This past week I attended the 2nd Beijing International Symposium on Computational Neuroscience at Tsinghua University. It was a one day conference all about computational neuroscience and it was a really great day.
You’re probably wondering what computational neuroscience is. I wasn’t exactly sure myself, but after a day’s immersion in it I feel able to describe it. Computational neuroscience is looking at things that brains do and trying to figure out a way that they might be doing them using computational models. For example, the last speaker of the day was examining how people learn to do a series of steps that results in varying rewards depending on the steps chosen. His models said that people, and animals, partially use a full and detailed prediction of each choice, the choices that the first choice makes available, and so on and logically say “well if I want such and such a result I can just follow this series of steps.” He asserts that with more experience animals begin to abandon the perfectly logical, but cognitively taxing, exact method and substitute a method of choosing wherein each possible choice at a given stage just has a sort of feeling attached to it that says “this choice seems to work out pretty well” and for each subsequent step the animal just picks each option that seems like it will work well, without explicitly planning. So he has this model and can simulate how animals would behave if they were following one or the other of these two methods or a hybrid of them and he then puts animals (people) through a test and sees whether he was able to accurately predict their choices using his computer model. And voila, he has done computational neuroscience. He now has a model that, while it is not really how the brain is operating, does a good job of approximately achieving the same result as a real brain in this particular circumstance.
So the conference was a whole-day affair. I got there in the morning for coffee and snacks and soon the festivities began. A professor from Tsinghua gave an opening remark that, summarized, went a little like this: “Welcome, we’re glad to have you all. BTW, Tsinghua is developing a neuroscience program and if any of you want to come here will pay you well and if you’re a new professor we’ll hook you up with some sweet stuff. Seriously, we’re really awesome.” It was almost obnoxiously smug sounding, but I would be smug too if I could actually offer that. The day consisted of 4 talks by notable people in the field and 3 periods of time to go look at posters. It was followed by an open question session and a buffet dinner.
I met a nice German masters student who I talked to a fair amount throughout the day as well as a number of Chinese students.
I found the whole affair very interesting as I didn’t know much about computational neuroscience to start with and had never been to a conference like that. It was nice to see people talking about receptive fields and other neuroscience things and then having a whole room of people seem to understand what was being said! I also found it humbling how often people just had to say “I don’t know” to curious questions from the audience. Also, during the discussion section at the end questions were raised like “if we can accurately model all the activity of the brain, does that mean we really understand the brain? Does it matter?” and “Is computational neuroscience going to be helpful in the near future?” and I was struck by how… simple the discussion was. These professional neuroscientists could offer no more credible or more solid arguments than I could and the whole discussion seemed to be pretty much about philosophies. Whether or not it is enough to simply be pragmatic and satisfied with a model that works and how hopeful you were willing to be about the applications of cognitive models to computer programming and robotics.
I really enjoyed the conference. It was great to spend a day hearing about things I hadn’t known about before and it was definitely a break in the routine. I look forward to more such events in the future!