Lecture on Face Perception, Methods, and Experimental Design

so to remind you we've been talking last week about kind of doing two things at once asking all set of questions of what we might want to know about face perception in the brain there are some questions but at the same time the agenda has been to consider the different methods available in human cognitive neuroscience and what kinds of questions each one can answer okay so last week we talked about a bunch of them and today we're going to wrap this up talking about tms and animal studies but first i just want to remind you very briefly i won't go through in excruciating detail we talked about behavioral methods which are great for characterizing internal representations as you saw with face inversion effects um and uh and some of the other behavioral data they have major disadvantages which is that with behavior you're just measuring the output it's pretty sparse and from that you have to infer all the stuff that happened in between the retina or whatever your sensory modality is and that would put all that internal mental stuff you have to infer just from the output so it's amazing that that works at all and you have to be really smart to do it and lots of people have been doing that for a long time but it's challenging so why not look inside and one of the best ways to do that of course is functional mri it has the best spatial resolution available for normal subjects but as you guys all seem to pick up on its temporal resolution is lousy and its ability to tell you whether the neural activity you're looking at is causally involved in behavior is like nil okay now at least one of you i only read a few of the assignments uh but at least one of you was confused about which causal role we're talking about and this is actually really important so let's take a moment to talk about this causality the idea of causality if x causes y that means essentially y wouldn't have happened without x or y happened more because x happened when x happens then when x doesn't happen okay so that's pretty basic right so that means if you want to test the causal rule of x on y you have to mess with x okay that's the key challenge okay so with that in mind here's this whole causal chain a stimulus lands on a retina a bunch of neural activity happens and some behavioral output happens there's a whole causal chain there now let's consider what kind of causality we're talking about there's one kind which is that the stimulus causes neural activity in the brain that's a kind of causality that we can absolutely test even if we're measuring that with functional mri because we can mess with the stimulus we can present different stimuli and produce different neural activity okay so in that case we can look at the causal effect of the stimulus on the neural activity no problem that's standard that's what we do pretty much in every um every experiment with erps or functional mri or so forth is that clear okay on the other hand if we want to know this kind of causality from some neural activity we measure in the brain to either a behavioral response or a subjective feeling reported by a behavioral response or something like that that's the challenging part that's the kind of causality that we can't infer from erps or functional mri everyone got that yeah okay it's sort of like obvious and not obvious okay so let's um let's talk a little bit more about that temporal resolution i know i kept saying the temporal resolution of functional memory is lousy but i had run out of time and skip through the key slides so let me back up and do that here this is the bold or mri response as a function of time this is an idealized version of it but it looks kind of like that and sorry these these things are tiny here but these are seconds five seconds 10 seconds so let me show you what that means if you're recording neural activity back here in the first stage of visual processing in the cortex coming up from the eyes which is where the occipital lobe yes what area primary visual cortex exactly okay so suppose that we stuck an electrode in my primary visual cortex and we flashed up a very brief visual display okay here's a visual stimulus on for maybe a tenth of a second bright flashing thing v1 loves v1 also primary visual cortex right okay the neural activity would happen in less than a tenth of a second super fast we know that from work in animals and even some work in humans okay so super fast after the stimulus okay it just goes straight up from the retina the lg and v1 boom there it is so all the neural activity happens right there and ends right there but the mri response is five six seconds later this big sloppy slow thing as the blood sloshes into v1 many seconds after the relevant neural activity okay so what's relevant here is not just that it's delayed but it's big and sloppy right and so both of those things are the reasons why functional mri responses aren't good for distinguishing what happens on a fine temporal scale of say events less than a second okay all right um okay in contrast as i mentioned when you glue electrodes on the scalp or stick these fancy magnetic sensors in the big hairdryer device around your head there you get beautiful temporal resolution but it's like you know the heisenberg principle of cognitive neuroscience you want time you don't get space right okay and similarly here we can measure the causal effect on um scalp response neural responses um but not not the causal role of those neural responses on behavior everyone clear with this okay um okay then i talked about um these rare cases where we can record um directly from the surface of the human brain with electrical activity where we now get both space and time at the same time and the key disadvantage there of course is that it's extremely invasive you have to take a big piece of skull off to get in there and of course that would only happen in the case of people who are already in pretty pretty serious medical circumstances okay so now when we have this incredible opportunity to record this amazing data from the center of the brain does that enable us to make this kind of causal inference from neural activity to behavior yes what do you think yes no isabel is that isabelle yes why are you shaking your head um because it just tells us what your parents are responsible for that's right it's it's cooler it's it's it's fancier it's more impressive than functional mri or erps but it's still the same deal we're just recording responses okay so we can do this causality from the from the stimulus to those neural responses but it doesn't tell us which of those responses are related to behavior yet right i showed you other methods that do but this one alone doesn't everybody got that all right um so then i talked about studying patients with vocal brain damage um and here you really can make a strong causal link between a bit of brain and a behavioral ability you lose that bit of brain you can no longer do that task that's a that's a really direct kind of causal role i talked about double dissociations i gave it short shrift but it's actually really important you should know it a double dissociation is when you have one patient who can do a but not b say recognize objects but not faces and another patient who can do b but not a say recognize faces but not objects and when you have in the literature two cases like that now you're in a really strong position to infer that there's something fundamentally different about face recognition and object recognition in the brain okay so that's really important the senses in which a double association is more inferentially powerful than a single dissociation okay more important means i'm sure to test you on it no it's also important whether i was going to test you on it or not okay and so of course in focal brain damage we we can absolutely infer causal role from a bit of brain to a behavioral ability lose that bit of brain lose the ability yeah okay um and um the the case that i showed you with that amazing movie of the guy getting stimulated in his fusiform face area and seeing percepts of faces on top of whatever he looked at that's a kind of quintessential beautiful example of the causal role of neural activity there we're basically directly manipulating neural activity we're injecting neural activity there electrically and looking at the behavioral and cognitive result that that occurs the guy sees a hallucinatory face okay okay now that is amazing data but as i mentioned we it's they're very rare we have no control over it when we get those data we celebrate and are all excited but mostly we don't get those data and so um and plus those people have serious problems with their brains that's why their brains are being opened up so is there any way to test the causal role of a particular part of the brain in a normal subject who doesn't have their skull open for neurosurgery and who has not had brain damage well there's one way and that's called transcranial magnetic stimulation okay so in transcranial magnetic stimulation you take a coil of wire about yay big that's you know tight wrapped coil of wire embedded in plastic connected to a ginormous capacitor and you hold it next to your head of course that's what you would do if you're a neuroscientist and you discharge and make an enormous current through that coil that's very uh very strong and very brief the whole thing lasts less than one millisecond and you guys know from 802 another case of the right-hand rule coming to our service you have a hell of a current going in a coil what's going to happen in brain tissue underneath increase the magnetic field electric field yeah exactly and so you'll get electric field perpendicular to the coil sticking right into the brain like that and what do you think happens if you stick a big huge transient electric field boom into your head like that is another magnetic field uh yeah you're right right hand rule's magnetic field i was thinking i was misremembering right electric current makes magnetic field right it was a long time ago i took 802. i did just a long time ago anyway for current purpose it doesn't matter either either would do it actually there's a variant of this where it's an electric field but it's debated how well that works okay anyway what happens is you affect neural activity and tissue right underneath the skull right okay um so um if you want to see a picture a video of that happening there's a video of me getting zapped with tms on my website you can check it out it's kind of ludicrous yes question what's the spatial resolution oh we're getting there we're getting there okay here's an early version of this to generate these very strong and brief magnetic fields they had these stacks of coils like this and they rotated them around it's a little crazy here's a more recent version it looks like a big torture device but it's actually no big deal the guy's just holding his head on a chin rest hold his head still and there's a person holding the coil next to his head like that and so that enables us to briefly and somewhat selectively disrupt a little patch of cortex there by sticking in this big random field uh now spatial resolution is not amazing maybe one two centimeters something like that okay it's better than you might guess for such an incredibly crude device like something people would have done 100 years ago and yet we still do it today you can also use a lovely method where you scan the subject with functional mri first find a particular functional region that you're interested in in that person's brain remember these things can vary in their exact location across subjects and then find a whole way to register externally on the on the scalp what is the closest spot to that region you found in their brain previously with functional mri and stick the coil right there and exactly titrate its location with reference to that brain image okay so that makes this whole enterprise more worthwhile so what can tms tell us about face perception well here's the problem here's my fusiform face area that guy right there it's a few centimeters in from the scalp from the skull so that's a drag unless we opened up my head we can't reach it there with the tms coil believe me the first time i had a chance to use a tms coil the very first thing i did was stick the coil there crank it to the max and try to see what would happen not a damn thing happened it was very disappointing i knew lots of friends who tried the same thing it was the most obvious thing it just doesn't work it's too medial yeah but there was a question over here a moment ago uh if you use tms like near someone's brain stem yeah that wouldn't be so smart luckily the brain stem is kind of deep in there so if you were really stupid and stuck it down i don't know way in here you might be able to cause trouble but mostly people don't stick it back there and actually the subjects won't let you anyway because there's a lot of neck muscles and it really hurts when you do tms over muscles and so if anybody had such a stupid ideas to try to set the brain stem the subject would probably like object immediately before they got very far with it because because what hurt and you guys are all probably wondering how safe is this um it's not totally clear there have been lots of studies in animals where they you know zap a rabbit you know a hundred thousand times or something like that and say well rabbit seems fine hops around