 So memory, to us, seems like a fairly homogeneous process. So from a point of view of an interface of a designer, it's a wonderful thing, because, you know, I remember some, you know, I think about something and then something comes back and I don't have to worry about what type of memory I use because the common theme of today's lecture will be that there are many different types of memories. There's iconic memory, the short-term working memory and there's long-term memory and there are very different types of memory. But that's all sort of hidden. This sort of interface is very, it's totally transparent and I don't have to worry about that. Of course, we all know from our own experiences, from comparing our experience with those of our parents, let's say, or of picture books or newspaper articles that memory is very selective and is highly edited, fortunately, unfortunately. And we only sort of selectively relive, well, edited, highly edited parts of our life. And I took this quote because I was really struck by it to some extent and there are few people who have no memory, no memory of this sort that without memory we're really, yes, we're conscious, it's going to be a theme of today's lecture, but we're really severely impoverished in terms of who we are, who we think we are, what we, you know, our notion of self and self-hood and personhood, all of that is really tied up to what we remember. So first of all, I mean, how do you define memory? And this is one definition I took from a book of an Israeli neobologist. Memory is the retention of experience-dependent internal representations over time. They have to be meaningful representation because otherwise if you just say, well, it's a change in your body or a change in your brain, it's function of experience, then sort of, you know, adaptation would be, injury would be as you grow or, you know, as you grow, as you change, you know, from a child to an adult, all of those would be memory and typically we don't consider those memory. So the DSA has some internal representation that gets updated as a function of your experience. There's a fundamental distinction one has to make, both in computers as well as in human or in biological organisms. The distinction is between activity-dependent memory and structural memory. So by activity-dependent memory, I mean memory that rely on ongoing electrical activity in your brain. So I'll show you a few slides but if I give you a number 3 or 33 and I'll ask you, you know, in a minute to tell me the number that I told you, you store that in a short term in something called working memory and the evidence we have since this show very strongly that you do that by elevating the firing rate of neurons. The firing rate of neurons is elevated for, you know, however long you keep the thing in working memory. And so this is a bit like DRAM where you retain in a computer, you retain your digital memories but as soon as the power goes off, the memory is gone. This is more like ROM where you have long-term memory which do not depend on ongoing electrical activity. So for example, when I tell you, remember the Statue of Liberty or, you know, think of Brad Pitt, when before I asked you that question there was no explicit firing, there was no neuron that fired explicitly in sort of to encode the Statue of Liberty or Brad Pitt, when I asked you this question you recalled you had to access some sort of structural memory which in turn gave rise to some neural activity. Probably for example in the case of Brad Pitt or in the case of the Statue of Liberty in your visual part of your brain because it involves imagery. We know, for example, from lots of cases where people with a brain activity goes essentially de facto to zero. So for example, circulatory arrest or when people do this now routinely in some types of major heart operation when the heart beats and you have to, for example, in fact Francis Crick had this couple of years ago, when he had a heart operation and when they opened the heart the realtor said they had to replace this with an alter, the inner tubing of the alter was something artificial because it's very difficult, people do it now but it's much more difficult to operate on the beating heart obviously while it beats. You cool down, you stop, you put the patient on a heart lung machine and in order to prevent damage to the brain you cool down the brain or you cool down the entire blood to something like 10 degrees, very chilly, 10 degrees Celsius, I don't know like 50 degrees Fahrenheit, quite chilly. You do it slowly over a couple of minutes and essentially what you see in the EG, I know this because we talked to the physiologist afterwards, the EG goes totally flat on all channels, flat EG, so in other words there's no over at large electrical activity left in the brain and then you typically have like between 30 and 40, 45 minutes left to do the operation because statistic shows of recovery of patient, if you wait any longer there will be serious cognitive deficits in the patient later on when they wake up. So in case of Crick he woke up and he had no trouble with his memory. So it's just an experimental proof, we also know this from people who are knocked out for short times, etc. You might not remember things that went on directly before the accident, like a couple of minutes before, so-called retrograde amnesia, but in general you have no trouble with remembering things over long term. Here of course we know a lot about the basis of this at least in simple animals, last year's Nobel prize was given among others to Eric Kandel for studying some of the structural and the molecular changes that underlie structural memory. So the best today's theory says, and there's lots of evidence in favor of them, that part of structural memory maybe a very large fraction is encoded in changes in synapses, so the contact points between neurons. The increase and decrease their weight is called synaptic plasticity and you have an increase in synaptic weight, it's called long-term potential or decrease called long-term depression and this is how memories are stored by manipulating synaptic weights. This by itself is an old idea, it goes back to heaven, in fact it goes even back to Freud and you can find even precursors before them. So the distinction between activity dependent memory and structural. Here if you power the brain down, the memory is gone here it's immune to that, at least for short times. There are many different forms of long-term memory. So long-term memory is not really precisely defined, it's probably all of these depend on structural changes in the brain and they go on from minutes to hours to your lifetime. And you