrong corner, 00:51:17.530 --> 00:51:20.920 the place cells are also in the wrong corner. 00:51:20.920 --> 00:51:23.230 OK, they systematically determine 00:51:23.230 --> 00:51:24.670 which way the animal will go. 00:51:27.610 --> 00:51:30.100 Oh, and also as I mentioned before, all those cells are 00:51:30.100 --> 00:51:30.640 in cahoots. 00:51:30.640 --> 00:51:33.140 They're all in sync going the same way. 00:51:33.140 --> 00:51:36.550 So when one of the cells rotates to the opposite corner, 00:51:36.550 --> 00:51:39.910 all the other ones rotate to the opposite corner. 00:51:39.910 --> 00:51:43.150 So it's as though somehow on trial to trial, 00:51:43.150 --> 00:51:45.970 the rodent thinks he's oriented in one way, 00:51:45.970 --> 00:51:48.320 he's actually 50-50 which way he's oriented. 00:51:48.320 --> 00:51:51.970 He's not using the feature cues, and his behavior 00:51:51.970 --> 00:51:55.240 according to where he looks for the food exactly follows 00:51:55.240 --> 00:51:59.170 that way he's oriented and so do all of his place cells. 00:51:59.170 --> 00:52:01.690 OK, that's that whole system goes together. 00:52:01.690 --> 00:52:04.870 That tells you that those place cells are relevant 00:52:04.870 --> 00:52:05.590 behaviorally. 00:52:05.590 --> 00:52:07.930 They are the system that either directly 00:52:07.930 --> 00:52:10.150 determines or is tightly linked to the system that 00:52:10.150 --> 00:52:14.770 determines which way the animal thinks he's facing. 00:52:14.770 --> 00:52:18.670 OK, I realize this is a little bit complicated. 00:52:18.670 --> 00:52:21.820 Does it make sense to you that as we've been talking about 00:52:21.820 --> 00:52:24.820 with reorientation, even though the animals should know 00:52:24.820 --> 00:52:27.430 from this stripe the difference between that corner 00:52:27.430 --> 00:52:29.770 in this corner, he doesn't know behaviorally. 00:52:29.770 --> 00:52:32.500 He's looking for food right there, and yet he goes 50-50. 00:52:32.500 --> 00:52:34.510 Weird and stupid, right? 00:52:34.510 --> 00:52:37.550 Place cells do the same thing. 00:52:37.550 --> 00:52:41.860 And further, the place cells and the behavior go together. 00:52:41.860 --> 00:52:42.427 Yeah, Sasha? 00:52:42.427 --> 00:52:45.010 AUDIENCE: So if you're reading information off of place cells, 00:52:45.010 --> 00:52:46.360 can you zap it? 00:52:46.360 --> 00:52:48.400 Can you [INAUDIBLE]? 00:52:48.400 --> 00:52:50.660 NANCY KANWISHER: Wouldn't that be nice? 00:52:50.660 --> 00:52:53.590 Turns out you can't for the reason someone over here 00:52:53.590 --> 00:52:54.940 asked a long time ago. 00:52:54.940 --> 00:52:57.890 You, I think. 00:52:57.890 --> 00:53:00.168 And that's because they're all interleaved together. 00:53:00.168 --> 00:53:02.710 And if you zap just one, you're not going to have-- one cell, 00:53:02.710 --> 00:53:03.860 you're not going to have an effect. 00:53:03.860 --> 00:53:05.943 And if you zap a whole region, you get all of them 00:53:05.943 --> 00:53:07.130 and you get muck. 00:53:07.130 --> 00:53:09.670 So you can't do that manipulation unfortunately. 00:53:09.670 --> 00:53:14.950 You need some kind of topography to do the manipulation. 00:53:14.950 --> 00:53:16.825 OK, so I just said how all this-- 00:53:16.825 --> 00:53:19.930 I got ahead of myself-- how it relates to behavior, 00:53:19.930 --> 00:53:21.530 but just to go through that quickly. 00:53:21.530 --> 00:53:23.110 So what we've done here is they've 00:53:23.110 --> 00:53:25.750 trained the mouse on this classic reorientation task. 00:53:25.750 --> 00:53:28.360 They disorient the mouse before each trial while recording 00:53:28.360 --> 00:53:30.340 from hippocampal place cells. 00:53:30.340 --> 00:53:34.780 As before, given cell flips 180 degrees from trial to trial, 00:53:34.780 --> 00:53:36.550 despite the fact that the stripes 00:53:36.550 --> 00:53:39.820 should disambiguate it and tell them which way he's oriented. 00:53:39.820 --> 00:53:42.280 And by the way, head direction cells and the grid cells 00:53:42.280 --> 00:53:47.410 also flip in the same way in cahoots with the place cells. 