The cerebellum is a really weird part of the brain in that it contains the majority of the particular type of cells that neuroscience has been focused on for the last 100 years… but is also the most ignored part of nervous systems as a whole. Even more than the brainstem, and the brainstem is like an autistic pyromaniac stepchild to most researchers. That is to say it has more neurons than the rest of the nervous system combined.
I recently got asked “how should we think about the cerebellum” and got really stumped! How do we describe the area of the brain with the majority of neurons?
First, the cerebellum is *really* under-researched mostly because it’s a huge pain in the butt and a lot of the tools that we use to research cerebral cortical areas simply aren’t available for the cerebellum. EEG in particular is almost worthless because there’s simply no easy way to apply electrodes that won’t get blasted by cortical activity. The cerebellum is also many times more neuronally dense than the cerebrum, so there’s not enough separation of signal to make it effective.
Also, due to the way the cerebellum functions (it “responds” to existing information), we don’t have the easy/instant action/response mechanic that most MRI work relies on to make assertions. The density of the metabolic interactions also makes it a big mess compared to cerebral imaging studies. Another big headache is that it’s so much *faster* cycle wise than cerebral cortical areas, meaning a lot of the existing tools simply don’t work/apply to cerebellar cortical work. And on and on, but it comes down to our tools being optimized to study cerebral regions.
Underneath all that though, the cerebrum and cerebellum are functionally *very similar*. They process nearly the same information just “inverted” of each other.
Going back to the model, I frequently refer to “dorsal” and “ventral” streams. It’s 100% appropriate to refer to cerebral cortical areas as the “notepad” for the dorsal stream, and the cerebellar cortical areas as the “notepad” for the ventral stream. And it is the interplay of these two streams which creates “smooth” behavior as they work to update and refine each other on an interdependent basis.
For those wondering why I’m not mentioning the limbic or DCN circuits, it’s because it’s a little more than the scope of this post – a lot of the limbic equivalent structures for the cerebellum are in the brainstem. And explaining those interactions requires a lot of other context.
So, going back to the interdependency between the cerebellar and cerebral cortical areas, it’s more than just “providing two different approaches”, the two structures are fully functionally cross linked by specific region, and cerebellar “updates” can *directly* “write” to cerebral regions, which then gets re-processed into the stream… downstream.
Let’s take a walk through how an information flow between the two structures would work. First, our first order receptors take in the information and it’s immediately processed by astrocytes local to the receptor end points. If no pattern match is found, it “flows” downstream. Think of local astrocytes gating the information and redirecting it at this stage only if a pattern is matched.
At the same time, the brainstem itself is generating a salience pattern (or “goal state” if you prefer). When incoming data matches the salience requirements, a signal is kicked upstream to start routing that particular information back down through the brainstem for a closer look. This “stream” decomposition to pick out relevant information is a large part of the function of the hippocampal complex (and might be the reason the dorsal CA2 is the way it is, “social” data is “predicted” rather than “observed” directly, but that’s a digression to be expanded another day).
The brainstem does local pattern matching and updates the salience (“goal state”) based on that pattern matching. In order to create the “update” to the pattern, it kicks it over to the cerebellum (or DCN more appropriately) which processes the data in a goal specific way and creates a “negative differential” or the “subtractive” response. Note that “subtractive” refers to the metabolic balance between the two streams (pretty roughly the “E/I” ratio for psychology), the *effect* of the inhibition is usually perceived as *additive* however.
Rather than duplicate *all* the data in the stream, it’s more metabolically efficient to create updates ONLY for the parts relevant to the updated goal state. To do this, the cerebellum strips out all the unnecessary information and leaves only pure “goal state differentials”. This pure goal state information is then re-integrated into the cerebral cortex along the dorsal/ventral axis of the brains functional units (e.g. limbic regions).
It’s probably safe to say that the cerebellum allows *efficient* updates of goal states.
I’ve previously talked about the “egocentric” vs. “allocentric” transform, however this is just one part of *how* it filters for goal state updates, rather than the primary function of the region. Ultimately it’s concern is creating efficient, smooth goal state updates to be processed down stream.
Okay, so let’s try to apply this back upstream to more expressed behavior. First, the most discussed function of the cerebellum historically has been related to movement, particularly in ataxias. Cerebellar degeneration under this model implies that smooth updates are not being generated and we see that obviously in movement disorders.
Switching context to something like “anhedonia” or “motivation” related behavioral shifts, we can apply these also generally to lack of goal state updates via the cerebellum. If we think of “feelings” as the salience pulse, “blunted” feelings are literally reduced metabolic effect of transmission of goal state.
This carries over to conditions like “depression” as well, where some or all goal state updates are being metabolically suppressed because integrating the goal state is too metabolically expensive (or the region doing the integration has inadequate metabolic capacity for the integration). And since “depression” and “anxiety” are the same thing, this applies in reverse there, either the goal state is being hammered out too “hot” or the regions doing the integration of the goal state are running too hot causing an imbalance between the streams.
Deeper into the psychiatric well, we could view “OCD” as a goal state update hammering really hot, so hot that it overwhelms the dorsal side of the integration regions which are then unable to provide feedback to the brainstem about current goal state.
tl;dr, the cerebellum functionally creates metabolically efficient goal/salience updates. Really effective cerebellar processing “feels like” “effortless motivation” and I’d argue maybe “happiness”? Ineffective cerebellar processing “feels like” depression/anxiety but also usually manifests as “personality” or “perception” disorders because it’s sending “inappropriate” to the stream integration process.
(will source this up later)