don't need to rehearse, I mean once you've learned it, it's in your brain. Of course it does decay, we all know that. And I don't really think, we certainly don't know at the neuronal level what actually triggers decay. We don't know, in fact there's so many things we don't know about memory. We don't know what the memory capacity is. We don't know what causes what causes what, I mean we know at the psychological level something about interference, etc. But we don't know why memories decay. You could say it's a generic argument, any hardware, it slowly goes down particularly since it's not digital storage, all or none, but it's analog storage but we don't really know the specific reasons. And we don't even know something as simple as what's our visual capacity. Remember in Johnny Mnemonic where he has his implant, he goes to 320 GB but we don't know what our capacity is. So there's some very simpler forms of memory. Simple as this light adaptation or after effects as we talked in class already. So right now when I'm looking at this bright light and then suddenly the light goes, well the other way, when you are at night adapt, let's say you want to do astronomy outside in your backyard, you adapt, you only have a red flashlight and it takes you up to 20 minutes really to dark adapt. So that's sort of a form of memory. After effects are habituations and priming. I think I have this. What do you see here? What did you see? Did anybody see anything? Yeah. Okay. You all see it? Okay now do you see it now? Okay if I now show this to you a week later, did you see this figure before? Okay. So if I now show you this, so it's a Jesus like face, bearded man, I mean most sort of. The claim is now, if I show you this in a week from now and a month from now, you'll still be primed in a sense that you'll still respond much faster and see a man's face there even though you might have forgotten explicit to remember it. So you might not explicitly remember that you saw this but you would still be primed. This is called visual priming. So again something has to change in the brain and has to be quite specific because you know this is priming for one specific type of image. And there are many illusion books are full of these. A famous one is the Dalmatian dog where if you look at sort of it's a black and white image you don't see anything but then after a while if you really sort of it comes to you that you're actually looking at a black and white picture of a Dalmatian under a canopy of shadow. There's a shadow place, there's a tree and there's the light, the sunlight comes through the leaves and strikes the dog so it's a very difficult texture to segregate. Where the shadow due to the leaves and where is the texture of the dog. But you see it and once you see it you recognize it much quicker. So this is called visual priming. Visual priming cross is probably quite important. It enables you to see things much quicker like traffic lights and anything in your environment like fonts. And of course it's so strong that you have expectations that even though it might actually not be the real figure because you are priming you expected something even much more likely to see it. So there can be quite powerful effects. Okay then there is a very famously classical conditioning. So that's the conditioning that Pavlov talks about that for which Pavlov, well Pavlov got a Nobel Prize actually I don't think he got it for conditioning. He worked at this in pre-communist in St. Petersburg under the Tsar although he also worked under the Communists later and of course he did the same experiment where he took dogs and the dogs would always, you know the dog would tend to learn to associate the particular person, the coming of a particular person was the fact that this person was going to bring them food and then they would salivate when they just saw the person because they bound, they associated two objects together. They associated the presence of the animal trainer with the future reward. They associated that and so they started to rule just when they saw the trainer. This is called classical associative conditioning. Associative conditioning because you associate two events, the CS and the US, the unconditioned stimulus and the conditioned stimulus and what you get is a conditioned response. So there are many many forms of conditioning and they play an important role in our lives although for the most part an unconscious role. Although this is part of now thesis research, some of this in order to acquire some of this you might actually need to be conscious. So some of the best studied as fear conditioning and eye-blink conditioning. So these are of course like usual, these are forms of conditioning that you can easily study in the lab. These may not necessarily be the most powerful one that operate in daily life although fear conditioning sure is important if you have it. So in fear conditioning you know the way you do it experimentally in a lab the way for example we do it, you shock people, you give them a light electrical shock and then you associate various tones with it. So there are different forms as simple conditioning and there is differential conditioning so typically what you do, so this is time, so you give a little, let's see tone, so this is called the CS plus, the condition, the positive condition stimulus and then let's see at the same time here or sometime later you give a shock. Okay and you do this, if the shock is strong enough you only need to do this once. Now in a lab usually you can't do this to humans the committee won't let you do that so you have to do it so in animals you can do, in mice it's typically done, on rats it's done when you shock the animal a couple of times or you do it in humans where it takes you know three or five or six depending on the strength and then you measure the response, so what response you measure in people you can measure different things like heart rate, like pupillary diameter, like skin conductance, changes in skin conductance, so this is of course at the basis of the lie detector, you have the autonomic nervous system, it responds and one of the way it responds is by increasing the skin conductance you start to sweat that's what you pick up in a lie detector test. Now of course there's a large amount of variability as now can tell you in skin conductance so which is of course a big problem with lie detectors which is why the scientific community, the psychologist said they do not believe that lie detector is a reliable way to measure and it's probably used when you talk to law enforcement or when you talk to national security guys, type people, the main reason I think it's used is low level