00:53:47.410 --> 00:53:49.870 But you can tell which corner the animal 00:53:49.870 --> 00:53:54.850 will go to by looking at what the place cells respond. 00:53:54.850 --> 00:54:00.730 And so when this place cell represents that location, 00:54:00.730 --> 00:54:02.560 the animal searches first there. 00:54:02.560 --> 00:54:06.430 And when it flips around, they search in the opposite corner. 00:54:06.430 --> 00:54:09.013 OK, so all of that just shows this really strong link 00:54:09.013 --> 00:54:10.555 between the place cells and behavior. 00:54:13.600 --> 00:54:16.750 So to recap, we've talked about four different kinds 00:54:16.750 --> 00:54:20.170 of cells involved in representing space 00:54:20.170 --> 00:54:22.030 and navigating around in it. 00:54:22.030 --> 00:54:24.220 Place cells that are like the "you are here," 00:54:24.220 --> 00:54:26.680 they respond when you're in a particular location. 00:54:26.680 --> 00:54:28.330 Direction cells that respond when 00:54:28.330 --> 00:54:32.080 you're heading in one direction, not in another direction. 00:54:32.080 --> 00:54:35.290 Border cells that fire when you're 00:54:35.290 --> 00:54:37.450 near a particular border in the environment. 00:54:37.450 --> 00:54:39.010 I have border cells going right now 00:54:39.010 --> 00:54:40.260 throughout this whole lecture. 00:54:40.260 --> 00:54:43.240 I've got a batch of border cells that are going. 00:54:43.240 --> 00:54:47.080 Grid cells that do this amazing thing of firing when 00:54:47.080 --> 00:54:49.270 the animal is in multiple different locations, 00:54:49.270 --> 00:54:51.160 and those locations that make it fire 00:54:51.160 --> 00:54:53.810 are arranged in a hexagonal grid. 00:54:53.810 --> 00:54:56.020 Think of it as a kind of ruler telling the rodent 00:54:56.020 --> 00:54:58.240 how far he's gone in this space, and those grid cells 00:54:58.240 --> 00:54:59.073 are like the rulers. 00:55:03.070 --> 00:55:04.810 Yeah, right. 00:55:04.810 --> 00:55:07.810 Those are the four kind we've talked about. 00:55:07.810 --> 00:55:10.390 So now, here's the cool thing. 00:55:10.390 --> 00:55:13.090 All this stuff-- navigation is awesome. 00:55:13.090 --> 00:55:16.180 We need it, it's important. 00:55:16.180 --> 00:55:19.480 All mobile animals need it for the reasons 00:55:19.480 --> 00:55:21.560 we've been talking about. 00:55:21.560 --> 00:55:24.280 But we can use this whole system for so much more 00:55:24.280 --> 00:55:25.400 than just navigation. 00:55:25.400 --> 00:55:27.580 Once you have this fancy system in your head 00:55:27.580 --> 00:55:29.650 to keep track of your location, to keep track 00:55:29.650 --> 00:55:32.308 of your direction, to keep track of where things are, 00:55:32.308 --> 00:55:33.850 how you're moving through that space, 00:55:33.850 --> 00:55:35.980 you can use that whole magnificent system 00:55:35.980 --> 00:55:37.150 in other ways. 00:55:37.150 --> 00:55:38.950 And in the last three or four years, 00:55:38.950 --> 00:55:41.200 there's just a huge number of studies that are really 00:55:41.200 --> 00:55:43.180 starting to take this very seriously, 00:55:43.180 --> 00:55:45.430 particularly the grid cells and thinking 00:55:45.430 --> 00:55:47.223 about how the grid cells-- 00:55:47.223 --> 00:55:48.640 I mean, probably the whole system, 00:55:48.640 --> 00:55:50.348 but people have been focusing on the grid 00:55:50.348 --> 00:55:55.220 cells and how they're used in multiple different situations. 00:55:55.220 --> 00:55:56.830 So here's one. 00:55:56.830 --> 00:55:59.590 OK, this is a cool study where what these guys did was 00:55:59.590 --> 00:56:01.690 they stuck a little device hanging around people's 00:56:01.690 --> 00:56:02.860 necks, the subjects' necks. 00:56:02.860 --> 00:56:04.720 Have a little camera aiming forward. 00:56:04.720 --> 00:56:07.180 It takes pictures at random intervals 00:56:07.180 --> 00:56:09.940 and records the person's GPS location. 