WTF Cerebellum - Little Brain, Big Deal

I did not until recently even consider the cerebellum when thinking about psychiatric conditions, but the more I read, the more I wonder why the cerebellum is not considered a potential important player in nearly every psychiatric disorder.  Although it can be said that all brain regions primarily function to make predictions, the cerebellum is especially active at refining impromptu predictions through short periods of time as sensory data changes to help us better navigate the world, not only in physical space, but our entire internal space.

Please leave feedback at https://www.psydactic.com.

References and readings (when available) are posted at the end of each episode transcript, located at psydactic.buzzsprout.com. All opinions expressed in this podcast are exclusively those of the person speaking and should not be confused with the opinions of anyone else. We reserve the right to be wrong. Nothing in this podcast should be treated as individual medical advice.

Transcript

WTF Cerebellum - Little Brain, Big Deal

Welcome to PsyDactic.  Today is Friday, May 3, 2024. I am Dr. O’Leary, a nearly full grown psychiatry resident in the national capital region.  What that means is that next month, I will graduate from residency, and aside from getting a big raise, I will also be able to be called “attending” or “staff psychiatrist.”  I started this podcast about halfway through my second year and this is the 58th episode.  I started it, thinking that it would be a structured resource for residents for learning various topics, but it became much more than that.  Everytime I try to do a deep dive into a particular topic, I get inspired.  There is so much we do not know about psychiatry and neuroscience, but there is so much that we are learning.  It can be hard to find all the information and make sense of it.  If you are willing to let me, I will try to do that for you.  I will read a bunch of papers, listen to hours of talks, and then try to say something smart about it.  I do not always succeed at being smart, but despite the fact that I often cringe at what I have said in the past, I keep doing it because for me, it has been the most rewarding learning experience of my life.  Forcing myself to do this podcast makes me feel a constant sense of renewal and also of accomplishment.  No one else has told me to do it, and I am the only one responsible for its content.  For new listeners, thank you for arriving.  For repeat listeners, thank you for coming back.  Today, I am going to discuss what just might be the most under-discussed part of the brain: the cerebellum.

As medical students, I was taught that the primary function of the cerebellum is to fine-tune our motor movements by responding to the sense of our body parts in space (which we call proprioception).  The classic clinic test for doctors is to have a patient touch their nose then have them try to touch the doctor’s finger that is held about 2 feet in front of the patient.  A classic sign that there is a cerebellar problem is that the finger has a hard time trying to stay on track, not due to some kind of tremor, but due to the brain’s inability to adjust the arm’s position with precision.  Especially as it gets further from the body, the finger deviates even more from the target and has to be reigned back in in a jerky fashion.

I did not until recently even consider the cerebellum when thinking about psychiatric conditions, but the more I read, the more I wonder why the cerebellum is not considered a potential important player in nearly every psychiatric disorder.  Although it can be said that all brain regions primarily function to make predictions, the cerebellum is especially active at refining impromptu predictions through short periods of time as sensory data changes to help us better navigate the world, not only in physical space, but our entire internal space.  

The cerebellum gets ignored quite often because it is hiding in the back of the skull, tucked away under a sheet of dura mater, and seemingly inconsequential for most of what we think and do.  When it doesn’t work, we can still move, we are just clumsy.  If I were to tell you that our cerebral cortex has about 16 billion neurons, how many neurons would you estimate the relatively diminutive cerebellum has?  Four?  Two?  The answer is about 69 billion neurons.

Have you ever heard that we use only 10% of our brains?  This false fact, in part, comes from the fact that, when we first started measuring the brain, we could only really measure what the brain could sense and what actions it could make our muscles do.  The prefrontal cortex for example, was thought to be an extraneous silent mass.  We now know that without the prefrontal cortex, we wouldn’t be able to learn about the prefrontal cortex.  The same is true for the cerebellum.