deterrents. Clearly you can train yourself, any good actors can do that, you can control yourself sort of to limit the extent control those emotions so when you lie you don't get terrible aroused and certainly if you're pathologically it's almost impossible to pick if you're a psychopath it's impossible to pick it up because you just don't have any of these normal reactions that you know if I don't lie and suddenly I lie I'm on a great deal of stress I might not reveal it from the outside but I'm going to get very anxious and start sweating in that something you can pick up. So when you do this pairing of the CS with the electric shock you do it a couple of times then the next time you hear this tone beep you automatically start to sweat. To a certain extent it's automatic. So for example what we do in mice we're doing this research also in mice, we use freezing so what happens to the mice when you shock the floor you do it three times or six times then every time they hear the tone they sort of totally freeze, you can see it for many, many seconds that you can just see very faintly, you can see and breathe but otherwise they totally remain emotionless. Now that's a typical reflex of a very small rodent, either they flee or they freeze depending on circumstances. That's not something you know big animal would do but that's very often what small animals would do for obvious reasons, freezing response. So that's another way to measure, like say you can measure a heart rate or you can measure a pupillary diameter these are different things you can measure. So now this is simple conditioning and you can do differential conditioning when you have a CS plus and a CS minus. So let's say you can also have other, so let's say this is a high tone beep and here you have a low tone this is called the CS minus and the CS minus is not associated with any shock so you randomly get this intermix, you get beep, beep, and every time you know during training every time I gave a beep I shocked you and so the idea is that the neural structures that do this conditioning they're clever, they can distinguish this from this, they know quote implicitly they know that they've learned that this one does not predict shock but this one does predict shock and so you start in the ideal case you'll start sweating here much more than in response to the CS minus. So the other common paradigm where people have done it particularly Dick Thompson and others is eye blink conditioning which is where you take a little puff you put a puff of air into the eye you blow it and of course particularly if you don't have contacts it's a quite annoying stimulus and you blink and this you have to do like 50 or 100 or 150 times but then you can quite reliable also get association with typically it's done in the auditory domain with various tones where in response to the tone you'll blink and so it's interesting you do this automatically and you can observe yourself that these tones and suddenly you blink. Of course you can perfectly well also blink reflex like but that has a different dynamics and you can pick that up. Key conditioning is probably terrible important for lots of pathologies like trauma, post-traumatic stress syndrome, when people had a crime was committed in battle, at some battle scenario and where people have these flash flashbacks and remember them and of course in the severe case like post-traumatic stress syndrome they might not be able to live a normal life because it interferes so much to such a large extent. Then what you can do, you can do extinction. So in extinction what you do you take this and you repeat just the CS plus for let's say 10-20 trials without giving the shock again. So the idea is that you learn a new memory you overlay a new memory into the old one the new memory says okay the CS plus isn't associated with anything the old memory says CS plus will be followed by a shock and so therefore you tremble or sweat or whatever but then when you do this extinction hopefully if you do this often enough you'll tend to the system that mediates this associated conditioning will learn that oh it isn't associated with anything bad anymore. Now of course a trouble it's very difficult to do this for these patients you know when you've been raped or in a war scenario you cannot really recreate this to any realistic extent in a lab or in a hospital so that's one trouble with post-traumatic stress syndrome. Then context dependent conditioning you might know this from school or from any place you learn it's much easier to recall things when you learn them when you retrieve them in the same location where you learned them it's a particular strong for fear conditioning so I mean I can see it in my animals but you can also see it in some people you know if a bad thing has happened to you in one particular location I say in a parking garage somebody tried to mug you you know you'll be very anxious and you won't know why I mean sometimes it might go hand in hand with an explicit memory of whatever happened there but very often it's just you feel sort of you know an antsy and nervous and anxious and why because well of context dependent for some fear conditioning you can clearly see this in mice when if you drop them in the same cage where they were shocked compare you take them in you put them in a new cage then they will also in the same cage where they were shocked although you don't do anything you just you know day later you put the cage in the same mouse it you know it has association I got shocked there so I might be shocked today so it immediately freezes. So again this you can of course also try to do extinction by putting people in the same context where where the bad thing happened to them. Context depending condition yet also help for recall just learning so if you know you're going to have to take the test here in principle it would make sense to actually learn over here I don't know how important it is for things like abstract learning I know it's terrible important for for fear conditioning I can see that with my dog where something happened in at this one door sort of a door probably fell onto one of my dogs and for the next four years the dog would never go near the door you know it's just a single event can be very very powerful the same is true for us very very strong for bad food right I mean very very it's very strong you know if you and you might remember when you were young and this can last here there can be an interval maybe up to 20 or 30 minutes between when you actually ate by the time you threw up and even though there's this long interval because it's a special purpose system you remember it and for the next many