is Matt Wilson he's a professor at the department of brain and cognitive Sciences at MIT Matt studies how animals learn and how they remember what they learn and today's talk is a hippocampal memory cognition and the role of sleep it's worse well good to be back here for the uh for the the summer session here what's whole um so I always like to you know give a talk from the perspective of I'm an engineer trained as an engineer uh from the perspective of an engineer who's trying to understand wha
t the brain can tell us about the fundamental nature of computation underlying human intelligence uh you know there's going to be an Ethics talk later we're talking about you know AI obviously there's a lot of interest in uh sort of you know Big Data driven you know approaches that can Transformer models uh that can kind of emulate intelligence and so there is a question you know what what does the brain have left to tell us if we can you know if we can emulate intelligence um what is it the bra
in has to tell us and I guess Ed Boyd's gonna be giving a talk later and we'll be talking about you know new technologies that give us even greater power to interrogate and understand the nature of underlying brain activity but you know I like to use the the model system that I'll talk about the the hippocampus to try to give some uh both mechanistic but also uh kind of a higher level insight to think about what is it the brain does that um the kind of challenges the ideas regarding intelligence
and I'll just I'll just give you the you know sort of the little spoiler you know we sort of think about we sort of think about intelligence as uh being fundamentally about creating uh rational systems that can take facts that can make you know appropriate judgments that you know make good decisions uh you know based on all the facts that are you know kind of smart and we sort of think of human intelligence as you know being the sort of the biological manifestation of precisely such a computati
on but I think when you really look at human intelligence and you look at the challenges that systems have you realize that the the the approaches that are used to achieve this sort of large-scale massive rational intelligence are not really available to biological organ isms live in very narrow in like small niches they have very limited access to data uh they you cannot rely on sort of repeating the past because the future is you know these niches change uh you have to produce adaptive behavio
r in novel context and that is that you have to do more with less you have very little data and yet you have to make decisions that reflect some broader understanding of the World At Large which means you have to make models very quickly you have to make and draw inferences from the limited data that you have and so and you have to do this you know quickly and efficiently and so the idea is that that the systems the biology has evolved and then systems to do exactly that now the consequence of t
hat you know making big decisions building models on very little data does not actually lend itself to this you know sort of uh uh a a rational grounded system rather it leads to systems which are heterogeneous and irrational and that is that you have many models individuals construct their own individual models based upon their Limited interrogation of the world and biological intelligence succeeds not because the individuals are so smart but largely because the populations as a whole are able
to uh to solve the problem collectively and so we have to think of biology and that is really smart individuals but as individuals that are part of a population that as a whole are able to do the best job they have they can with the limited data that they have so that will be that that will be the challenge how does biology kind of extract uh and uh data from limited uh experience in the world and then build models draw inferences based on that and so the the my talk will be in two parts uh the
first part I sort of think of as the uh interrogating the world as the interaction with the world this is this is sort of the weak sleep model and the waking state animals interacting with the world and looking at how the brain represents processes information during act this this active State and then there's the offline status I like to refer to it that would include sleep but also periods of quiet wakefulness or inattentiveness where we see the brain switching from a mode uh rather than takin
g information from the outside world involved in internal processing so there are these sort of two functions take in data and then build model and do inference and then these two states the Wake state sleep and quiet wakeful State and we'll look at the uh the sort of the the biology the nature of brain activity in these two conditions and then try to you know extrapolate their computational function and relevance to this larger this larger problem of you know building models with very little da
ta so the system that I'm going to be talking about here this is uh this is a picture of a rat brain the overlying neocortex removed and here you see the hippocampus which lies in the medial temporal lobe the hippocampus sort of this curve structure the temporal lobes here on the side so if you sort of think of the hippocampus you just go into the middle of your brain you've got these two curve structures and in cross section if you take a slice through the hippocampus the hippocampus is compose
d of this very fundamental circuit it's an older kind of Cortex and it has a simpler kind of structure so this arche cortical structure a three layered structure it's in which you have a layer of cell body so here for instance if you take these uh the the kind of the waypoints in the circuit that starts from the adjacent cortex the enterinal cortex the fibers from the internal cortex come into the hippocampus and go through the so-called tri-synaptic Loop they make three hops through the three t
hese three primary sub-regions or subfields of the hippocampus the dentate gyrus and then cells and then take gyrus synapse cells in areas ca3 cornubomanus three from ca3 to ca1 ca1 to the subiculum and out so it's the if you if you want to count the three synapses there's one two three and then out ca1 has a direct output but the sebiculum also has is a primary output structure you might notice that uh I don't know what's you know what's up with these biologists they uh you know they didn't tak
e like you know sort of uh first grade the arithmetic they go from one to three what happened to ca2 ca2 is indeed a structure in fact there's also ca4 these are sort of these kind of intermediate in the rodent the smaller less you know less obvious less prominent structures uh one interesting thing about ca2 is ca2 is similar to ca3 in terms of its architecture it uh it's it's a little bit different in that it doesn't actually it it doesn't enjoy the kind of direct connectivity with the Dente g
yrus but it does have one of the sort of the kind of the signature Hallmark properties of the ca3 region which is strong recurrent connectivity a lot of it's like a very dense associative Network the other interesting thing about ca2 uh which doesn't lend itself to a rodent talk but would lend itself to a human talk ca2 is the one subfield of the hippocampus that has evolutionarily uh expanded the most so in humans ca2 is is enormous so you have a massive ca2 and rodents you have a very small ca
2 so you can kind of ask the question what is it that you know the hippocampus itself and the overall architecture has not really changed evolutionarily it's just the sum of the circuits have changed and so you kind of think what is it that ca2 does that uh that might be particular to the kind of conditions information or processing that are unique to humans and uh so we can kind of we can come back to that but that's because I won't have any CO2 data here so we're just going to have to focus on