you know and the best they can do in animal studies when i first used tms around 20 years ago i i read a few basic safety studies and i thought god i don't know and i and i but i also realized that if you look at the papers the initials of the subjects were the same as the authors so i called them up and i said hey you know tell me honestly like did you guys ever notice any ill effects from getting zapped and the guys talked to said yeah i've been zapped about 10 000 times and i never noticed anything except for one thing after a whole hour of getting zapped it gave me a hell of a craving for ice cream so i decided okay i can do i can live with that we got it through the human subjects committee and we do a you know not a lot but some tms in my lab and i'm probably but now been zapped at least as many times as that kind and i guess you guys can judge for yourself so you don't have the before conditions so it's a little hard anyway as far as anybody can tell it's a perfectly safe yes i would say there are some contraindications if you are prone to seizures medications so if you ever sign up for tms study read the five print good good point yep okay so back to this um it will be lovely to zap that guy uh but it's too hard to reach okay so then this guy david pitcher came along and he had a very good idea and my um you know 1970s synopsis of his idea paraphrasing stills is if you can't zap the region you love love the region you can and so uh pitcher said hey what about that other guy there we haven't talked a lot about it it's sometimes called the occipital face area i think of it as a kind of crappy version of the ffa it's kind of face selective it's not as face selective it's more variable so it's not as fun to study but it's there in most people i have a damn fine one i have to say many people do and it is right out there next to the scalp just asking for it right okay um so here's what david pitcher did he gave subjects a uh you need a behavioral task right because in this case we're testing the causal role of a bit of brain on behavior so we're going to measure behavior and so what is our task okay um so here's his task sorry it's a little tiny but this is this is one trial time is going this way you present a face there's a brief interval you present another face and the task is just are those two faces same or different okay it's kind of your basic face perception task but then what you can do is you can zap the occipital face area at different time in you know during presentation of that second phase and you can do it at different time intervals remember its effect is very brief the actual magnetic change is less than a millisecond okay so here's what david pitcher found in that study this is accuracy at this same different matching task when you stimulate the right occipital face area versus vertex that means you stick the coil up here which is pretty far away from face regions it's kind of a controlled condition not a perfect one but better than nothing by the way tms usually doesn't hurt unless you stick it over muscles you stick it over the frontal lobes and i don't know every time i try to disrupt my language abilities it hurts too much because there are muscles up there but most places like the top of the head there aren't muscles and it doesn't really doesn't really hurt but it still makes a loud cracking noise and it's kind of like somebody went like that so you might imagine you need a control condition right because if people are uh that also has a tms pulse right if you bang somebody on the head when they're trying to do a task you probably disrupt their performance so you need to bang them somewhere else to see if it's specific to that location okay so um okay so here's a little effect of accuracy it's not a huge effect size okay so here it's going from you know 85 correct to 70 percent correct when you zap occipital face area compared to vertex everybody get what's going on here so that's good that tells us something zapping here messes up face perception more than zapping here okay okay so that tells us something about causal role but what else would you want to know that's a beginning but having just learned what i told you about tms what else could you do that would tell you more yeah do you see faces ah well that's a good question it wasn't what i was fishing for but it's a very good point so this shows disruption but i showed you with that video before that if you electrically stimulate the ffa you see a face well unfortunately nobody has reported that when you zap a face area you see a percept of a face boy that would be fun if true but it doesn't work and there's much debate about why it probably has to do with the fact that your ability to target just that region is less good than it is with direct stimulation there are many reports and many published studies where if you zap v1 you see a flash of light okay i don't see the damn flash of light i've tried and tried and tried and you know people in my lab who i trust promise me they actually see it isn't just bs but i don't know i don't see it anyway so probably the question of when you get disruption and when you get a positive percept is a very interesting complicated one i think it will ultimately have to do with how those batches of neurons not only respond to faces or light or whatever but how they code for that information such that when you put a big um kind of artifactual non-biological signal in there will it have any meaning that the subject can interpret i don't know if that's helpful it's not i i don't i think nobody really understands that when you'll get a positive persecutor but i hope you can at least understand that at least if you mess with it and muck it up you can disrupt that logic is clear when you will be able to actually stick in a signal and get a positive coherent percept as a is a is a more subtle thing okay okay what else would you want to know yes whether it was object perception or not absolutely absolutely all we're showing here is it's messing up face perception maybe the guy can't see here right maybe he's just globally blind maybe he'd have the same problem with object perception absolutely the assigned reading for wednesday shows exactly that experiment okay what else would you want to know remember a tms pulse lasts less than a millisecond that enables us to ask a whole interesting kind of question what else could we find out yeah there's the x-axis yeah yeah i'm sorry i don't mean to be insulting your intelligence you're probably sitting there saying this is too obvious that's all i'm talking about you can zap at different times and ask when is information going through there when is that region playing a causal role in behavior and here is a very beautiful data that david got and there's basically no effect at any point other than that interval between uh 8 60 and 100 milliseconds okay so that's cool tells you that's when that region is likely engaged in processing make sense okay um and is it char dual yes already made the point i was going to ask you guys does this tell us this region is specifically involved in face perception absolutely not we'd have to test other things right could affect everything you every visual percept right okay so you can read more about that all right just so to collect all the advantages it gives you strong causal evidence that a particular part of the brain is involved in perception or behavior uh it has good temporal information unlike studying patients with vocal brain damage and it is the only disruption method that can be used in normal humans okay so that's why even though it's so crude and rudimentary we use it because it's the only thing that kind of fills that niche a couple other unimportant things spatial resolution isn't as good as we'd like but it's surprising how much you can learn nonetheless um and doesn't reach very far below the scalp although ed boyden the amazing ed boyden is working on a crazy new version of it that might okay so where have all this this kind of menagerie of methods gotten us to i won't go through all this in detail we listed all these questions before i gave you some of the answers from previous methods the ones we've just talked about show for example that the fusiform face area or the occipital face area in the case of tms are causally involved in face perception apparently not in object perception pending the paper you're going to read and so that's important because it says when we try to come up with you know theories of how face recognition works we might think about having a different theory for face recognition from our theory of object recognition right okay so this is all magnificent and wonderful but i finessed this list of questions so that the methods would be able to address them at least a little bit and i sneakily left off a whole suite of other questions that are extremely important arguably more important that those methods don't address okay so we want to know not just that a region responds to faces we want to know exactly what is represented in that region or other regions that respond to other things we want to know what is the neural code for faces okay we want to know what are the actual computations that go on in a given region how do they unfold over time and how do those computations produce the representations and behavioral abilities that we measure we want to know what are the actual anatomical connections i showed you that little occipital face area right nearby but discontiguous from the fusiform face area i've wanted to know for 20 years whether those damn things are connected anatomically shockingly we still don't know that okay we want to know what is the causal role of each region in perception i showed you a few ways that we get little bits of data kind of sort of but there's a lot of cases where we don't so um and we and we want to know how does all this stuff get wired up over development right what is the role of experience do you need to see faces to wire up the face region or is it there at birth before you ever see a face right the sad truth is for the most part we don't have good methods to answer these questions in humans okay so that's just a big bummer but it's true most of these questions can only be answered by research and animals or can be best answered by research in animals okay so i'm going to take a moment to talk about ethical issues in animal research just to note that i think there is an issue and i'll say that it's not unreasonable if you have qualms i noticed an earlier lecture i started talking about recording from animal brains and i didn't have time at that moment to mark this but i do think it's important if it makes you uneasy that's totally legitimate you should think about that and respect that and think hard about whether that's you know what you make of that unambiguously causing animals pointless suffering is just completely unacceptable okay so i think we can all agree on that and i think there's a very difficult trade-off between avoiding suffering in animals and research that has saved countless lives people can legitimately come down on different sides of this uh but many lives have been saved including mine based on animal research that enabled treatments that were life-saving and a few things to think about to help you inform how you handle that trade-off first of all know that animal research in the united states is very heavily regulated okay so um animals receive excellent vet care shockingly better than probably lots of citizens of this country that's another topic um also there's a very major emphasis on avoiding pain right so i think it's probably generally true that it's infrequent that lab animals suffer a lot of pain researchers and vets are very careful to avoid that so the bigger issue is not so much are the animals physically suffering from pain per se but what kind of life is it to live in a lab and be a lab animal and i think that's a legitimate question um for monkeys at least where i at least and maybe maybe in my speciesist bias being more sympathetic to similar species i don't know if that's legitimate but it's a natural increasing um they're increasing efforts to improve the quality of life for monkeys in in labs many monkeys are now housed in social groups where they can hang out with their families and that certainly improves their quality of life many monkeys basically play video games all day into carlo lab they're studying visual perception and what do they do they get the monkeys in there basically doing visual tasks in in exchange for juice rewards not all that different from what probably lots of you guys do now maybe they'd be happier in nature probably much of the time they'd be happier in nature but i think that's complicated too nature can be pretty nasty so it's not totally obvious that quality of life in a random lab is worse than quality of life in nature the third point i'd say is that the benefits of research are forever you discover something major about how brains work that's forever right so you got to amortize whatever cost of animal suffering there is against the foreverness of that insight um and so in my view not that you need to agree but in my view animal research is vastly more justifiable than things like eating meat or buying leather right which is just transient entertainment or convenience right um so anyway you guys i encourage you all to think hard about this and to come to different conclusions i just