I asked permission to sit in on the Journal Club that’s held regularly by our neuroscience doctoral students today. It was a real treat. They are a bright bunch.
A key point of discussion concerned the sort of operational definitions that are fairly common in behavioral neuroscience (e.g. habit learning, spatial learning). These definition are extremely important in the design of experiments and in the interpretation of results and they evolve over time: model-based and model-free were the two relevant contexts for today’s paper from Redish’s lab at University of Minnesota.
Central to these neuroscience approaches is the holy grail of dissociating the different types of learning that occur between regions of the brain. The problem of course is that many brain regions participate in any one kind of learning. Further, any experimental design, no matter how excellent (in my opinion, the late David Olton was among the very best) is likely to have any single learning experience confounded by multiple types of learning.