Last semester I sat down with one of my professors and talked through my interests in the psychological sciences. After I’d described what I was interested in, he told to give PSY-351 – Computational Modeling.

I’ve been to a couple classes now and I’m coming to better terms with what exactly a computation model is and isn’t. To be a computational model, you need not only a “what,” but also a “why.”

Take, for example, collaborative filtering that I worked on this summer. This is a statistical model. There’s nothing in it about why people like things, only the assumption that people tend to like similar things, so by looking at bunches of things that people like, we can pick out trends.

A computational model is actually something like the semiotic model I was playing around with which not only describes a method for determining what people like but does so by positing a structure for why they like those things.

Why is a computational model desirable? Well, consider physics. We could have a model that tells us how fast things fall out of the sky. It could get the rate of 9.8m/s² and describe motion well. That is a useful model. Having a model of gravity and universal attraction gives us a “why” things fall and in doing so allows us to predict things about falling objects on other planets and orbits and all sorts of things.

Computational models, as opposed to statistical models or empirical curve-fitting models, because they provide a theoretical handle on to why the process takes place can be tested much more rigorously and ultimately, if proven reliable, may well provide predictions into areas where observation is currently impossible.

To date all that we have been discussing are models of object categorization, but our final project is to implement a computational model from a field related to our research and implement it.

I’m in a slightly difficult position in that the models that are most closely related to the research grant supporting my studies (interfaces for soldiers using robots searching for IEDs) would likely be computer vision and I don’t particularly want to learn those in depth. I’m going to go meet though with the professor and hopefully we’ll work something out. Maybe I’ll end up doing something related to preference modeling.

I went to the Methodist Church this morning to get back in touch a bit with my childhood roots.

To some extent I was comparing it to the Unitarian Universalist Church I went to last weekend.

Accessibility: This one went to the Methodists hands down. They’re right by Vandy, and what is more important they’re accessible by bus. To get to the UU Church I had to walk through the lawns of a string of $500,000 houses because they’re half a mile down a busy road with no sidewalk from the closest bus stop. (I’m giving my car to MPP.)

Friendliness: This was what bothered me the most at the UU church. I stood around for a long time and almost no one said “hi.” How a community treats strangers is important to me. In the Methodist Church I had at least half a dozen people say “mornin’” to me or introduce themselves. In fairness though, I was at the morning service at the UU church, and a good percentage of the folks there were guests themselves.

Study: This one I’m not really fit to judge. The UUs have a meditation room open between the early and late services when the Methodists have Sunday school. Sunday school was somewhat interesting, but we didn’t really dig into anything. I didn’t try the meditation room. I’m really wanting to talk to people and understand their positions as of late. I just want them to be a little older than me (which the average college student isn’t).

Continue reading →