wanted to note that these are um issues that um that are worth thinking about um that said the methods in animal research are breathtaking and they get more and more breathtaking every day in this building people are constantly inventing astonishing new ways to answer all kinds of questions and i wanted to give you just a gist of some of the kind of stuff that you can do to answer that list of questions that i said we can't really answer in humans so just very briefly this used to be a whole lecture but i just decided to cut it to one slide very briefly about 10 plus years ago these two amazing people doris tau and vinric freyvault who mark my words will get a nobel prize someday or at least they should and they might they popped a monkey in a scanner and did the very same experiment that we do on humans okay so here's a monkey brain again the cortex is unfolded so you can see the whole surface the dark bits are the bits that used to be inside a fold the little yellow patches are the patches that respond more to faces than objects just analogous to the ffa in humans but there are six little patches in monkeys okay so that's kind of so far that's like okay fine monkeys have them too that's cool but the thing is because it's a monkey you can then stick electrodes straight into that region right there and you can record from hundreds of neurons in that region and you can record the response of each of those hundreds of neurons to hundreds or thousands of stimuli you can characterize the neural code for faces in monkeys in a way that you just can't for humans okay in fact doris sao published a paper last year called the neural code for faces based on a decade of this research it's quite breathtaking okay um second you can watch those representations change those neural population codes change over time you can see you know at one time point what what the what the code seems to be saying here and then here and then here and you can watch that those codes change over time in each of those regions and you can see different representations in each of those regions it's quite breathtaking you can answer this question of what are the anatomical connections between these regions with a whole bunch of different methods that i won't go through here but you can actually answer what you know what's connected to what and what these guys have found is that all of those yellow face patches are connected to each other by long-range connections that go through the white matter underneath the gray matter those regions are not connected at all to the intervening other patches of cortex so that set of six little regions is like a kind of computational unit with different hubs that that talk to each other okay and you can see all that in monkeys in a way that we still don't know in humans you can electrically stimulate or disrupt with other methods any one of those patches one at a time you can disrupt them for 50 milliseconds here 200 milliseconds there whatever you like and you can study this whole system over development how does it change from shortly after birth to uh monkey adolescence and you can control experience during development you can raise monkeys without ever letting them see faces and ask whether they're seeing a face is necessary for the development of that region we'll talk about that in a few lectures my point is just that with animal research you can answer vastly uh richer more sophisticated questions than you can ever answer in humans and that's just life yes what's your name i'm esther esther hi so assuming these experiments they showed the monkey faces right back done all different ways remember monkeys see other monkeys but they see a lot of humans too and monkey face patches respond pretty similarly to human faces and monkey faces human faces respond pretty similarly to human faces and monkey faces too even if you don't like work in a monkey lab okay so just to say that there are loads of other methods and we'll get these later in the course okay so that snake assignment i hope that seemed i hope that i thought you guys for the most part did did very well and did kind of exactly the kind of things that we had in mind um and i just want to go through a few bits of terminology because i realized some of you who messed up the wording i hadn't really fully explained what the different words mean okay so first of all there's this like incredibly boring words of independent variables independent variables and frankly i didn't know which was which until i started teaching this stuff a few years ago but the concept is really important right an independent variable that's a factor that you the experimentalist manipulate and change so that you can then measure what effect it has on a brain or behavior the effect you measure is the dependent variable okay it's the independent variable is called the independent variable because you the experimentalist get to mess with it get to manipulate it okay the dependent one you're measuring its dependence on the independent one okay this is just basically any experiment you muck with something in in in the world and you measure the consequences the thing you muck with is independent variable the mucky the thing you measure the effect on is the dependent variable make sense okay all right so for example the bold response that's a dependent variable and pretty much all the experiments we'll talk about here all right the hypothesis most of you got that the hypothesis is the the statement about the world that you're trying to figure out if it's true in your experiment okay um a prediction most of you got this but let me just say prediction is supposed to be extremely precise it's the exact statement of what you will see when you measure your dependent variable if the hypothesis is true what is the crucial thing you have to look for in the data you measure that tells you if the hypothesis is true or not and the prediction is what you will find if the hypothesis is true okay confound we haven't talked about this yet a confound is a difference between your conditions that you're manipulating other than the one you intend to manipulate okay and hence confounds give you alternative accounts case in point we compare the response in the brain when people look at uh faces versus when they look at a bunch of random objects the fact that the faces have more curvy surfaces or are animate or are more interesting those are all confounds with respect to the hypothesis that that region is responding specifically to faces everybody got that okay it's very common amid undergraduates to use confound to mean anything bad about an experiment that's not right a compound is a very particular thing it's another dimension that co-varies with the thing that you care about it's like a nuisance variable that's correlated with the thing you're manipulating and hence is giving you a difficulty inferring a clean inference from your data okay all right a contrast we talk about activations in the brain like those little yellow patches i showed in monkey brains a moment ago that's a that shows you the bits that responded more in functional mri when that monkey was looking at faces than objects the contrast is faces versus objects right it's like it's a looking for a higher response in one condition than another makes sense okay these should all be fairly clear i just know that not everybody got this okay now the point of a contrast is to isolate a mental process okay so let's talk about that for a second um so how do we decide what contrast to use okay well first thing you have to do is get clear about your hypothesis state it explicitly most of you guys did that really well often your hypothesis with functional mri at least will concern a particular mental process that you're studying okay like face recognition okay now remember importantly i said this briefly way back functional mri can only tell you about differences between two conditions right the absolute number you're going to measure the mr signal intensity in one condition say when people are looking at faces and it's going to be something like 726 and it's totally meaningless okay that's just how strong the mri signal is from that point it doesn't mean a damn thing on its own okay but then if we also measure in that same part of the brain uh the mr signal intensity when the subject is looking at objects and at 720 then now we're in business okay all right so everything is a difference so that means that in any imaging experiment you'll need to compare two or more conditions okay one condition will never get you anything okay and if you want to isolate a particular mental process you need to turn that mental process on or off or you need to vary how strongly it's turned on okay this is all in the surface of how we're going to decide what contrast to use that's our goal is to turn on or off one little thing okay and here's the problem if i told you okay look at my face and don't don't process you know low level visual information and don't think about what i'm saying just see my face like it's like what you can't do that right there's a whole processing chain you can't just do one little mental process at a time okay and so that means we can't just turn we can't just have a task where you do only mental process x and a task where you don't do mental process x if you're not doing other stuff okay so what that means is we need to choose two tasks each of which has lots of mental processes but that differ in only one okay so and then we can compare those two so this is called subtraction logic and it comes from work over 100 years ago in cognitive psychology and people who are just measuring behavior this dude donders he's a dutch physiologist and he invented the subtraction method to measure reaction times in humans way back and so with functional mri we're doing the same thing so we're going to come up with two different tasks which involve the whole suite from input to mental processing to output and yet we're going to try to make them differ in just one particular mental process everybody put the program here okay all right um so what you aspire toward in the contrast that you choose is something called a minimal pair right so the idea is we're gonna have these two tasks that are identical in every respect except for that one thing we care about okay so here's a task and here's a task this one involves snake perception and this one is identical to this one except for snake perception okay that's what we want okay uh and if you get those two things that's called a minimal pair and this is the single most important thing in experimental design all the other stuff like you know how you arrange your stimuli over time and all that kind of stuff okay it matters a little bit but this is the crux of the matter what are those conditions and are they the right kind of minimal pair okay and you guys got the gist but i felt like most of you didn't really engage okay what exactly were those non-snake conditions right so that's really the crux of the matter okay okay so the most common problem with imaging experiments is not that the scanner wasn't as fancy as it could have been or they didn't use the latest cutting edge analysis method the most common problem is that people's contrast their conditions were not designed beautifully enough to isolate a single mental process okay okay that is that the conditions were not minimal pairs any other difference between the two conditions other than the one you intend is a confound okay all right so let's engage on this now if we ran a whole experiment um only on male subjects is that a compound no why not as well because it's not a difference between the two experimental visions yeah it's just a bad design feature or something that limits your ability to draw inferences again like you know sub-optimal design it's not the same as a confound content is this very particular thing okay if all the snake pictures have grassy backgrounds and all the non-snake conditions do not is that a compound yeah exactly a confident right okay so i just said all this i'll stop pouring you okay and the reason that the the grassy background thing as a compound is it gives you an alternative account of that contrast maybe it's grassiness not snakeness that's the key difference you don't know okay all right okay so all of that said minimum pairs are kind of like a platonic ideal of experimental design they're what you aspire toward but you can never really do it right if the two conditions were identical um except for this one little thing they'd be identical right you can never totally pull it off okay but you can track the little ways in which you fail and you can test them one at a time in later experiments okay all right good all right so here's what we're going to do we're going to break into groups and you guys are going to think how to take the kind of designs that you already put together and turn them into actual experiments which is going to require deciding on a whole bunch of other things and then we're going to discuss the things you come up with okay what are the exact conditions you'll run in your experiment so we could spend a whole class talking about this so i'd love to hear like your best ofs but i don't want to engage on that for a whole class like you know a lot of the keys some of you guys had very clever non-snake conditions uh to test you know to get close to minimal pairs i want to hear about those um but then beyond that