Author: jlolds

Native Angeleno, professor of Neuroscience and Public Policy at George Mason University

Air Force Office of Science Research moving to the Buckeye State?

Apparently the scientific community is up in arms about the notion, story here. Note that my home town, Arlington Virginia is the current home of AFOSR. Interestingly NSF is apparently leaving their current HQ in Arlington for a further out location in Alexandria in several years. Might DARPA and ONR be next?

So I think the real story is about the nexus of these current federal research agencies dispersing. For years, they’ve essentially been ‘across the street’ from one another. Arlington has thrived, and many universities (including my own) have presences in Arlington to be close to the action.

The Two Cultures are alive and well….

William Dereseiwisz pans Michael Suk-Young Chwe’s ‘game theory’ theory of Jane Austen here. Cutting right to the chase, here’s the money quote:

There is only one problem with this approach: it is intellectually bankrupt. Actually, there are a lot of problems, as Michael Suk-Young Chwe’s abominable volume shows. If this is the sort of thing that we have to look forward to, as science undertakes to tutor the humanities, the prospect isn’t bright. 

Robert Shiller on neuroeconomics….

From yesterday’s NYT, here. This year’s Nobel laureate in economics is of the firm opinion that the field of neuroeconomics is going to become increasingly important because humans do not behave rationally. I agree with the basic point but I would add two things: first, at least in the US (probably less so in China) fMRI studies are often statistically underpowered. I think this is generally a challenge for all of behavioral economics. Second, in real economic ecosystems (with millions of human agents) complexity plays an important role in emergents (such as market prices for goods). To capture these complex adaptive systems we really need the tools of computational social science and especially the tools of agent-based models. These models have the strength of accommodating millions of agents (in silico) and potentially provide insights and predictive power to the above types of emergents.