How Well Does Brain Structure Predict Behaviour?

By Neuroskeptic | September 13, 2016 12:39 pm

To what extent does brain structure correlate with different psychological traits? An interesting new paper from Massachusetts General Hospital researchers Mert R. Sabuncu and colleagues uses a new method to examine what the authors call the ‘morphometricity’ of various behaviours and mental disorders.

Sabuncu et al. define morphometricity as “the proportion of phenotypic variation that can be explained by macroscopic brain morphology” – in other words, the degree to which people with similar brains tend to be similar in a particular behaviour. Morphometricity is somewhat analagous to the concept of heritability from genetics.

Using FreeSurfer software and the statistical technique of linear mixed-effects (LME) modelling, the authors examined over 3,800 structural MRI scans, pooled from 9 studies. Sabuncu et al.’s analysis was based on calculating an anatomical similarity matrix (ASM) across the individual brains. The ASM represents the “global morphological resemblance between pairs of individuals in the sample”. Essentially, the ASM represents how similar two individuals are in overall brain structure. Sabuncu then calculated the morphometricity of each trait by comparing the structural similarity to behavioural similarity.

The results showed that Alzheimer’s disease is almost perfectly morphometric, with an estimated value of 0.94–1.00 (where possible values range from 0 to 1). Schizophrenia was moderately morphometric (estimate 0.55), with autism coming in slightly lower at 0.38. Perhaps surprisingly, Parkinson’s disease had a much lower morphometric value of just 0.20.


Other, non-disease-related traits, such as IQ and level of education, were highly morphometric too, with values above 0.8. In fact, IQ was slightly more morphometric than sex (IQ 0.95 vs. sex 0.93), while age was perfectly morphometric (1.00).

The authors conclude that

In the dawning era of large-scale datasets comprising traits across a broad phenotypic spectrum, morphometricity will be critical in prioritizing and characterizing behavioral, cognitive, and clinical phenotypes based on their neuroanatomical signatures. Furthermore, the proposed framework will be significant in dissecting the functional, morphological, and molecular underpinnings of different traits.

This is an important paper, but we shouldn’t rush to over-interpret the results. For instance, whereas Sabuncu et al. say that the high morphometricity estimates for disorders such as autism and schizophrenia “unequivocally point to a neuroanatomical substrate for these clinical conditions”, this really doesn’t follow.

Consider the schizophrenia data in this study. The MRI scans came from a database called MCIC. The problem is that 86% of the patients in this study were taking antipsychotic medication, which might well effect brain structure. 29% of the schizophrenia patients also had a history of alcohol or drug abuse, which could leave an impact on the brain as well. So we can’t say whether the high morphometricity of schizophrenia is driven by the syndrome itself, or by exposure to various substances.

We should not conclude that high morphometricity means that brain structure causes a particular behaviour.

In terms of the morphometricity measure itself, I note that the authors say that it “does not require cross-validation, which is often the technique used in machine learning to gauge prediction accuracy”. That’s because their method “exploits the entire dataset to fit the model and estimate the unknown variance component parameters, and in turn morphometricity, in an unbiased fashion.” I would prefer if some validation had been performed, e.g. by calculating the morphometricity of a randomly generated ‘trait’ or by permuting the trait data.

ResearchBlogging.orgSabuncu MR, Ge T, Holmes AJ, Smoller JW, Buckner RL, Fischl B, & Alzheimer’s Disease Neuroimaging Initiative (2016). Morphometricity as a measure of the neuroanatomical signature of a trait. Proceedings of the National Academy of Sciences of the United States of America PMID: 27613854

  • gwern

    A rare use of variance component estimation outside genetics! I wonder if we’ll see the idea expand to some other areas with a similar problem of being thought to potentially explain a lot of a trait but are too data-poor to fit a linear model and directly predict all of the trait, like gut microbiome studies?

    > So we should not conclude that high morphometricity means that brain structure causes a particular behaviour.