here's something that probably none of you mentioned it's understandable i don't think i said much about it what are subjects doing in the scanner are they just lying there and the stimuli are just flashing up and they're going dum dee dum dee dum are they doing something with the stimuli go think about what you would want to have happen okay so what is the task third some of you mentioned baseline conditions but didn't really say what they are what would a baseline condition be and do you want them or is it a waste of scan time think about that okay okay next suppose you get to scan 10 subjects for one hour each now think about how that design is actually going to go are you going to assign different conditions to different subjects so these five people will see all the snake images and these five people will see all the non-snake images or you're going to have snakes and non-snakes within each subject next it's nice to to not make the subject do their task non-stop for an hour we usually give subjects breaks so we break an experiment into pieces of three to ten minutes or whatever i wrote yeah um and so those are called runs so think about how you wanna allocate those conditions to runs okay and how many runs will you include and then think about what's going to happen within each run so if you're going to have multiple conditions within a run are you going to stick all the snake conditions in the first half and all the non-snake conditions in the second half if not why not um and if there are multiple conditions within a run yeah are you going to clump them all together or interleave them randomly and what are the trade-offs there and what is the order of conditions okay it within a run uh and we won't get to number 10 for the moment okay so we're going to break you guys into four groups and you're gonna talk amongst yourselves and try to come up with your best answers to these in like five ten minutes something like that and then we're gonna pull your thoughts on this when we get back okay okay so part of my agenda in doing this is just to break up the monotony of me going blah blah blah because experimental design is like it's important but it's not the most riveting thing the other thing is experimental design is basically just organized common sense and so most of this stuff you guys just answered all these questions just by thinking about them you need to know a few things about the the methods but really in experimental design the the biggest the best guideline the best way to think about design is think about okay you're the subject you're lying in the scanner you're doing that does that work are you actually going to be doing what you're supposed to be doing are you going to be you know selectively turning on and off like this one little mental process you care about are you doing a million other things like falling asleep and getting bored and all of that and predicting what's going to happen next and all that kind of stuff okay all right so let's just take a few of um a few examples what were some of the what were some good kinds of control conditions that is non-steak snake stimuli that are good to compare uh to snakes that maybe aren't perfect minimal pairs but that get part way there i saw a few just in the few papers that i looked at yeah i gotta i'm sorry i've asked you your name like six times but i'm gonna on my trusty sheet tell me again how you say it okay okay okay uh so for ours we compared snakes to horns to worms yeah because they have like really similar shapes awesome and they're both both animate that's great love it what else who else had a good controlled condition or who had an interesting control condition yes sorry your name is lauren lauren yes okay yep our group had pretty much like the same baseline background and we would just superimpose images of different objects on it so that roommate like remained consistent throughout uh-huh and the background was like what forest floor uh-huh okay so you stick you know a toaster on the forest floor or something like that versus a snake or something yeah the idea was more like other animals or like um stuff that would make more sense that's good that deals with the grass confound problem right absolutely very good what else david you had interesting ideas well we we were talked a lot about animate versus inanimate things so like comparing it to a garden hose yes garden hose love it i actually ran this experiment a bunch of years ago and we used a garden hose or a bunch of garden hoses coiled up in the grass we tried to make them slither and all that anyway but garden hose is great say more you had other good ideas in your yeah we also we talked about uh so my body feels like looking at videos like with motion why would you say why oh because um uh you know when you see the snake it kind of slithers and has this very distinctive thing that is that it feels like the motion is what creeps me out when i see this thing totally me too and if you if you had a rigid thing that was looked like a snake but it was just sort of sliding rigidly then wouldn't really creep me out exactly this is a key insight right so think about you know if we're interested in how you perceive snakes we want to we want to know not just how you do it in some weird lab environment we want to know how you'd actually do that the whole reason to choose snakes is it seems like something that could be biologically relevant there might be special hardware when i'm out hiking and i see even a curved stick i like jump and shriek before i can censor myself it's horrible i find it very embarrassing it's not consistent with my self-image but i have no control over it just happens and so i thought i've thought for a long time there's some damn bit of my brain that's making me do that and it pisses me off and i'm going to find it while we looked and didn't find it but anyway you go from those intuitions right often your own introspections are very informative and your i think your intuition is exactly right there's a very characteristic motion that snakes have and it could be that that's the cue so then the trick is you have slithery motion versus what i don't know it's hard what other kinds of motions could you have right so you could have just different rigid motions where it's not slippery what is that like changing shape like it's just sliding or rotating right right exactly anyway all those are all good ideas good um so what should the subject do in the scanner should they lie there and go dum dee dum dee dum should they do a task if so what task oh if you guys don't volunteer i'm going to start calling on people even though as the jenkins study showed it's nearly impossible to look at these damn photographs and figure out who's who um david in the back task task or no task what tasks yes we talked about having the subjects like find like find a way to indicate that they're you know paying attention yeah and not just dosing off right so one idea was to have them you know like essentially indicate if they saw us make the dinsius thing but that brought up problems like well we don't want them thinking about snakes the entire experiment also if they're going to tell you they're seeing a snake maybe by pushing a button then on the snake trails they're pushing a button and the non-snake trials are not well on the non-stick trails they will push another ah well then they have two different motor responses so then we would have them you know run the experiment again but good good smart very nice ideally we might just have to perform a task that's completely unrelated to looking at snakes or thinking about snakes just so that there's nothing affecting you absolutely and the point you made before is a good one if you're looking for snakes all the time maybe even if it even if it's apples and dogs you're thinking is it a steak is it a snake and maybe you're using that region and it's like ah that's a mess right absolutely yeah all right so this is a common challenge in experimental design and these things don't have clear right answers what i want you to do is just see the trade-offs you know on the one hand just passive viewing lying there it's kind of good in a way the things are just impinging on your sensoria and it's doing whatever it will do but the downside is subjects fall asleep and get bored and you don't know if they're awake right so that's a problem all right okay and but the key thing is whatever the task is you don't want the task to engage asymmetrically with a stimulus condition because then you're building in a confound right right so in the group i was in we were talking about well you could have people well we were talking actually about faces and objects in that case you could have people name the things um but you know if they're naming snakes versus non-snakes it's not very good if they're going snake snake snake snake snake you know dog toast or apple right that's like you know one is easier than the other and more repetitive there are all kinds of problems there all right baseline conditions i didn't really say what a baseline condition was sorry about that um what i meant by a baseline is different from a control condition the control condition would be like the non-snakes contrasted with the snakes right baseline tends to be like a minimalist condition that's supposed to you know turn the brain off okay can we turn the brain off no of course not right but we can aspire toward it we can go far part way out there we can say okay if we're studying vision let's minimize activity in the visual system as best we can okay so you could just have a blank screen that feels like a pretty minimal thing you could have people close their eyes the reason that envisioned experiments people tend to have fixation where there's a tiny dot and subjects are supposed to hold their eyes on it is that in natural left to their own devices people move their eyes a lot several times a second and moving your eyes produces all kinds of activity and lots of neurons and so it's a very kind of active visual thing even if there's nothing on the screen and so staring at a dot is kind of closer to shutting off your visual system even though it's not shutting it off okay so given that most of the contrasts we've talked about are like faces versus objects or snakes versus non-snakes and all the activations that i've shown you guys are contrast between um an experimental condition in a controlled condition why are we bothering with baseline it doesn't even figure in that contrast yes jimmy well like if that region is really selected for complete stakes you can use the baseline as the nsx like control because like you compare the other controller to it if it's really selective for snakes then that non-snake object should respond the same as the minimal awesome everybody get that so that's that was exactly right and this is like a i think a very interesting point so suppose we have remember with mri you just have two numbers so here's the snake response and here's the non-snake response if we don't have a baseline that's all we have two numbers okay and that's fine you know if we run enough subjects that could be significant but now let's think what else we know if we have a baseline suppose we have a baseline of staring at a dot and that's down here we'll call that fixation are you impressed and you run enough subjects so that's significantly different are you impressed exactly why because then if the results are higher up or if the fixation is the second one that you just drew then the response to a snake is twice as much as nonstop exactly does everybody see how yeah that might be significant but kind of who cares right some tiny little ratty ass effect right you know versus if it's like here it's or or even this is the case jimmy was talking about like that no response at all more than staring at a dot to the non-snakes right and yet this response to the snakes that'll be even more impressive they're different degrees of selectivity right not just does it respond differentially but how selective is it oh boy i'm going way over time i'm sorry um so um you guys did great kind of thinking through these things and of course i didn't get halfway through my lecture that's okay we'll roll over the best parts for later and the ones that aren't that fun will just go by the wayside um i will put notes on the rest of some of these things but i think all of you guys pulled out you know just thinking hard about it and using common sense you can see that a lot of experimental design is common sense all right see you guys on wednesday you