00:56:09.940 --> 00:56:12.265 OK, so you send them off for a few months 00:56:12.265 --> 00:56:14.140 with this little device, and you do something 00:56:14.140 --> 00:56:15.340 to protect people's privacy. 00:56:15.340 --> 00:56:17.173 I don't know exactly how they maneuver that, 00:56:17.173 --> 00:56:18.817 but I'm sure they found a way. 00:56:18.817 --> 00:56:20.650 And so then they get this set of photographs 00:56:20.650 --> 00:56:24.340 taken from this person's front view of wherever they were 00:56:24.340 --> 00:56:26.380 over several months as they went wherever 00:56:26.380 --> 00:56:29.380 they went in their lives with a little GPS 00:56:29.380 --> 00:56:31.090 tag for each photograph. 00:56:31.090 --> 00:56:34.210 OK, so then what they do is they bring the subjects in 00:56:34.210 --> 00:56:35.710 and pop them in the scanner and show 00:56:35.710 --> 00:56:36.877 them some of those pictures. 00:56:40.380 --> 00:56:43.110 And they asked people to relive the experience 00:56:43.110 --> 00:56:46.830 that they had when they were looking at that thing. 00:56:46.830 --> 00:56:49.560 Put this on me, it'd be my monitor like all the time, 00:56:49.560 --> 00:56:52.290 and I wouldn't know which experience to relive, 00:56:52.290 --> 00:56:56.730 but I guess these people had richer lives let's hope. 00:56:56.730 --> 00:57:00.720 OK, so now what they do is they use multiple Voxel pattern 00:57:00.720 --> 00:57:05.490 analysis in the hippocampus while people are reliving 00:57:05.490 --> 00:57:07.860 those experiences in the scanner by looking 00:57:07.860 --> 00:57:10.545 at those images taken from their front facing cameras. 00:57:13.170 --> 00:57:15.970 And then they asked, is the pattern of response 00:57:15.970 --> 00:57:18.095 in the hippocampus, like some bunch of voxels, here 00:57:18.095 --> 00:57:20.730 is some pattern, is it more similar 00:57:20.730 --> 00:57:26.100 for events the subject remembers that were nearby in space? 00:57:26.100 --> 00:57:29.130 OK, so you do this for me, it's like yes, I occasionally 00:57:29.130 --> 00:57:32.280 go to the Stata cafeteria, and I occasionally 00:57:32.280 --> 00:57:37.055 go to the Koch Center cafeteria, and I spent a lot of time 00:57:37.055 --> 00:57:38.430 at home, and those two things are 00:57:38.430 --> 00:57:40.140 closer to each other than my home thing. 00:57:40.140 --> 00:57:43.050 Are the patterns more similar for nearby locations 00:57:43.050 --> 00:57:47.160 than for more distant locations? 00:57:47.160 --> 00:57:48.480 And they were. 00:57:48.480 --> 00:57:54.390 So this is the distance on a log scale between two patterns that 00:57:54.390 --> 00:57:56.850 result from the subject looking at two different images, 00:57:56.850 --> 00:58:02.380 and this is the similarity of the pattern in the hippocampus. 00:58:02.380 --> 00:58:04.320 Now some of you might be wondering, 00:58:04.320 --> 00:58:05.747 and in fact, I wonder this too. 00:58:05.747 --> 00:58:07.830 I think this is a cool study so I'm presenting it, 00:58:07.830 --> 00:58:10.080 but it doesn't make sense to me because everything 00:58:10.080 --> 00:58:12.870 we know about the hippocampus is those place cells 00:58:12.870 --> 00:58:14.580 are pretty interleaved. 00:58:14.580 --> 00:58:18.780 So how you manage to get a pattern response reading out 00:58:18.780 --> 00:58:22.440 a systematic location out of the hippocampus is a mystery to me. 00:58:22.440 --> 00:58:26.100 So they can't be fully-- there must be some kind of structure 00:58:26.100 --> 00:58:29.490 in there to the layout of those cells 00:58:29.490 --> 00:58:31.380 to enable them to get this information. 00:58:31.380 --> 00:58:33.750 OK, does everybody get how it's telling you 00:58:33.750 --> 00:58:39.690 that the hippocampus is remembering and reliving 00:58:39.690 --> 00:58:42.030 some representation of the locations 00:58:42.030 --> 00:58:45.210 of where you had those experiences? 