    Like in epigenetics. Since brain structure is not determined randomly at conception like genes, the usual causal inference problem remains. One can say, though, that if one believes that changes in brain volume are a cause of these traits, there should be a high morphometric correlation, and a failure to find those would be evidence against.

  • leoboiko

    Can behavior cause brain structure?

    • Jayarava

      Both experience and behaviour can result in new synapses forming, so the answer is yes, but only at the micro-level. You aren’t going to grow a new lobe or anything.

      • SEASanders

        I understood some studies have shown that trained musicians have a greater physical connection between the two sides of their brains (which might be of the scale that could be measured in this study?) which would seem to indicate that musical training can affect brain morphology. I suppose a genetic predisposition to learn music might also account for such a result, or demonstrate a paired effect between genetic predisposition and physical practice of a skill that requires a high degree of coordination between right and left sides of the body like playing a violin,

  • Jayarava

    “To what extent does brain structure correlate with different psychological traits? ”

    Unless we have grossly misunderstood the brain at some fundamental level, the answer is 100% ± 0%. Every mental phenomena is 100% produced by the brain and thus correlated to brain structure. You know this.

    The interesting thing here is that some kinds of apparently similar behaviours appear not to be strongly correlated to similarity of brain morphology.

    What you miss in your critique is that In schizophrenia and autism the aetiology varies *massively* across individuals (hence the autism “spectrum”). The singular labels belie the complexity of both syndromes. I would not expect all sufferers of schizophrenia or autism to have similar brain structures, simply because they might have different causes and manifest in completely different ways across individuals.

    By contrast the aetiology of Alzheimer’s shows much smaller variations by comparison. So again, the result is what we would predict.

    Parkinson’s is interesting, but then we don’t know what causes it – probably a range of genetic and environmental factors. We do know that it affects quite a small part of the brain. And the method in the articles seems to compare whole brains. So maybe that’s it?

    • John Malato

      “Every mental phenomena is 100% produced by the brain and thus correlated to brain structure.”

      What is your evidence? This seems like an assumption, not a statement of knowledge.

      • Jayarava

        It’s the definition of a mental phenomena. If the phenomena is not produced by the brain it is not a mental phenomena, but some other kind of phenomena.

        But never fear you can easily disprove this conjecture by showing us a counter-example. I.e. a mental phenomenon which is not produced by the brain or which is not correlated in any way with brain structure.

        • John Malato

          I asked you for evidence for the claim that every mental phenomena is produced by the brain, and you did not provide any. Instead, you restated your assumption by providing a “definition.” But definitions are accepted meanings of words. In this case, your “definition” is not accepted in any but the emergentist theory of mind.

          The philosophical obstacles associated with emergentism notwithstanding, your definition also excludes the possibility of multiple realizability, unless you define “brain” as anything that produces mental content. In that case you’re committed to the idea that mental states are just things produced by brains, and brains are things that produce minds. Which, needless to say, is extraordinarily vague.

          So there’s no need for me need to “disprove your conjecture” because, as you’ve aptly characterized it, your “definition” is just a conjecture. You are forming a theory of mind based on incomplete information about the nature of consciousness. You’re suggesting that all mind states are produced by brain states using induction. There are two problems with this.

          1. The burden is on you to show why “all mental states are produced by brain states” is a reasonable inductive inference on the basis of the much less robust observation that some mental states are correlated with brain states.

          2. Because you’re conflating metaphysical questions about mind-body interaction for physical questions about mind-body correlation, your proposal for my only avenue of rebuttal is inadequate. That every mental state is *correlated* with a physical state is a physical claim. But even if every mental state is correlated with a brain state, the claim that “the brain *causes* the mind” is a metaphysical claim which is not a matter of deduction from the physical facts.

          • Jayarava

            I reject dualism and I reject positivism. So I cannot respond within the framework you are insisting on. I feel no obligation, certainly no “burden”, to have this discussion on these terms.

          • John Malato

            I take your commitments to be:

            1. The brain causes the mind
            2. Monism

            If you’re a monist, then you’re either a physicalist, idealist or panpsychist. If you’re a panpsychist then you don’t need brains to have minds. If you’re an idealist, then the mind causes the brain. So you must be a physicalist.