so to remind you we&#39;ve been talking last week about kind of doing two things at once asking all set of questions of what we might want to know about face perception in the brain there are some questions but at the same time the agenda has been to consider the different methods available in human cognitive neuroscience and what kinds of questions each one can answer okay so last week we talked about a bunch of them and today we&#39;re going to wrap this up talking about tms and animal studies but first i just want to remind you very briefly i won&#39;t go through in excruciating detail we talked about behavioral methods which are great for characterizing internal representations as you saw with face inversion effects um and uh and some of the other behavioral data they have major disadvantages which is that with behavior you&#39;re just measuring the output it&#39;s pretty sparse and from that you have to infer all the stuff that happened in between the retina or whatever your sensory modality is and that would put all that internal mental stuff you have to infer just from the output so it&#39;s amazing that that works at all and you have to be really smart to do it and lots of people have been doing that for a long time but it&#39;s challenging so why not look inside and one of the best ways to do that of course is functional mri it has the best spatial resolution available for normal subjects but as you guys all seem to pick up on its temporal resolution is lousy and its ability to tell you whether the neural activity you&#39;re looking at is causally involved in behavior is like nil okay now at least one of you i only read a few of the assignments uh but at least one of you was confused about which causal role we&#39;re talking about and this is actually really important so let&#39;s take a moment to talk about this causality the idea of causality if x causes y that means essentially y wouldn&#39;t have happened without x or y happened more because x happened when x happens then when x doesn&#39;t happen okay so that&#39;s pretty basic right so that means if you want to test the causal rule of x on y you have to mess with x okay that&#39;s the key challenge okay so with that in mind here&#39;s this whole causal chain a stimulus lands on a retina a bunch of neural activity happens and some behavioral output happens there&#39;s a whole causal chain there now let&#39;s consider what kind of causality we&#39;re talking about there&#39;s one kind which is that the stimulus causes neural activity in the brain that&#39;s a kind of causality that we can absolutely test even if we&#39;re measuring that with functional mri because we can mess with the stimulus we can present different stimuli and produce different neural activity okay so in that case we can look at the causal effect of the stimulus on the neural activity no problem that&#39;s standard that&#39;s what we do pretty much in every um every experiment with erps or functional mri or so forth is that clear okay on the other hand if we want to know this kind of causality from some neural activity we measure in the brain to either a behavioral response or a subjective feeling reported by a behavioral response or something like that that&#39;s the challenging part that&#39;s the kind of causality that we can&#39;t infer from erps or functional mri everyone got that yeah okay it&#39;s sort of like obvious and not obvious okay so let&#39;s um let&#39;s talk a little bit more about that temporal resolution i know i kept saying the temporal resolution of functional memory is lousy but i had run out of time and skip through the key slides so let me back up and do that here this is the bold or mri response as a function of time this is an idealized version of it but it looks kind of like that and sorry these these things are tiny here but these are seconds five seconds 10 seconds so let me show you what that means if you&#39;re recording neural activity back here in the first stage of visual processing in the cortex coming up from the eyes which is where the occipital lobe yes what area primary visual cortex exactly okay so suppose that we stuck an electrode in my primary visual cortex and we flashed up a very brief visual display okay here&#39;s a visual stimulus on for maybe a tenth of a second bright flashing thing v1 loves v1 also primary visual cortex right okay the neural activity would happen in less than a tenth of a second super fast we know that from work in animals and even some work in humans okay so super fast after the stimulus okay it just goes straight up from the retina the lg and v1 boom there it is so all the neural activity happens right there and ends right there but the mri response is five six seconds later this big sloppy slow thing as the blood sloshes into v1 many seconds after the relevant neural activity okay so what&#39;s relevant here is not just that it&#39;s delayed but it&#39;s big and sloppy right and so both of those things are the reasons why functional mri responses aren&#39;t good for distinguishing what happens on a fine temporal scale of say events less than a second okay all right um okay in contrast as i mentioned when you glue electrodes on the scalp or stick these fancy magnetic sensors in the big hairdryer device around your head there you get beautiful temporal resolution but it&#39;s like you know the heisenberg principle of cognitive neuroscience you want time you don&#39;t get space right okay and similarly here we can measure the causal effect on um scalp response neural responses um but not not the causal role of those neural responses on behavior everyone clear with this okay um okay then i talked about um these rare cases where we can record um directly from the surface of the human brain with electrical activity where we now get both space and time at the same time and the key disadvantage there of course is that it&#39;s extremely invasive you have to take a big piece of skull off to get in there and of course that would only happen in the case of people who are already in pretty pretty serious medical circumstances okay so now when we have this incredible opportunity to record this amazing data from the center of the brain does that enable us to make this kind of causal inference from neural activity to behavior yes what do you think yes no isabel is that isabelle yes why are you shaking your head um because it just tells us what your parents are responsible for that&#39;s right it&#39;s it&#39;s cooler it&#39;s it&#39;s it&#39;s fancier it&#39;s more impressive than functional mri or erps but it&#39;s still the same deal we&#39;re just recording responses okay so we can do this causality from the from the stimulus to those neural responses but it doesn&#39;t tell us which of those responses are related to behavior yet right i showed you other methods that do but this one alone doesn&#39;t everybody got that all right um so then i talked about studying patients with vocal brain damage um and here you really can make a strong causal link between a bit of brain and a behavioral ability you lose that bit of brain you can no longer do that task that&#39;s a that&#39;s a really direct kind of causal role i talked about double dissociations i gave it short shrift but it&#39;s actually really important you should know it a double dissociation is when you have one patient who can do a but not b say recognize objects but not faces and another patient who can do b but not a say recognize faces but not objects and when you have in the literature two cases like that now you&#39;re in a really strong position to infer that there&#39;s something fundamentally different about face recognition and object recognition in the brain okay so that&#39;s really important the senses in which a double association is more inferentially powerful than a single dissociation okay more important means i&#39;m sure to test you on it no it&#39;s also important whether i was going to test you on it or not okay and so of course in focal brain damage we we can absolutely infer causal role from a bit of brain to a behavioral ability lose that bit of brain lose the ability yeah okay um and um the the case that i showed you with that amazing movie of the guy getting stimulated in his fusiform face area and seeing percepts of faces on top of whatever he looked at that&#39;s a kind of quintessential beautiful example of the causal role of neural activity there we&#39;re basically directly manipulating neural activity we&#39;re injecting neural activity there electrically and looking at the behavioral and cognitive result that that occurs the guy sees a hallucinatory face okay okay now that is amazing data but as i mentioned we it&#39;s they&#39;re very rare we have no control over it when we get those data we celebrate and are all excited but mostly we don&#39;t get those data and so um and plus those people have serious problems with their brains that&#39;s why their brains are being opened up so is there any way to test the causal role of a particular part of the brain in a normal subject who doesn&#39;t have their skull open for neurosurgery and who has not had brain damage well there&#39;s one way and that&#39;s called transcranial magnetic stimulation okay so in transcranial magnetic stimulation you take a coil of wire about yay big that&#39;s you know tight wrapped coil of wire embedded in plastic connected to a ginormous capacitor and you hold it next to your head of course that&#39;s what you would do if you&#39;re a neuroscientist and you discharge and make an enormous current through that coil that&#39;s very uh very strong and very brief the whole thing lasts less than one millisecond and you guys know from 802 another case of the right-hand rule coming to our service you have a hell of a current going in a coil what&#39;s going to happen in brain tissue underneath increase the magnetic field electric field yeah exactly and so you&#39;ll get electric field perpendicular to the coil sticking right into the brain like that and what do you think happens if you stick a big huge transient electric field boom into your head like that is another magnetic field uh yeah you&#39;re right right hand rule&#39;s magnetic field i was thinking i was misremembering right electric current makes magnetic field right it was a long time ago i took 802. i did just a long time ago anyway for current purpose it doesn&#39;t matter either either would do it actually there&#39;s a variant of this where it&#39;s an electric field but it&#39;s debated how well that works okay anyway what happens is you affect neural activity and tissue right underneath the skull right okay um so um if you want to see a picture a video of that happening there&#39;s a video of me getting zapped with tms on my website you can check it out it&#39;s kind of ludicrous yes question what&#39;s the spatial resolution oh we&#39;re getting there we&#39;re getting there okay here&#39;s an early version of this to generate these very strong and brief magnetic fields they had these stacks of coils like this and they rotated them around it&#39;s a little crazy here&#39;s a more recent version it looks like a big torture device but it&#39;s actually no big deal the guy&#39;s just holding his head on a chin rest hold his head still and there&#39;s a person holding the coil next to his head like that and so that enables us to briefly and somewhat selectively disrupt a little patch of cortex there by sticking in this big random field uh now spatial resolution is not amazing maybe one two centimeters something like that okay it&#39;s better than you might guess for such an incredibly crude device like something people would have done 100 years ago and yet we still do it today you can also use a lovely method where you scan the subject with functional mri first find a particular functional region that you&#39;re interested in in that person&#39;s brain remember these things can vary in their exact location across subjects and then find a whole way to register externally on the on the scalp what is the closest spot to that region you found in their brain previously with functional mri and stick the coil right there and exactly titrate its location with reference to that brain image okay so that makes this whole enterprise more worthwhile so what can tms tell us about face perception well here&#39;s the problem here&#39;s my fusiform face area that guy right there it&#39;s a few centimeters in from the scalp from the skull so that&#39;s a drag unless we opened up my head we can&#39;t reach it there with the tms coil believe me the first time i had a chance to use a tms coil the very first thing i did was stick the coil there crank it to the max and try to see what would happen not a damn thing happened it was very disappointing i knew lots of friends who tried the same thing it was the most obvious thing it just doesn&#39;t work it&#39;s too medial yeah but there was a question over here a moment ago uh if you use tms like near someone&#39;s brain stem yeah that wouldn&#39;t be so smart luckily the brain stem is kind of deep in there so if you were really stupid and stuck it down i don&#39;t know way in here you might be able to cause trouble but mostly people don&#39;t stick it back there and actually the subjects won&#39;t let you anyway because there&#39;s a lot of neck muscles and it really hurts when you do tms over muscles and so if anybody had such a stupid ideas to try to set the brain stem the subject would probably like object immediately before they got very far with it because because what hurt and you guys are all probably wondering how safe is this um it&#39;s not totally clear there have been lots of studies in animals where they you know zap a rabbit you know a hundred thousand times or something like that and say well rabbit seems fine hops around you know and the best they can do in animal studies when i first used tms around 20 years ago i i read a few basic safety studies and i thought god i don&#39;t know and i and i but i also realized that if you look at the papers the initials of the subjects were the same as the authors so i called them up and i said hey you know tell me honestly like did you guys ever notice any ill effects from getting zapped and the guys talked to said yeah i&#39;ve been zapped about 10 000 times and i never noticed anything except