If I remember correctly, at the time of Hammurabi, teachers had to take responsibility for all their students, including everything their students did, with very severe penalties when they did bad things. I never quite imagined that there would be this many people that I would have to take responsibility for. Thank you, Mitch, for the delightful introduction. I knew this was going to be an after or intra-dinner discussion, so I thought it was appropriate that we started with a culinary item. Now, we have to do a little housekeeping first, and that is to tell you that I have no disclosures currently that I need to reveal. Some of the primary sources are the books that are listed there, and then the other sources will be articles that will be cited in their appropriate places. Now, it all begins with a frog's leg, and it begins in the latter part of the 18th century. It was an experiment done by Luigi Galvani, and it was this experiment which basically established that there was a relationship between the nervous system and electricity. What happened was that he suspended a frog leg on a copper hook on an iron frame, and he observed that when he attached a wire—it happened to be an iron wire—between that frame and the animal, that the leg jerked. He interpreted this as animal electricity. It wasn't what was thought at the time. It wasn't the soul or animal spirits or, for that matter, fluid that was responsible for the nervous system function. Now, he was immediately assaulted, and this is sort of the history of stimulation mapping, that immediately there's controversy with each step of the way. He was immediately assaulted by his contemporary at the other end of the Po River Valley, Alessandro Volta, who said, no, you haven't shown animal electricity. What you've shown is that this is electrochemical electricity, and then each one of them went on to do the definitive experiment. Volta got rid of the animal and showed that he still had electricity, and then Galvani went on and got rid of the iron frame and copper hooks and showed that he still had animal electricity. That didn't stop the controversy, of course, but the result was that each one went on to establishing two very major features of the then nascent study of electricity. Volta's was established much faster. By 1800, he had a battery that he could show. It took until about the middle of the 19th century for the idea of animal electricity to be established, and that was mostly the work on the nerve muscle junction by DuBois-Raymond. But that's the beginning, really, of electrophysiology and the idea that electricity had something to do with the nervous system. Now, there was, of course, an antecedent, and it happened that the last half of the 1700s was a period of great interest in electricity, or at least what we would now call static electricity. The thing that many of you may have heard of in the past were the experiments of Benjamin Franklin with flying a kite in a thunderstorm and trying to electrocute himself. In the process of that, a group of individuals experimented with the effect of electrical stimulation on the decapitated heads of criminals. And the effects were, as are quoted here, so grotesque that the spectators fled, and I was unable to find any more specific discussion of what this was like. But it was nonetheless bad enough that the entire business was forbidden, so stimulation mapping, before it got out of the box, was already legally forbidden. There was one significant finding for the future here that was found at that time, and that was that one of the individuals stimulated through a trephon that he placed in an awake patient and showed that at least wherever his trephon was, the dura was insensitive to pain. Well, insensitive to stimulation, at least. Now, for the next half century and a little more, not much happened with stimulation mapping. And the main problem was that they simply couldn't control the current. The devices they had available for generating electricity were ones that were ill-understood and ill-regulated, and their currents were so large that when in experimental animals they tried to replicate some of this, they got current spread directly to the external muscles, they got some local effects, and they got seizures. And so it was all confounded together and couldn't be sorted out. But in that period, there were two significant advances that are important to what we're dealing with today. One of them was this, and this is the work of Rolandi, because it was a whole period of the flowering of a neuroanatomy, and this is, I think, the first accurate description and naming of the cortical gyri. And this is the work of Rolandi in the first quarter of the 1800s, and he described the vertical posterior frontal and anterior parietal gyri that we all know as motor and sensory, and that were talked about earlier this afternoon, and for which the common name became Rolandi cortex. Now, of course, this so often happens when you make claims of firstness. This was wrong. He had been preceded nearly about 40 years earlier by the neuroanatomist Vic de Jure, but unfortunately Vic de Jure's studies were omitted from his publications due to a printer's error and didn't surface until well after Rolandi's work. Now, at that time, however, as far as function was concerned, the prevailing view was that the cortex had nothing to do with it. The cortex was the location of the common sense, and the common sense was uniform across the whole cortex. There was no cortical motor role and no focal cortical localization, and this remained the general view of the scientific community until, as we'll see, about the 1860s. Now, there were a few people who didn't believe that, and so the next advance during this same period was the concept of cortical localization. It appeared, however, in a rather unfortunate form. This was the work of Franz Joseph Gall in the early part of the 1800s, and he thought that the facilities of the mind were separately localized in gray matter, so he not only recognized that they were in the gray matter of the cortex, but that they were localized. And during that time, the generally accepted view was the only localization was in the basal ganglia and not in the cortex. But what he localized was a little different than what we have been interested in. We're pretty sort of prosaic in our interests. He wanted to know about where hope or justice or sublimity or acquisitiveness or moral and religious sentiments were localized. And that was, in fact, on his original map of cortical localization. There is not a mention at all of motor or stimulation effects. This became enormously popular during the first part of the 1800s. It became popular as phrenology because he had the not totally—that is, Gall and his colleague Spurtheimer— had the not irrational idea that if a piece of brain was more functional in this particular individual, like a muscle developed in someone who was athletic, it got bigger. And we know, at least in kids, that the skull shape is determined by expansion of the brain, and so why wouldn't it locally expand in that