you know on what we have and that is so this circuit ca3 to c o n s so ca1 is interesting in that if you look at the cells the neurons in ca3 and ca1 they're they're almost identical they look very similar the individual cells the real difference is the connectivity ca3 has strong recurrent connectivity ca1 has almost known so ca1 it's like a feed forward Network so you have ca3 is this classical recurrent associative Network ca1 as a feed forward Network and in the early 1970s John O'Keefe who
was a behavioral electrophysiologist placed electrodes into ca1 recording the activity and of course you're probably familiar with the fact that you see one the Nobel Prize for this discovery he discovered that individual neurons when you take animals and you allow them to explore space these individual neurons Express spatial receptive Fields as they fired at certain locations you know the the story I always like to sort of tell the story as both kind of a sort of the history of science but al
so a way of thinking about an approaching science so people have been studying the hippocampus for you know decades prior to uh O'Keefe's electrophysiological studies Behavior electrophysiological studies and you know in large part because of its interest in in the field of memory and human memory of course you're also familiar with the case the seminal case the patient hm we wonder what is surgical procedure involving the bilateral resection of the medial temporal lobes cut out of this Parts po
rtions of uh went in kind of cut out portions of the hippocampus and the adjacentronic cortex that treat an intractable epilepsy and following that lost the ability to form any new episodic or autobiographical memories so here was you know a part of the brain that was specifically and perhaps uniquely involved in the formation of autobiographical memory which was contrary to a lot of theories of brain function up to that point which sort of positive the brain is kind of you know like the hologra
phic Theory theories of mass action the idea that the brain is a distributed memory system that there's no one and there's no specialization it's just the brain does processing uh probably most sort of famous famously advocated by Carl Ashley who did you know these sort of very important experiments in which he went in and just kind of scooped out different Area locations in the brain and different amounts of brain and then looked at the impact of behavior and his conclusion was you know the mor
e brain you scoop out the more the greater the impact of behavior and it doesn't make any difference what part of the brain so you scoop out any parts of the brain it's just it's it's the it's the quantity of brain that matters so this was really uh very contrary to that notion and Drew a lot of attention the hippocampus there was a lot of work electrophysiological work people using rodent models putting in electrodes a lot of these models use classical learning you know Theory approaches classi
cal conditioning uh but all of them had one thing in common for you know experimental necessity and expediency they involved head fixing the animal and why do you have to do that if you put electrodes put little tiny Electro electrodes that are small enough to record from Individual neurons than you know the sort of the tolerances in that recording are very uh you know very small so you have to be within tens of microns of cells you can't really tolerate a lot of movement instability so everythi
ng needs to be fixed or secure so all the experiments at that point had been done with head fixed animals uh now what O'Keefe did was say you know there's a kind of a behavioral I mean a sort of a behavioral psychologist but also electrophysiologist to really understand what the brain does you know in natural conditions you really really need to study it in Freely behaving animals so he devised the methods that allowed for fine wire electrodes to be placed in the animal it you know with that wit
h a tether allowing the animal to move around and that's sort of the simple that sort of simple step of allowing animals to do what they naturally do revealed this fundamental property of hippocampal coding and they really sort of changed the overall view of the hippocampus uh you know in part because it revealed this novel correlate the spatial correlate but it also fit with a larger sort of uh kind of theory or approach to the study of uh you know cognition and intelligence and that is the thi
s is the contrast between the so-called the behaviorists theory of you know of learning and cognition and that is the uh the idea that in all learning is is just kind of a is based on simple principles of associative learning stimulus response for war and so that is if you just like pair of stimulus with the response and you reward it you reinforce that it's kind of like you know the heavy and the sort of you know the heavy and synapse is a is a you know a synaptic version of that rule it's like
reinforcement learning and you know the argument was well that's how we learn everything so everything is learned it's all Acquired and it's all tied to you know to associative reinforced associative learning but there was another school of thought that said that that posited that that actually learning is you know we don't learn everything that we actually know a lot before we learn and that was we start with internal models and we use our interactions with the world to both to both tests but
also modify that in those models and that is that they're sort of their internal cognitive models this is the sort of the you know the cognitive approach uh and the demonstrations of that one of the you know sort of the classical demonstration of that was the phenomenon known as latent learning latent learning was uh you know sort of classic behavioral Theory you take an animal you put in a box you know you you know your flashlight you give it some food to you know to go to a certain location to
learn to repeat that and so animals learn what they experience uh in latent learning if you take two animals you put one in a box and you just let it wander around you you know it doesn't do anything the other animal you leave in the cage and then the next day you come out and you train both of them so now they're going to learn a task but in one case the analysts had passed a latent exposure to the environment the other case it is not and it turns out the animals that had latent exposure to th
e environment learn better so they learned something by just exploring the space there was no reward they weren't you know they weren't learning any particular Behavior and so this idea that there was latent learning building based on a cognitive model and then O'Keefe's discovery of place cells and this fit into this notion that the hippocampus yes question is it was the question what do you mean by latent learning exactly no external reward so it's essentially Learning Without reward that was
the you know that was sort of the challenge to the behaviors Theory and so the idea is look I'm wandering around here doing things I'm not learning anything until I get reward if I get rewarded for something then I increase the frequency of that behavior and behaviors that don't get rewarded I are decreased in frequency and so this is the idea that's how we learn we just do things we're like cheap that you know sort of you know follow the you know the the brightest light the you know the sweetes
t you know uh you know reward that that's how that's how behavior is driven as opposed to whatever we do you know we're constantly learning we're learning by uh sort of again testing and forming and refining internal models but what are those internal models and one of those models that O'Keefe uh in a you know along with uh with Lynn nadell in a very influential book the campus as a cognitive map the idea is that that we learn on a foundation of spatial representation or maps and that is we hav