many years you'll tend to avoid that you'll tend to avoid that place and you'll tend to avoid that food a very powerful versus conditioning effect now there are two forms of them okay so there are many different forms and they're all by different neural structures even with the eye blink of fear conditioning there are two sub forms one is called delay and the other one is called trace so it's a weird nomenclature but essentially in delay there's the essentially co-terminate so here's the tone and here's the puff of air or the eye blink conditioning or the shock we do shock so the important thing is that they terminate together or they basically overlap or trace conditioning you have this interval between the offset of the tone and the electrical shock versus here they go together so what I imagine is a more robust form of conditioning than this because here sort of you know the things go at the same time or here there's an intervening event and of course in this intervening interval in this case it's short as second all sorts of other things could have happened and for example you could have maybe distracted the amel or the person in this short interval and then it turns out in fact as we and other people show now that you can interfere with in this paradigm than in this paradigm now as I mentioned for food this can be very very long for food it's 20 or 30 minutes but that's because it's a special purpose thing because obviously when you eat something usually you don't immediately throw up it takes you some time you know in order to develop nausea etc. there's a famous paper a couple of years ago by people down here in San Diego Clark and Squire I won't go into the details who essentially said that and this goes back 30 years these sort of ideas the threat that trace conditioning the more complex form of conditioning requires attention and awareness because you have to put things sort of into short term into some sort of working memory because you get the don't eat and only then you know two seconds later or a second later whatever you get the air shock and that is a sort of more sophisticated version and they could show using a questionnaire so they ask people so you're watching this movie it's a silent movie you're being told it's a memory experiment which of course it's not really and you should really pay attention to the movie and then at the same time you get these you get these different noises and tones and occasionally you get a puff of air and the claim is that when you're really engaged by the movie when you're really looking into the movie you don't pay attention to clearly you hear their tones you hear their noises you hear the puff of air but you don't really pay attention what's the temple order of one or the other and so the claim is that if you don't pay attention you don't develop trace conditioning if you're not aware of the relationship of the so-called contingency relationship of the relationship between the C's and the U.S. the tone and the puff of air then you don't then you're not conditioned well for delay conditioning when there's no when here when they co-terminate this form you don't need to be aware that's their claim that's exceedingly interesting because that allows us I was really electrified by this because this allows us in principle an experimental test that we can try in animals because now we in principle we can do this in animals and we have so it was together with David Anderson's lab here in biology and two postdocs C.J. Hahn and Colin O'Tooleake we've done this now in mice and the ideas are following I mean first of the rational why are we doing this in another in you know in humans and why are we also doing it in mice well ultimately you want to discover the NCC now clearly doing that in humans it's not going to be possible because functional imaging is great but you know it's very crude so even in monkeys you know you don't for ethical reason for practical reasons you cannot really interfere at the circuitry level where you want you cannot really take apart the circuitry right now we just don't have the technologies but we can begin to do that in a genetic model and particularly in mice because it would be even better if we could do it in flies and flies do have trace and delay conditions but they are so much further removed from us it's much more difficult to make an analogy an argument by analogy but here if I can say so I have two phenomena for phenomena in humans and in humans one of these phenomena is associated strongly with the other one not and they have a certain characteristics and if I can reproduce the same phenomena in mice of course I don't know about awareness but if I can show similar things in mice as I can in humans then I can argument well maybe something very similar is going on in humans and so here what we have we have the same thing we have similar to what Squire and Clark showed for eibling condition here we do it for fear conditioning so we have tones long tones and then we have shocks two second shocks this is delay this is trace and then here at these times we show them distracting light so the mouse is in this cage there are tones here and then sometimes the electric floor we give these brief electric shocks to the floor here and then at this time here I flash lights randomly this is pretty dark and suddenly you know it's like if I suddenly turn the lights in this room on and off the expectation is that I would interfere with it might interfere and it does very nicely it interferes with trace conditioning but not with delay conditioning very specifically so in other words we can we can take animals and we do trace conditioning with them we do delay conditioning with them using this freezing and it works fine the animals learn this but then if I flash while I learn I flash a sort of a house light I dim the house lights repeatedly the animals was unable to do or has a much greater difficulty doing trace but it doesn't interfere with delay it also doesn't interfere with context dependent conditioning so that's potentially very interesting because here you have a now mouse model that you can use to study attention and awareness and why is it interesting well for example things like this is a Henry Leicester here professor in all symbology so what they can do now they can use specific silencing genes they can introduce genes into the animal either by a viral vector or by a transgenic animals that will specifically and that will under control of some external factor knock out that will silence neurons it only fires here but then you add this magic bullet I won't talk about the neurons it's silent you can do this transiently in a time scale of a couple