region? And then when you felt the skull through the scalp, you would see bumps there. And that was phrenology. So it wasn't entirely illogical at a time when knowledge was so limited. But it became very popular. It was quickly taken over by a variety of quacks and charlatans. One of the popular maps of it was the ones that are depicted on this slide. And in that particular map of the Fowlers in, I think, about 1840, there is language. And that's that red circle that's coming right out of the orbit. Now it turns out that that's the one piece of this whole map that actually has a foundation in scientific observation. The first, apparently, first case report trying to link a brain lesion to a language disturbance was by the French neurologist Bouillot and was an orbital frontal tumor in a patient who had what we would call now Broca's aphasia. Now, of course, this is more than a half century before Broca that that case was reported. So despite the fact that this was another rather unfortunate predecessor of stimulation mapping, the whole phrenology becoming this scam of the quacks that took it over, there was a little bit of real science behind a little bit of it. Now, nothing happened much otherwise with stimulation during that first half century. They couldn't control the currents. The general prevailing view was that localization of things in the cortex was irrelevant. That was probably an accident of the fact that the experimental animal that those studies were based on happened to be a pigeon, which, of course, doesn't have localized cortical function like mammals do. Until the works of Hewling Jackson in the 1860s and 70s. Jackson was not a localizationist initially. He believed the standard view that the cortex was not differentiated. But his studies of patients with different patterns of seizures arising in what we now call motor cortex convinced him that there was focal localization in the cortex. He worked out the whole, from just the observation of the patients, he worked out the whole homunculus that we now describe in humans. And this opened, this excited the community enough, the scientific community enough, that there were now a number of, in the next few years, a number of studies in experimental animals that showed localized evoked motor activity from cortex. First by Edvard Hitzik and his colleague Frisch on the evoking movements from dog cortex in 1870. Three years later, in a much more carefully done study in monkeys, David Farrier showed quite detailed localization of motor activity. Now, neurosurgeons were not at all to be left behind. And neurosurgery so rapidly jumped into this new way of localizing function within the cortex. And the first stimulation of an alive human brain, what is said to be at least, was the work of Robert Bartholow from Cincinnati in 1874, who reported this case, the one that's pictured on the left. What Bartholow's patient had a lesion, it was described as an epithelioma, if I remember, that had eroded through the skull and scalp and had exposed the skull at the vertex. So the sagittal sinus was in this and both the medial parts of both hemispheres in this unfortunate patient. I assume it was probably a meningioma that had just grown and grown and grown. But in any event, that was the situation. And in this unconsenting patient, Dr. Bartholow took a probe and stuck it into the brain and turned on the current and evoked contralateral motor movements. And he stuck it more posteriorly and turned on the current and evoked painful sensations in the contralateral body. And he stuck it back in the motor area and got a seizure. And the patient died three days later. And this was not well received by the public. So now the third attempt to establish stimulation mapping as some kind of scientific activity also has ended badly. However, neurosurgeons were not to be quite so easily deterred this time. And also they were much more able to control the current. And so during the next pretty much 50 years or a little longer, the last part of the 1800s, first part of the 1900s, is really the era of motor mapping. Almost all significant major neurosurgeons in the various European and American countries reported experiences with this kind of motor mapping. And that included Harvey Cushing. And that's what you see on this slide is Cushing's paper in 1909. It is a report which has been said to be the first time sensory changes were evoked from human cortex. It is interesting from a bibliographic sense for two reasons. One is that you could get published in Brain for just reporting two cases. It takes a little more than that now. And the second is simply an interesting little note. Some of the younger of you probably don't know that in the days before electronic libraries and the Internet, the way scientists used to exchange information was to send reprints to each other. And the way you did that is you took a penny postcard. There really were such things quite a while ago. And you filled out the article you wanted and sent it off to the author. And he would send back to you a reprint often with a little dedicatory comment to you about it. And the copy I happen to have is this one which Cushing sent to Wilder Penfield with the little comment that you see up there reminiscing about an experience that the two had had together long before. The peak of motor mapping is probably in the works of Ockfried Förster. Förster was a neurologist. He was self-trained as a neurosurgeon. He is also known as the person who became Lenin's private physician and spent several years in the Soviet Union. But in the period immediately after World War I, he had an extensive practice in dealing with patients who had had war injuries, particularly from shrapnel, with focal injury to what we would now call a Rolandic cortex, motor sensory cortex. And he used stimulation mapping as his technique for identifying the sites that he wished to excise in these patients. This is the sort of pattern that had been initiated by the person whose picture I showed you and forgot to mention, which was Victor Horsley. Horsley had actually done the first epilepsy operation in 1868. It was a patient with focal motor epilepsy who had been localized by Hughlings Jackson based on Jackson's clinical observations. And he excised the damaged area of motor cortex to try and control the patient's seizures. And that was what Cushing was trying to do when he did his stimulation mapping, and it was what Forster did in great extent in his large practice after World War I. And this is the paper that Forster wrote with Wilder Penfield, who trained in part with him, with their stimulation with a current, it was a phoretic current, so it was an oscillating current, above the seizure threshold. So their goal was to evoke the beginning of a seizure. So it's a little different than we use motor mapping now, which is something we kind of try not to do. But this was not what they do. All of them, pretty much everyone in that period, was trying to use the stimulation to replicate the patient's aura and initial characteristics of their seizures. And he had an extensive period with it, and he reported that. And this paper is from a lecture of his summarizing their stimulation effects. And what's important for our discussion is that those are the portions of brain in which they could evoke some kind of activity that the patient reported. All of these things, of course, were done under local. That the patients reported either motor or sensory or, in the case of auditory cortex, illusional changes. And all of the rest of it, well, everything else that's white there were areas that, despite using these rather large currents, they didn't get anything. They didn't experience anything, and they didn't get anything. So now, at this point, we have a technique for stimulation mapping of motor and sensory phenomena that we can use in neurosurgery, as we do now, but we couldn't do anything in the rest of these areas of cortex. Now, it was Wilder Penfield who did two things. One is that he introduced the use of EEG, or, in his case, electrocorticography, in the operating room as well as in the evaluation of patients with epilepsy. And with this, the identification of seizures of temporal lobe origin and with an indication to do temporal lobe resections. And this was then a stimulus for extending stimulation mapping as a technique to identify language. And so he introduced, then, the techniques that we use to identify language, which was this difference between the primary cortices, motor, sensory, auditory, visual cortex, where you can evoke a response, and we heard about that earlier today, and the rest of cortex, where you just don't get evoked responses. What you get is interference with activity, and this is the way that we now map and that Penfield perceived mapping language cortex. And he introduced that technique somewhere around 1940. At least that's what Herb Jasper said. But he says in his publication, Penfield's publication, that he started to use it extensively, at least, in the mid-1940s. So it's relatively, by the standards of the time course that I've been talking to you about, relatively recent. And he demonstrated that this gave a different pattern of localization than what was the traditionally accepted model at that point, which was the classic Broca-Wernicke model based on lesion studies that we all learned in medical school. And what he found was, and that's illustrated on the right side of that picture, what he found was that he could evoke, he could produce this interference from a much wider area than the classic models had predicted. However, Penfield then went on with his publication, his publication is the Penfield and Roberts book, to make this statement. He said, but curiously enough, now that we have carried out this test, that is mapping the speech cortex, some hundreds of times, we find it necessary to use this much less frequently. This is because we have a growing ability to predict speech area limits. And the area limits, he said, are the ones illustrated on the right, which are, of course, the classic speech areas. But he somehow didn't recognize any of the implications of this for planning surgery or for the differences in individual cases. He cites a whole bunch of individual cases, and if you read them in detail, you can get a feeling that there might have been discrete locations, localization in those patients, but he didn't recognize it. And stimulation mapping wasn't very much used by his Penfield trainees. My mentor in all of this was Art Ward, who'd been a Penfield resident in the early 40s, and he used it a little bit in the course of his epilepsy surgery. We always mapped motor cortex. He'd usually map anterior speech, and then that was pretty much it. Ted Rasmussen, who succeeded Penfield, believed in positive responses with stimulation, but did not believe in negative ones. It's interesting that as this has evolved, the recent paper from Mitch's group, it's exactly the opposite now. Perfectly happy to believe in purely negative ones and plan our resections accordingly. And so it all sort of wasn't very extensively used until we got interested in this in the last half of the 1900s, and that's when there was a period eventually of the renaissance of cortical stimulation mapping, which has resulted in conferences like this. And there are a couple of things that we did that led to this, and this was the first one. And that was to show the focal nature of individual subject maps and the variation of that from one subject to another. And this is a series of—these are all epilepsy cases, because that's all I was using it for. And the only tumors that I ever did were tumors who had refractory epilepsy. And you can see on this that all of the open circles are where we had stimulated, and the closed circles are where we evoked interference. We made two technical changes. One is that we started to use a constant current stimulator, which gave us much better control of our currents, and I think much smaller currents than Penfield was using. Excuse me. And the second was that we standardized our measures of language. We used object naming, which was the same measure that Penfield had used, but we used it in a very standardized way, both in terms of time of exposure, the kind of items we exposed, and the fact that we introduced control trials in which the patient didn't—or where we didn't stimulate the patient, and that we could then know that the sites that we identified as significant were sites of repetitive errors, which were unlikely to be just random errors on a statistical basis. So this standardized both the behavioral side and the current-level side. We continued to use 60 hertz current, and I used that throughout my whole career. That was simply inertia. But we used biphasic constant current pulses. And then we did a second thing that Penfield, for a variety of reasons I guess that I don't know, actually never did, which was to show that—at least to some extent— to show that this really made a difference, that it predicted pieces of brain, the stimulation interference predicted pieces of brain that you needed to leave, and that it wasn't just a matter of anatomy. And these are two studies. The one on the left is on epilepsy patients, and it's not a very extensive study, but it's a black-and-white study. One month after we'd done our resections, Carl Dodrell, who was our neuropsychologist, gave these patients a standard aphasia battery. It's a battery known as the Webman battery. It's not used now, but it's a pretty sensitive aphasia battery. And we compared the patients who had some kind of change in that, in whom we had stimulated close to the site where our section went, close to the sites of stimulation interference, to the patients where it did not. And we defined closeness as within two centimeters along a contiguous gyrus. And this is, as I say, a black-and-white study. If we got close within those two centimeters, there was evidence that even out at one month there was a significant decrement in the test, though not necessarily clinically, and that if we didn't, we didn't. And it wasn't related to their preoperative verbal IQs. It wasn't related to the extent