00:58:45.210 --> 00:58:47.220 Everybody get how this shows us? 00:58:47.220 --> 00:58:49.320 But then they asked another interesting question, 00:58:49.320 --> 00:58:52.727 and they said, oh, does it also represent time? 00:58:52.727 --> 00:58:54.810 So we've been talking about space for the last two 00:58:54.810 --> 00:58:57.143 lectures, but now we're going straight off the deep end. 00:58:57.143 --> 00:58:59.520 And our first step, not even near the deep end 00:58:59.520 --> 00:59:03.217 yet is, does it do not just space but time? 00:59:03.217 --> 00:59:04.800 So they can take all those photographs 00:59:04.800 --> 00:59:06.870 and say, OK, how far apart in time 00:59:06.870 --> 00:59:09.060 were these two photographs taken? 00:59:09.060 --> 00:59:11.640 And they can do the same graph, and yes, they 00:59:11.640 --> 00:59:13.500 get a relationship with time as well. 00:59:13.500 --> 00:59:17.350 The farther apart in time people saw those two patterns, 00:59:17.350 --> 00:59:20.910 the more different the patterns in the hippocampus. 00:59:20.910 --> 00:59:22.770 Isn't that cool? 00:59:22.770 --> 00:59:26.520 OK, so that's one example showing 00:59:26.520 --> 00:59:30.510 that the hippocampus holds some kind of large scale 00:59:30.510 --> 00:59:34.600 representation of not just space but also time. 00:59:34.600 --> 00:59:37.830 And so there's a lot of work on how this gives structures 00:59:37.830 --> 00:59:42.240 to our memories for distances over the range of 100 meters 00:59:42.240 --> 00:59:45.267 and times between 15 hours and a month. 00:59:45.267 --> 00:59:47.100 I'm going to run out of time, so unless it's 00:59:47.100 --> 00:59:49.500 a clarification question, I'm going to keep going. 00:59:49.500 --> 00:59:50.030 Yeah, OK. 00:59:50.030 --> 00:59:51.750 AUDIENCE: Are different states confounded for time? 00:59:51.750 --> 00:59:52.860 NANCY KANWISHER: Yeah, yes. 00:59:52.860 --> 00:59:55.277 If you just did it like that-- so you have to do something 00:59:55.277 --> 00:59:58.290 to pick out time things that aren't confounded with space. 00:59:58.290 --> 00:59:59.902 You have this big sample of pictures, 00:59:59.902 --> 01:00:01.860 and you take a subset where you balance for it. 01:00:01.860 --> 01:00:02.580 Absolutely. 01:00:02.580 --> 01:00:04.170 They would have to do that. 01:00:04.170 --> 01:00:08.400 I can't actually remember, but they must have done that. 01:00:08.400 --> 01:00:10.158 Imperfect as peer review is, you'd never 01:00:10.158 --> 01:00:11.700 get through peer review if you didn't 01:00:11.700 --> 01:00:14.130 take care of that problem. 01:00:14.130 --> 01:00:16.240 OK, so that's the first thing. 01:00:16.240 --> 01:00:19.890 Here's another even more radical example. 01:00:19.890 --> 01:00:23.280 So people have shown that grid-like representations-- 01:00:23.280 --> 01:00:24.990 and I'm skipping over most of the details 01:00:24.990 --> 01:00:27.240 here to give you the gist because actually the details 01:00:27.240 --> 01:00:29.460 are a bit complicated. 01:00:29.460 --> 01:00:32.550 But they've shown that people seem to use their grid cell 01:00:32.550 --> 01:00:38.070 system when they are thinking about conceptual spaces, not 01:00:38.070 --> 01:00:39.840 just physical spaces. 01:00:39.840 --> 01:00:42.930 OK, so there's one classic experiment 01:00:42.930 --> 01:00:46.470 in which these guys taught subjects a conceptual space. 01:00:46.470 --> 01:00:49.830 They taught them about different kinds of birds, 01:00:49.830 --> 01:00:52.320 and these birds differed on two dimensions. 01:00:52.320 --> 01:00:55.920 They could vary in neck length or in leg length. 01:00:55.920 --> 01:00:57.840 And these things were orthogonal varied, 01:00:57.840 --> 01:00:59.550 so they made some artificial birds 01:00:59.550 --> 01:01:02.130 that filled up that space. 01:01:02.130 --> 01:01:03.660 And so here are some of the birds. 01:01:03.660 --> 01:01:06.390 This one has short legs, and OK, here's 01:01:06.390 --> 01:01:08.280 one with short legs and a long neck. 01:01:08.280 --> 01:01:12.210 And here's one with a longer neck and shorter-- 01:01:12.210 --> 01:01:12.930 wait, let's see. 01:01:12.930 --> 01:01:15.240 Longer legs and shorter neck right there. 