The cerebellum is connected to every other brain region extensively.  It has more dedicated computing power in its pinky than the cortex can muster with every region on overdrive.  It needs this computing power because it appears that the cerebellum is constantly building impromptu models, not only of the recent present, but simultaneously of the near future.  How else could it keep that finger on track without wandering off in some random direction.  Based on the current trajectory, it needs to be able to predict in milliseconds how to adjust our motor output, and not just calculate that, but communicate that to our premotor and motor cortex and our striatum.

It can do this for far more than fingers, and it can do it while simultaneously rebuilding its model every time an error is detected.  The cerebellum optimizes our movement, AND it also optimizes our self-referential thoughts, our immediate and future plans, our social awareness, and in short increases the precision of the predictions that every other part of the brain is making.  Even cooler is the fact that the neurons throughout the cerebellum are arranged in a universally preserved motif, which means that the function they perform is basically the same (whether you are moving or responding to a joke).  The same kind of function is useful for everything we think and do, similar to how a transformer like those used in AI can use the same basic architecture to output probabilities about just about anything and help select the next best response.

I am so impressed with the cerebellum right now.  It is my new favorite neuronal mass.  For those who have listened to this podcast before, you know I nerd out over Bayesian reasoning, and have characterized the brain as more or less a giant bayesian calculator, but it wasn’t until I started reading about the cerebellum that I finally started to understand how powerful this bayesian calculator can actually be.  For those who need a really quick primer, Bayesian reasoning is a way to use conditional probabilities to predict the future.  Let me put that into some necessarily mathy English:  It can calculate the probability of some outcome given the current state of the world.  Calculating the outcome relies on knowing that the current state it is inferring is true.  So before I inspire you to turn off this podcast let me talk a little about how this matters.

Let’s say someone says this sentence, “You're such an asshole,” and then they laugh.  This could mean many different things.  They could be ridiculing you, or criticizing you, or accusing you.  They might also be ribbing you, or cajoling you, or sharing some kind of inside joke.  Almost certainly, when something like this is said by someone, and they have never said it to you before, your brain throws an error signal.  Suddenly, a lot of processing power is recruited to figure out what this means.  It gives you pause.  During that pause, your cerebellum is going wild trying to help you calculate the most probable meaning of this statement.

In the 1990s, the first well supported hypotheses about the cerebellum's role in psychology began to appear.  For example, a 1998 paper by Andreasen, Paradiso, and O’Leary (no close relation of mine that I am aware of), discussed an idea called “Cognitive Dysmetria.”  See the show transcript at PsyDactic.Buzzsprout.Com for the references.  Think about the example of the ass-hole statement I just gave.  There are many directions that any response could go.  A person could not respond at all.  They may laugh.  They may act indignant.  They may verbally attack the person saying it.  All of these directions are decided in your brain based on the probability that any of them is the best action.  Your affective state, sensory information, working memory, present goals, and available memories (among other things) are all contributing to the calculation of the best response, and your cerebellum is doing a lot of this math.

When the cerebellum is doing math related to your movements, it has direct sensory fibers to the body feeding it information and it is also getting information from the motor and premotor cortex.  When it is doing math about social situations, it is also getting a load of information about what is happening in the cortex and trying to figure out what is the most important information.  It is easier and more obvious for us to see the cerebellum’s effects on positioning our body in space than it is to see its effects on everything else the brain does.  It is also easier to do.  The cerebellum is comparing direct signals about our body’s position in space when we are moving or trying to stay still (but still moving a little bit), and processing the errors between the predicted and actual placement of our body.  In this way, it sends signals to the thalamus and cortex that make us immediately adjust.  How does it adjust our position in social space? Or temporal space? Or affective space?

This brings me to Cerebellar Cognitive Affective Syndrome, sometimes referred to as Schmahmann syndrome after Dr. Jeremy Schmahmann, the Professor of Neurology at Harvard Medical School who has described it.  Let me give you an abbreviated quote from the paper The Cerebellar Cognitive Affective/Schmahmann Syndrome: a Task Force Paper from the journal Cerebellum, published in 2020.  Here I summarize what parts of the cerebellum have been consistently associated with different cognitive and affective states.

