[Music] so we may start this discussion in english by asking you to to react maybe at your um we should keep this short i think but maybe i can launch the discussion with a question to our foreign speakers and to everybody who spoke at this meeting there's one aspect we didn't speak about which is whether it makes sense to scan multiple subjects at the same time because after all economics is about social interactions and so uh i wonder whether any of you want to react to uh this concept which may be just a hype of hyper scanning scanning multiple brains together and seeing how they synchronize or react to each other's signals does everybody know what happens with these hyper scanning experiments so we you'll take two scanners and two subjects and you'll have subjects playing against each other and they'll both be being scanned at the same time and you want you might have you might get different signals from the brain of a trustee versus an investor for example of various different um situations um so when i've spoken to reed about it i i i can't quite make out it seems to me that he saves a lot of time doing it that way and he saves having to have a lot of uh fake confederates but i'm not sure that i have i can work out quite what else he he gets out of it even when when speaking to him because any information that is conveyed between that one brain has about the other brain can only come through the stimulus and so it seems to me that you can't that analyzing one brain's activity in terms of the other brain's activity must by definition be worse than analyzing it with reference to the stimulus that's come from the other brain i mean i don't know do you i that that's so it seems to me logically impossible that you would get any more information by writing the two pieces of brain activity together than you would by going through the intervening stimulus and if you're going to go through the intervening stimulus there's no need to have them in the both in the scanner at the same time you might have one in the scanner with a confederate and then reverse the situation and have the other in a scanner with a confederate and so it seems to me that what you gain is that you gain time essentially but unless unless you can imagine a situation where where somebody makes a very unlikely decision and you want to and and only one subject in 10 might make that decision but you want to image the results whenever people do i don't know but that seems to be well i think i logically agree with you actually on this point uh the vast majority i don't think there is much that can be done that could not be done by having a projection of the subject to just one person sitting in the scanner however i would not exclude that there might be a few surprises especially in the synchronization between two different brains i don't know katrin if you wanted to react yeah just briefly these are experiments that are running in the lab at the moment this is not related to neuroeconomics as such but to imitation so we have two subjects and we record their eg at the same time and then the real question is what happens when subject imitate each other when they breach imitation that just do things with their hands in particular that are not related to each other or when when one is leading and the other is imitating on where they are acting simultaneously and in this particular case of social interactions i think it really makes sense to record the brain activity from the two subjects at the same time so i think that it's definitely a different situation in eeg when you have so many more stimuli about the individual that you have access to because you can see them they're not in a scanner but i but i still don't quite understand why so so what analysis will you be doing that will inform you about um that will be informative that will relate the activity of one brain to the activity of another brain given that that second brain itself does not have access to the to the activity patterns from the first you know it's just a matter that people are reacting one to the other i wouldn't admit that you could do that with a video with someone watching the video uh but then if it's a video then there is not this idea that you are interacting with someone okay but but you could also just interact with somebody but only one of them have eeg on on their head at the same time and you and you you can look at the eg of the emir t or the imitator separately why do you need to call it the brain activity what's the interesting thing that social interactions are is not something stable it's something that the one that imitates is the one that will be leading uh the action um two seconds later so if you really want to understand the dynamics of social interaction it's interesting to have the two brains scanned at the same time when i don't because it's not it's not a stable state that's fine so i can play exactly the same game and this person is sometimes the imitator sometimes being imitated it's fine i can analyze them i can analyze that brain depending on the behavior i just i just i don't understand why you need to know the relationship between the brain activity in one person and the brain temperature another person given that the brain doesn't know that the brain in one person doesn't know the activity in the other the only information he has is the stimuli that he sees it's just a matter of um being better at understanding for instance if you see um a change in the brain patterns of both subjects before a change in the social interactions for instance it suggests that people uh just that one brain has access to the productivity of another no no no it's just that people are able to analyze uh information that are much more subtle that one do what the experts uh that uh rate the uh imitation patterns are doing that's just the point that people are much more subtle than want is just analyzing the body language so you would agree that if you're not past it you would agree that