of minutes you can turn this on or off essentially you have these neurons and you add some drag to this system it activates a promoter and it does all sorts of things but function where it does it turns off these neurons for let's say you know 20 minutes a half an hour and so it's it allows me to begin to interfere very specifically with the system okay let's close several parenthesis and go back to memory so this is one form of memory a classical conditioning and the interesting thing it is some interesting relationship to awareness which might allow you to test these things at the level of animals now there's also a very important model system very important memory is called procedural memory now this is essentially sort of a talk about zombie systems in most of the zombie systems of this ilk so you know biking and climbing and you know learning how to speak and learning how to tie knots and doing all the other things that you do in your life you know brushing your teeth all the other things you do those are all instances of skills and habits you need to acquire that you acquire them by doing them over and over and over again then after you do them you do an effortlessly you don't have to think about them fortunately these are very resistant to decay so unlike you know even Alzheimer patients you might still you know they might have lost a large fraction of the explicit memory but they still have you know their skill and habits now all of these are instances of implicit or what psychologists call non declarative memories because you don't have an explicit memory you might have an explicit memory for when you learned or where you learned these you know if you learned let's say a new skill like you know snowboarding you might have a memory of oh yeah I remember I was you know at Lake Tao that's where I learned it but if I ask you tell me the details about how to snowboard you don't have access to that it's very very difficult you do not have access to that this in comparison was declarative or explicit memory that's of course typically what people mean by memory when you know when somebody talks about memory that's what they talk about now of course you saw so I don't have to say a lot about this memento the fact is they are patients the most famous one being HM even more dramatic one is Clive wearing so HM was a patient who had temporal lobe epileptic seizures and in order to help them what the doctors did what they occasionally still do they take out the part where the what the force originates which is hippocampus now if other hippocampus was already damaged so then he had bilateral hippocampal damage and that made him essentially unable to form new memories and he already hard because of course he was already diseased he had already great trouble with previous memories but really I mean so it's not that he could remember everything perfectly up to the moment of operation and nothing afterwards even before he had pretty bad memory but certainly afterwards this is now in 1954 the patient still alive he had no explicit memories it was just like in memento with Lenny you know you walk into the room you introduce yourself you talk to the guy the guy looks perfectly normal and conscious he talks to you you know you ask him how he feels he gives meaningful answers you ask him what he had for breakfast of course he has no ideas guessing you ask him what time of the day is you know he might look outside and say it's noon you ask him what year it is he might say you know it's 1954 and then you walk back out and you come back in two minutes later and you know you go through the same story again. A very dramatic case which in a sense is even more disconcerting because this is something that could happen to everyone. Clive Waring was a highly gifted Renaissance musician especially in Renaissance Italian composer anyhow and also conductor and then he worked a lot he worked a lot at this period he worked a lot didn't go to sleep a lot was highly stressed got sick with a viral brain infection almost died due to that infection I mean without modern medicine he would have died he survived large parts of both temple loaves were destroyed permanently and that's a movie made about him quite a famous BBC movie and it's really pathetic to see this person because he for example once he fills his diary which is filled page after page which says I've finally woken up for the first time next page I've finally woken up next page I've woken up and I'm alive for the first time and he goes through these things again and again and again he talks to his wife he has no idea who his wife is but he says well I guess I must know because he has a feeling of familiarity with her and he has procedural memory for her and how she moves etc so he has some big feeling that he knows her but he doesn't really know who she is he doesn't in this case it's really this is the most dramatic case he has almost no memory of himself while lobbyist patient for example Oliver Sachs has a description of the guy who was at the hippie the eternal hippie or something like that was called because the person remembered everything up till the early 70s and he was a grateful dad fan and that's what he knew and what's what he talked about he was hippie at the time but now of course 25 years later he still lives in the past Clive Waring has no past he's truly a person like a Greek actor he has no past whatsoever he's all present there's nothing but present in him he has procedural learning so this was the part of why HM is the most famous patient in sort of neuroscience is he had a specific you know you take part of the brain you have the specific losses inability to firm long-term memories also you could show in this patient that he had he could still learn for example mirror of writing or tower of annoy other games so he you know mirror of writing which is you know quite tricky to do so he could learn that yet he had no recollection of learning that so again this is the case of a patient where you have a dissociation yes he learns you can still learn skills because that's done somewhere else but he doesn't constantly remember doing them those skills now the point for us is none of these patients including this fictitious patient Lenny here are clearly unconscious clearly you can talk to them they have feeling they have sensation no question about it so in fact you don't need any of this I mean you need some of this to be a human to live in the world and to do anything but you don't need you probably don't need any of this to be conscious now then there is a short-term memory which is again sort of a bevy of different processes again short-term memory means anything on