QUOTE “[D]eficits in linguistic, visuospatial, and executive function were held to result from the disrupted connectivity between the posterior… lobe … and cerebral association areas, especially prefrontal cortical areas in relation to executive control, parietal cortical areas with respect to visuospatial function, and frontotemporal regions in relation to linguistic function. Affective-emotional disturbance was seen as associated with lesions in the “limbic cerebellum,” associated with the vermis and fastigial nuclei connections with the reticular nuclei in the brainstem, intralaminar and anterior thalamic nuclei, the hypothalamus, as well as with the hippocampus, septum, amygdala, ventral tegmental area, periaqueductal gray and mammillary bodies, cingulate gyrus, and pregenual, retrosplenial, and paralimbic neocortical re- gions”

I encourage you to go to YouTube and search for Schmahmann Syndrome or Cerebellar Cognitive Affective Syndrome because I cannot do it justice here.  He reports that children born without a cerebellum actually have relatively minor motor problems compared to substantial deficits in intellect and language, visual spatial processing and executive function.   The posterior lobe of the cerebellum in particular appears to be integral in language, cognition, spatial learning, executive functions, and affective learning.  The problems with language in persons with cerebellar injuries, lesions, or developmental problems are not aphasias.  They can speak, but have particular difficulty with things like ambiguity, sarcasm, implied meaning given the context, metaphors, and inferences.  Additionally, the cerebellum appears to be necessary for theory of mind.

Dr. Schmahmann also describes multiple patients with Chiari malformations (these are malformations of the skull that allow the cerebellum to hernia or move into a position that causes it to get compressed and malfunction.  The possible manifestations of Chiari malformations are extensive, which reflects how it affects the entire brain.  He describes patients who are acting in ways that might get them labeled as having a psychotic disorder or personality disorder or obsessive compulsive disorder or impulse control disorder, et cetera, whose symptoms dramatically and permanently improve once surgeons repair the Chiari malformation.

Remember that to execute movements, relatively small portions of the cerebellum are activated.  These regions compare sensory feedback with the feedback it predicted it would get.  Then it can find mismatches and send signals to the motor cortex to correct the path or the force used.  How does this relate to social learning?  If a joke I tell does not result in laughter does the cerebellum light up?  If the parts of the cerebellum that are supposed to process errors in social communication are not correcting errors, does this result in merely an absence of awareness of the social situation or does it result in clumsy social interactions?  What happens for more complex feedback, like facial expressions or disembodied vocalizations?  If the cerebellum is malfunctioning, does this help explain why someone with autism might be able to recognize that someone is angry, but not connect it to the context, and react either with withdrawal, dysphoria, or even aggression?

There cerebellum seems to be a module for unsupervised learning where it has to figure out relationships without being given any explicit instructions.  People who lack cerebellar function in certain areas may be able to compensate for this by more supervised type learning, where they have others specifically tell them how to act in certain social situations, what certain gestures or expressions by others mean, etc. Things like social stories and rehearsals can be helpful for this kind of learning.  It is understandably more concrete and inflexible, but it is certainly better than nothing.  When someone says, “Hello, how are you,” you say, “Hello. I am ok, how are you.”

One of the researchers in this field is Dr. Anila D’Mello.  She is Assistant Professor and Jon Heighten Scholar in Autism Research in the Department of Psychiatry and O'Donnell Brain Institute at UT Southwestern.  Her research has linked language regions of our left frontal cortex to their corresponding regions of the cerebellar cortex, especially in processing semantic information.  For example the cerebellum lights up when presented with grammatically correct but implausible sentences.  Her example was “The donut sweeps the steps.”  I wonder what the cerebellum is doing when children read Alice in Wonderland for the first time.  In short, subjects with greater cerebellar activity during tasks involving identifying semantic errors were also better at the task.  In a similar study, when presented with very predictable sentences and ask to choose the next best word like “Two plus two is Four or Boat” only a very small region of the cerebellum lights up, but when presented with something like “The man looked at the… Seat or Birds,”  the outcome is less predictable, so more of the cerebellum lights up.  When presented with language that is completely non-predictable such as “bait, ask, best, rent” and asked if the next best word is “act” or “sand” the cerebellum goes wild.