e maps of the world and then we learn things on top of that but the map essentially is you know it's there we can refine it we can kind of you know kind of tweak it uh but the the substrates for formation of maps is there and here's comes the hippocampus there are these recordings you have these cells that look like they you know encode spatial locations now I always like to point out you know encoding is like gets tossed around it's like a sort of a little computational buzzword oh you know neu
ral codes for this codes for that but you know codes are something and we can also talk about representations codes have a very particular you know meaning and requirement and that is sort of codes are directed transformations that are intended to communicate information to a receiver and that is that a code is only meaningful if it gets decoded by someone simply transforming information does not mean it's a code and so the challenge for the hippocampus is both to point out a is there a spatial
representation is there transformation the spatial domain well you think by the way oh look I see like Place cells isn't that the spit you know there's that's that's like a it looks like a transformation well it could be if the spatial information wasn't already there in the input if it's already there in the input it's not being transformed uh and if it's not being transformed then you have to ask the question oh you know what's the you know what what would be the coding imperative if you alrea
dy had the information input so uh to understand the nature of the capital code you kind of want to know what's going on on the input the anteriorinal cortex and so this is where the companion prize the Nobel Prize that John O'Keefe received for the discovery of play cells was given to Edvard and migrant Moser who discovered for the discovery of so-called grid cells in the enteronal cortex and that is a representation of what appeared to be a largely kind of euclidean even like cartesian-like ri
d-like regular grid-like structure in which sort of metric distance information was represented and then these grids were conveyed in the hippocampus and so the hippocampus as we'll see forming representations of places in space and so we can again think about the difference between a location in space a spatial representation like a map and then a place representation in the hippocampus now those two things are obviously going to be related we think about you know the this room as enjoying both
a a place representation and a spatial representation you think about the spatial representation it's like the geometry the coordinate system you know where is this located on if I took I pulled out a map of MBL where is it located but the place representation so place is you can think of this as being the combination of location and context now what is context context is sort of this nebulous this sort of nebulous term it's the thing that makes this the experiences that occur in a particular s
pace uh unique and memorable and so you've been in this location you've been sitting probably as I always like to point out whenever I you know teach classes on this uh like in the first class the one thing I tell the students you know one thing I'm pretty sure is going to happen the next time we have this class is that everybody's going to come back and they're going to sit in exactly the same place you probably come back and you sit in the same location if there's sort of a tendency to it's ju
st you know I don't know you just feel more comfortable well and yes question so so this is if you change your plate you know the location so let's say you come in and you move to another location in this room the the cells that carry that information I'll say encode that information will be different and that is that by looking at the hippocampus I'll be able to tell that you actually move to a different location great question exactly so there was a lot of work in sort of you know after the di
scovery of place cells it was a lot of work trying to understand exactly what you know what control place sells the obvious thing would be well it's like the cues it's you know you come in you see certain things it's the configuration of cues and there's a distinction between uh referred to as local or proximal and distal cues distal cues would be things like you know off in the distance if you're out you can think of buildings landmarks proximal cues would be exactly this things like you know c
hairs the you know the design on the you know on the flooring and manipulations of those two cues demonstrated that they both have like different impact on this like on the spatial coating I don't want to go into too much detail on that but uh you can think of there being sort of two properties of a map one is like the actual layout the you know sort of the geometry the metrics then the other of course is like the orientation and as you have to you know you have to orient a map in order for it t
o be useful and while proximal cues had a lot of impact on you know exactly what the distribution of cells would be and you could like by shifting cues you could kind of distort the distribution of of the fields the distal cues would have great impact on the orientation of the mountain that is you used it to align the so-called directional system so there are cells that carry information Compass head directional information and those are combined with this spatial geometry to give you the layout
but the larger questions of this notion of place so what does it mean place what is context and you know some interesting experiments so you can take the same cells bring the animal bring you know you come in you know today come back tomorrow morning record the cells and large variety of body of work uh you know pointed to the stability of these Place cells same cells fire same location the map is preserved with repeated exposure but there are other things that could change or so or so-called r
emap those cells you come back into the same space but I change some condition now what would that condition be that I changed so for instance I could change the time of day let's say you come in here in the morning you have lectures and then you come in in the evening same space but you know sort of different talks maybe it's different topics maybe it's like another course and what you find is that that is sufficient to induce remapping a different set of cells and so you can ask the question w
hat you know how can you have a different place in the same space time of day that can be a that can be a you know a sort of a determining Factor it could be the content let's say you have uh you know instead of me up here talking you know you're gonna have Ed Boyden up here he's going to be talking and the question would be are you going to have the same cells or the same hippocampal place cells going to be active in the same way for a slight difference slight change in the queue that is just t
he one you know just one person everything else the same one person and so this idea that you need to that you need to have you need these Maps that are associated with context but context is not something that you can you know you can tie to any one property it's not like oh it's like if you know the color changes then we change the context it's anything that might be relevant that might require some kind of Novel encoding if I if I have new information I need a new you know I need a new map an
d so that we can call context and so this circuit and there's a lot of work a lot of interest in developing computational models that that try to confer the properties of these circuits or these these computational functions onto the properties of these circuits and so this idea of making very small changes very small changes of the input leading to very large changes in the output as like a different map for like you know morning and evening one speaker versus the other uh and the idea is that