the order let's say a minute or less so there's now the theoretical construct by cognitive psychologists Alan Bradley particularly at Oxford called working memory and it has sort of substructure there's a central executive and then there's a visual buffer called a visual scratch pad or buffer and then in the same thing in the auditory domains called the phonological loop this can be measured in a semantic domain by the digit span so I wanted to try this to see how your digit span is so what I'm going to do I'm going to read slowly with a heavy accented boys I'm going to read you numbers okay they're just numbers between zero and nine and let me see how do we do this yeah I'm just going to read them once and you should try to remember them don't write them down that's cheating right that's using an external memory just to tell me what your digit span is and then just I'm going to read them once and then I'm going to read them again and just see how many digits you remember okay so it's 0, 4, 3, 9, 7, 5, 2, 4, 7, 9, 8, 3 okay so it's 0, 4, 3, 9, 7, 5, 2, 4, 7, 9, 8, 3 no wait shit it's not going to work up I'm being stupid here we have to do this we have to write them down sorry let me do this again so I think what I have to do I say go and then you write them down okay I mean not these let's try with some new numbers okay do you all have paper and a pen but only write them down at the end right otherwise 1, 5, 7, 4, 3, 2, 5, 0, 5, 7, 9, 4 okay so should I read them out 1, 5, 7, 4, 3, 2, 5, 0, 5, 7, 9, 4 does anybody got no 2, 4, 6, 8, 10, 12 so it's has anybody gotten like 10 or 7 yeah because otherwise you just write down 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 it should be in order anyhow I mean it's just a toy example so you might all have heard the magic number 7 plus minus 2 so it's a very famous paper probably it's famous for as much for its rhetorical device using 7 plus minus 2 the magic number as it is for the actual data but the claim is that essentially you remember 7 plus minus 2 object so the plus minus depends on you know your alertness you know how interesting the you know the material is for you how well you're rested you know all sorts of others particularly your attention your alertness on all sorts of other factors and this working memory is and so this measures a digit span of working memory here's the actual distribution from one paper so this I've here they waited for you can see this is percent correct so correct means of course the correct digit at the correct location there's several special effects so the last number of course you will tend to remember or the last two remember the reasons the effect you'll tend to remember much better than of course things in the middle and you very often also tend to remember the first number better than you remember the other numbers yeah so the claim amnesiacs are people like HM you know people who have amnesia but usually this amnesia doesn't extend to shorter numbers so here clearly they can certainly remember you know three or four numbers or five numbers and they get pretty bad working memory relates very strongly to IQ as measured by IQ test for whatever that's worth so it's a very strong correlation your digit span how many things you can store online it's a very strong correlate of your IQ and the claim is you need your IQ you need working memory for doing all the tasks that you do constantly like you know when you navigate from here to somewhere else or when you add two numbers in your head when you compare, when you draw, when you for example some of you take notes when you look at here and then you write things down on your notebook clearly you need working memory so for all those things you know when somebody talks to you you try to make sense of what they say in German for example there's almost can be 200 words between the beginning of a sentence and the actual verb that gives the entire sentence meaning so of course you have to store all of that somewhere and then the word comes and you know you can finally process the sentence so for all those sorts of things you need working memory anytime you read anything sort of moderately moderately complex you need to remember what just went on to understand a movie you need to remember what went on a minute or two or three ago so for all those things you need working memory so it's really at the heart of of our ability to act in the world it's chunk so people talk about chunks here what they mean by that for example the sequence I gave you there was a chunk 432 and of course 432 you can you know and if it's even longer 432 1 that's much easier to remember than for random digits when you know when I gave your social security you know when I give a number that's your social security your birthday on you know 9-11 those are all special symbols they've been encoded for other things and so of course you can make use of those and so you can use them to encode so for example this doesn't only extend to verbal memory extend for example chess figures when I show you when your chess player and I show you random figures you know I put you've done this experiment when I show you a chess position I show you two chess position A I take a chess board and randomly put pieces on the board I ask you to remember those then you're bad and you're as bad as an expert chess player or as good as an expert chess player and the capacity will be similar to this or I take the figures the chess figures from particular games these are all legal configuration what you see then you will be again will be pretty bad you know will have whatever limit it is the chess player, train chess players can do much better because now they see these chunks and say oh yeah this is a fear and chateau this is a queen scam they remember the configuration because you know he has sort of memorized these chunks and so he remembers not the individual elementary you know king on R2 but he remembers you know a configuration of four figures so that's called chunking now there are some patients who have severely impaired spam these are not amnesia these are other people very often they also have language difficulties and they are severely impaired so they might only remember one thing or maybe two things yet they're clearly still conscious so they might they are difficulty drawing and they're less difficulty reading and comprehending in all of that so very often they have language problems which makes them difficult to investigate but you know as far as you can tell they seem fine they hear fine they sense fine they cannot do some psychophysical experiments so they cannot do a two alternative choice when you have to compare this person with