<TITLE: Genetics and Pathology of Oscillatory Brain Activity
ACADEMIC DOMAIN: medicine
DISCIPLINE: neurology
EVENT TYPE: lecture discussion
FILE ID: ULECD110
NOTES: continued in and continuation of ULEC180

RECORDING DURATION: 21 min 33 sec

RECORDING DATE: 29.3.2007

NUMBER OF PARTICIPANTS: 13

NUMBER OF SPEAKERS: 4

S1: NATIVE-SPEAKER STATUS: Finnish; ACADEMIC ROLE: junior staff; GENDER: male; AGE: 31-50

S2: NATIVE-SPEAKER STATUS: Danish; ACADEMIC ROLE: junior staff; GENDER: male; AGE: 31-50

S4: NATIVE-SPEAKER STATUS: Finnish; ACADEMIC ROLE: research student; GENDER: female; AGE: 24-30

S5: NATIVE-SPEAKER STATUS: Finnish; ACADEMIC ROLE: junior staff; GENDER: male; AGE: 31-50

SU: unidentified speaker

SS: several simultaneous speakers>


<S1> it's my pleasure to introduce you to doctor <NAME S2> erm , he has worked in in helsinki for quite a few years er er in in human EEG and M-E-G recordings erm , most of us study the brain with a bottom-up approach we look at mechanisms and try to understand how the system works by knowing how its constituents work er <NAME S2> here presents another view a a a top-down approach that that understanding how the system works by by er finding out er rules concerning the system that that even may be independent of the mechanisms underneath so er <S2> yeah </S2> go ahead <NAME S2> [@@] </S1>
<S2> [yeah @@] </S2>
<LECTURE ULEC180 by S2>
<S1> so i have right for the first question <S2> [@yeah@] </S2> [@@] er first er well thank you it was it was really interesting <S2> (thank you) </S2> er first question er is on on the memory impairment in <S2> [mhm yes] </S2> [alzheimer's er] , do you have a a correlation of individual DFA exponents with individual memory scores </S1>
<S2> yeah , and why do you think i didn't show it @@ no okay so er i think i do and it bothers us that we don't see yet @@ a good correlation , <CHECKING SLIDES> erm but why i'm going back is because , what this figure shows is indeed that if you only have 18 alzheimer's patients as we do and they are randomly selected from these genotypic variants my experience with pearson correlation coefficient of only less than 20 subjects is that you don't need many outliers before you don't see the correlation and vice versa er you don't need many outliers <COUGH> before you get a significant one so i think it is very if , yeah so i mean if if an alzheimer's patient when being normal would have a very high score of DFA and then we get this MMSI score from the clinic saying that okay he has alzheimer's a- at that level so it's mi- mild al- alzheimer's yeah then this DFA could have dropped er this much you know or this much it goes this way but but still he could still have a more severe alzheimer's but a higher DFA in (abstract) units compared to someone who is down here and has mild alzheimer and DFA's only has only moved a little bit so i think yeah one of the things we have learnt from the twins is that . yeah that you you would have to somehow be able to er see the changes growth an- and amplitude and and DFA measures relative to what their normal is but right now we don't know what their normal er values should be </S2>
<P:04>
<S1> other questions <P:06> okay erm @i have two more@ <S2> that's good </S2> erm , do you see scaling behaviour also in behaviour , neuroscience centre uses mhm mice behaviour (lab) quite a bit do you think these kind of analyses could be used on the behavioural data as well </S1>
<S2> i er i think so because i mean that it has been used now quite a lot erm as a psychologist in the state er states called <NAME> who has published er at least one science paper on on variability in in reaction times and erm so when you study behaviour in humans actually very often what you see is a normal distribution of of variability and because this is a normal distribution people have viewed it as random variation but if variation in some trait let's say reaction times is normally distributed for some errors the magnitude of some error is normally (xx) that's not necessarily random because you can distribute those events in time in a highly non-random fashion <WHISPERING> excuse me </WHISPERING> and he has shown that that there are also these long-range er correlations there so i don't see why we wouldn't have something like that in in in mouse behaviour , i mean it's also known erm actually most of this analysis er is mostly used in heartbeat analysis er i think that there you also see that the heartbeat intervals show very complex fluctuations over time and there they're looking at timescales up to hours to find ti- er scaling behaviour er up to hours in in humans that might , yeah so if you would er just do that for instance in a behaving animal and assuming that er that there's a relationship between the behaviour an- and the heart rate variability could quite straight-forwardly connect there </S2>
<S1> so so DFA could be a useful er piece of the phenotyping pattern </S1>