01:01:15.240 --> 01:01:18.000 OK, so you've got every possible combination. 01:01:18.000 --> 01:01:20.220 They didn't show people a space like that. 01:01:20.220 --> 01:01:24.540 They just taught them things about these different birds. 01:01:24.540 --> 01:01:27.880 They had to remember their names and various facts about them. 01:01:27.880 --> 01:01:30.840 And so the idea is that when people learn about those birds, 01:01:30.840 --> 01:01:34.170 they mentally construct a 2D space. 01:01:34.170 --> 01:01:36.330 Because, in fact, those birds were 01:01:36.330 --> 01:01:39.210 generated from a 2D space, varying neck length 01:01:39.210 --> 01:01:40.155 and leg length. 01:01:42.770 --> 01:01:45.710 And so then when they scan subjects, 01:01:45.710 --> 01:01:50.030 they found essentially a neural signature 01:01:50.030 --> 01:01:54.870 of a grid system representing that 2D space. 01:01:54.870 --> 01:01:57.620 So even though the grid system presumably 01:01:57.620 --> 01:02:00.950 evolved to enable us to navigate around in a 2D space 01:02:00.950 --> 01:02:04.100 and keep track of where we are in that 2D space, 01:02:04.100 --> 01:02:06.830 it seems like it's now getting co-opted and being 01:02:06.830 --> 01:02:10.290 used for all kinds of representations of 2D spaces, 01:02:10.290 --> 01:02:14.990 including extremely abstract, artificial learned 2D spaces 01:02:14.990 --> 01:02:17.635 that you weren't even taught explicitly as a 2D space. 01:02:17.635 --> 01:02:19.010 You were just taught these birds. 01:02:22.530 --> 01:02:24.720 So that's pretty amazing. 01:02:24.720 --> 01:02:30.090 In another recent study, they had subjects do a role-playing 01:02:30.090 --> 01:02:32.298 game while in the scanner. 01:02:32.298 --> 01:02:33.840 And in the role-playing game, they're 01:02:33.840 --> 01:02:36.210 interacting with virtual characters. 01:02:36.210 --> 01:02:40.440 And those virtual characters had different kinds 01:02:40.440 --> 01:02:42.930 of social power and different affiliations 01:02:42.930 --> 01:02:44.950 with other individuals. 01:02:44.950 --> 01:02:49.530 So here's another-- the social space 01:02:49.530 --> 01:02:53.070 that was invented by the experimenters, 01:02:53.070 --> 01:02:55.530 and the subjects are playing this game, 01:02:55.530 --> 01:02:57.600 interacting with other virtual individuals 01:02:57.600 --> 01:03:04.680 who vary in social dominance and their affiliation to others. 01:03:04.680 --> 01:03:08.100 And they find place cell activity 01:03:08.100 --> 01:03:13.080 that seems to echo the position of another person 01:03:13.080 --> 01:03:14.820 in that social space. 01:03:14.820 --> 01:03:16.860 I mean, that's extremely abstract. 01:03:16.860 --> 01:03:20.370 And yet again, parts of the navigation spatial system 01:03:20.370 --> 01:03:22.278 are being co-opted to do this. 01:03:22.278 --> 01:03:24.570 I'm not giving you the details on how all this is done. 01:03:24.570 --> 01:03:26.465 I'm just telling you that studies 01:03:26.465 --> 01:03:27.840 have shown that these systems are 01:03:27.840 --> 01:03:30.930 being co-opted for other uses. 01:03:30.930 --> 01:03:34.980 Here's another very charming non-spatial use-- well, 01:03:34.980 --> 01:03:37.050 sort of spatial use-- 01:03:37.050 --> 01:03:38.260 of place cells. 01:03:38.260 --> 01:03:42.300 So those bats, turns out, are extremely social organisms. 01:03:42.300 --> 01:03:44.307 They have very sophisticated social structures, 01:03:44.307 --> 01:03:46.890 and they care a lot about each other and who's related to whom 01:03:46.890 --> 01:03:49.110 and who's doing what to whom. 01:03:49.110 --> 01:03:53.490 And it turns out that there are social place cells in bats. 01:03:53.490 --> 01:03:56.220 That is, cells in this bat's brain, 01:03:56.220 --> 01:03:59.580 if I were a bat, that would be representing your location, 01:03:59.580 --> 01:04:01.170 Jack. 01:04:01.170 --> 01:04:03.930 OK, so not the usual thing where my place cells are just 01:04:03.930 --> 01:04:06.090 saying where am I. I'm watching you, 01:04:06.090 --> 01:04:09.750 and my place cells are telling me, where are you. 01:04:09.750 --> 01:04:13.260 Something social organisms care a lot about, including bats. 01:04:13.260 --> 01:04:17.790 So they have an observer bat here hanging upside down, 01:04:17.790 --> 01:04:20.820 and he's watching this bat fly over to there and back. 01:04:20.820 --> 01:04:23.370 And then in this experiment, he subsequently 01:04:23.370 --> 01:04:24.540 flies that same path. 01:04:24.540 --> 01:04:27.180 That's how we know that he's watching that path because he 01:04:27.180 --> 01:04:30.510 has to mimic the bat's path that he just observed. 