something that happened ten seconds ago that they can do but they seem to have all the sensations that a normal person has and also if you if you if you just cast your mind back a couple of minutes ago when I asked you to remember those numbers I don't think certainly when I introspect I certainly don't recall all those numbers in my working memory at the same time I mean seem more like there's a scanning process that I can consciously remember the last two digits or the whenever digits I go through but I don't sort of have a clear vivid picture of all seven of them or eight or six in my mind so you probably need this for consciousness but it's difficult to say because in the limit if you don't have working memory it's very difficult to talk to people because if I ask you okay what did you just see well you know if you don't have working memory you might have had a perfectly perception of blue but if you you know don't have working memory you might not be able to answer me that question so it becomes difficult to test this in the limit of really having no working memory at least using current sort of current technologies and certainly there's a very close relationship so working my memory might not be strictly necessary but for sure there is a very close relationship between working memory and consciousness because working memory is such an integral part of our day-to-day life what do we know about its neural correlate well people have done a lot of experiments let's say in the visual domain they record memory cells I'll show you one down here as you would expect in the neural pathway the anterior part of the infotemple cortex where you have neurons that respond to high level things like faces and other animals and you know high level figures and then also in the lateral prefrontal cortex here you might remember so you have these two streams remember the dorsal stream vision for action the ventral stream vision for perception they sort of reconverge at the level of this area and in detail you find neurons that seem to encode the memory both for aware location as well as for what the object the first has done a lot of work here in this domain so this for example here records in this part of prefrontal cortex and this for example is a typical there's a whole baby a whole zoo of different working memory neurons this is one here where the animal gets a cue doing this time that the neuron basically suppressed and then this is the same neuron doing five different trials then here the monkey had to remember something for 35 seconds it's called a delay trial so you let's say you the two you flash up an image no you flash up two the two images next by each other and one image is highlighted briefly but then both images remain remain there and only you know 35 late seconds later can you actually push a button for example push on the on the location of that one object so you have to keep it in your mind and you have these neurons that respond with a you know high fine rate for as long as the animals are required to hold this memory and then as soon as the memory sort of as soon as the monkey can then you know reach out and push the button when he doesn't have to remember anymore those the fine of the neuron goes to zero the other experiments people now do more sophisticated this is at MIT this is called a delay match to sample you fix it you get a drawing of a bell now remember these drawings you should remember they're probably meaningless to the monkey because the monkey you know doesn't know what a bell is it's never touched a bell probably never seen one in this cage so these are just arbitrary line wiggles on a line for the monkey so it sees a bell fixates then sees now two choices a bell and a letter box and now has to remember so here the monkey has to remember well during this period here the monkey has to remember what it was namely bell information to access information to access facial information so here the monkey now to remember this location to do this delay the monkey has to have a where memory and then finally has to make an eye movement right so here here doing this but he has to store a particular type of object in this period he has to store a particular type of location and then there are cells that respond to both that respond to both the location as well it was an elevated fine grade but here's the first stimulus here's the second stimulus and here for example if the monkey will respond to one object it's a good object let's say it could be the bell in this case and will not respond to the poor object let's say the in the letter box this is what I meant by active memory you can see at a fine grade let's say from 30 hertz to you know 70 hertz or something they almost double and probably people have made nice computational models this probably involves working memory sort of reverberatory circuit a local feedback circuit it's like a latch up that sort of you know once you trigger it keeps on firing until it's actively reset by the fact the monkey okay now I can forget about it because I can make my eye movement and then we know from working memory you know you can rehearse for as long as you rehearse you can keep things in working memory as long as you're not interrupted remember that one critical location in the movie when he's frantically looking for a pencil to write down that you know the girl from there is that she herself did this and then you know he's interrupted and then he forgets about it so same thing here you can the monkey can remember this for long long times and the cell will fire as long as the monkey keeps on rehearsing last memory component this is called iconic memory this is probably the one that's most critical for memory so iconic memory is let me demo it to you now how many letters of that that you see we can do that again just fix it at the cross and then how many letters do you think you can recall okay which four well I guess I can't freeze by going out of here so this is a typical recall experiment and people have in housing I think in the 19th century you flash up things like this and then I ask you to remember and typically what you get is exactly what you said is like three to four letters and this of course depends on all sorts of factors relating to attention so usually you know I would imagine you get those corner letters better than the ones in between and some letter like X might be more salient than another letter like I don't know W with some letters of course more easy to confuse you know with each other etc so you can do all this sort of research the fact is you only get a small number now how about George Spurling then it's a famous psychologist and he did this