Jonathan D. Charlesworth„ Timothy L. Warren, & Michael S. Brainard (2012). Covert skill learning in a cortical-basal ganglia circuit Nature DOI: 10.1038/nature11078
This hump day, we at Sick Papes bring you an exclusive heart-to-heart poolside chat with Jonathan Charlesworth, the author of a hot-off-the-press pape that was described by one of our Sick Papes interns as, “so-sick-it’ll-make-your-antiobiotic-resistant-staph-infection-feel-like-a-sensuous-spaghetti-squash-rubdown.” Currently AOP in the journal Nature, Jonathan’s pape investigates the neural mechanisms of trial-and-error song learning in the adorable Bengalese Finch. Along with his eminent co-authors, Tim Warren and Michael Brainard (both reportedly expert tele-skiers), Jonathan found that LMAN, a region of the bird brain which was previously thought to contribute to online behavioral exploration, can still drive song learning even if its output is inactivated during training. Confused? Just read the goddamn interview:
(SP): In your pape, you demonstrate how blocking the output of the anterior forebrain pathway (AFP) prevents gradual song learning, but when AFP output is unblocked, the learned behavior suddenly appears. That’s a fantastically trippy and completely unexpected result. Did you design the experiment to test this possibility, or did you just stumble across it?
(JC): We designed it to test this, but we expected the opposite result (i.e. no learning).
(SP): It seems intuitive that motor learning develops gradually over time, building upon previous learned experienced. For example if I wanted to become a bad-ass skateboarder, I would have to learn how to stand on the board without falling over before I could pull off a kick-flip. But your data suggests that gradual online learning is not necessarily required. Does this mean I should dive directly into the half-pipe? Or do you think that the effect you describe applies only to the initial stages of very basic motor learning?
(JC): Our results suggest that (if the output of your basal ganglia is blocked), you can learn to select the best of whatever behaviors you initially could generate. So, if you are always terrible at skateboarding (as I suspect you are), then you will not improve. However, if you are sometimes terrible but occasionally do well, then you might be able to learn to produce those good performances more consistently. In our birds, they were already very good at song but performed that song in slightly different ways; after training with the output of their basal ganglia blocked, they learned to produce the “best” version of that song more consistently (where we define best based on our experimentally-imposed threshold for receiving reinforcement). We at the Brainard lab in no way endorse diving into half-pipes or inactivating the output of your basal ganglia at home.
(SP): It was shown in a previous sick pape that when LMAN (the output nucleus of the AFP) is inactivated, song variability decreases dramatically. This suggested that LMAN is actually injecting noise into the song production pathway, then monitoring the outcome in order to fine tune the song. How does your result that song learning occurs independent of constant LMAN output affect this interpretation?
(JC): Our results suggest a more detailed model. Our results indicate that another source (other than the AFP) is injecting noise into song that can also be used for learning. We think that this source is within the song production pathway itself. Our results suggest that LMAN monitors the outcome of both sources of noise in order to fine tune the song.
(SP): All of your experiments were done in adult birds. What do you think would happen if you silenced the output of LMAN throughout development, and then turned it back on during adulthood?
(JC): Since we think this is limited to trial-and-error learning (where they can only covertly learn the best of what they already were able to produce - i.e. their initial range of variation), and early juveniles never perform decent song, they would never perform good song. However, such an experiment in older juveniles might lead more noticeable learning.
(SP): A lot of other birdsong groups use zebra finches, while you guys study Bengalese Finches. Other than the fact that your entire lab is from the Ganges Delta, why do you prefer the Bengalese?
(JC): Although this is irrelevant for the current study, Bengalese finches allow us to study both syllable sequencing and syllable structure (e.g. fundamental frequency) in the same birds.
(SP): What, if any, implications does your work have for dudes like me who are training to be professional MMA fighters/Gamelan singers/ forensic anthropologists? Might it be possible for me to disconnect my LMAN during practice to convince all my competitors that I was not improving, then one day reconnect it and just blow all those crummy bastards away?
Indeed our results provide hope for sandbaggers and hustlers everywhere.