was like separating patterns small change input large stages in the output and that was a function that was ascribed to the dentate gyrus now on the other hand as you point out okay you come in here you know sometimes maybe some people are different sometimes you have a coffee cup there are lots of changes many changes to uh you know in the context that are not relevant that don't indicate that the end of the you know a sort of new representation needs to be formed so in that case you need to be
resistant to changes large changes in the input leading to no changes in the small change output and that's the so-called pattern completion and so this pattern separation function I you know and maybe after we can sort of talk about what the you know the properties of these circus dentate gyrus that that were argued to confer this property and then ca3 largely the recurrent associative connectivity getting this pattern completion-like function and so that was like go say Okay pattern separatio
n pattern completes this is like getting the context right so it's like okay I got a set of it's like you know it's like pulling out the right notebook if you're taking notes it's like you know I gotta if I'm taking notes I'm gonna have the right notebook to write my stuff down right I don't want to write pull out the notebook for you know the evening lecture and start taking notes that everything gets scrambled up so it's a way of establishing these uh sort of buckets for encoding information n
ow while you're in that context that's just the starting point so you kind of think about that oh like that's you know that was the you know that was the big problem it's storage and retrieval pattern encoding representation storage and retrieval a lot of work in associative memories focus on exactly that how do you use these circuits to encode and retrieve information they talked about capacity Fidelity all these things but when you really think about it what is the information that you're putt
ing into these notebooks that's just kind of ensure that you've got the right context you pulled out the right note but what you actually put into those notebooks are it's the time series of events the experiences that occur within that context and so you have to have these two functions available to you the ability to have a stable notebook a context or representation a framework on which you can code information but then you have to have the ability to encode time sequence information episode
called episodic information uh in a in instance unique way and that is like one trial learning episodic memory is you and you know you're able to encode One-Shot memories things that occur exactly wants and so we'll see how the hippocampus you know tries to achieve that spatial representations contextual dependence on you know this a place and then the ability to encode time sequence information using the Dynamics of the campus yeah follow-up question or no yeah great question right so this ques
tion you know what you know you know what triggers remapping a lot of people have studied this I just had a you know post-doc who sort of you know recently published you know paper on this uh you know sort of taking a Bayesian inference framework and that is that you know that you're trying to do inference that you don't you never really you don't know right is this a different context is it not is this change important is it not and so you know as you accumulate you know evidence that in fact t
he information that you're that you're getting actually forms more than one it doesn't fit well into one distribution so you need to have like you know multiple distributions and so at that point when you decide and you know I need a new map that's the the trigger and there's no one you know there's uh there's no one rule but it it the idea is that it depends on your experience the more experience you have the more evidence that you know it really does seem like you know I come here in the morni
ng I come here in the evening and you know the the the talks they seem different they really don't seem I can't really lump them together and so as you gain more experience you decide actually these are like two you know I thought these were just you know one lecture series actually two lecture series I'm going to split those in so at that point you actually get the splits so this is what you find that that when you you know if you add put an animal into an environment with uncertainty as that u
ncertainty is resolved there can be triggers that will lead to precisely that this sort of precipitous you know a some kind of sudden remapping uh so there's also um you can you describe you know mapping there's like a sort of a total remapping so-called global remapping global remapping is like complete reconfiguration of the place cells and the spatial representation uh then this is what's referred to as rate remapping and that is that the cells fire in the same location but the firing rates v
ary and so you can kind of think about this as like you know sort of in a vector space Global remapping is like completely changing the you know the angle of the you know that Vector rate remapping would be sort of changing the length or you know slightly the you know the angle in you know you know in one space uh and then the other is partial remapping which is sort of what you're referring to and that is you don't have to change everything or in the case of rate remapping you don't really chan
ge you know you don't change anything you just sort of change the uh a light kind of scalar properties of the vector partial remapping you can think of decomposing this larger Vector space into smaller subspaces and so you can change some parts but not other parts and what this what this suggests is that the overall you know representation is not like one monolithic representation one map you can think of as being composed of smaller sub Maps this this compositional notion and so the question is
there a logic to the way in which you might want to break down a map how do you break it down you know if you're in a conventional map I mean you break it down and it's like a geometric or kind of spatial breakdown right I have maps of adjacent locations and largely the breakdown of those you know you know how I uh you know decompose that is based on the uh the relative Independence of the information within each of those maps and so they're going to be things that are going to be you know that
are going to be sort of sufficiently described within that individual map and then they're going to be transitions or interactions between these Maps which could be interchangeable so maybe I go from this room to you know there might be you know five other rooms and so uh you know I might want to be able to kind of Swap and swap parts of the map in and out and so this idea there can be generalization this is one way of thinking about a generalization is sort of taking the rules and representati
ons from one experience and then translating them into another the idea that there's some equivalence even though the context seems to change seems to have changed there's still stuff that I can use from prior experience it's all these ideas the ideas of remapping global rate remapping partial remapping the idea of potentially compositional representations that might reflect the use of these you know cognitive maps to serve a larger function and the larger function that I would argue the system
is trying to in the hippocampus contributed to is essentially kind of building building a generalizable model of the world that's what animals are trying to do limited data I need to build a model of the world once I have a model then I can use that model you know in like an in a generative uh uh you know fashion to both predict and direct Behavior even though things change that is it's going to be you know the model is going to work even if there are uh uh you know minor changes to the context
