Studying for LSAT alters connectivity in the brain’s reasoning circuits

(… Or as some will say, “makes you smarter”. Hmmm…) Image: View from where I'm sitting right now. Fish oil, magnesium, water and LSAT.

The brain never stops changing and adjusting – in structure and connectivity – to new learning and life experiences. Think about the last time you learned a new motor skill. What began as difficult and awkward movements slowly transformed into something automated that you don’t even need to think about. Thank your brain’s malleability (aka “neuroplasticity” in neuro-lingo) for that.

To some, the fact that your brain changes in response to motor learning may seem almost, well, intuitive. What about more abstract types of learning? Something central to intelligence, like the ability to use logic? How would learning to reason change the brain?

Mackey et al (2013) Intensive reasoning training alters patterns of brain connectivity at rest. Journal of Neuroscience, 33(11): 4796-4903

Mackey et al (2012) Experience-dependent plasticity in white matter microstructure: reasoning training alters structural connectivity. Frontiers in Neuroanatomy, 6 (32): 1-9

In search of a real-world setting, researchers turned to LSAT participants. Since LSAT determines the caliber of law school students can get into, researchers reasoned that they would be more motivated to study for the LSAT than a similar laboratory task. For those unfamiliar with the law school admission test, it evaluates a person’s ability to identify logical fallacies, strengthen or weaken arguments, identify assumptions and reading comprehension.

Researchers recruited 25 prelaw students enrolled in an LSAT course and imaged their brains at rest with fMRI both before and after 90 days of intensive reasoning training. As controls, researchers also include 24 age- and IQ-matched prelaw students intending to take LSAT in the future. The “resting state of the brain” is actually a misnomer. The brain is never completely silent; spontaneous, synchronized fluctuations in blood oxygenation levels occur even when the person is not actively “thinking” about a task. Resting-state fMRI tries to find correlations in the activity patterns of different brain parts, which can tell us how functionally connected these areas are. When connected, the activity spreads like “waves” from one brain area to the other.

Not surprisingly, the scores of most LSAT-takers improved after training, especially in the reasoning domain (not so much reading comprehension). At the neural level, researchers found significantly increased connectivity between the parietal and prefrontal cortices and the striatum, both within the left hemisphere and across hemispheres.

Left image, purple and indigo: Prefrontal cortex; left image, light blue: parietal cortex; Right image, yellow: striatum et al (aka "midbrain")

What do these brain regions DO? In one theory, strictly with regards to reasoning, the parietal cortex acts to keep in mind individual relations between tidbits of information (A->B, B->C), and the prefrontal cortex compares or integrates the available information to reach a conclusion (A->B->C so A->C!). The striatum -best known for its involvement in motor skill learning and addiction–encodes errors in reward prediction (oh, why didn’t I get this right?) and helps with flexible problem solving (maybe I should change strategies). Note that these are broad generalizations, and each brain region is involved in many, many more functions.

Overall, it seems that intensive and targeted logic training increased synchronized activity between different parts of the reasoning circuit, both within the left-brain and between hemispheres. Many people think that reasoning is largely left-brain dominant - although this is a gross simplification - as similar regions in the right-brain can also be recruited to support complex reasoning.

Thicker lines means more increase in connection. L: left-brain, R: right-brain.
PFC: Prefrontal cortex, Str: striatum, Parietal: well, you know.

The researchers next asked which connections were most related to improvement on the LSAT. Looking across 231 different connections, the increase in activity coupling between the left parietal and right prefrontal cortex, as shown above, most strongly correlated with increased performance. The data also suggested that how much a student’s test score improved correlated with how much change went on in their reasoning circuit – however, this correlation did not survive a more stringent statistical test.

To strengthen their findings even further, researchers used diffusion tensor imagining (DTI) to look at white matter microstructure in the brain before and after LSAT preparation. DTI measures how water diffuses through neural tracts and gives info on the integrity and (to some extent) speed of neural transmissions. As with fMRI, the largest changes were found within the parietal cortex and between the right parietal and left prefrontal cortices.

In sum, 3 months’ practice of reasoning skills is enough to change connectivity within the brain, and this change may correlate with how well a person improves in his/her test score. Given that the study looked at a real-world scenario, it’s pretty incredible that the authors found consistent and significant changes in a cohort of only 25 people. It’s cool that the authors looked at resting state connectivity as opposed to brain activity during a reasoning task. This way, they show fundamental changes in the brain at baseline.

However, I would like to know if the change in brain connectivity (“strengthening” of reasoning circuits) actually translates to better performance on a new reasoning task. For example, can "priming" the brain with logic games result in quicker learning of chess? I would also love to know how long these changes last – given the fact that a delay between studying and taking the exam can drastically decrease test scores, I’m guessing that these connectivity changes aren’t there to stay. You’d probably have to regularly flex those parietal-prefrontal muscles to maintain high reasoning abilities.

I’d like to leave you with this question: given the findings of this study, do you think LSAT scores measure a person’s cognitive potential (how well they can reason given the training), or do they simply reflect his or her cognitive past (how much they’ve reasoned previously)?

For my housemate JL, who ends up herp-derped every night due to intensive LSAT cramming.

Note: This post generated quite a lot of discussion on reddit/r/neuro. The user Carcel in particular outlined the limitations of this type of "real-world" fMRI imaging research astutely. I think he/she is spot on, and I'll share the comments with you below.

I was really just critiquing the article, but honestly I'm not a huge fan of this genre of research either. I might catch some heat for this, but the "pick anything" and scan it method has always seemed somewhat scattershot to me. As such it has all the strengths and weaknesses of an exploratory approach. Sometimes you find something you didn't expect, e.g. chess grandmasters have recruited parts of the fusiform face area to analyze chess board states, but often you are just casting too wide a net.

Don't get me wrong, tools such as fMRI are huge advances, but there is a tendency to ignore their limitations. Research like this seems to come with so many caveats that their conclusions just aren't all that conclusive.

Between the relatively low spatial resolution and poor understand of functional localization there are a lot of claims being thrown around about "emotion circuits" and "reasoning circuits" and "LSAT circuits" that are really unsubstantiated. Worse yet, when you pick a complex task such as LSAT studying, or any other "normal activity", there is no assurance that every participant's experience has been the same.

So to sum it up, no there is nothing inherently wrong with this research as long as we are moderate with conclusions and acknowledge the potential errors. The more specific the choice of task and more well understood the regions they are focusing on the better the research is likely to be, but even then there are significant issues that need to be considered.