of the resection, which means it wasn't related to just a matter of anatomy. And it wasn't related to the degree of seizure control. It was how close we had gotten to the stimulation sites. And then the study on the right is in tumor patients. It was done a little later by Mike Hagland and Mitch and myself. And that's clinically significant aphasias. And that basically showed that if we got within seven millimeters of one of the sites of stimulation interference, we had a deficit that didn't necessarily go away. If we got within less than seven millimeters, or more than seven millimeters but less than a centimeter, we had a transient deficit. And if we stayed away more than that, we really just didn't have permanent deficits at all in those patients. So now we had some evidence that stimulation mapping, these patterns in individual subjects, told us what we could or couldn't safely take out. Now, like Penfield, we did find that this was variable across the patient population and that it did involve pieces of brain that were not the classic Broca and Wernicke areas. And many of you have seen this particular map. We had 117 dominant hemisphere cases, again, all in the context of epilepsy surgery, where we had undertaken stimulation mapping. The little numbers in each of the zones that we've divided the cortex into that aren't in a circle are the size of the sample that had an example of stimulation effects there. And the numbers in the circles are the percent of that sample where we had interference. And the only spot where you find it most of the time is that piece of motor cortex just in front of face motor cortex, a piece of inferior frontal lobe just in front of face motor cortex. Everywhere else throughout the whole classical Wernicke area, the best we could do was about a third of the patients. And if we collapsed everything across the whole superior temporal gyrus, we could find any kind of interference in only about two-thirds of the patients. So the classical anatomic model didn't tell us, wasn't likely to tell us on a population basis, where language was at all. Now, we were also able, though, then, to take the negative situations where there wasn't language interference within the classic Wernicke or Broca area and do operations like this. And this is the exposure of cortex in a patient who had focal seizures that began with aphasic arrest. The orientation is the classic surgical one, so the top of the head is at the bottom and the front is to your right. And the little V arrows mark out Sylvian cortex. And the numbers in the 12, 5, 12, 14 mark out Rolandi cortex. And the circled sites are sites where we evoked language interference. And as you can see, we had a frontal one, number 10 up there. And then we have the two posterior sites. And we've just jumped without my doing anything. So let's go back. And we have done a resection of his focus, which we'd worked out with extraoperative mapping through a grid, or extraoperative recording through a grid of his seizures and some mapping, and then used intraoperative mapping to get even greater detail and test his language while we did a resection right smack in the middle of what should have been Wernicke's area. And this patient did not have a postoperative aphasia. And he was seizure-free afterwards. And he returned to being a teacher. So that is an example of what one could do, taking this individual map with localized sites that were different from one patient to another. And if you happen to have evidence that they didn't have function within the area of brain where the focus was, even if it was in the middle of classic language areas, you could safely do an operation. Now, the major motivation for me for doing many of these studies was also to learn something more about the neurobiology of language. And so we looked at a number of things. We had verbal IQs on almost all these patients. And so we looked to see if there were any difference based on that. And there was. We could identify areas that were larger, and in particular locations in patients with a lower IQ. We looked for gender differences, and there were. And it turns out the gender differences had been identified in stroke patients already, with the females being less likely to have posterior language sites in dominant hemisphere than males. And then some years later, when we got a little larger series of young patients, we could identify differences related to age. We eventually had a series in which the youngest patients with dominant mapping were at age four and our oldest at age 70. And across that, we could show some difference in patterns in the younger patients versus the adolescent patients versus the adult group of patients with a general expansion of language sites with the increased age. Now, the last piece of our renaissance, Mitch has already told you a little bit about. He joined our faculty in the middle 80s. He was interested in applying this to tumors. I was continuing with my epilepsy surgery practice. He asked me if he thought we could do this together. We worked together for several years, and then he popularized this to the oncologic neurosurgeons. This is just a case that we happened to do during that time, a large tumor in the inferior frontal area extending into face motor cortex. And you can see the sites of language that we localized there and then planned our resection based on that to avoid those areas but then resect as much of the tumor as we could. Now, for the last half of this talk, I wanted to give a little historical background for four things that are features, it seemed to me, of the discussion. Some of these we've already had discussion on earlier today. One was the relation to other modalities for localization. That is, other modalities of assessing cortical function. The relationship to identifying subcortical pathways. The evidence for plasticity. I think we're going to talk about that a little more tomorrow. And what I think is a fairly significant issue of how we might want to, or why we might want to address what kinds of functions to map. Everything so far I've told you has really involved mapping of, basically, language as assessed by object naming. All of it, except for the last little bit, the work with Mitch, has been in the context of epilepsy surgery and our initial experience with tumor surgery. So let's look at a little bit of historical background for these issues. First is relation to other modalities for localization. Well, I was taught that if you just kept your resection in the middle of a tumor, you didn't have to worry about any of the other function. That was the gospel when I was trained and the experiences that what I had been doing. But eventually, we collected a series of cases that showed that that was simply wrong. That is, there were a group of patients who had functioned in the middle of gross tumor, and that is illustrated in both the tumor on the left and then the mapping of that patient and the sites of language disturbance are identified by the red circles. Within two months of our publishing that paper in neurosurgery, a paper came out from Washington University in St. Louis. This immediately generated much confusion. Too many Washingtons, too many universities, and it turned out that my son was the lead author on the other patient on the Washington University paper, so too many Hojomans. But they both showed the same thing. They were both case series showing function within the middle of gross tumors. The tumors were gliomas, and it, I guess, shouldn't have been surprising because if you look at the pathology of low-grade gliomas, it's not uncommon to find some axons extending right through the middle of the tumor. So you can't base it on gross tumor, to stand in gross tumor. Now much to my surprise, the same thing applies to dysplasias, and as we've had more experience with dealing with dysplastic cortex, we have found that despite the fact that this is supposed to be a lesion that appears during the course of development of cortex, and so you would think if any time that you were going to have plasticity in the brain, it would be during dysplasia, and you would displace the functions to other areas of cortex. But that doesn't happen, and this is an example. This is an open lip schizencephaly. You can see the MR on the small top left picture, and you can see the localization of the cortex there, and the arrow points to the open lip schizencephaly, and A, B, C, D outlined were the extent of the epileptic focus that I wanted to take out, and as you see immediately on the anterior lip in the circle of stimulation sites there is language, and immediately, in fact, in the middle of the posterior lip is language arrest from orophageal motor movement. And it turns out, particularly in schizencephalies, that motor function is very commonly on one lip of the schizencephaly. So even something as occurring as early life as that does not necessarily displace cortex. Now what about other ways for investigating brain function? We've heard a little bit of this earlier, and I've always thought that the techniques we have for investigating cognitive function in human cortex basically fall into these three categories. There's the brain regions that we make the lesion by inactivating that brain, or make the relationship by inactivating that brain lesion, and that identifies sites that at least at that point in time are essential for that function, and that's lesions. That's the classic way. Electrical stimulation mapping, which presumably inactivates the piece of cortex for a temporary period, and also using intravascular neuroparalytic drugs, although that hasn't been much exploited for investigating cognitive functions. Then both of the other ways of doing this basically are simply correlative. They tell us not what's crucial, but where activity is participating as identified by a correlation. There are the groups that look for a metabolic correlate, and that's fMRI that we heard about earlier this afternoon, and PET imaging, and then a technique that's often called optical imaging, which involves subtle neuronal swelling, probably from ionic shifts and blood volume shifts. And then there are the electrophysiologic correlates, which we also heard a bit about, the electrocorticogram, microelectrode recording of single neuronal activity, and the scalp EEG. And there's no a priori reason why any of the correlative ones should necessarily be the same as the ones that tell you what's essential, and so that just becomes a matter that we need to determine from just showing whether there's a difference, rather than from assuming that they would give us the same information. So let's look at a few of those, at least based on some of the past. What's the relation to fMRI? Well we heard a very nice discussion of that this afternoon. At the time that I was still active, I looked at this a good bit, and at that time the literature showed basically that there was not a very close relationship between what you got with stimulation mapping and what you got with fMRI, and we of course had shown the stimulation mapping predicted what we could do with resection, so now it's an empiric issue as to whether the fMRI would too, and those studies really haven't, at least at that time, had not been done. And it was the consensus at that time in a consensus conference that some of you attended in I guess now almost quite seven years ago, that in fact at the moment the only reliable way of identifying the crucial areas to plan an operation was to use stimulation mapping, although fMRI might be useful as telling you areas that you wanted to sample with stimulation mapping. I think that's an area where quite clearly you heard today there's continuing new data and continuing controversy. All right, what about corticography? Well you don't really get an evoked potential in that, in the area, the language areas. You get induced activity. If you stick a microelectrode in, you find that you don't get a nice evoked response. What you get is an induced change in activity, but not one that's time locked closely necessarily to your stimulus. Early on, we were interested in looking at a corticographic change that was identified at the sites where stimulation interfered with language, which appeared during the same language task, and which weren't there when you used a control task. And the control task we used was to give the same visual stimuli, but with a spatial matching task that had to be done with it. And we identified two changes in the corticogram that met those criteria. Now this study was done in the late 70s and published in the early 80s. And at that time, we were still chopping off the EEG at 20 hertz. And that was a heritage from the days of EEG machines that used tubes. And when you used them in the operating room, if you didn't chop it off, then it was way too noisy to make any sense out of it. So this study was done not where you could look at what we would look at today, which would have been 70 hertz gamma activity, but activity up to 20 hertz. And in order to identify the change that we thought was there, and that's illustrated on the patient whose recordings are shown on the right, we thought we saw from the frontal sites a rather prominent slow potential. So that was presumably a premotor potential of some kind. And from the temporal area, what was then known as desynchronization, a loss of the alpha type activity and an appearance of low voltage fast activity of unknown frequency. And so we said, well, we'll simply quantitate the 7 to 12 hertz alpha activity. And we want to see if that disappears. And we were able to show that statistically that disappeared at the language sites, and there not during the spatial task. And if we compared the language sites to the sites around where we didn't interfere with language, it disappeared at the language sites and not the surrounding sites. So loss of this kind of activity, which I think subsequent data has suggested might well be activity in the 70 to 120 hertz or so range, although there is at least one animal study that has said that the loss of the 4 to 7 hertz activity isn't necessarily linked directly to the appearance of this higher