01:04:30.510 --> 01:04:36.780 But while he's watching that bat fly on that path, what you see 01:04:36.780 --> 01:04:38.410 is here's a cell right here. 01:04:38.410 --> 01:04:44.250 Here is the path flown by the bat, like out and back. 01:04:44.250 --> 01:04:48.190 And this is when the bat is flying out and back himself, 01:04:48.190 --> 01:04:51.180 and this is when the other bat is flying out and back. 01:04:51.180 --> 01:04:52.540 That blurred a little bit. 01:04:52.540 --> 01:04:56.170 Here's the place field for self and the place field for other. 01:04:56.170 --> 01:04:57.150 They're not the same. 01:04:57.150 --> 01:04:59.850 A given cell doesn't represent the same location 01:04:59.850 --> 01:05:01.230 when it's me who's there and when 01:05:01.230 --> 01:05:05.400 it's the person or bat I'm watching who's there. 01:05:05.400 --> 01:05:07.290 But they're place fields in both cases. 01:05:10.620 --> 01:05:12.615 Social place cells. 01:05:12.615 --> 01:05:14.490 I'm going to keep going because otherwise I'm 01:05:14.490 --> 01:05:17.910 going to run out of time, but I'll hang around after. 01:05:17.910 --> 01:05:23.880 OK, so this whole system is used not just 01:05:23.880 --> 01:05:27.690 for representing social status. 01:05:27.690 --> 01:05:31.950 What kind of bird this is in this abstract bird space, 01:05:31.950 --> 01:05:35.200 but actually for making decisions, for thinking. 01:05:35.200 --> 01:05:40.613 OK, so as rats run in mazes, you can record-- 01:05:40.613 --> 01:05:42.030 we've shown this already-- you can 01:05:42.030 --> 01:05:45.090 show multiple hippocampal place cells. 01:05:45.090 --> 01:05:47.340 And can you guys imagine that if we were recording 01:05:47.340 --> 01:05:49.230 from several different hippocampal cells 01:05:49.230 --> 01:05:51.960 at the same time, we could read out those cells 01:05:51.960 --> 01:05:54.570 and make a guess about where the rat is in its location. 01:05:54.570 --> 01:05:57.660 It's just like MVPA but done across neurons. 01:05:57.660 --> 01:06:01.620 So we have a pretty good sense of where the rat is. 01:06:01.620 --> 01:06:05.640 So now we have a rat navigating around in this maze, 01:06:05.640 --> 01:06:07.650 and what I'm going to show you, the white circle 01:06:07.650 --> 01:06:09.990 is where the rat actually is. 01:06:09.990 --> 01:06:12.300 And the little color thing is telling you 01:06:12.300 --> 01:06:15.090 where the simultaneous readout from several place 01:06:15.090 --> 01:06:19.715 cells in that rat's hippocampus would predict where the rat is. 01:06:19.715 --> 01:06:21.090 Like can we tell where the rat is 01:06:21.090 --> 01:06:22.677 by looking at its place cells? 01:06:22.677 --> 01:06:24.510 OK, so right here they're in the same place. 01:06:24.510 --> 01:06:25.135 It makes sense. 01:06:25.135 --> 01:06:27.300 The rat is right there, and we're reading it out. 01:06:27.300 --> 01:06:29.530 OK, so far so good. 01:06:29.530 --> 01:06:33.120 But now what we're going to do is watch what that place cell 01:06:33.120 --> 01:06:36.840 location does as the rat moves around in his environment 01:06:36.840 --> 01:06:38.655 and makes decisions about where to go next. 01:06:42.200 --> 01:06:44.330 OK, so what we're going to see is the rat 01:06:44.330 --> 01:06:47.690 is going to come up to an intersection of the maze-- 01:06:47.690 --> 01:06:49.033 I think it's right here. 01:06:49.033 --> 01:06:51.200 And he's going to decide, am I going to go this way? 01:06:51.200 --> 01:06:52.760 Am I going to go that way? 01:06:52.760 --> 01:06:55.880 And as the rat stays there deciding which way to go 01:06:55.880 --> 01:06:59.030 and the white stays there as he sits there thinking, huh, 01:06:59.030 --> 01:07:00.420 should I do this? 01:07:00.420 --> 01:07:01.580 Should I do that? 01:07:01.580 --> 01:07:03.980 You could call it neural deliberation. 