for one thing after a whole hour of getting zapped it gave me a hell of a craving for ice cream so i decided okay i can do i can live with that we got it through the human subjects committee and we do a you know not a lot but some tms in my lab and i&#39;m probably but now been zapped at least as many times as that kind and i guess you guys can judge for yourself so you don&#39;t have the before conditions so it&#39;s a little hard anyway as far as anybody can tell it&#39;s a perfectly safe yes i would say there are some contraindications if you are prone to seizures medications so if you ever sign up for tms study read the five print good good point yep okay so back to this um it will be lovely to zap that guy uh but it&#39;s too hard to reach okay so then this guy david pitcher came along and he had a very good idea and my um you know 1970s synopsis of his idea paraphrasing stills is if you can&#39;t zap the region you love love the region you can and so uh pitcher said hey what about that other guy there we haven&#39;t talked a lot about it it&#39;s sometimes called the occipital face area i think of it as a kind of crappy version of the ffa it&#39;s kind of face selective it&#39;s not as face selective it&#39;s more variable so it&#39;s not as fun to study but it&#39;s there in most people i have a damn fine one i have to say many people do and it is right out there next to the scalp just asking for it right okay um so here&#39;s what david pitcher did he gave subjects a uh you need a behavioral task right because in this case we&#39;re testing the causal role of a bit of brain on behavior so we&#39;re going to measure behavior and so what is our task okay um so here&#39;s his task sorry it&#39;s a little tiny but this is this is one trial time is going this way you present a face there&#39;s a brief interval you present another face and the task is just are those two faces same or different okay it&#39;s kind of your basic face perception task but then what you can do is you can zap the occipital face area at different time in you know during presentation of that second phase and you can do it at different time intervals remember its effect is very brief the actual magnetic change is less than a millisecond okay so here&#39;s what david pitcher found in that study this is accuracy at this same different matching task when you stimulate the right occipital face area versus vertex that means you stick the coil up here which is pretty far away from face regions it&#39;s kind of a controlled condition not a perfect one but better than nothing by the way tms usually doesn&#39;t hurt unless you stick it over muscles you stick it over the frontal lobes and i don&#39;t know every time i try to disrupt my language abilities it hurts too much because there are muscles up there but most places like the top of the head there aren&#39;t muscles and it doesn&#39;t really doesn&#39;t really hurt but it still makes a loud cracking noise and it&#39;s kind of like somebody went like that so you might imagine you need a control condition right because if people are uh that also has a tms pulse right if you bang somebody on the head when they&#39;re trying to do a task you probably disrupt their performance so you need to bang them somewhere else to see if it&#39;s specific to that location okay so um okay so here&#39;s a little effect of accuracy it&#39;s not a huge effect size okay so here it&#39;s going from you know 85 correct to 70 percent correct when you zap occipital face area compared to vertex everybody get what&#39;s going on here so that&#39;s good that tells us something zapping here messes up face perception more than zapping here okay okay so that tells us something about causal role but what else would you want to know that&#39;s a beginning but having just learned what i told you about tms what else could you do that would tell you more yeah do you see faces ah well that&#39;s a good question it wasn&#39;t what i was fishing for but it&#39;s a very good point so this shows disruption but i showed you with that video before that if you electrically stimulate the ffa you see a face well unfortunately nobody has reported that when you zap a face area you see a percept of a face boy that would be fun if true but it doesn&#39;t work and there&#39;s much debate about why it probably has to do with the fact that your ability to target just that region is less good than it is with direct stimulation there are many reports and many published studies where if you zap v1 you see a flash of light okay i don&#39;t see the damn flash of light i&#39;ve tried and tried and tried and you know people in my lab who i trust promise me they actually see it isn&#39;t just bs but i don&#39;t know i don&#39;t see it anyway so probably the question of when you get disruption and when you get a positive percept is a very interesting complicated one i think it will ultimately have to do with how those batches of neurons not only respond to faces or light or whatever but how they code for that information such that when you put a big um kind of artifactual non-biological signal in there will it have any meaning that the subject can interpret i don&#39;t know if that&#39;s helpful it&#39;s not i i don&#39;t i think nobody really understands that when you&#39;ll get a positive persecutor but i hope you can at least understand that at least if you mess with it and muck it up you can disrupt that logic is clear when you will be able to actually stick in a signal and get a positive coherent percept as a is a is a more subtle thing okay okay what else would you want to know yes whether it was object perception or not absolutely absolutely all we&#39;re showing here is it&#39;s messing up face perception maybe the guy can&#39;t see here right maybe he&#39;s just globally blind maybe he&#39;d have the same problem with object perception absolutely the assigned reading for wednesday shows exactly that experiment okay what else would you want to know remember a tms pulse lasts less than a millisecond that enables us to ask a whole interesting kind of question what else could we find out yeah there&#39;s the x-axis yeah yeah i&#39;m sorry i don&#39;t mean to be insulting your intelligence you&#39;re probably sitting there saying this is too obvious that&#39;s all i&#39;m talking about you can zap at different times and ask when is information going through there when is that region playing a causal role in behavior and here is a very beautiful data that david got and there&#39;s basically no effect at any point other than that interval between uh 8 60 and 100 milliseconds okay so that&#39;s cool tells you that&#39;s when that region is likely engaged in processing make sense okay um and is it char dual yes already made the point i was going to ask you guys does this tell us this region is specifically involved in face perception absolutely not we&#39;d have to test other things right could affect everything you every visual percept right okay so you can read more about that all right just so to collect all the advantages it gives you strong causal evidence that a particular part of the brain is involved in perception or behavior uh it has good temporal information unlike studying patients with vocal brain damage and it is the only disruption method that can be used in normal humans okay so that&#39;s why even though it&#39;s so crude and rudimentary we use it because it&#39;s the only thing that kind of fills that niche a couple other unimportant things spatial resolution isn&#39;t as good as we&#39;d like but it&#39;s surprising how much you can learn nonetheless um and doesn&#39;t reach very far below the scalp although ed boyden the amazing ed boyden is working on a crazy new version of it that might okay so where have all this this kind of menagerie of methods gotten us to i won&#39;t go through all this in detail we listed all these questions before i gave you some of the answers from previous methods the ones we&#39;ve just talked about show for example that the fusiform face area or the occipital face area in the case of tms are causally involved in face perception apparently not in object perception pending the paper you&#39;re going to read and so that&#39;s important because it says when we try to come up with you know theories of how face recognition works we might think about having a different theory for face recognition from our theory of object recognition right okay so this is all magnificent and wonderful but i finessed this list of questions so that the methods would be able to address them at least a little bit and i sneakily left off a whole suite of other questions that are extremely important arguably more important that those methods don&#39;t address okay so we want to know not just that a region responds to faces we want to know exactly what is represented in that region or other regions that respond to other things we want to know what is the neural code for faces okay we want to know what are the actual computations that go on in a given region how do they unfold over time and how do those computations produce the representations and behavioral abilities that we measure we want to know what are the actual anatomical connections i showed you that little occipital face area right nearby but discontiguous from the fusiform face area i&#39;ve wanted to know for 20 years whether those damn things are connected anatomically shockingly we still don&#39;t know that okay we want to know what is the causal role of each region in perception i showed you a few ways that we get little bits of data kind of sort of but there&#39;s a lot of cases where we don&#39;t so um and we and we want to know how does all this stuff get wired up over development right what is the role of experience do you need to see faces to wire up the face region or is it there at birth before you ever see a face right the sad truth is for the most part we don&#39;t have good methods to answer these questions in humans okay so that&#39;s just a big bummer but it&#39;s true most of these questions can only be answered by research and animals or can be best answered by research in animals okay so i&#39;m going to take a moment to talk about ethical issues in animal research just to note that i think there is an issue and i&#39;ll say that it&#39;s not unreasonable if you have qualms i noticed an earlier lecture i started talking about recording from animal brains and i didn&#39;t have time at that moment to mark this but i do think it&#39;s important if it makes you uneasy that&#39;s totally legitimate you should think about that and respect that and think hard about whether that&#39;s you know what you make of that unambiguously causing animals pointless suffering is just completely unacceptable okay so i think we can all agree on that and i think there&#39;s a very difficult trade-off between avoiding suffering in animals and research that has saved countless lives people can legitimately come down on different sides of this uh but many lives have been saved including mine based on animal research that enabled treatments that were life-saving and a few things to think about to help you inform how you handle that trade-off first of all know that animal research in the united states is very heavily regulated okay so um animals receive excellent vet care shockingly better than probably lots of citizens of this country that&#39;s another topic um also there&#39;s a very major emphasis on avoiding pain right so i think it&#39;s probably generally true that it&#39;s infrequent that lab animals suffer a lot of pain researchers and vets are very careful to avoid that so the bigger issue is not so much are the animals physically suffering from pain per se but what kind of life is it to live in a lab and be a lab animal and i think that&#39;s a legitimate question um for monkeys at least where i at least and maybe maybe in my speciesist bias being more sympathetic to similar species i don&#39;t know if that&#39;s legitimate but it&#39;s a natural increasing um they&#39;re increasing efforts to improve the quality of life for monkeys in in labs many monkeys are now housed in social groups where they can hang out with their families and that certainly improves their quality of life many monkeys basically play video games all day into carlo lab they&#39;re studying visual perception and what do they do they get the monkeys in there basically doing visual tasks in in exchange for juice rewards not all that different from what probably lots of you guys do now maybe they&#39;d be happier in nature probably much of the time they&#39;d be happier in nature but i think that&#39;s complicated too nature can be pretty nasty so it&#39;s not totally obvious that quality of life in a random lab is worse than quality of life in nature the third point i&#39;d say is that the benefits of research are forever you discover something major about how brains work that&#39;s forever right so you got to amortize whatever cost of animal suffering there is against the foreverness of that insight um and so in my view not that you need to agree but in my view animal research is vastly more justifiable than things like eating meat or buying leather right which is just transient entertainment or convenience right um so anyway you guys i encourage you all to think hard about this and to come to different conclusions i just wanted to note that these are um issues that um that are worth thinking about