frequency gamma activity. And I think that's another issue that is deserving of further study. Now what about subcortical language representation, which we're going to hear about even more tomorrow I think. And there have been several observations that I think are germane to that. Surface stimulation predicts the effects of a resection that includes buried cortex. And that has always surprised me. You can work, plan your anterior temporal resection. You can push it back to close to the stimulation identified sites. You do a resection that includes the superior temporal gyrus all the way down, including the opercular part of it that we heard about earlier today. And you don't get language disturbances. Frontally, you can extend your frontal lobe resection back to the area where you've identified the beginning motor changes and you don't get disturbances. You can then, if the surface stimulation sites extend little or at all into the sulci, you simply don't find, or at least I didn't find, isolated sites buried in the sulcus that I didn't know about when I did my resection. And we time and again would be able to stimulate the crown of a gyrus, evoke interference, stimulate with the same current, down on the side of, after we'd resected the gyrus below, subpeely, on the buried part, and get nothing. Go back and stimulate the surface, get an effect, stimulate on the buried sulcal portion of the cortex, and get nothing. So for there's some difference between the crowns of the gyri and the portion of the gyrus that goes down into the sulcus, I asked Arne Scheibel, who was a well-known neuroanatomist who studied the morphology of individual neurons, if there was any difference in neurons between crowns and sulci and gyri, and he said, yes, there is, and that was something I didn't know. But apparently the configuration of the dendritic trees is different, so that the pattern of folding of the cortex is not a random thing, and at least for our purposes, is not random in terms of functional localization. We were able to show insular involvement, as you heard earlier today. We were able to show some involvement in the plenum temporale, which is an area of interest in language neurobiology, largely because it's asymmetric, it's bigger in the dominant hemisphere than the non-dominant on average, and that's because the sylvian fissure does not have the same configurations on average in the dominant and non-dominant hemispheres. But we found that when we stimulated it, and this is some work that actually we did with Mitch when we'd taken out an inferior parietal tumor, we found only a little extension of a surface language site onto the plenum. On the other hand, when you undercut a surface language site, then you can get effects from the buried cortex, from the subcortical areas, and exactly what fibers that we're interfering with there are, I think, not entirely identified. Now, other stimulation effects have suggested that, in fact, the subcortical pathways are very complicated, and I think we got a little idea of that from one of the talks this afternoon. Penfield made much in his book about the subcortical pathways that went from the thalamus to the different cortical language areas. His only problem was that there was no evidence at that time that there was language function in the thalamus, so it was all hypothetical. About 10 years later, my stimulation mapping career began not by stimulating cortex, but by stimulating the thalamus in the course of stereotaxic operations, and showing that we could indeed, in left thalamus, evoke different patterns of language interference, and those are shown on the right side, effects we didn't get from the right thalamus, and that sort of corresponded to the thalamocortical localized pathways that went from these various thalamic nuclei, out lateral thalamic nuclei, out to the cortical surface, and those pathways are, of course, very well known. They've been described for a long, long time. They're monosynaptic, both directional pathways that go from the thalamus out to the cortex, and, in fact, there's some degeneration, both if you have a cortical lesion, the thalamus degenerates a bit, its target area, and conversely, so there's a lot of reason, I think, to think that there are these thalamocortical pathways. They're not very well shown on the current DTIs, so there's pathways there that we can identify as likely to be involved in language, but we haven't really considered them, and I think they're rather important. I think many of the times we get deficits when we undercut the surface language areas, it's interrupting these connections that are responsible for it, not the connections through the dorsal or ventral longitudinal pathways. Sorry, how did that happen? What happened? I pushed the button to get the next slide, and they decided that I should have to be quiet. I don't have very much more to say, but I guess I'd like to say this last bit. Okay, next one, and the next one, we're going backwards, okay, one more. This is the button that I pushed, and it came out badly. That's it. Thank you. And in the course of the studies I was doing on the thalamus, these were done in NIH with John Van Buren, and we were using multi-contact electrodes, and some of the contacts went through the white matter pathways that were just lateral to the splenium of the corpus callosum, and again, we were able to evoke language interference from stimulation of those white matter pathways. In this case, though, remember the thalamic effects were all lateralized to the dominant side, but in this case, we could get the effects from both sides, and those are the ones identified there by the little red arrows, so that it is possible to map at least these pathways with stimulation. We've assumed that these were the pathways that got language information from the non-dominant hemisphere to the dominant one, and are the ones that are interfered with with a posterior cerebral infarct that knocks out the dominant hemisphere visual inputs, and also gives you anomic aphasia because you can't get the visual information coming across. Okay, we're just about done. Now the other study that actually hasn't been published, it may eventually, suggests again that there is relatively little plasticity in the cortical language sites, at least in the epilepsy population. I'm not sure that applies to the tumors. This is a study that we, it turns out that when you've been in practice as long as I was that your operations don't always work, and you have a moderate number of patients who come back for a second go, and so we were able to identify 22 adult patients who had a second mapping from one to a bit over 20 years apart. The mean was about eight and a third years. We worked out a way of aligning the two cortical maps using the sulcal patterns, the surface vessels, and rolandic cortex. That was a bit of a challenge. 