and so that's what animals you could say are trying to do and then the question is how are they actually doing that how's that reflected in the uh in the activity of the hippocampus is there any other questions if that's sort of the big that's sort of the top down view so we had definitely spatial navigational deficits uh uh you know sort of part of the uncertainty as to you know what that told us about the hippocampus is that the the lesions also included the adjacent in torontal cortex so the
anteronal cortex you can think of as if it was critical for providing the spatial information at the campus that could make it difficult to form Maps but there was it was a lot of work you know a lot of arguments about that sort of the necessity of the hippocampus in navigation because hm highly intelligent was able to function you care in a conversation so you know all those other cognitive capacities seem to be intact it was simply the loss of the ability to you know form uh kind of short-term
anterior grade new memories and then there was also this loss and limited loss in retrograde memory and that is sort of for relatively recent right well in the case of hm and in humans it could be years past but couldn't remember what he did you know yesterday or last week but if you asked questions about did he remember things about his childhood he could relate you know sort of tell you things that he remembered from his childhood so this this sort of gave rise to this idea it's like this sor
t of you know dual memory system the idea that we have uh you know system performing short-term memories and then those shorter term memories get Consolidated into longer term memories and that the locus of those two memories is different hippocampus is the site where you form you know short-term or immediate memories and then those get translated and transferred into other parts of the brain let's say into the neocortex so there was this idea that you have you know memories being you know slowl
y gradually shifted to other uh you know parts of the brain and this was one of the models for sleep sleep was okay sleep is the time that you transfer memories from the hippocampus to neocortex and we can talk about that yeah yeah I always like to sort of point that out I think it's really I think it's really sort of an important uh uh you know observation and insight and that is the damage to systems that are involved in forming memories of the past also impact our ability to imagine the futur
e and so it's this idea that you know Imagining the future involves a generative process that relies on this memory system so what is it about the hippocampus that would tie like past and future and so that's kind of the segue into the sort of the talk about the electrophysia what actually goes on the hippocampus and what we'll see is that while that initial observation of place sells and you know spatial representation hippocampus was correct it was a follow-up observation also made by John O'K
eefe in the early 1990s that involved the encoding of temporal order information and so what he observed was that place cells didn't just fire when the animals had a location that the timing of cells with respect to an internal oscillation the Theta oscillation carried temporal order information and so when you look at when you decode the full population what you see is the hippocampus is actually it you know you know every you know 100 milliseconds or so 10 times a seconds it's repeatedly repre
senting and reflecting uh your experience going from the immediate past to the immediate future and that is a sort of tying past present and future together and so you're doing this constantly you're you know you're remembering what I just did you know tying that to where I am now and then you can say imagining or you know projecting predicting what the future state is going to be so the hippocampus by its nature is constantly tying together in this temporal domain past present and future now th
e question okay how do you take that limited very limited and that you know the time scale of that you know kind of past present future uh extrapolation is on the order of a few seconds of you could say like the kind of cognitive behavioral time you're imagining well you know you're remembering where I was a few seconds ago imagine where I'll be a few seconds from now that's not an absolute time frame but it's you know it's still it's relatively fixed so thinking about well how would you go beyo
nd that and so that's where we can kind of bring in some of these other mechanisms that suggest well the hippocamp is able to in fact take you know information about time-ordered events and then use them to both you know sort of recall things in the past but also project you know forward into the future using some of the properties that I'll I'll point out one is this encoding uh kind of time sequence encoding using the Dynamics of brain oscillations and then the use of these different brain Sta
tes in particular the quiet wakeful state to kind of Express these discrete State transitions that uh would have the ability to kind of transcend the real-time constraints of just a few seconds so you could like imagine link together events I could think oh I'm here I could you know get to my car and then I could get to Boston so from here I can imagine getting from here to Boston like three steps and so that would be the kind of you know sort of temporal you know order linkage that would be pos
sible given the properties of the hippocampus uh so you know this happens pretty frequently I think actually let's say I got the slide one sometimes more than I you know I get to to kind of get just to the intro to kind of introduce the basic topic and we'll take a break you know at 10 30 so but the first half of the the talk really was to try to fill in a lot of the details of what I just went through right and so this idea of the hippocampus involved in memory and then the observation of spati
al encoding the link between spatial memory and you know episodic memory it's not an obvious link if you think about why would you know damage that became why would impact these two and it does impact this even in hm and in humans it was definitely a spatial navigational deficit you know show them how to get to the bathroom you come back 10 minutes later he will not be able to get to the bathroom right you just can't you know would not be able to do that simple sort of function uh and then the a
rgument is that they sort of one critical you know feature of these two forms of memory is that they involve encoding of of sort of temporal order temporal sequence information you have to remember events in the order in which they occurred in one you know and you know that's sort of fundamentally important you think about it generally because sort of time order from time order and this is my sort of you know you know what why and how model of you know uh sort of cognitive function the three-ste
p so this three-step program to intelligence the first thing you have to do just like the encoding step like what happened in your you know evaluation of the world you have to take in information you have to know what actually happened uh but you also then from that you know that experience you have to try to infer why things happen as you have to infer causality and inferring causality requires that you get the order right and once you get the order right once you've once you've identified the
causal agents what are the critical things a actually causes B I mean there's a lot of stuff out here most of it is irrelevant some things are relevant because you know they have predictive power now I can you know with this refined model I can invert that and use this in a generative way to determine how I want to see so now if I specify an objective I can use the model to figure out how I would achieve that objective so there's the what first and that's where the encoding comes in so first thi