01:07:03.980 --> 01:07:09.470 We will see what his place cell activity shows you. 01:07:09.470 --> 01:07:11.720 So here we go. 01:07:11.720 --> 01:07:16.730 Rat starts there-- whoops, how do I play this here? 01:07:16.730 --> 01:07:19.370 OK, so rat is heading up there, and so are his place cells. 01:07:19.370 --> 01:07:21.290 He comes up to the intersection. 01:07:21.290 --> 01:07:24.380 He stays in one place, but look what his place cells are doing. 01:07:24.380 --> 01:07:26.570 Should I go over there? 01:07:26.570 --> 01:07:28.400 I'm just interpreting what this means, 01:07:28.400 --> 01:07:32.150 but it sure looks like neural deliberation to me. 01:07:32.150 --> 01:07:33.380 And that's what he decided. 01:07:33.380 --> 01:07:36.350 Everybody get what we just saw? 01:07:36.350 --> 01:07:38.300 While he's standing there, he's in one place, 01:07:38.300 --> 01:07:40.670 but he's clearly deciding where to go next. 01:07:40.670 --> 01:07:43.220 And while he's deciding, those place cells 01:07:43.220 --> 01:07:46.430 are essentially apparently running simulations 01:07:46.430 --> 01:07:49.920 of where he might go next. 01:07:49.920 --> 01:07:52.680 OK, so we started with this big long list of things 01:07:52.680 --> 01:07:55.260 you need to know to navigate around in the world. 01:07:55.260 --> 01:07:58.440 And the neural basis of all this is really not understood yet, 01:07:58.440 --> 01:08:01.800 but I've shown you what I think are a bunch of tantalizing 01:08:01.800 --> 01:08:03.360 snippets. 01:08:03.360 --> 01:08:06.180 Which lead to the idea that our best current guess 01:08:06.180 --> 01:08:08.770 about the neural locus of these things, 01:08:08.770 --> 01:08:11.250 which is very far from the actual understanding of how 01:08:11.250 --> 01:08:14.220 they work, is that the perception 01:08:14.220 --> 01:08:15.960 of the layout of space around us, 01:08:15.960 --> 01:08:20.250 the PPA and the occipital place area are very involved in that. 01:08:20.250 --> 01:08:24.160 Also in saying for an unfamiliar place what kind of place 01:08:24.160 --> 01:08:24.660 is this. 01:08:24.660 --> 01:08:26.785 I didn't show you those data, but you can, in fact, 01:08:26.785 --> 01:08:29.340 decode whether you're looking at a scene or a beach 01:08:29.340 --> 01:08:33.180 by looking at the pattern of response and the PPA. 01:08:33.180 --> 01:08:36.210 We talked about the idea that the retrosplenial cortex 01:08:36.210 --> 01:08:38.819 may be involved in recognizing familiar locations-- that's 01:08:38.819 --> 01:08:41.609 a bit of a question mark. 01:08:41.609 --> 01:08:44.010 That the idea that your map of the world 01:08:44.010 --> 01:08:45.720 is represented in your hippocampus 01:08:45.720 --> 01:08:47.670 by way of place cells, which also say 01:08:47.670 --> 01:08:49.710 where you are in that world. 01:08:49.710 --> 01:08:51.417 That your heading direction in humans-- 01:08:51.417 --> 01:08:53.250 I didn't give you all the evidence for this, 01:08:53.250 --> 01:08:55.167 but in humans, there's quite a bit of evidence 01:08:55.167 --> 01:08:56.935 that retrosplenial cortex is very involved 01:08:56.935 --> 01:08:57.810 in heading direction. 01:08:57.810 --> 01:08:59.727 I guess I did give you evidence-- patients who 01:08:59.727 --> 01:09:03.420 have had damage there and can recognize places but not know 01:09:03.420 --> 01:09:05.310 how they're oriented there. 01:09:05.310 --> 01:09:08.580 That planning routes around boundaries in your environment 01:09:08.580 --> 01:09:10.470 involves the occipital place area 01:09:10.470 --> 01:09:12.720 and the parahippocampal place area 01:09:12.720 --> 01:09:15.060 and that this business of reorientation 01:09:15.060 --> 01:09:18.479 seems to particularly involve heading 01:09:18.479 --> 01:09:23.220 direction cells in humans, most likely in retrosplenial cortex. 