Question: When doing MRI scans, and observing the brain during speech, is it possible to find if they are lying, their emotions and aspects like that?

Carolyn McGettigan answered on 21 Mar 2010:

Great question. MRI is a very powerful technique, and has given us so much more insight into how the brain works. However, like all technologies, its capabilities are limited. In my experiments on speech, I can find out which parts of the brain are likely to be involved based on how I design my experimental design. I could get people to listen to speech and some other types of sound in the scanner, and then look at what bits are more active for speech basically by doing a subtraction (i.e. identifing where speech > other sounds). For every study, I will have a hypothesis about which brain areas are likely to be involved. If I see unexpected areas of activation then I could look at other published studies that have identified this area too, and see whether what they were investigating is related to my study. So, once I saw the amygdala activated in one of my studies where people were looking at videos of someone speaking. The amygdala has been strongly associated with fear and processing of fearful stimuli. But I knew my participants weren’t afraid (!), and when I did some more reading I saw that many studies have also implicated this brain area in processing faces. If I’d just looked at the results and said “the amygdala’s active, therefore these people were afraid”, that would have been inaccurate. So, there’s a strong interpretative element to working with MRI that is related to setting up very controlled experiments and comparing the results with our particular hypotheses.

So if you consider these aspects, it means that we can’t get someone to speak freely in the scanner and then perform complicated interpretations about their emotional state or whether they are telling the truth. This kind of thing happens all the time in TV shows like House, and it’s very misleading. To study emotion, we’d need to set up specific conditions. For example, we could show someone happy faces, frightened faces and neutral faces, and then compare activity across these to see the brain areas more associated with each of the emotions ‘happiness’ and ‘fear’. For lying, it’s even more complicated. The act of lying is behaviourally quite complex, and people probably do it in very different ways. Currently, the only MRI studies that have looked at this have asked volunteers to do **really** simple ‘lying’ tasks, like pretending they haven’t seen a picture before when they actually have been shown it. There is some limited evidence that analyses can tell apart activation patterns for ‘lying’ and ‘telling the truth’ in these simple examples with volunteers, with some errors. But trying to use MRI for anything more complicated, like lie detection for criminal investigations, is currently not possible and a long way off. Despite this, people are already using it in this way – there are companies in the US that offer MRI lie detection services, and a similar approach using EEG (which records electrical activity from electrodes on the scalp) has been used in a criminal court in India. This is very irresponsible, and it’s important for the public to be aware that current applications of neuroscience for these purposes is not justified.