<S2> yeah i i think so <S1> yeah </S1> yeah yeah so what what we're doing in CNCR in some of these er shared projects where we look at the same mouse strains so the idea here is that we will screen network oscillations in the hippocampus in vitro er from 80 different in-bred mouse lines and the same 80 mouse lines are going to be screened behaviourally with up to one hundred different behavioural traits so they're developing boxes of different types where they the mouse er meets different challenges novel objects suddenly and they have a computer monitoring system that then quantifies how the different mouse lines react to these different cages and then they make discriminating functions on that actually they have much more success than we have with the oscillation business so er different mouse lines have very different behaviour it measures (xx) as like how much time they spent in the middle of the cage if it's empty versus how much do they walk around the walls er how impulsive are they how curious are they how s- anx- anxious are they towards novel objects and such and erm i think yeah there they have continuous tracking on how they behave and i could i- imagine that temporal analysis on on many of these measures would be a sensible sensible er idea <P:09> i think there's a great hesitation just because it seems like a no- noisy signal and and digging information out of such a noisy signal with an abstract algorithm that always has some er scepticism and that's why for me it was important to see that there was a strong genetic component this (suggestion) to me is a firm proof that it's a biological trait and not just some uncontrolled experimental variable </S2>
<S1> yes yes , the last questions is on on on the spatio-temporal dimension <S2> uh-huh </S2> er ne- ne- correlations in neuronal activity both in space and time are are critical in the development of nervous system </S1>
<S2> (xx) </S2>
<S1> the- they are critical for the de- development <S2> yeah </S2> and and architecture of the on nervous system <S2> yeah </S2> er , do you know if if er the emergence of scaling behaviour or or long-range correlations in space or in time er er when and how do they emerge during the development </S1>
<S2> okay there's er then we do come back to a slide er i have to show <CHECKING SLIDES, P:06> mhm , i mean th- th- that is er . i i'm changing a little bit view on that erm because i- i do nowadays realise that that scale-free fluctuations as for the complex temporal structure that you see to a large extent reflects actually that you have gotten a fractal type of connectivity structure in the network and that is likely to emerge er erm during development so what , so what this theory of self-organising er complex systems says is that if you drive a spatially distributed er system where the units interact non-linearly and often locally if you drive that with an unc- correlated input then you change the activity but because it's a non-linear system it it becomes non-linear because you change the structure a little bit and once you have changed the structure a little bit then the path of the next activation will be a little bit different er and then you get this interaction between activity changing structures and structures bias in where the activity should go and that is then sort of the essence of this complexity view it is that if you have this process going on and that has now been shown in many computational models that n- it's not very sensitive what the exact rules are for how activity should change the structure and it's not sensitive to what whether it's correlated or temporally co- uncorrelated or spatially uncorrelated er energy that is used for driving the system it just intrinsically organises itself into a spatially fractal structure and because of this activity being biased by that spatial structure you get temporally correlated output so that's the story that was , oops , <COUGH> here that if this interaction between space and time makes the system self-organise into such a complex connectivity then at any given point in time when you activate this system you sort of see reveal its developments or history er . but yeah there are still a lot to be done in understanding this because i can see that during development this type of patterns can emerge but it may also emerge (xx) time scales <COUGH> and you can you can the neuron is is very active and gets a calcium imbalance it can suddenly enter period of several minutes <COUGH> where this er has a very reduced excitability and so that you also can get this type of erm spatial pattern not in terms of (hardware) excitability but er or (hardware) connectivity but er spatial pattern of regions that have er high excitability versus low excitability </S2>
<S1> wouldn't this apply also to metabolic pathways or any in- intercellular pathways <S2> yeah </S2> but now- nowadays this pat- the- these pathways are in- investigated as static er <S2> yeah [yeah] </S2> [er] sequences or or trees </S1>
<S2> is that so still i mean er <P:05> i i'm i must admit that i'm not aware about the literature o- of what could maybe be done in that area er </S2>
<P:05>
<S1> maybe somebody here could say if i (xx) the audience is is more knowledgeable than us <S2> @yeah@ </S2> @on on this field@ </S1>
<S2> yeah , erm <P:13> yeah i wouldn't know what to </S2>
<S1> that that that that would at le- at least be a a thing to consider in in in mapping the pathways <S2> yeah </S2> so that activation changes the structure <S2> yeah </S2> and and <S2> [yeah] </S2> [which] changes the activation </S1>
<S2> yeah yeah no but i mean if that theory is still influential but it is er it is cooling off a little bit in the physics society er because they feel that now we have understood the basic principle and now it's it has been passed onto applications so now as you know in f- this is such a generic principle that almost anyone doing science could think of a way that this could be relevant to them so i i think i could almost i i still view the theory as of having broad validity 'cause who's who's not in this situation that there's some spatial inhomogeneity that could influence the temporal evolution of some measure from there erm <P:12> okay [yeah] </S2>