um that said the methods in animal research are breathtaking and they get more and more breathtaking every day in this building people are constantly inventing astonishing new ways to answer all kinds of questions and i wanted to give you just a gist of some of the kind of stuff that you can do to answer that list of questions that i said we can&#39;t really answer in humans so just very briefly this used to be a whole lecture but i just decided to cut it to one slide very briefly about 10 plus years ago these two amazing people doris tau and vinric freyvault who mark my words will get a nobel prize someday or at least they should and they might they popped a monkey in a scanner and did the very same experiment that we do on humans okay so here&#39;s a monkey brain again the cortex is unfolded so you can see the whole surface the dark bits are the bits that used to be inside a fold the little yellow patches are the patches that respond more to faces than objects just analogous to the ffa in humans but there are six little patches in monkeys okay so that&#39;s kind of so far that&#39;s like okay fine monkeys have them too that&#39;s cool but the thing is because it&#39;s a monkey you can then stick electrodes straight into that region right there and you can record from hundreds of neurons in that region and you can record the response of each of those hundreds of neurons to hundreds or thousands of stimuli you can characterize the neural code for faces in monkeys in a way that you just can&#39;t for humans okay in fact doris sao published a paper last year called the neural code for faces based on a decade of this research it&#39;s quite breathtaking okay um second you can watch those representations change those neural population codes change over time you can see you know at one time point what what the what the code seems to be saying here and then here and then here and you can watch that those codes change over time in each of those regions and you can see different representations in each of those regions it&#39;s quite breathtaking you can answer this question of what are the anatomical connections between these regions with a whole bunch of different methods that i won&#39;t go through here but you can actually answer what you know what&#39;s connected to what and what these guys have found is that all of those yellow face patches are connected to each other by long-range connections that go through the white matter underneath the gray matter those regions are not connected at all to the intervening other patches of cortex so that set of six little regions is like a kind of computational unit with different hubs that that talk to each other okay and you can see all that in monkeys in a way that we still don&#39;t know in humans you can electrically stimulate or disrupt with other methods any one of those patches one at a time you can disrupt them for 50 milliseconds here 200 milliseconds there whatever you like and you can study this whole system over development how does it change from shortly after birth to uh monkey adolescence and you can control experience during development you can raise monkeys without ever letting them see faces and ask whether they&#39;re seeing a face is necessary for the development of that region we&#39;ll talk about that in a few lectures my point is just that with animal research you can answer vastly uh richer more sophisticated questions than you can ever answer in humans and that&#39;s just life yes what&#39;s your name i&#39;m esther esther hi so assuming these experiments they showed the monkey faces right back done all different ways remember monkeys see other monkeys but they see a lot of humans too and monkey face patches respond pretty similarly to human faces and monkey faces human faces respond pretty similarly to human faces and monkey faces too even if you don&#39;t like work in a monkey lab okay so just to say that there are loads of other methods and we&#39;ll get these later in the course okay so that snake assignment i hope that seemed i hope that i thought you guys for the most part did did very well and did kind of exactly the kind of things that we had in mind um and i just want to go through a few bits of terminology because i realized some of you who messed up the wording i hadn&#39;t really fully explained what the different words mean okay so first of all there&#39;s this like incredibly boring words of independent variables independent variables and frankly i didn&#39;t know which was which until i started teaching this stuff a few years ago but the concept is really important right an independent variable that&#39;s a factor that you the experimentalist manipulate and change so that you can then measure what effect it has on a brain or behavior the effect you measure is the dependent variable okay it&#39;s the independent variable is called the independent variable because you the experimentalist get to mess with it get to manipulate it okay the dependent one you&#39;re measuring its dependence on the independent one okay this is just basically any experiment you muck with something in in in the world and you measure the consequences the thing you muck with is independent variable the mucky the thing you measure the effect on is the dependent variable make sense okay all right so for example the bold response that&#39;s a dependent variable and pretty much all the experiments we&#39;ll talk about here all right the hypothesis most of you got that the hypothesis is the the statement about the world that you&#39;re trying to figure out if it&#39;s true in your experiment okay um a prediction most of you got this but let me just say prediction is supposed to be extremely precise it&#39;s the exact statement of what you will see when you measure your dependent variable if the hypothesis is true what is the crucial thing you have to look for in the data you measure that tells you if the hypothesis is true or not and the prediction is what you will find if the hypothesis is true okay confound we haven&#39;t talked about this yet a confound is a difference between your conditions that you&#39;re manipulating other than the one you intend to manipulate okay and hence confounds give you alternative accounts case in point we compare the response in the brain when people look at uh faces versus when they look at a bunch of random objects the fact that the faces have more curvy surfaces or are animate or are more interesting those are all confounds with respect to the hypothesis that that region is responding specifically to faces everybody got that okay it&#39;s very common amid undergraduates to use confound to mean anything bad about an experiment that&#39;s not right a compound is a very particular thing it&#39;s another dimension that co-varies with the thing that you care about it&#39;s like a nuisance variable that&#39;s correlated with the thing you&#39;re manipulating and hence is giving you a difficulty inferring a clean inference from your data okay all right a contrast we talk about activations in the brain like those little yellow patches i showed in monkey brains a moment ago that&#39;s a that shows you the bits that responded more in functional mri when that monkey was looking at faces than objects the contrast is faces versus objects right it&#39;s like it&#39;s a looking for a higher response in one condition than another makes sense okay these should all be fairly clear i just know that not everybody got this okay now the point of a contrast is to isolate a mental process okay so let&#39;s talk about that for a second um so how do we decide what contrast to use okay well first thing you have to do is get clear about your hypothesis state it explicitly most of you guys did that really well often your hypothesis with functional mri at least will concern a particular mental process that you&#39;re studying okay like face recognition okay now remember importantly i said this briefly way back functional mri can only tell you about differences between two conditions right the absolute number you&#39;re going to measure the mr signal intensity in one condition say when people are looking at faces and it&#39;s going to be something like 726 and it&#39;s totally meaningless okay that&#39;s just how strong the mri signal is from that point it doesn&#39;t mean a damn thing on its own okay but then if we also measure in that same part of the brain uh the mr signal intensity when the subject is looking at objects and at 720 then now we&#39;re in business okay all right so everything is a difference so that means that in any imaging experiment you&#39;ll need to compare two or more conditions okay one condition will never get you anything okay and if you want to isolate a particular mental process you need to turn that mental process on or off or you need to vary how strongly it&#39;s turned on okay this is all in the surface of how we&#39;re going to decide what contrast to use that&#39;s our goal is to turn on or off one little thing okay and here&#39;s the problem if i told you okay look at my face and don&#39;t don&#39;t process you know low level visual information and don&#39;t think about what i&#39;m saying just see my face like it&#39;s like what you can&#39;t do that right there&#39;s a whole processing chain you can&#39;t just do one little mental process at a time okay and so that means we can&#39;t just turn we can&#39;t just have a task where you do only mental process x and a task where you don&#39;t do mental process x if you&#39;re not doing other stuff okay so what that means is we need to choose two tasks each of which has lots of mental processes but that differ in only one okay so and then we can compare those two so this is called subtraction logic and it comes from work over 100 years ago in cognitive psychology and people who are just measuring behavior this dude donders he&#39;s a dutch physiologist and he invented the subtraction method to measure reaction times in humans way back and so with functional mri we&#39;re doing the same thing so we&#39;re going to come up with two different tasks which involve the whole suite from input to mental processing to output and yet we&#39;re going to try to make them differ in just one particular mental process everybody put the program here okay all right um so what you aspire toward in the contrast that you choose is something called a minimal pair right so the idea is we&#39;re gonna have these two tasks that are identical in every respect except for that one thing we care about okay so here&#39;s a task and here&#39;s a task this one involves snake perception and this one is identical to this one except for snake perception okay that&#39;s what we want okay uh and if you get those two things that&#39;s called a minimal pair and this is the single most important thing in experimental design all the other stuff like you know how you arrange your stimuli over time and all that kind of stuff okay it matters a little bit but this is the crux of the matter what are those conditions and are they the right kind of minimal pair okay and you guys got the gist but i felt like most of you didn&#39;t really engage okay what exactly were those non-snake conditions right so that&#39;s really the crux of the matter okay okay so the most common problem with imaging experiments is not that the scanner wasn&#39;t as fancy as it could have been or they didn&#39;t use the latest cutting edge analysis method the most common problem is that people&#39;s contrast their conditions were not designed beautifully enough to isolate a single mental process okay okay that is that the conditions were not minimal pairs any other difference between the two conditions other than the one you intend is a confound okay all right so let&#39;s engage on this now if we ran a whole experiment um only on male subjects is that a compound no why not as well because it&#39;s not a difference between the two experimental visions yeah it&#39;s just a bad design feature or something that limits your ability to draw inferences again like you know sub-optimal design it&#39;s not the same as a confound content is this very particular thing okay if all the snake pictures have grassy backgrounds and all the non-snake conditions do not is that a compound yeah exactly a confident right okay so i just said all this i&#39;ll stop pouring you okay and the reason that the the grassy background thing as a compound is it gives you an alternative account of that contrast maybe it&#39;s grassiness not snakeness that&#39;s the key difference you don&#39;t know okay all right okay so all of that said minimum pairs are kind of like a platonic ideal of experimental design they&#39;re what you aspire toward but you can never really do it right if the two conditions were identical um except for this one little thing they&#39;d be identical right you can never totally pull it off okay but you can track the little ways in which you fail and you can test them one at a time in later experiments okay all right good all right so here&#39;s what we&#39;re going to do we&#39;re going to break into groups and you guys are going to think how to take the kind of designs that you already put together and turn them into actual experiments which is going to require deciding on a whole bunch of other things and then we&#39;re going to discuss the things you come up with okay what are the exact conditions you&#39;ll run in your experiment so we could spend a whole class