How much of Cognitive Science will be true in 10 years?
Razib Khan asked today which results from Cognitive Science are robust enough that we still would believe them in 10 years, a propos a paper on ability to believe and tendency to see faces. (Go read. Interesting stuff). And, now that I’m immersing myself in understanding power analyses, and effect sizes, and replications, I’m really not sure. I think there are some things that if you squint enough to filter out the fine grained stuff will be robust. Like generativity/constructivity of memory. The attentional filter. Some version of dual processes. In social psychology things like groupiness, and the type of cues to persuasion that is discussed in Cialdini’s “influence”.
I just taught my first entire course in the philosophy of science/psychology/cognitive science. I started doing Cognitive Science just about right away when I got back to school (1990), and had spent time reading cog sci for a few years prior to that, so I have lived at least part of this development. The research on mind is so very much driven by metaphor – something that has been well known anyway. (I’m not the first one pointing this out, but I can’t for my life remember where I read this the first time). It has been the wax tablet, the enchanted loom, and the computer. Useful metaphors, but as all metaphors, they can become strained. Some of the research driven by this approach is bound to turn out being kind of like research on phlogiston or the ether.
I spent my summer holidays reading through an edited volume by Todd and Gigerenzer (ecological rationality – link is not to book but their home page), recommended by Jason Collins. And, some of the chapters there call into question things like over-confidence, and problems with meta-cognition which seem very well established. The issue is that the researchers assume an underlying normal distribution (hence, it is unreasonable that way more than half of all drivers consider themselves better than average), whereas they suggests that humans may be quite sensitive (intuitively) to non-normal distributions. As in with drivers, the vast majority of accidents are caused by a rather small proportion of crappy drivers, and the majority of drivers are better.
I also recall, I believe from David Hull’s lovely “Science as a process” that scientific work is very Darwinian. There will be lots of attempts, and very few survivors. Which ones, we don’t know.
Which reminds me, I have to get Samuel Abersmans “The half life of facts”