ng I have to do is I've taken information and I have to preserve temporal order otherwise I can't do the next step the causal inference step right uh and so the approach is going to be simple we're going to take you know this is O'Keefe you know uh you know used a variant of this putting in little fine wire microelectrodes into the brain the shafts are insulated the tips are exposed allowing you to pick up electric Fields generated by nearby neurons and just sort of given the size and the geomet
ry of these electrodes so-called tetrodes because they have four channels you're able to pick up the electrical discharges of spiking activity from very small dipoles that are generated here for small dipoles you have to be like really close to cells so you can get really close to cells what this means is you can see the discharges the electrical discharges from a bunch of cells and the proximity of this the Tetro geometry means you can resolve the spatial location which means you can pull out i
ndividual sources distributed around this so you can get lots of cells from a single 4-chain electrode you put in a bunch of those electrodes into a little array you put the array on the animal's head you stick this you know you know you put the you know the tetros down into the hippocampus and into one of the properties of the campus I mentioned this is this older simpler architecture archet cortical structure where you have a primary set one primary cell layer where all the cells are packed in
so that's great it means it looks like hey all the cells are there so if I put the electrodes down where all the cells are I'm going to get good recordings other parts of the brain cells are a little bit more distributed so it's harder to you know take advantage of that because again the electrodes are really just going to pick up stuff that's near the electrode so if you you know the the density and distribution of cells differs it's not going to be as effective also the nature of electrophysi
ology requires that sort of dipoles be oriented in a particular way which means if you have good layered structures it's easy to it's easier to pick up to position the electrodes to pick up the spiking from lots of cells the cells are oriented in different ways like in some nuclear structures it can be harder so in a sense the hippocampus is the ideal structure to get lots of cells and so when you do this this is just an illustration of actual data each point here is a is an action potential whe
re it's just where the axes here are the amplitude the peak amplitude of a spike here it's four channels so it's four dimensions you're just seeing two of those Dimensions so you pick two channels and you know the nature of the electric Fields produced by these things is that you know bigger Fields closer the cell is bigger the the field is so stronger the field and so when you see for instance here A bunch of points which means a bunch of spikes that have a large amplitude on channel one and a
small amplitude on channel two that means the cell is like close to channel one and far from channel two and so different cells are going to have different amplitude profiles and of course this is in four dimensions so if you do this you know you pull out a dozen a few dozen cells from each electrode you put in a dozen 20 you know 30 electrodes you can get a few hundred cells and so you know back in the day and Ed this course is going to be talking about new technologies and a lot of people in c
leaning our Labs you know using new technologies for doing large-scale Imaging which is great I mean this really is the future uh but the one thing that the electrophysiology and use of wires upwards is that it's actually looking at one thing it's looking at the electric field generated by electric currents that are the result of opening and closing a particular channel so you're looking specifically at electrical activity and uh you know unlike other methods where you have to Divine you know so
rt of sensors devise and divine sensors that will report different states of the cell you know and the most popular of which are these kind of secondary you know indicators of activity like calcium which are not directly reflecting you don't see actual spikes you see the consequence of that and so this Remains the most direct way of looking at the actual spiking activity until and Ed will be talking about the new generation of you know sort of voltage sensor genetically quoted voltage indicators
voltage centers that could provide the same kind of you know uh the the same signature of electrical activity and so that's certainly in the is the future but armed with just this basic What I Call Garage Neuroscience because you could literally go into your garage if you have the right wires you could make these things you can make these devices and you could do this at home because this doesn't require you know it doesn't require a cyclotron not a wire you know it doesn't require a 7t magnet
it's just you need some wires you need steady hands and um yeah and once you do that you see you can kind of see this activity and one nice thing about this approach these wires you know once you put them in because they're flexible wires they remain relatively stable over time days weeks months so you can carry out chronic long-term recordings with electrodes in animals that are moving freely just as O'Keefe had done in those seminal experiments in the early 1970s and so when you do that here t
his is just an illustration of the property of the phenomenon of place cells this is just showing each individual Point here is an action potential a spike they're color-coded sort of based on this you know color coding and so the idea is that when you see Spike amplitude distributions like this you can go in and in this case sort of manually identify regions or clusters which would uh which would sort of identify these things as units and careful not to refer to them as cells because you don't
really know their cells just says these spikes seem to have come from like a you know a location a fixed location in space right they came from a particular region the amplitude profile reflects the the you know the relative distance from those electrodes but you see there are sort of variations in that amplitude and that's because the spike amplitudes themselves can vary there was a question question no question no okay uh and so you know these could be you could say well you know units what el
se could they be if they weren't if they weren't cells they weren't actual neurons right neurons the only those are the only units there are right the little ball and you know well yes and no and so like many of this you know the neurons in the brain uh while we kind of think of action potentials and spiking activity is coming from the cell bodies or the you know the initial segments going out the axons the dendrites or the receiving structures are also capable of producing Action potentials and
so you can get spikes that are generated in the cell bodies that you don't see in the dendrites spikes that you see in the general and dendrites that you don't see here so really you know the cell itself is kind of a more complex computational unit and this you know you'll be able to see that and as you can have spikes here that are that might be generated at different points but that's not because they come from different cells they just come from different parts of the cell and that the you k
now one one way in which you would see that would be for instance here this sort of change in amplitude come could come from actual change of the amplitude of the spikes or a shift in the location that extra potential is actually generated from different locations oh this is like a change in the relative location uh and so all of that is actually going on you're seeing that you can see both dendritic and somatic Action potentials here amplitude so this is like the actual amplitude like the point