for his experiments I'll be telling you about for his PhD research at Bellab and Harvard he was struck by the fact that yes I can only get I mean apparently I can only recall three or four but I can sure see many more I have a feeling that I can see them all I can clearly see them all right you should all I mean you probably all have that feeling I mean I certainly have that feeling that I can see all letters right but I only remember three or four then he used what is what's called a partial recall technique so I'll demonstrate that to you now I'll show you the same display but now there's going to be an arrow the arrow is either going to be against the upper row the middle row or the bottom row and depending where it is I would like you to recall just those letters okay okay now okay that's XWZ so if you now do this then he did it actually with tones depending you know high pitch medium pitch low pitch and this is called partial recall and now what happens people in this condition can recall like three letters or 3.2 letters which tells you come back to that a second what does it tell you well it tells you that since I didn't tell you ahead of time which of those rows I am going to which of those rows I'm going to cue you on and you can on average if I do this for different rows on average you can get something like 3.2 letters that tells you that you had to have at least 3 times 3 9.6 letters accessible at least for short time but then that very quickly decays and he studied the time cost of this decay by giving this cue you know the tone of the arrow at different times so either I give it immediately when the image is on or I wait 100 milliseconds you know the image on disappears then I wait 100 milliseconds 200 milliseconds 300 milliseconds etc what you could show a very rapid decay curve that once I wait more than half a second or 700 milliseconds then you're essentially back to the original performance but if I do it earlier you have access to most of the content of that row so the implication is there's a very fast memory an image memory in this case but the same thing is true for auditory domain there's a very fast image memory that you have access to but it decays very quickly let's say within a second or so it's essentially decayed and this image memory if you quickly read things off there you can read off the different aspects so the fact that you can only have limited access to them tells it's because it decays so rapidly it's there everything is there but you only have access to a very limited part because by the time you read it off all the other components have decayed so if you read off the middle row by the time you're done the up and the lower row has disappeared yeah now people study iconic memory they manipulate the images and they show you images or they show you letters and they do all sorts of experiments so iconic memory has different components it has a pre-iconic component so iconic icon means a pre-categorical component in a sense that some aspect of it you don't store as abstract letters like the letter A or the face of Albert Einstein but you just store it as a bit patterns almost and then some past aspects of it are post-categorical that you know that you process it to a high abstract categorical level where you say well it's a letter A or it's the face of Albert Einstein the way I think about it these components are probably neuronal afterglow supplemented by some active firing in other words we know that if you just show a brief input, a brief stimulus to the retina that we have this wave that goes through the system and even if it's just 10 milliseconds input it activates neurons and the neurons will file at least 400 milliseconds and of course the retinal neuron will fire for that long and it goes into V1 those neurons will fire for 50, 60, 80, 100 milliseconds etc and then probably amplify that afterglow, that neuronal afterglow is probably amplified by local circuits that keeps this up maybe for half a second and a second and it's difficult to imagine there's no patient who doesn't have this sort of iconic memory and it's difficult to imagine what life would be without this very, very brief form of visual memory this is also the one that if I blink although we constantly blink that I can use to because until I just point this out to you you're of course perfectly oblivious of the fact that you blinked I mean you know it abstractly but so there the image goes dark for 80 milliseconds, the time it takes for your eyelid transition over the eye yet it's not that you have, for 80 milliseconds it's just blank so that sort of interpolation process probably relies on iconic memory this is the one that I think to me is the one closest to consciousness I cannot imagine how you can do without, working memory I might imagine you might be able to do without it although you would be severely impacted but this one I don't see how you can do without there's very little research going on iconic memory, most of the works on working memory because working memory can see clear neurons that do and that don't fire in the absence of a stimulus, so if you do this ex-bum in v1 some of those visual ex-bums I just showed you, the monkey has to remember the location of something or the intended of something the image goes off, 80 milliseconds later the cells are stopped firing there's no question about it, while those high level neurons in IT or in prefrontal part of the brain some of them they will fire as long as the monkey has to remember that here there's no, there doesn't seem to be any simple type of neuron that's involved in any one neuron in iconic memory and I suspect it's something that goes on in different stages in the neural way and we know from psychologists that the different aspects of iconic memory they probably map onto these different stages I think that's it these are the three different types of memory, long term memory long term memories, this entire set of them working memories in iconic memories so next week we'll finish, we'll talk about some split brain patients, that's rather interesting, experimental interference with consciousness, we'll talk about some high level ideas about are you actually conscious of your sort of the highest level of information processing machinery in your head or does consciousness only access some intermediate level theories, we are thinking explicitly that of course consciousness is sort of the most elaborate part of information processing machinery in our head, but that's probably not true there's quite a bit of evidence against that and then sort of we'll summarize the sort of review where we have been and where we have gone okay, any questions?