if you're not if you're not passing all that stimulated information but if you're passing only the the the play of the individual then then there can't be any point in doing it but if you but if you if the individuals can actually see each other and they might have access to more information than we as an experimenter have access to then you might see evidence of that in their brain signals so that's what you're saying i think i think we should give the microphone perhaps to peter who wants to add something on have you changed your mind on some issues next question i think should i respond to the hyperspanning scanning question and let me first of all because when we replayed the markets financial markets in the scanner we actually were um thinking about initiative whether we actually needed a market running concurrently with the scanning because we were really concerned whether whether it makes a difference whether the subject is looking at past data which is what we did now as opposed to concurrent where nobody knows neither the experimenter nor the other people in the market nor the person in the scandal where things are going i think that's also relevant for games replaying uh you know putting a subject in a scanner and telling that this person is going to uh play against people who have played in the past may be different from um actually uh concurrent play right yes you mean in a different scans right two people two people in uh but i'm happy that it's different to play against a real person but why do you have to scan that same that person at the same time yes so so go to go to now the market setting right so you uh you you insist on playing a market putting somebody in the scanner concurrently unlike what we did now you may be interested actually in looking at two different persons in that market who played the two different roles like one is a specialist and the other or you know but but i i need to run the market concurrently right so there is a so if you insist on so you you're referring to games okay so you have actually i could say a three-person game right so um if you now insist to play a three-person game and you want to make sure that all the people are are there when one of the persons is scanned i can imagine you may want to scan a second person of the three persons in the game right and then you need it there may be a different concept here yeah i think my view is kind of like stanislaus that um if so first of all the efficiencies are often very substantial um fixing suppose you want to study 40 people trading in markets which may have bubbles in them um you're going to pay for you're going to pay for 40 subjects and if you need groups of five people at a time you know basically all the subjects outside the scanner are typically wasted although we we've we've done some studies for example some that i showed you today where one person is in the scanner and they're playing another person who's outside and one advantage of that is that you can see if the behavior actually changes as a function of being inside a scanner versus not being a scanner which goes a fairly long way toward addressing concerns about people getting tired or it's loud or distracted or something like that so far i don't think i've seen any results reported in which in within the scanner behavior and sort of controlled i call them extras like we're in los angeles and hollywood and the people outside the scan are kind of extras you know who are just there to support the main star which is the person's brain you're looking at and usually the behavior looks very similar which is kind of reassuring probably that's not going to be true for every task and so forth but so far it looks good the way i think of it like at a statistician is if you scan two people versus one you have two time series of data and the the value of hyperscan if there is much besides this efficiencies which we kind of agree on um is if you know the combination of the two time series enables you to do something that you can't you know looking at one time series at a time which in kind of experimental psychology terms means there's some kind of interaction effect of the the time series um i don't have any great ideas of what that might be it could sometimes it could be that simply aggregating it just like we do in um in event related design aggregating up lots of signals might give us better predictions than simply having one time series if they're synchronized you know again it seems kind of second order given how expensive things are to do the other thing would be a case in which the collective behavior really is not just sort of the sum of the individual behaviors so think about something like having a conversation you know if if i'm talking and you're paying attention the only way an outside observer well one way an outside observer will be able to understand see that would be to see joint activation in language regions of the brain or something like that now it could be that self-report would do quite well in that you know or recall measures afterwards you know one person's talking in the scanner and the other person is listening outside and later you ask them to recall parts of the conversation if they can't you say they probably weren't participating in the conversation but to me that's sort of the general recipe you need something which having two or more time series can predict some dependent variable better than a cheaper method and so far nobody's had a really knockout great result but it would be um you know i i tell reid that this is and students you know this is a big prize available for you to thinking of a really cool result and then it would energize people that maybe this is a technique worth worth using but i think it's reasonable to be skeptical until somebody has a really terrific kind of killer app that's that's necessary thank you uh [Music] this is my question in french it's easier um [Music] uh a foreign [Music] [Laughter] is is [Applause] you