s here you just you're recording the electric Fields is just the voltage the local voltages when you when the voltage reaches the peak you pull that point off and you drop it on here so channel one this would be the voltage on you know you know one channel voltage on channel two and so it's just the relative channel one channel two that's what you're looking at here it's a different you know cells generating action potentials will produce different profiles you know big here maybe small here and
it's that profile that's giving you this clustering of points so you break down these clusters you color code them and when you do that you get something looks like this and this is something that was tested on a so-called linear maze so the animal in this case is running in two directions gets a food reward each and runs here and alternates going back and forth like this and what you see when when you constrain the animal's trajectory or path in space to follow these sort of repeated these you
know these repeated sequences uh and that is that the cells now are not simply you know reflecting the location on the maze but also the direction they acquire what's what's referred as directionality so this yellow cell for instance will fire as the animal runs through here you get this each one of these spikes this is now accumulated over like 10 minutes so this is like the average activity but in each and every pass when the animal goes through here you'll get this like and then it'll go thr
ough and then animal comes through now an animal goes back in this direction you don't get the same response now the you know so the cartoon version is it only fires in this direction and then it doesn't fire this indirect in this direction in reality you sort of get a difference there's like a different firing rate higher you know much stronger response in One Direction the other great question so this is can you get Place feels without reward that was the fundamental premise of the latent lear
ning experiments you put an animal in a box no reward and they learn something you put an animal in a box with no reward and you get Place Fields Place Fields form and so they form very rapidly within like you know just a few you know a few passes like one pass through a location is enough to lead to the expression of a place field that then over you know a period of you know sort of repeated passes seconds to minutes will get stronger but nonetheless the map itself is established like one shot
that's the idea no reward what does reward do there have been a lot of experiments recent experiments some older experiments that suggest that reward it's like a gain Factor so it can increase the density and firing rate of cells you get more cells around locations of a reward and that the firing rate of those cells is modulated so you get a stronger response at reward location you over represent reward locations but you don't need it to you know to form play sales so it points out two things on
e is there's just like this again this kind of the spatial layout but then there's the stuff that happens in space and so in this case the stuff that happens in space is like moving in a particular direction and so in the animal you know runs in this Direction versus this direction now you say that's actually like do I need a different code for the two directions because different things might happen this direction might be different might be sort of functionally or behaviorally you know distinc
t and that that I might have to do different things in this Direction versus that direction so that was the idea of context so in this case you could say direction of movement serves as like a sort of a contextual signal for remapping the cell now doesn't just fire at this location only Fires at this location and this direction uh yes question it's the it was is a question that had been raised early on you think oh this directionality maybe it's just because of differences in cues right the anim
als pointing in this direction sees one thing and that actually was sort of an early Theory this is called local view Theory and the idea was yeah this the directionality is because in this direction you're seeing one set of cues this direction you see another set of cues and there was the idea that place cells form based on the configuration of cues right it's like an associate of learning thing so I have a cell here that's only going to fire when I see you know that pillar and that speaker in
a particular right with a sort of a particular geometric Arrangement so it becomes dependent on cues and the arrangement of cues and the div you know the two directions obviously different configuration of cues now there's sort of a you know one simple challenge to that idea is oh yeah it's just the you know it's the configuration the stink you know visually polarized configuration cues well you just turn off the lights if you turn off the lights when Amazon is doing this task you know what happ
ens to these Place fields almost nothing it's like still they still fire same location same everything's the same you don't you definitely don't need visual cues in order for these cells to fire the other thing what's interesting is so you can do this if I take this maze in this case and I just have like a track that goes up like a really strong u-shaped track what you find is there will be Place cells that are directional but now it's for the turn direction right they'll fire when you go around
around this curve this turn this way right counterclockwise and not clockwise now that sort of challenges the you know the view hypothesis at all you know and the idea of specific configurations of cues as driving it because in the one case I'm seeing all the Q I'm seeing both directions right I'm just seeing them in a different order this way or this way you know when I go around this way and then this way this way it's the it's the same it's you know it's basically the same cues it's not one
set of cues it's not one view both views are actually part of both trajectories and the difference is it's this you know the clockwise counterclockwise so it's like yeah you know it can't be some like static property it really does have to be something about the time sequence it's something about the time sequence that you know must be relevant 100 and all of these things will be used uh in just a little yeah just a little side note say you know rodents kind of get short shrift when it comes to
this idea it's like oh yeah mice rather blind you know I like primates primates you know uh and for you know for rodents it's all about smell right smell first you know touch you know it's not a sensory second then vision is like way down there when you actually do the psychophysical experiments and then you put them in environments where they you pit visual you know olfactory and somatosensory information much like primates vision is actually a it's a it's a dominant sense so they use Vision ov
er other cues if it's available the difference is they don't they don't require that and and also the nature of visual information indeed for like sort of distal cues you know the vision is relatively low Acuity but for very proximal cues in fact cues in which you can combine like rebrissel you know tactile and visual information they can do very high Acuity proximal object recognition just like you know just like primates it's just that you have to put everything you got to put everything very
close and you know maybe when we get to it I'll talk about this experience actual recording in the visual cortex where we found it was very much like you know the the sort of the primate Centric view when you record the visual cortex put a you know rodent in a large room with distal cues you know relatively you know uh impoverished environment you find Visual responses are terrible very few cells fire they're not really strongly you know uh kind of cue dependent but if you make one small change
and that is instead of putting the cues out on the wall you put the cues down on the floor and close to them right proximal visual cues now all of a sudden now you get most of the cells fire they're very selective and you get all you know I'll talk about like the reactivation during sleep so you get the real visual code for rodents is something that is you know tied to the thing that's most ecologically relevant to them and that's you know it's proximal cues because that's where their world is p