01:09:23.220 --> 01:09:24.899 So you don't need to memorize all. 01:09:24.899 --> 01:09:27.149 I mean, I don't care that much about the locations. 01:09:27.149 --> 01:09:28.566 What I want you guys to understand 01:09:28.566 --> 01:09:31.067 is what are these problems that are involved in navigation 01:09:31.067 --> 01:09:32.609 and what kinds of things can we learn 01:09:32.609 --> 01:09:35.178 with different kinds of behavioral and neural measures. 01:09:35.178 --> 01:09:37.470 And you may have noticed in the last couple of lectures 01:09:37.470 --> 01:09:39.540 that I presented lots of behavioral data, 01:09:39.540 --> 01:09:41.939 because actually so far, the richest 01:09:41.939 --> 01:09:44.790 insights about how the system actually works still come-- 01:09:44.790 --> 01:09:48.240 or many of the rich ones come from behavioral data. 01:09:48.240 --> 01:09:50.620 OK, quiz is in two minutes. 01:09:50.620 --> 01:09:53.114 Does anybody want to ask me a question before the quiz? 01:09:57.530 --> 01:09:58.107 Yeah? 01:09:58.107 --> 01:09:59.940 AUDIENCE: Do you know that the [INAUDIBLE]?? 01:10:03.170 --> 01:10:05.870 NANCY KANWISHER: Yeah, so you have to do lots of controls 01:10:05.870 --> 01:10:07.190 to work that out. 01:10:07.190 --> 01:10:10.773 And I didn't show you any of the details of the data. 01:10:10.773 --> 01:10:12.440 But yeah, these guys are pretty careful, 01:10:12.440 --> 01:10:15.170 and there are all these things-- there are many different ways 01:10:15.170 --> 01:10:21.080 in which people are watching hippocampal neurons 01:10:21.080 --> 01:10:24.770 and decoding trajectories from hippocampal neurons. 01:10:24.770 --> 01:10:27.020 You may have heard about replay, which is 01:10:27.020 --> 01:10:28.780 a big thing in this department. 01:10:28.780 --> 01:10:30.770 Tonegawa and Wilson labs study this, 01:10:30.770 --> 01:10:35.450 where you have a rodent moving around in one trajectory 01:10:35.450 --> 01:10:36.950 during the day, and then you record 01:10:36.950 --> 01:10:38.450 from those neurons at night, and you 01:10:38.450 --> 01:10:41.270 see replay of the trajectories that the rodent went through 01:10:41.270 --> 01:10:42.540 in the previous day. 01:10:42.540 --> 01:10:44.962 And so there you have to be really careful to say, OK, 01:10:44.962 --> 01:10:46.670 there's a lot of data and a lot of noise, 01:10:46.670 --> 01:10:48.295 and is this really more than the noise? 01:10:48.295 --> 01:10:52.050 And it is, but it takes a lot of statistical work to show that. 01:10:52.050 --> 01:10:52.880 Yeah? 01:10:52.880 --> 01:10:57.930 AUDIENCE: [INAUDIBLE] scenario be like a place that I know. 01:10:57.930 --> 01:11:01.080 With the neurons, the same neurons 01:11:01.080 --> 01:11:03.930 would fire if I go back to that environment, I assume, 01:11:03.930 --> 01:11:05.100 or like the place? 01:11:05.100 --> 01:11:08.343 NANCY KANWISHER: Yeah, yep. 01:11:08.343 --> 01:11:10.260 AUDIENCE: But then you can't really have that, 01:11:10.260 --> 01:11:12.750 or you can't reserve neurons for those [INAUDIBLE].. 01:11:12.750 --> 01:11:13.470 NANCY KANWISHER: It's a good question. 01:11:13.470 --> 01:11:14.910 How do we have enough neurons? 01:11:14.910 --> 01:11:17.430 Yeah, especially for some place we go every six months. 01:11:17.430 --> 01:11:19.763 Are they sitting around waiting for us to go back there? 01:11:19.763 --> 01:11:22.500 No, there's some recycling of neurons across very different 01:11:22.500 --> 01:11:23.440 locations. 01:11:23.440 --> 01:11:25.890 So within that location, they'll be consistent. 01:11:25.890 --> 01:11:27.390 But yes, you do recycle. 01:11:27.390 --> 01:11:30.300 So the same neuron will have one place cell in this environment, 01:11:30.300 --> 01:11:31.980 and it may or may not have a place 01:11:31.980 --> 01:11:33.330 cell in another environment. 01:11:33.330 --> 01:11:35.270 That's a good question.