talking about this so i&#39;d love to hear like your best ofs but i don&#39;t want to engage on that for a whole class like you know a lot of the keys some of you guys had very clever non-snake conditions uh to test you know to get close to minimal pairs i want to hear about those um but then beyond that here&#39;s something that probably none of you mentioned it&#39;s understandable i don&#39;t think i said much about it what are subjects doing in the scanner are they just lying there and the stimuli are just flashing up and they&#39;re going dum dee dum dee dum are they doing something with the stimuli go think about what you would want to have happen okay so what is the task third some of you mentioned baseline conditions but didn&#39;t really say what they are what would a baseline condition be and do you want them or is it a waste of scan time think about that okay okay next suppose you get to scan 10 subjects for one hour each now think about how that design is actually going to go are you going to assign different conditions to different subjects so these five people will see all the snake images and these five people will see all the non-snake images or you&#39;re going to have snakes and non-snakes within each subject next it&#39;s nice to to not make the subject do their task non-stop for an hour we usually give subjects breaks so we break an experiment into pieces of three to ten minutes or whatever i wrote yeah um and so those are called runs so think about how you wanna allocate those conditions to runs okay and how many runs will you include and then think about what&#39;s going to happen within each run so if you&#39;re going to have multiple conditions within a run are you going to stick all the snake conditions in the first half and all the non-snake conditions in the second half if not why not um and if there are multiple conditions within a run yeah are you going to clump them all together or interleave them randomly and what are the trade-offs there and what is the order of conditions okay it within a run uh and we won&#39;t get to number 10 for the moment okay so we&#39;re going to break you guys into four groups and you&#39;re gonna talk amongst yourselves and try to come up with your best answers to these in like five ten minutes something like that and then we&#39;re gonna pull your thoughts on this when we get back okay okay so part of my agenda in doing this is just to break up the monotony of me going blah blah blah because experimental design is like it&#39;s important but it&#39;s not the most riveting thing the other thing is experimental design is basically just organized common sense and so most of this stuff you guys just answered all these questions just by thinking about them you need to know a few things about the the methods but really in experimental design the the biggest the best guideline the best way to think about design is think about okay you&#39;re the subject you&#39;re lying in the scanner you&#39;re doing that does that work are you actually going to be doing what you&#39;re supposed to be doing are you going to be you know selectively turning on and off like this one little mental process you care about are you doing a million other things like falling asleep and getting bored and all of that and predicting what&#39;s going to happen next and all that kind of stuff okay all right so let&#39;s just take a few of um a few examples what were some of the what were some good kinds of control conditions that is non-steak snake stimuli that are good to compare uh to snakes that maybe aren&#39;t perfect minimal pairs but that get part way there i saw a few just in the few papers that i looked at yeah i gotta i&#39;m sorry i&#39;ve asked you your name like six times but i&#39;m gonna on my trusty sheet tell me again how you say it okay okay okay uh so for ours we compared snakes to horns to worms yeah because they have like really similar shapes awesome and they&#39;re both both animate that&#39;s great love it what else who else had a good controlled condition or who had an interesting control condition yes sorry your name is lauren lauren yes okay yep our group had pretty much like the same baseline background and we would just superimpose images of different objects on it so that roommate like remained consistent throughout uh-huh and the background was like what forest floor uh-huh okay so you stick you know a toaster on the forest floor or something like that versus a snake or something yeah the idea was more like other animals or like um stuff that would make more sense that&#39;s good that deals with the grass confound problem right absolutely very good what else david you had interesting ideas well we we were talked a lot about animate versus inanimate things so like comparing it to a garden hose yes garden hose love it i actually ran this experiment a bunch of years ago and we used a garden hose or a bunch of garden hoses coiled up in the grass we tried to make them slither and all that anyway but garden hose is great say more you had other good ideas in your yeah we also we talked about uh so my body feels like looking at videos like with motion why would you say why oh because um uh you know when you see the snake it kind of slithers and has this very distinctive thing that is that it feels like the motion is what creeps me out when i see this thing totally me too and if you if you had a rigid thing that was looked like a snake but it was just sort of sliding rigidly then wouldn&#39;t really creep me out exactly this is a key insight right so think about you know if we&#39;re interested in how you perceive snakes we want to we want to know not just how you do it in some weird lab environment we want to know how you&#39;d actually do that the whole reason to choose snakes is it seems like something that could be biologically relevant there might be special hardware when i&#39;m out hiking and i see even a curved stick i like jump and shriek before i can censor myself it&#39;s horrible i find it very embarrassing it&#39;s not consistent with my self-image but i have no control over it just happens and so i thought i&#39;ve thought for a long time there&#39;s some damn bit of my brain that&#39;s making me do that and it pisses me off and i&#39;m going to find it while we looked and didn&#39;t find it but anyway you go from those intuitions right often your own introspections are very informative and your i think your intuition is exactly right there&#39;s a very characteristic motion that snakes have and it could be that that&#39;s the cue so then the trick is you have slithery motion versus what i don&#39;t know it&#39;s hard what other kinds of motions could you have right so you could have just different rigid motions where it&#39;s not slippery what is that like changing shape like it&#39;s just sliding or rotating right right exactly anyway all those are all good ideas good um so what should the subject do in the scanner should they lie there and go dum dee dum dee dum should they do a task if so what task oh if you guys don&#39;t volunteer i&#39;m going to start calling on people even though as the jenkins study showed it&#39;s nearly impossible to look at these damn photographs and figure out who&#39;s who um david in the back task task or no task what tasks yes we talked about having the subjects like find like find a way to indicate that they&#39;re you know paying attention yeah and not just dosing off right so one idea was to have them you know like essentially indicate if they saw us make the dinsius thing but that brought up problems like well we don&#39;t want them thinking about snakes the entire experiment also if they&#39;re going to tell you they&#39;re seeing a snake maybe by pushing a button then on the snake trails they&#39;re pushing a button and the non-snake trials are not well on the non-stick trails they will push another ah well then they have two different motor responses so then we would have them you know run the experiment again but good good smart very nice ideally we might just have to perform a task that&#39;s completely unrelated to looking at snakes or thinking about snakes just so that there&#39;s nothing affecting you absolutely and the point you made before is a good one if you&#39;re looking for snakes all the time maybe even if it even if it&#39;s apples and dogs you&#39;re thinking is it a steak is it a snake and maybe you&#39;re using that region and it&#39;s like ah that&#39;s a mess right absolutely yeah all right so this is a common challenge in experimental design and these things don&#39;t have clear right answers what i want you to do is just see the trade-offs you know on the one hand just passive viewing lying there it&#39;s kind of good in a way the things are just impinging on your sensoria and it&#39;s doing whatever it will do but the downside is subjects fall asleep and get bored and you don&#39;t know if they&#39;re awake right so that&#39;s a problem all right okay and but the key thing is whatever the task is you don&#39;t want the task to engage asymmetrically with a stimulus condition because then you&#39;re building in a confound right right so in the group i was in we were talking about well you could have people well we were talking actually about faces and objects in that case you could have people name the things um but you know if they&#39;re naming snakes versus non-snakes it&#39;s not very good if they&#39;re going snake snake snake snake snake you know dog toast or apple right that&#39;s like you know one is easier than the other and more repetitive there are all kinds of problems there all right baseline conditions i didn&#39;t really say what a baseline condition was sorry about that um what i meant by a baseline is different from a control condition the control condition would be like the non-snakes contrasted with the snakes right baseline tends to be like a minimalist condition that&#39;s supposed to you know turn the brain off okay can we turn the brain off no of course not right but we can aspire toward it we can go far part way out there we can say okay if we&#39;re studying vision let&#39;s minimize activity in the visual system as best we can okay so you could just have a blank screen that feels like a pretty minimal thing you could have people close their eyes the reason that envisioned experiments people tend to have fixation where there&#39;s a tiny dot and subjects are supposed to hold their eyes on it is that in natural left to their own devices people move their eyes a lot several times a second and moving your eyes produces all kinds of activity and lots of neurons and so it&#39;s a very kind of active visual thing even if there&#39;s nothing on the screen and so staring at a dot is kind of closer to shutting off your visual system even though it&#39;s not shutting it off okay so given that most of the contrasts we&#39;ve talked about are like faces versus objects or snakes versus non-snakes and all the activations that i&#39;ve shown you guys are contrast between um an experimental condition in a controlled condition why are we bothering with baseline it doesn&#39;t even figure in that contrast yes jimmy well like if that region is really selected for complete stakes you can use the baseline as the nsx like control because like you compare the other controller to it if it&#39;s really selective for snakes then that non-snake object should respond the same as the minimal awesome everybody get that so that&#39;s that was exactly right and this is like a i think a very interesting point so suppose we have remember with mri you just have two numbers so here&#39;s the snake response and here&#39;s the non-snake response if we don&#39;t have a baseline that&#39;s all we have two numbers okay and that&#39;s fine you know if we run enough subjects that could be significant but now let&#39;s think what else we know if we have a baseline suppose we have a baseline of staring at a dot and that&#39;s down here we&#39;ll call that fixation are you impressed and you run enough subjects so that&#39;s significantly different are you impressed exactly why because then if the results are higher up or if the fixation is the second one that you just drew then the response to a snake is twice as much as nonstop exactly does everybody see how yeah that might be significant but kind of who cares right some tiny little ratty ass effect right you know versus if it&#39;s like here it&#39;s or or even this is the case jimmy was talking about like that no response at all more than staring at a dot to the non-snakes right and yet this response to the snakes that&#39;ll be even more impressive they&#39;re different degrees of selectivity right not just does it respond differentially but how selective is it oh boy i&#39;m going way over time i&#39;m sorry um so um you guys did great kind of thinking through these things and of course i didn&#39;t get halfway through my lecture that&#39;s okay we&#39;ll roll over the best parts for later and the ones that aren&#39;t that fun will just go by the wayside um i will put notes on the rest of some of these things but i think all of you guys pulled out you know just thinking hard about it and using common sense you can see that a lot of experimental design is common sense all right see you guys on wednesday you

Transcript for:Lecture on Face Perception, Methods, and Experimental Design

Transcript for:
Lecture on Face Perception, Methods, and Experimental Design