<S4> [just] er you (maybe) said that in future plans you have er you're planning to correlate the DFA with some psychological phenotype of human <S2> uh-huh </S2> er have you already done anything like this or </S4>
<S2> yeah but i have to admit that to kick-start this i arrived at CNCR and was given data and that's why i have kept on working with eyes closed eyes open data <S4> [mhm-hm] </S4> [erm] because it just takes if someone gives you that in an hour it's very tempting to analyse that before you go measure 400 @subjects yourself@ <S4> yeah </S4> with another paradigm but i would like to move on to different paradigms <S4> yeah </S4> er </S2>
<S4> sounds interesting </S4>
<S2> i i think <NAME S1> has data on er two working memory er tasks that vary in how long they have to keep something in memory in order to do the performances <S4> [mhm-hm] </S4> [it's a] continuous working memory task [and i think we are going to look at the data on w-] </S2>
<S4> [have you heard about attenti-] </S4>
<S2> and we are going to look at the data with er working memory task </S2>
<S4> [now] </S4>
<S1> [mhm maybe] [@@] </S1>
<S2> [@@] </S2>
<S4> have you heard about attention concentration test </S4>
<S2> [there was a] </S2>
<S5> [actually that's] my field but she's <S4> [i'm @@] </S4> [my wife and] pushing me to ask you some questions <S2> @@ </S2> okay </S5>
<S2> (A-D-A-C) or </S2>
<S5> erm there's a attention concentration test it's prolonged er er working task where you need to sustain your attention long period at a time <S2> uh-huh yeah  </S2> and i was thinking that perhaps this oscillation measures could also reveal if there's a differences </S5>
<S2> well i think so i mean so the idea is that we will we want to er , they also have data already from children that <S4> mhm </S4> have been scored for their attentional abilities <S5> okay </S5> actually all these 400 the problem is that yeah the data from when they were 17 was in a physical form of which that i can read by my computer <S5> [mhm-hm] </S5> [whereas] the EEG data that were from when they were five were on tapes and we have to find a tape recorder that can er <SS> [@@] </SS> [@convert this data@] into something readable er @@ and that's why we haven't analysed it but that would be o- <SU> @@ </SU> an obvious er way to go </S2>
<S5> yeah because i see that this method could kind of validate the data of the other tests <S2> yeah </S2> with some </S5>
<S2> yeah yeah no that wo- that's definitely erm i mean the effect of attention and working memory is always difficult to pull apart <S5> [mhm-hm] </S5> [er] but it's a very robust phenomenon that parietal alpha oscillations are elevated during the whole retention interval of a working memory <S5> [mhm-hm] </S5> [task] i've just it's disputed whether that is functional inhibition and er something that is not directly r- relevant for for maintaining these items in in working memory or whether it is just er making sure that your actual representation is not disturbed so it functionally inhibits more er (information) to go through the doors of er stream and and to reach the frontal cortex those are the two er options for this effect they would both predict that it will stay elevated but the good thing is that for this type of question the relationship to the dynamics of of your ability to stay <S5> [su- yeah yeah s-] </S5> [focused it does] not really matter whether one or the other is true because it's still a mirror of the neuronal operations [correlating with that effect] </S2>
<S5> [okay yeah (in a way)] </S5>
<S2> er so i i'm curious to hear maybe afterwards <S5> [yeah] </S5> [which] which exact test are you doing because people who study attention are still very much biased towards stimulus evoked paradigms like focus on this one press when there's this happening to the left and then there's suddenly a- an atten- attention catching er deviant stimulus and then they look at the peak 300 and <S5> yeah </S5> things like that and that's how people usually study [attention] </S2>
<S5> [yeah] this is like a over-simplified task you need to you know kind of learn it by heart so it's very simple very very boring you just need to <S2> [mhm-hm] </S2> [keep on] going <S2> yeah </S2> so prolonged working task <S2> yeah [i've] </S2> [so and we have] noticed that when the variance goes you know its vari- (xx) variance of the er of the you know this well test it's getting higher you know this is a kind of indication of the attention deficit <S2> yeah  </S2> and i would s- i would think that you know the same kind of thing is going on here </S5>
<S2> and but you s- do mainly behavioural studies or do you have er your own measure as well </S2>
<S5> er behavioural studies i'm coming from psychology </S5>
<S2> yeah yeah yeah okay you're not combining it with scalp EEG <S5> no </S5> okay , well , well i definitely think i mean do you have a continuous measure of their behaviour somehow <S5> yeah </S5> or is it just what what measure is that </S2>
<S5> so it's a how i would say so there's a like er reaction times rea- reaction time measures with continuous and then it is er translated to the logarithm which describes this <S2> uh-huh </S2> er er this whole sequence of the performance </S5>
<S2> yeah okay , oh but i i i definitely want to move along those lines and yeah i think it's a . yeah </S2>
<S1> perhaps we can continue this discussion [afterwards] </S1>
<S5> [yeah] </S5>
<S2> [yeah yeah] i think we should just stop now and let those two people @who <SS> [@@] </SS> [are curious about this thing]@ shall we do so <S1> yeah </S1> yeah </S2>
<S1> okay so thank you </S1>
