[SOUND] Neuropsychological studies that had actually considered the idea of language switching after Putzel and started to try to identify the locus that led to a problem with switching were mixed. One study had found that the supramarginal gyrus, roughly in the parietal lobe, was involved in switching. Another patient had some kind of motor problem, and the suggestion there was that there was a frontal involvement and the frontal lobe was involved in switching. So we had no real idea at the time that we started these studies that, in fact, there was one area that might be involved or a set of areas that might be involved in language switching. The first study to look at language switching with neuroimaging methods was performed by Cathy Price, David Green, and their student, Von Studnitz, the one who had published this paper on behavioral switching. The finding from their study was that there was an effective translation in certain brain areas. And that there was also an effect of switching with increased activity in the supramarginal gyrus and the inferior frontal gyrus as well. So we see some of evidence that the supramarginal gyrus is involved, and some evidence that the inferior frontal gyrus is involved. Two subsequent studies that we did in our lab explored this further using this cued picture-naming task adapted for the MRI scanner. Instead of having people listen to a cue, we ask them to look at a visually presented word, say [FOREIGN] and then imagine they were saying the word in their head, saying the name of the picture in their head. Now, of course, you might wonder, well, what were they really saying? Well, we tested them outside of the scanner, but one of the problems at the time was that we really had no reliable way of gathering responses inside the scanner. So we asked people to use covert naming. The other problem as well is that when people say things, so if they're going to say table, chair, car, they tend to move their heads. And when people move their heads, that's not good for brain imaging. We need them to be as still as possible, so we used covert naming. And across two different studies, we found that in fact the dorsolateral prefrontal cortex, this area that's involved in task switching that has been observed in subsequent studies, was active for the language switching when people had to switch between languages, relative to when they only had to say pictures in one language. Other studies have also found that the dorsolateral prefrontal cortex is involved in switching. In one study, they asked a group of depressed patients to undergo treatment with transcranial magnetic stimulation. Transcranial magnetic stimulation involves a small pulse, which is passed through the skull resulting in interruption of brain activity. And in this case, the idea would be that it would help with the depression. What the patients found was that, in fact, there was more intrusion from the second language. They thought of more words in the second language after this treatment than they would normally. And the authors then attributed this to the fact that they have somehow affected an area that's normally involved in switching. Now, the dorsolateral prefrontal cortex is not the only area involved in language switching. Other studies have found a network of areas that seem to be involved. So, studies have found that the caudate nucleus is important. In one case, a trilingual patient was tested and found that the caudate nucleus showed effects or damage in a patient who was doing involuntary language switching. Abutalebi and Green have identified a circuit of areas that are involved in switching and modulating language use. The prefrontal cortex, an area thought to be involved in executive functions, decision making, response selection, inhibition, working memory. The anterior cingulate cortex involved in attention, monitoring conflict, detecting errors. The caudate, that we just talked about, involved in language selection, set switching, language planning, lexical selection. And then the inferior parietal lobule, which is involved in maintenance of representations and working memory. And the idea is that when someone has to switch between languages, in this case, specific to language, there is a wave of activity, and initial stages of that involved having to come up with the right representations, figuring out which language is going to be used, sort of a language context, if you will. Then there's a set switching or a selection of a language, a planning. After that, there's the idea of, is this the right language or the wrong language, to monitor for any errors. And then finally, the selection, the inhibition necessary to execute that particular language response. So you could think of language switching as involving a wave of activity across these different areas as people go from the general context of what they're trying to do down to the specific response, down to making sure that they got the correct response. In the study I'd done in 2001, we had found and replicated this effective language switching, but we hadn't found an effect in another task. We'd asked people to switch between saying the name of a verb and saying the name of a noun. So we'd show them, for example, an orange, and we'd produce the cue, the, and we'd ask them to produce the noun, orange. Or we'd produce, we'd show them a picture of a door, and we would then produce before that, we'd present them with the cue, to, so they'd get to, and then a door. And the idea is that they would have to say shut, right? And what we found was that, in fact, there was no real increase when they switched between responses, or when they stuck to to or the, just the single nouns or verbs. And we were somewhat surprised by this. We expected to find a task-switching effect as well as a language-switching effect, but we didn't find that. And at the time, that was somewhat puzzling, but we wondered why this was. Why was this true? What had happened? And in fact, it's interesting because it leads us into the possible finding that task switching may be diminished in bilinguals.