20 years apart, your techniques for getting operative photos and a few other things are pretty different. The cortical surface looks a little different, but not that much. Of course, the cortical veins are great fiducials. They move with the cortex. They have a configuration that is very much like your fingerprint, and it's different for every patient. You can use them as fiducials to relate to your stimulation mapping sites. At the second mapping, we found that 86% of the sites related to language at the first mapping were within one and a half centimeters, a radius of that site. That was in 20 of the 22 patients. 54% were within one centimeter, which I view as the limit of resolution of that study. That is, I don't think our reconstructions are any more accurate than that. At least in these adult patients, even though they'd had a resection, mostly anterior temporal, between the two operations, there wasn't really any evidence of plasticity. We do know that there can be acute changes with plasticity. These are changes with learning. This is something that gets overlooked, although there's quite a bit of evidence for it. That is, a much larger pool of neurons is active when you first learn something, and that can be shown with stimulation mapping. It's been shown with single neuronal recording, and it's actually been shown with fMRI, too. This shrinks as you acquire skill in a particular task. We used this Washington University verb from noun task that was alluded to a little bit earlier here. We simply had one list that they had practiced, and one list that was novel, and simply compared our mapping. The control items were pretty much the same, but the two different lists we were able to interfere with from a much wider area in the novel than the practiced ones. Yes, there is plasticity acutely early on, but at least in the epilepsy population, not once it's been overlearned. Okay. Now, finally, what function to map? About three months ago, a patient that I'd operated on back in the 1980s found me on social media. What he said was this. He said, thank you for taking care of my seizures. I've been seizure-free for 30 years. It's been great, and I don't mind the fact that you took away my ability to read. Now, we'd done him with our standard technique at that time. We'd done him with carticography. We'd done him with mapping of object naming. I don't know that he can't read. I haven't seen him or checked it out, but my guess is from other patients that his reading had become very effortful for him, and he just didn't bother to do it. What it indicated is that the localization of reading, and this is reading of simple sentences or reading of single words, has a differential localization to some extent in many patients. This is 55 patients in which we compared the localization with object naming to the localization with reading of these simple sentences. In the height of those vertical bars shows you the proportion of the sample, again, that had changes in one or the other. As with object naming, there's a lot of individual variability in the localization. As with object naming, individual patients have very localized sites, but the sites are sometimes separate. As you can see from the particular patient shown on the left, the black are the areas that were common to both, and that's sort of parasylvian, but around it particularly, both anterior-temporally and superior-frontally or middle-frontally, are sites where you interfere in only one of the two tasks, and in those cases, mostly not in naming, but in the reading tasks. Some of the reading changes were quite interesting because they involved syntax rather than semantics of the simple sentences that they were reading, and those were in sort of the anterior-temporal or anterior and middle-temporal and frontal sites. The same thing's true of multiple languages. If you use the same naming of the same pictures in different languages, you get different patterns of localization. Now, they aren't grossly different, but they're subtly different. The patient illustrated on the left taught English to Spanish-speaking students, and we mapped both English and Spanish. The filled circles are where we interfered with English. The filled triangle is where we interfered with Spanish. Spanish was fine at the English sites. English was fine at the Spanish site, and we have 22 patients proficient in seven different languages that we have mapped. They have different first and second languages. Some of them were people who came to the United States, and therefore, English was not their first language. Others were patients who had grown up here but used the foreign language commonly for their grandparents. And at 21 of those 22 patients, we found some sites that were related to only one language. In nine, we had site-specific sites for both languages. It wasn't related to proficiency. It wasn't related to language type. It was related to localization. L2, the second-language sites were underrepresented in the classic Broca and Wernicke areas compared to monolingual adults, and interestingly enough, to the sites where language is localized in the children who are very young, which is basically a parasylvian area. And we did a research study where we looked at the localization of a whole series of different language functions. The ones shown up there in the red at top, we had this patient's primary language was English, and the filled circle at the top was the site where we interfered with English naming, but not anything else. The patient had Greek as a second language that she used with her grandparents. That's the little square, and you see there were sites where we interfered with Greek, but not anything else. There was sentence reading, and there's a site up there in the parietal operculum where we interfered with sentence reading, but not anything else. And we had a measure of recent short-term verbal memory encoding, and there was a site where we interfered with that and nothing else. And it's interesting that the encoding test for the memory measure is basically the same thing that they do when they name. But when they name, we tell them, just name it, and when they encode it in memory, we say, name it, and then encode it, and then remember it. And that one instruction changes the pattern of localization. And then we did this in a series of 14 patients, and you could then construct an indication of where you were likely to find these different changes. The green and blue identify the changes where we got language interference. We had also measured two other things. We had measured oral facial movement and the ability to discriminate phonemes. You heard a little bit about the phoneme discrimination test earlier today when they talked about categorical perception. And that was the one function which wasn't localized in separate things. Oral facial movements, mimicking the movements that you need for speech, but simply a pure motor task. And discriminating phonemes were localized essentially to the same sites. The people who believe in the motor theory of speech perception were delighted at that finding. I don't know that that's necessarily the basis for it. But that's the green areas and the blue area. The yellow identify sites where we interfered with the encoding phase of verbal memory. And between those two