rimates yes they rely more on high Acuity distal information but it's not a vision thing it's not like rats are not blind basically that's it right there and so if the visual cues are available they would use them but here it's just the time sequence information you turn the lights off and they will use you know even in the absence of all of these cues olfactory you know a somatosensory visual they can still do this using the integration of self-motion information or so-called idiotic informatio
n and that idiosthetic or self-motion information can include vestibular information they can that you know detect linear and angular you know sort of accelerations it can also include like motor efferents they can keep track of like their footballs the kind of strategies that insects use for instance they don't have a insects don't have a you know in you know inner ear no vestibular system but they can keep track of you know uh of uh you know heading and um and distance by County footballs rode
nts will do the same thing and so uh which is why you can have for instance you put a rat in a running wheel and you still see play cells I mean it's like oh they're not moving but they're still running so they can use that the motor efference uh you know together and all together that as sort of has you know argued that the hippocampal system the spatial representation can be sufficiently driven by what's referred as the path integration system that they're using a sort of a system for like Com
puting spatial location by integrating motion signals to figure out where they are and that's how you build the map through path integration yeah it's a you know a little little uh segue but uh so I I was uh yeah I was wondering what I think it would jump on the running wheel so they're like these there were these running wheel experiments that were that were really designed initially to challenge the idea that the representation of the hippocamps were really spatial and the argument being well
you put an animal on a running wheel and they're not moving their location is not changing so if there's any non-stationary in the hippocampal representation can't be space right and so when you do this uh uh initially like your basaki found this and you know Howard I can bomb followed with it this idea that sells in the hippocampus seem to be able to produce non-stationary responses that are more correlated with time elapsed than elapsed location okay and so they referred to these cells as time
cells rather than Place cells and this is like there's a whole history going back right from the beginning of you know sort of a Keith's Discovery sort of the challenge of the notion that the hippocampus is primarily or some would argue exclusively involved in representing space the so-called non-spatial hypothesis like yeah another hippocampus there's nothing special about space space just happens to be the thing that is always there and so of course you always see it so there's a great effort
to devise experiments which took space out animals doing things without spatial correlates and you know the short takeaway from that experiment is like three decades more or up to like five Decades of research have only really strengthened the argument that there is something fundamental about space it's never it was ever the case that O'Keefe argued that a space and only space his argument really was that there was something special about space that whatever else you include include in the hip
pocampus representation space is always there and that largely has been the let's say the conclusion from most of the experiments designed to try to you know remove space as a as a sort of a relevant correlate and that is you don't need space a tasks that don't need space and yet the hippocampus still encodes it so that's you know uh to me suggests yes there's something special about so and what would that you know why would space be space be special I think is an interesting question now of cou
rse there are many other functions that the hippocampus has been tied to that don't seem obviously spatial a lot of interest for instance social cognition social cognition is something that uh you know you could argue and has been argued well that's like a ca2 thing ca2 might be you know more specifically involved social cognition it's elaborating humans doesn't involve space and yet in different parts of the hippocampus everything I've talked about right now there's this long axis of the hippoc
ampus the so-called dorsal and ventral access in the rodents or anterior posterior in humans so the hippocampus in humans it's kind of yeah you know as you evolve you get lazy hippocamps kind of slides down because you know it's gonna and so what's dorsal in the rodent is posterior in humans so in the human literature talking about anterior hippocampus is really in the road literature you would talk about as ventral hippocampus that's the and ventral hippocampus has a different kind of connectiv
ity connects the prefrontal cortex it's been associated with like emotional learning and you know social you know sort of social cognition yet when you look in in the ventral hippocampus you still find Place cells a place representation is still there it's different Place feels tend to be larger they may have certain kind of symmetries but place is still there so that was always the O'Keefe argument whatever else is being encoded place is there now why would place still be there why would you us
e let's see the idea of place and again I kind of come back to this notion of a path integrator so a path integrator is largely I I construct simple equation of motion it's my it's the x dot x dot idea right if I have x dot some something that's changing as a function of time State changing a function time I can integrate that to get my state space is an obvious one the x dot is just again it's like self idiophatic self-motion information but x dot X can be anything it doesn't have to be space r
ight and so it's the idea that path integration the path can be a path in in you know literal space and semantic space and social space and so it's the idea of using of the you know the hippocampus using this derivative information it's representing change in state in some way and then use the same principle the same principles that work in space to do path integration so the path integrator Works in space you can generalize to anything else where you have an x dot and so that's the and it kind
of comes back this well we'll see we'll see if we get this idea of of of utilizing gradient information and in particular uh the inputs to the hippocampus reflecting gradients or these you know like X dots being able to compute the derivative of those things being kind of fundamental to the computation that the hippocampus is going to perform to try to keep track of time sequence information could be a sequence of locations could be a sequence of you know words or letters could be a sort of a se
quence of social interaction so it could be a sequence but you need the x dot and the you know the nature of the of the representation that can exploit that is the will be the common you know sort of be the common property that's the and so that's really the that's the argument so you know I tend not to get to bog down in this is it space is it not space is it you know uh and then the idea of time or time cells clearly time is the critical is the one common thing that's why in the opening Slide
the slide one that actually got to that was the argument it really is you know it's the it's the encoding of temporal information how is that done so iposite is like time sequence information but I think more fundamentally it really is this sort of x dot time varying time-bearing state
Comments