She also studied cerebellar activation in social learning in a task that involved tossing a ball to different players who were supposed to toss them back, but in some cases were unreliable.  Players quickly learned who the bad players were, and stopped tossing the ball to them.  Regions of the cerebellum not involved in motor functions appeared to be activated while having to make this decision.  They then used TMS to inhibit the regions that were responsible for this change in behavior and found that players no longer differentiated between those who threw the ball back and those who didn’t.  It appears that their ability to predict the future was impaired.

D’Mello also used TMS to model motor versus cognitive associated regions of the cerebellum and looked at their effect on speech.  For the motor regions, syllable articulation was affected, but semantic fluency was preserved.  This relationship was opposite when cognitive association regions were stimulated.  Articulation was preserved, but fluency was impaired.

They also demonstrated in separate experiments that the cortical regions associated with regions receiving excitatory TMS experience greater connectivity.  This is similar to what we expect makes TMS of the dorsolateral prefrontal cortex effective for depression.  It is not the DLPFC alone that is thought to change our mood state, but the downstream effects on things like the anterior cingulate and the nucleus accumbens that appear to make the difference.

One more honorable mention that D’Mello talks about is that, through some method I didn't quite understand, they were able to demonstrate that the activity in the cerebellum was not being driven by activity in the cerebral cortex nearly to the same degree that activity in the cerebral cortex was being driven by the cerebellum.  The cerebellum was the thing making things happen.  Another fascinating fact (I know I said “one more” already, but I can’t help myself) is that while damage to the cortex early in life is often able to be overcome by the brain's ability to repair itself and its plasticity in functioning, damage to the cerebellum early in life results in persistent deficits that the brain cannot overcome.

Now consider autism.  Take any group of humans with autism.  Because the features of autism relate to complex social behaviors and personal interests this group is necessarily a very heterogeneous mix.  Neurodivergence is a relatively recent word that attempts to capture this.  As the saying goes, “If you know one person with autism, then you know one person with autism.”  It makes sense then, that since the cerebellum is so widely distributed in its influence, it could play a central role in the features of neurodivergent humans.  Think about what the cerebellum does. It fine tunes our cognitive gymnastics, our social ballet, our semantic flexibility, our choice of behavior, and our adaptability to changing sensory stimuli.

All of these are disorders of prediction.  Now let me give a shout out to Sinha et al who published a paper titled “Autism as a Disorder of Prediction” in the PNAS in October of 2014.  In this paper they discuss how many of the phenotypic features of autism including insistence on sameness, sensory hypersensitivity, difficulties interacting with dynamic objects, and difficulties with theory of mind are all problems with making predictions about the immediate or near future.  Additionally, what they call islands of proficiency including mathematics, static form coherence, visual search, block design tasks, calendar calculations, musical performance, and drawing abilities, which are on average better accomplished by those diagnosed with autism may be this way because they represent rule based phenomena where there is less uncertainty.  If you know the rules, you can perform the task.  This combined with a tendency to restrict interests gives autistic individuals areas of particular strength.

They also make some particular predictions based on this hypothesis, including that individuals with autism will be found to have hyperplastic interests.  This means that, because their impaired predictive abilities (other than strictly rule based ones) are impaired, they will experience more of the world and novel, which means they may respond to things we think are mundane as particularly salient, in an interesting or avoidant way.  They also hypothesized that individuals with impaired predictive abilities in general would have reduced appreciation of humor.  This is because one of the bases of humor is violating what is expected.  Someone with autism may not be anticipating a comedian to act in a certain way and therefore would not be impressed when she violates a normative expectation.  They may instead, be either uninterested or confused.  Another predicted consequence of impaired ability for prediction is reduced motor anticipation.  They define this as reduced ability to adjust to things like uneven terrain with the same speed as someone without autism.  It may also manifest as poor postural control, meaning that they would be more likely to be off balance or would experience a greater degree of sway of their midline when standing still.  There are studies that appear to support this.