            If you’re a physicalist, and a monist, then you’re a reductive physicalist. You can be of some different types. Type A is essentially an elimitivist position. Since you hold that brains cause minds, you hold that minds exist. So you’re not type A. Type B is the idea that the mind and the brain are epistemically, but not ontological, different. The mind is another way of knowing things about the brain, but it isn’t categorically different. Type C suggests that this epistemic gap can be closed in principle. I’ll lump type C with B and just call it B. I take you to be type B.

            We can restate your commitments:

            1. The brain causes the mind
            2. Type-B physicalism

            You’re suggesting that because you’re not a dualist you don’t need to defend the claim that the brain causes the mind. That’s not entirely right. You don’t need to defend that specific claim because you’re a type B physicalist. That is, I’ve shown that you can be a monist (an idealist or panpsychist) and hold that the brain does not cause the mind. But even with that correction in mind, since you’re relying on type B physicalism to defend the claim that the brain causes the mind, you would need to defend type B physicalism.

            Of course, you don’t feel the need to have that discussion or defend your position, so I don’t expect you to actually argue for type B physicalism. This was just an FYI.

          • Jayarava

            My considered philosophical position is this. In terms of ontology, I agree that substance reductionism describes what the world is made of. However the stuff so described is made into structures that are *real*. So I combine substance reductionism with structure antireductionism. I think that it is beyond reasonable doubt (see my epistemology below) that reality is mind-independent, immanent, and natural. I am persuaded that causality is metaphysical and therefore cannot contribute to definitions of reality.

            My epistemology acknowledges that the individual is limited by transcendental idealism, but I argue that collective empiricism (observing and comparing notes) allows us to infer accurate and precise knowledge about reality and that we have accumulated a good deal of such knowledge to date. I call this collective empirical realism. Causality plays a part in our epistemology because of how we interact with the world and the definitional role interaction plays in constructing knowledge.

            Given these starting points, I can see a place for an epistemological distinction between mental phenomena and physical phenomena. But I can see no basis for an ontological distinction. Therefore I don’t have a mind-body problem and I don’t feel compelled to adopt a solution to a problem that does not exist.

            I’m confident that any accurate and precise description of mental states that eventually emerges will be consistent with this ontology, but we don’t have that description as yet. If it turns out that we find a description that is inconsistent with this ontology, then I will change my mind. A Bayesian approach suggests this is pretty unlikely and the most likely scenario is where I direct my efforts until I have reason not to. Being a tooth-fairy agnostic is just a fig leaf I don’t feel compelled to wear.

            So really I don’t fit into the narrow categories of your aggressive pigeon-holing exercise. I am not motivated to defend myself against strawman arguments.

            You clearly have some academic notions of philosophy, but no real love of wisdom. Like many amateur philosophers you can argue, and are motivated to win arguments, but you *cannot think*. And sadly you don’t seem able to express yourself without your intense self-hatred showing. So I will be blocking your account shortly. And will not give you a second though afterwards. You are nothing to me. Nothing.

          • John Malato

            “Given these starting points, I can see a place for an epistemological distinction between mental phenomena and physical phenomena. But I can see no basis for an ontological distinction. ”

            This is word for word the Type B physicalism I described and into which I’m apparently “pigeon holing” you. Unfortunately, this doesn’t dissolve the mind body problem. No philosopher actually thinks it does.

            In general, your philosophical notions appear to be folk intuitions dressed up with mostly misapplied jargon. Your argumentative style is rude and presumptuous. I would advise against attacking people in the future, stick to their arguments.

    • Ariel Perez

      I agree!

  • Ulrich Weihler

    So maybe they should do fMRIs and laboratory for diagnosing people and giving them medication having influence on their behavior and brain size. Why are they not regulated or nobody accuses them like criminal drug dealers.

  • Pingback: Alzheimer’s Patients And Dolls Studies Weekly | Alzeimers Treatment()

  • Pingback: Violent abstractions (2/2) - Bodhisat()



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar