Robosapien says:

Perception and action are inseparable!

Robosapien says:

Cognition is for Action!

Recently I have read a much-cited paper by NYU professor Thomas Nagel titled What is it Like to Be a Bat? I was reading it on the advise of the professor supervising my undergraduate independent research project on connectionism and embodiment. The paper is interesting but also interesting is what I found when I searched the web for the title.

There exists a band called What is It Like to be a Bat? that does some really wild techno-punk music. The band leader is also a professor. The website has a lot of very confusing descriptions of what the band is and has done (high-art pop??). Still it really is an interesting concept.

What can I say: In my private Idaho there are professors forming strange bands named after the strange papers of other professors.

I began University studies in 1988, enrolled in engineering. It is now 17 years later and it looks as if I just might complete my undergraduate studies this year. Now, before you try to ball-park the size of my student loan debt after 17 years of studies, you have to realize that I did take 12 years away from my studies to establish a career as an information technology consultant.

I started in 1988, at the University of Alberta, in engineering because everyone told me to take engineering. I wanted to be a physicist but everyone told me to be an engineer when I was in high school. I didn’t even know what an engineer was! When I started taking the courses, I discovered that I didn’t like it and it didn’t like me. So 1989, after one horrible year, I switched to studying psychology for no particular reason.

I really didn’t like psychology that much either. Some of it was interesting but personality psychology made me ball my fists in frustration. How many times can you listen to all those “life is like an onion” theories that Freud and Jung made up? Those guys violate or confront my world view I guess. I liked cognitive psychology and perception courses but I eneded up bored and spent all my spare time working with computers and modems. I would have liked to study computers, but there was no way I was going back to the engineering department.

A few years after that I realized that there was actually a completely separate computing science program, so I enrolled. After a single year of computing science, I ran out of money and had to drop out. I became a cook for a year and then, with the financial support of my parents, I returned to complete my computing science degree.

Everything went well until my final year: 1994. In my first term, just weeks before exams in my second last term, my best friend was killed. I was devastated the University was unsympathetic and would not let me defer my exams. Later a friend of mine got the ombudsman involved and I was finally allowed to take the exams. I was bitter at the University for the way I was treated, and still suffering from the shock of losing my best friend, and besides that I was losing my enthusiasm for computing science. I dropped out.

In the winter of 1994, just three courses short of a computing science degree, I dropped out and was looking for work. I had, in fact, started my consulting career the summer before and it was going well. For a while I had been obsessed with Unix, the Internet, and the very new World Wide Web. While studying computing science I had skipped a lot of classes to play with unix in the labs. I was budding sysadmin and had a lot of skills that seemed obscure at the time but that became the key “in demand” skills during the Internet boom. I didn’t need to look far for work.

I hooked up with a company called AfterMac to start an ISP business and build an online store. My career took off, and it seemed like I didn’t need that degree anymore. I certainly had no problem getting clients, respect, and money without a degree. Once, just on a whim, I submitted an application for re-admission to the computing science program and was rejected. Who needs them anyways, I thought.

For the last three or four years I have had clients that have me visiting the University of Alberta campus frequently. Last year I decided that, since I was always on campus anyways, I might as well take some courses. Knowing that I would not be readmitted to computing science, I applied to general program and declared a major in Psychology. Much to my surprise, they both readmitted me and gave me credit for all the courses I had taken a decade earlier. I was just nine courses short of a degree! I found that by switching to a double major of Psychology and Mathematical Sciences, which includes computing science, mathematics, and statistics, I could reduce the requirement down to just eight courses for graduation.

Last year I completed four courses. My initial idea was to take the easiest courses I could, and take the path of least resistance to graduation. But that turned out to be impossible; three of the courses I choose randomly turned out to be extremely compelling. I ended up in three of courses taught by the same instructor, more or less because they fit into convenient time slots. I expect to do the least work possible to get the best grade I could. But before I even began the course I found myself compelled to study hard and learn more.

The courses were all cognitive science classes, offered by the psychology department, taught by Dr. Mike Dawson. I bought my text books a month before classes started, and one sunny afternoon I sat out in my hammock and started reading Understanding Cognitive Science. Many hours, four chapters, and a sun-burn later I was hooked. I started reading my other course materials ahead of time too. My classes rapidly became more important that my work, and were the highlight of my every day.

The courses I took last year were as follows: An introduction to cognitive science, a course on embodied cognitive science and synthetic psychology, and a course on connectionism and synthetic psychology. I also took a course on advanced perception but I just wasn’t exciting compared to the others. I got three A+ grades (and an A- in the perception class).

As a brief aside: It was a weird experience being 34 years old and an undergraduate. I am older than most teaching assistants and maybe even a few professors. As a consultant I am used to be treated as an “expert” who is expected to speak freely, ask questions, and offer ideas. In my role as student I found it hard to open my mouth for fear of embarrassing myself. It was very weird. Weirder still is that on a day to day basis I work with scientists and other academics, as a technology expert. The people I work with are professors no different than those teaching me in class. But my perception of and attitude toward such roles change as I switch role from expert to student. A very weird feeling.

The textbook for the introduction to cognitive science class was great (written by Dr. Dawson). In the course on embodiment we undertook a project to build a robot with Lego Mindstorms. Fun and exciting but not nearly as exciting as all the extra readings I did (seriously!). The course on connectionism was probably the most compelling; we focused on interpretating representation that emerge in trained neural networks and what that implies for a synthetic psychology.

I went back to University looking for a free-ride and ended up finding myself compelled to work every bit as hard as I have in my career. I now stand almost exactly where I was in the fall of 1994. I am just four courses short of my degree but this time I am standing on a mountain of enthusiasm for what I am studying.

One problem I have is that I have taken all the courses offered on the topics I am interested in. So, I applied to take an independent research course supervised by the professor who taught the three courses that I found so compelling. I start that research this fall and I am very intimidated. We agreed on the broad topic of “Advanced Neural Networks” but I need to decided what I want to work on. My problem is that I want to work on far too much. I tried a lot of obvious things like reading about what other people are doing; re-reading the professor’s papers to identify which aspects I liked, etc. Everything I came up with sounded too complicated for a short 4-month research project.

But just yesterday I think I had a break through. I decided to start broad and narrow things down. I picked two broad potential sub-categories of “Advanced Neural Networks” that I can work my way down from. First, there is network interpretation and second, exploring the intersection of connectionism and embodiment.

The first topic is the kind of work done by my supervisor and the connectionism course I took involved interpreting the internal representations formed by neural networks. I would very much like do more of this. It is hard and interesting and fun. I have a lot of ideas for network problems to explore, but none that I can describe as a research project.

The second topic would amount to a literature review and result in a paper. I already have a long list of readings that could go toward this. But taking this route would probably preclude any hands on experimentation with neural networks. There probably wouldn’t be time (you never know… four months can be a long time or a short time depending on how things turn out).

I have to refine my ideas by the end of the month (sooner would be ideal) so I can contact the professor and find out what requirements there are to completing the course. I do not actually know how these independent courses are supposed to work. I feel more intimidated than ever… and it just feeds my enthusiasm. That has to be some kind of pathology.