I have only scratched the surface of what we now know about cerebellum and humans have only scratched the surface of what can be known about the most densely packed processor in our brain.  I could go on, but I need to stop.  Wait,  I am not going to stop.  One more thing.  I mean it this time.  So, cerebellar activation has consistently been shown to be less in individuals with autism than controls.  However, we know that individuals with autism often have very specific and narrow interests, which is not atypical of other children.  It is the amount of time neurodivergent people spend with their particular interest that is abnormal.  D’Mello and colleagues also looked at cerebellar activation in children with autism versus controls, and while they found that for stories that were not related to an autistic person's particular interests, they did have very little activation in the cerebellum.  However, when a story containing their interest was introduced, their fMRI started to look like control children who also were given highly interesting stories.  This is very cool, because it suggests that highly rewarding stimuli may be able to make an autistic person’s cerebellum function more like a more typical cerebellum.

Thank you for listening.  I am Dr. O and this has been an episode of PsyDactic.

Ilaria Carta et al. ,Cerebellar modulation of the reward circuitry and social behavior. Science 363, eaav0581 (2019). DOI:10.1126/science.aav0581

Vandervert Cerebellum & Ataxias (2016) 3:10 DOI 10.1186/s40673-016-0049-z

Seidler RD, Kwak Y, Fling BW, Bernard JA. Neurocognitive mechanisms of error-based motor learning. Adv Exp Med Biol. 2013;782:39-60. doi: 10.1007/978-1-4614-5465-6_3. PMID: 23296480; PMCID: PMC3817858.

Anna Nakamura, Yukihito Yomogida, Miho Ota, Junko Matsuo, Ikki Ishida, Shinsuke Hidese, Hiroshi Kunugi. The cerebellum as a moderator of negative bias of facial expression processing in depressive patients. Journal of Affective Disorders Reports. Volume 7. 2022. https://doi.org/10.1016/j.jadr.2021.100295.

https://www.youtube.com/watch?v=NVsrexn3pT8&t=2372s (Cerebellum Anatomy and Neuroscience)

https://youtu.be/3nYFbuzceDI?si=0j-S0vRIQfMZ5b3b 

https://www.youtube.com/watch?v=P-UNk7idQ9Y&t=205s

https://www.youtube.com/watch?v=J4hrxkA5lVc (The Cerebellum in Cognition, Development and Disorders with Anila D'Mello, PhD)

Argyropoulos GPD, van Dun K, Adamaszek M, Leggio M, Manto M, Masciullo M, Molinari M, Stoodley CJ, Van Overwalle F, Ivry RB, Schmahmann JD. The Cerebellar Cognitive Affective/Schmahmann Syndrome: a Task Force Paper. Cerebellum. 2020 Feb;19(1):102-125. doi: 10.1007/s12311-019-01068-8. PMID: 31522332; PMCID: PMC6978293.

Ferrari, C., Ciricugno, A., Urgesi, C., & Cattaneo, Z. (2022). Cerebellar contribution to emotional body language perception: A TMS study. Social Cognitive and Affective Neuroscience, 17(1), 81–90. https://doi.org/10.1093/scan/nsz074

Andreasen NC, Paradiso S, O'Leary DS. "Cognitive dysmetria" as an integrative theory of schizophrenia: a dysfunction in cortical-subcortical-cerebellar circuitry? Schizophr Bull. 1998;24(2):203-18. doi: 10.1093/oxfordjournals.schbul.a033321. PMID: 9613621.

Phillips JR, Hewedi DH, Eissa AM, Moustafa AA. The cerebellum and psychiatric disorders. Front Public Health. 2015 May 5;3:66. doi: 10.3389/fpubh.2015.00066. PMID: 26000269; PMCID: PMC4419550.

Sinha, Pawan et al. Autism as a disorder of prediction. PNAS. October 6, 2014 111 (42) 15220-15225. https://doi.org/10.1073/pnas.1416797111