Over 100 years ago, neuroscientists showed that the surface of brain is carved up into different territories that each subserve specialised neurological functions. There are motor areas, pre-motor areas, visual areas, and somatosensory areas. Subsequently, brain imaging confirmed these findings and enabled scientists to build “connectivity maps” that highlight the relative strengths of the connections linking between different areas and this we ascribed to things like behaviour and IQ. But, like all these things, a new study shows that the devil’s in the detail and the concept might now need a rethink. Chris Smith hears how...
Janine - I am Janine Bijsterbosch, and I work at the University of Oxford. In the past years there's been a lot of interest in mapping out the strength of the connectivity between different regions in the brain, and this is typically done using an MRI scan where we measure a movie of the brain activity in different brain regions while subjects are just lying there and doing nothing in particular. And several studies have shown that this pattern of connectivity differs from one person to another person in a way that is meaningfully related to behaviour. And for this study we wanted to ask whether it is really these patterns of these connectivity strengths that are linked to behaviour, or whether there are other aspects of the MRI data that may be even more important.
Chris - And how did you approach that. What did you do?
Janine - What we did was to make use of a method that allows us to decompose or essentially summarise the data from one person's scan into a number of different features. So specifically we can estimate three different measures. The first one is the pattern of connectivity strength between the different brain regions, like I mentioned before; the second summary measure is the strength of the MRI signal: so the size of the signal in different brain regions. And the last measure describes the spatial layout of the brain regions. So essentially the exact shape and the location of where a region is in the brain.
Chris - And where did you get the data from?
Janine - So this was data from the human connectome project - so we used data from just over 800 individuals.
Chris - Men, women, young, old?
Janine - Men, women, all relatively young. So this is a young healthy population.
Chris - Okay so basically you take a large corpus of data that's already been generated and you're analyzing it in this new way where you can fix one of the things and ask how that varies when compared with the averages of the other two. And you work your way through those three features?
Janine - Yes that's right. So once we have those three different features that I described, we can fix two out of the three of them to the group average, so that they don't vary from one person to the next, and only allow one of the three features to vary across people, and that then allows us to ask the question of which of these three features of brain data varies the most across individuals and is most strongly linked to behaviour?
Chris - Because previously we would have said thats the connectivity but now you can actually analyse it in this new way, what did you find?
Janine - So we found that in fact it was not the connectivity patterns that were the most strongly linked to behaviour but instead it was the spatial organisation of the brain regions. So this means that the exact shape and the location of the regions in the brain is very strongly linked to behavioural factors such as intelligence and life satisfaction and drug use.
Chris - That's incredible isn't it. Why should that make such a difference. One would think that it doesn't matter where they are or necessarily how big they are but if they're really richly connected that was the prevailing thinking that should make all the difference, so why this totally 180 degree turn around?
Janine - Yes well that is one of the big questions that now follows up from our research and we're currently doing work to try and understand this better. There's a few different reasons why it could be; it could be something to do with the underlying anatomy of how the brain is organised between these different individuals. Or it could be something to do with how we represent how these regions interact with each other. So, for example, it might be that there are particularly overlapping regions that contribute to multiple networks in the brain that might explain these findings but this is an area that we're currently delving into further.
Chris - So what are the overall implications then? Do we have to sort of tear up our book of knowledge on how connectivity and behaviour go hand in hand and begin again, looking at it through the lens that you've created here?
Janine - Yes well I I think these findings are important for a number of different reasons and the first one is, like you suggest, that the methods that we currently use to estimate these connectivity patterns are actually quite strongly driven by the spatial layout of these brain regions. And this is quite important because typically we interpret these changes in the strength of connectivity measured with MRI in terms of the amount of communication that that exists between the different brain regions. However of course if those connectivity patterns are to some degree driven by the spatial layout, then that interpretation is incorrect so I think one of the implications of our results is that it really highlights the importance of separating out these different features from the MRI data more precisely in our analysis so that we can draw appropriate conclusions based on different measures.
Chris - And if we consider, just in finishing, people who are not neurotypical - people with Asperger's for example - where classically neuroscientists have said "look there's a connectivity difference in the brains of these individuals compared to what we would regard as normal," if you were to apply your techniques to those individuals what do you think you'd see?
Janine - Yes that's a really interesting question and a lot of studies are looking into whether we can use connectivity measures as predictors for quite a wide range of different psychiatric and neurological disorders. And our findings show that in healthy people the strongest link is actually between these spatial layout information and behaviour and so I think that's a really good opportunity to study whether this spatial information can also be used as a meaningful indicator in patient groups.