Sep 29 2015

Study Correlates Brain Connections to Intelligence

One marker of a good scientific study is that it provokes more questions than it answers. That thought kept occurring to me as I read a recent study: A positive-negative mode of population covariation links brain connectivity, demographics and behavior. The study essentially correlates specific patterns of brain connectivity – the degree to which different parts of the human brain talk to each other – with a suite of what are generally considered positive traits, such as education, income, and self-control.

Let me describe the study and then we’ll try to unpack the many layers of what such a study might mean.

The authors are among the first to avail themselves of the Human Connectome Project (HCP), part of the NIH funded BRAIN project. The HCP seeks to perform high definition function MRI scanning on 1,200 individuals while also gather large amounts of data about those same individuals, such as IQ testing, personality profile, and a host of demographic and historical information. The HCP has already released its data for the first 500 subjects to be scanned, so that other scientists can use the data for their own research.

The authors of the current study did just that. They used the data from 461 individuals to create a map of the strength of connections between many different regions of the brain. They then did a massive statistical number crunching to search for any correlation between the strength of any particular connections and specific traits of those subjects. They found:

“We identified one strong mode of population co-variation: subjects were predominantly spread along a single ‘positive-negative’ axis linking lifestyle, demographic and psychometric measures to each other and to a specific pattern of brain connectivity.”

In other words, if you ranked the 461 subjects on one linear scale of classically positive to negative traits (for the sake of this article I won’t get into the philosophical question of what makes these traits positive or negative), there was a strong correlation with with strength of a specific pattern of connections – the more positive traits a subject had, the stronger their connections among these specific regions.

This is the point at which the study provokes a long list of further questions. The bottom line right now is that we don’t know what all this means, but it is intriguing on many levels.

First (always first) – is the correlation real?  This finding was the result of a massive data mining project. The correlation is statistically strong, but it could be a fluke of looking for many possible correlations. Fortunately there is an easy way to replicate the study – wait for the HCP to release data from the next 500 subjects and then repeat the process on that data, specifically looking for the alleged correlation.

Assuming the correlation is real, and it holds up to replication, the next question for any correlational data is – what is the direction of causation? The standard causal hypotheses include: do the strong brain connections cause people to be intelligent and successful, does a lifetime of study and discipline strengthen those particular brain connections, or do success and the strength of those connections both result from some other third factor?

This kind of data will not definitively answer the causal question. For that we will need to look at people at different ages, to see what comes first – the strong connections or the positive traits. Ideally we could do an experiment in which we try to impose certain positive traits (like education, or discipline) and then see if it changes the strength of those connections. I also think we will need to have a deeper understanding of the modules and networks in the human brain to develop a coherent understanding of what those brain connections are doing.

At its heart, the causality question comes down to the old nature vs nurture debate. Are we born with our brain function or is it largely a product of our environment. Usually the answer to this question turns out to be  – both. We are born with talents and predispositions, but these are altered by environment, nurturing, education, training, and effort.

These kinds of studies may help us learn more about the nurturing part of the equation. What environmental factors that we can alter correlate with strong brain connections that in turn correlate with positive life outcomes? Of course, we can eliminate the neuroscience middle-man here and just correlate environmental factors to desirable outcomes, but this is tricky business. Having an objective marker in the brain might strengthen this type of data, and also enable us to see benefits through altered brain connections without having to wait decades to determine life outcomes.

The next question raised by this data is another old neuroscience question, sometimes referred to as the “g-factor”  – is there a module, network, or series of networks in the brain that correlate with “general intelligence?” Of course, we then have to operationally define “intelligence” but for now let’s just use the simplistic definition, intelligence is whatever is measured by IQ tests.

The “g-factor” question is this – is there a network in the brain that is central to overall cognitive function and therefore by itself correlates to a long list of positive intellectual traits? Another way to state this question is this – are the various measurable cognitive abilities (such as verbal skill, math skill, music aptitude, self-control, abstraction, problem solving, pattern recognition, etc,) each completely independent variables (anyone can be great at one and terrible at another), or do they all benefit or suffer from a general cognitive ability (the g-factor)? Do people who are smarter in one area tend to be smarter in other areas, and why?

The current study does not resolve this century-old debate, but it does seem to put one pebble on the g-factor side of the scale. The researchers found that one pattern of connections does correlate to a long list of positive traits.

Finally, this study made me wonder about what the ultimate potential of the HCP is. How much can we learn about brain function from studying patterns of connections within the brain? I would think quite a lot.

However, the study also highlights how much we have to sort out. We are in that phase of studying the connectome where as we gather more information we are learning more about how much we don’t know, so it seems that our ignorance is growing. Essentially we are finding that the brain is at least a couple of levels more complex than perhaps our initial guess. Specifically, it does not seem that for much of the brain specific brain areas have one specific function. There are exceptions to this – the primary visual cortex, for example, seems to be dedicated to processing visual information in a very specific way.

As we consider the “higher” brain functions, however, the simple correlation between brain regions and function breaks down. It seems that each brain region can participate in many networks of connections. Those networks seem to correlate with specific tasks or functions, but there also there may not be a simple one-to-one correlation. Networks interact with other networks. Everything ultimately may be context dependent – modules and networks in the higher cognitive areas of the brain may be dependent upon what else is going on in the brain at that moment.

This doesn’t mean we can’t map the brain and assign real functions to different patterns of connections. It just means we are finding that such correlations are complex and we have to be thoughtful about how we ask questions. We may not yet be asking the right questions, and the ultimate result of current connectome research is that it will teach us what are the right questions to ask. (For example, what are the basic cognitive functions and how do they result in what we observe as behavior?)


The current study is exciting not for what it answers but for the questions it provokes. This is not our first glimpse into the complexity of the connections of the human brain, but it does mark a new chapter, utilizing the largest and highest definition data set yet collected.

The HCP gives scientists access to a massive (and growing) data set of information about brain connections and cognitive traits. I expect to see many studies coming from this data, slowly teasing apart the complex questions raised by this and other existing studies.

9 responses so far