While it’s not advisable to point out your own failures, I do like upcycling my losing contest entries. And I do like sharing opportunities with fellow budding writers–whether it’s the opportunity to use the contest-holders as a resource, or to learn from the winners. To that end, I’ll admit that this was a losing entry in Oxbridge Biotech’s 2014 Science Writing Competition. They’ll be showcasing the finalists in the upcoming weeks, and the first of them is here. Let’s all learn from them, shall we?
For my own piece, I chose to write about a weird result that kicked off dozens of follow-up studies in our lab. The silver lining of not being a finalist is that I could tweak it before posting it here (read: add profanity where it helps). I hope this provides a helpful mnemonic for remembering what the cerebellum does. In short, it is the honey badger of brain structures: it doesn’t give a shit. This is its story.
We’ve known since George Stratton wore prism goggles around town in the 1890s that if you alter someone’s sensory input, flipping their vision upside-down or simply shifting their vision some amount to the right or left, the brain adapts to this so that the person can function in the world again. It’s even possible to learn to ride a bike with these goggles on. The brain doesn’t overcome the disturbance by thinking about what’s happened and concentrating hard on counteracting it. It happens naturally, and we think it happens in a part of the brain called the cerebellum, mainly because people who suffer from cerebellar damage are really bad at adapting to wearing prism goggles.
The cerebellum a mystery to most neuroscientists. It’s an evolutionarily older part of the brain. It has more neurons than the whole rest of the brain combined. Its neurons are organized into a strikingly uniform and gridlike architecture. And it is involved in seemingly everything from low-level motor control to high-level functions like attention to the pathology of disorders like schizophrenia and autism. Its architectural regularity has led researchers to hypothesize that the cerebellum must perform a single function, and its heavy interconnectedness with other brain regions has dictated that this function must be some computation common to all thought.
Solving this mystery, then, is a tall order. The cerebellum is consistently active during tasks that require remapping between sensory inputs and motor outputs, like adapting to prism goggles. The closest we’ve come to consensus is this: it compares the predicted consequences of our actions to the actual feedback we receive. When you throw something and miss, the cerebellum detects the mismatch between your expectations and reality and signals to you that you made a mistake. This error signal tunes up the motor circuits that caused the mistake, making it less likely that you’ll repeat the mistake in the future. But what we are uncertain of is how our explicit knowledge (knowing that) and implicit knowledge (knowing how) interact as we learn from such mistakes.
In 2006, Pietro Mazzoni and Jon Krakauer, then at Columbia University, published the results of an experiment designed to take apart the effects of an explicit strategy and implicit adaptation. They first had people reach for 8 targets arranged like numbers on the face of a clock, one at a time. After a while, their cursor was suddenly rotated (In clock terms, imagine reaching for your 12:00 and seeing your cursor headed for 2:00), creating a mismatch between where they expected the cursor to go and where it did go. People gradually shifted their arm movements in the opposite direction (10:00) and managed to resume hitting the target accurately with the cursor.
In a second experiment, the experimenter fed people the winning strategy: if your cursor is headed towards 2:00 instead of 12:00, try aiming for 10:00 to counteract the rotation. Easy, right? Well, people actually did worse. What was the problem? Reaching for the 10 should have been just as easy as reaching for the 12. And yet, trial after trial, people saw their cursor drifting away from the target. Despite possessing the recipe for success, the people’s hands were adapting to the rotation just like the people who hadn’t been given the strategy, and it was ruining their performance! The researchers hypothesized that the cerebellum must have kept chugging right along, tuning up those circuits to eliminate the mismatch between where the person was aiming and where the cursor went–even though the person was aware that they were intentionally aiming elsewhere.
Jordan Taylor and Rich Ivry, at UC-Berkeley, wondered if, given enough time, people could halt this rampant drift. And it turns out they could. It just took a while. Now they really had a puzzle–there seemed to be two processes, the “smart” strategic one that was harmful if handed out for free, but seemed to also be responsible for people’s ability to stop drifting and turn their performance around, and the “dumb” one that adapted people’s movements, sometimes even in spite of the strategy. This makes sense–try to remember whether your front door lock turns left or right. It’s tough to do with the smarter, “thinky” parts of your brain, and yet the next time you approach that lock, key in hand, your hand will know what to do. They reasoned that there must be two kinds of error signals–one that says “hey, you didn’t hit the target” and another that says “hey, you didn’t go where you were aiming.” Usually, there’s no difference between these two. But when you have to aim for 10:00 to get your cursor to hit 12:00, they’re divorced. And nobody told the cerebellum. The cerebellum works in blissful ignorance to reduce your aiming error, minimizing the difference between your expectations and reality. There must be some other part of the brain, they thought, that senses the growing target error and enacts a change in strategy to help you reach your real goals.
The prefrontal cortex, for all its love of strategies, rules, and goals, fit the bill. To test this, Taylor and Ivry rounded up patients with damage to either the cerebellum or the prefrontal cortex. They had them reach for targets, and when their feedback was first rotated, they fed them the winning strategy: to reach for 10:00 to hit 12:00. The cerebellar damage patients, unlike the healthy controls, were able to reach for 10:00 without their hand drifting inexplicably away over time. Their performance remained perfect, never suffering from the unintentional drift that a working cerebellum would have caused. This is one of the juiciest results a scientist can hope for: a patient population that is actually better at something than healthy controls. The prefrontal damage patients, on the other hand, drifted away from the target like runaway trains, and never managed to get back on track like the healthy controls could. Lacking the part of their brain that could represent the goal of hitting the target, the cerebellum kept chipping away at their aiming error, unaware of their massive mounting target error. They couldn’t access the strategy required to turn their deteriorating performance around.
So, then, it seems that we have strong evidence to localize explicit and implicit knowledge in the prefrontal cortex and cerebellum, respectively. And no brain scans, electrodes, or lasers were involved in the gathering of this evidence. In an age where the public perception of science is fixated on new technologies that can produce pretty pictures, behavioral neuroscientists display a great deal of moxie. They must be both detectives and storytellers. They engage the brain (intact or otherwise) in a game of 20 questions. They set behavioral traps. They ask humans to perform just the right task that will reveal the boundaries of their brain’s capabilities. And when they ask nicely in just the right way, the data tell a story not only about behavior, but about which parts of the brain shape that behavior.
Here, we learned about both. It seems that the cerebellum just wants to make our expectations match up with reality, regardless of whether we have smarter goals represented somewhere in the prefrontal cortex. Like the honey badger, the cerebellum doesn’t give a shit. So, for all the pride humans take in our evolutionarily newer machinery, there’s a humbling message here about the forces that shape our behavior whether we like it or not. They hum along beneath our intelligent veneers. Despite our best efforts at thinking things through, not all of our faculties are designed to guide us to the right answers in all contexts. Seeking out these liminal contexts may reveal or own suboptimal or idiosyncratic behaviors and thought patterns. Finding scenarios where we fail (or would fail, if not for the intercession of compensatory efforts on the part of other, smarter brain parts) makes us all the more aware of what makes us human. We are driven by smart and dumb forces alike. And that’s a good thing. Most of the time, anyway.