sites, two regions where those predominate, are the specialized sites where we interfered only with naming or only the syntactic aspects of reading. And that's the orange ones. So you have this model of areas of temporal and parietal cortex that are involved and posterior frontal that are involved with recent memory, recent verbal memory. Areas around parasylvian area that are involved with motor primary expressive language functions, motor control of that. And then the specializations for other aspects of language that seem to develop between them. Now finally, you can of course map non-dominant hemisphere. We didn't do very much of that. Mostly because the patients didn't complain of any deficits they might have after our non-dominant resections. And so it didn't seem to have major clinical application. But we did a little bit of it. This is a report from back in the early 80s when we mapped a series of visual spatial functions. The square is a site where we interfered with the ability to map, to match faces and figures. The stars is a site where we had a measure where the patient looked at a face and then had to label the series of actors who had different facial expressions. They had to look at that actor's face and label the expression. And they were pretty clear ones. And so in the control trials they were very reliable. But when you stimulated at this place where the star is, you simply got a different label put on this facial expression. And the filled sites were a memory for these visual spatial tasks was interfered with as well. And the open circles, of course, we had no effect. So there is localized function within the non-dominant hemisphere. We haven't shown any specific relationship to that. And this has been done only in a limited and experimental way. I've had a lot of this, of course, are all multidisciplinary studies. We have neuropsychologists, other neurosurgeons, neurologists, electrophysiologists all involved with us. These are many of the people who have been my collaborators in the various studies that I've done in this setting of exploiting the opportunities that come along with the surgery of epilepsy to examine how cognitive processes are organized in the human brain. And then to extend those kind of findings to clinically applicable things like stimulation mapping that can improve operative approaches. Thank you. That was a tour de force, a magnificent career, George, and a lot of pearls in that. And I'd like to open it up to the group. Before I open it, I want to ask you one question. I think the most striking thing that I've seen in my career, and you taught this to me very clearly, is the tremendous variability of organization of language. Whereas the motor cortex is fairly standard. We can find it. Sensory cortex is the same. But variability is just one of the features that sort of brings us to this point of having to use stimulation mapping. I wonder if you could comment on what you think that is related to. Well, you know, the anatomy of cortex isn't nearly as uniform as you would have thought maybe from some of the discussion earlier today. If you look at this for parasylvian cortex, this has been fairly well worked out at a gross level. The sylvian fissure has a different configuration in the dominant hemisphere than it does in the right on average. But with individual variability as well, this changes the size, say, of the supermarginal gyrus on one side and the plenum temporale compared to the other. And there's some recent structural MRI studies that have been very interesting where they have compared the size of the different gyri of one hemisphere to the other. Done it on a much more detailed mathematical basis. There is a lot of variability, particularly in the parasylvian area, frontal lobe. And its variability, which is much greater, at least one paper, was done in a series of mono and dizygotic twins. And it was much greater in the dizygotic twins. And there was relatively little variability in mono-zygotic twins. So the anatomic variability apparently has a major genetic component. Now, the functional variability is a lot more than the anatomic variability. And I don't really know what's responsible for the functional variability. I can only say that things don't seem to move very much, at least in the epilepsy population. So whatever it is, at least in adults, pretty fixed. But how it gets there, we don't know. And in fact, the reality is we don't quite know why it is that there are these crucial areas. We know that if you stick a microelectrode in either hemisphere, either temporal lobe at least, you can find neurons that correlate very nicely with the language tasks. So it isn't that there's just activity there and nowhere else. And in fact, the proportion of neurons that you get is about the same in the two hemispheres, even when you know that they're all left dominant. So, but exactly what it is that makes these sites crucial, we don't know. There is a theory out there that these sites are sort of like the index of a computer. They tell you where, or tell the brain, where it has to go to find the networks that are needed for the language function, maybe. But we don't really know that. Any questions for Dr. Ojibwe about his findings? Maybe from some of the more senior folks who have had lots of experience? I've put them all to sleep. No. No, I think it's very clear. Yes, George. Well, first of all, it's a great honor to be here, to Dr. Ojibwe in person. I would like to ask you, we know that there are quite a different language pathways, quite different tracks from ILF, SLF, and SINETA. When was the first time that you actually started to stimulate and test these tracks? How did you decide which test to select? Reading, edition, writing, how did you come up with the ideas? Well, our subcortical mapping started with motor pathways. And this is, again, what Mitch and I did. And it was identifying the internal capsule. And it was mostly in tumors that extended into the temporal stem, where that's often on the medial bank of what you want to resect. And I might, our experience was that using bipolar stimulation was a very accurate way of being able to assess that, how close you were to that pathway. I have not, my strategy for when I was in practice, you need to recognize that it's been a little over a decade since I stopped doing clinical work. But my practice then was to, when I had to undercut, was to have the patient engage in a language task. And I usually used a mix of naming and reading, single word reading, and looking for interference as I did my resecting. And sometimes would also look with stimulation. But there's, I mean, that's a problem, because you just don't have time enough to assess everything, and yet you do know that you run a risk of potentially interfering with something once you start to undercut. Okay, George. Thanks again very, very much. Thank you.

Dr. George Ojemann

stimulation mapping

language function

human brain

Broca and Wernicke areas

object naming

reading

language deficits

corticalography studies

subcortical pathways

neurobiology of language