Given my upbringing, I grew up feeling that getting a PhD was the natural goal of my life. Which was fine when I went to Caltech wanting to be the next Feynman, like half the other frosh. By the time I graduated, though, I was pretty burned out, and didn't see myself using the pretty little planetary science degree. So I went off into the Real World, as we called it at Tech, even if much of the rest of the world might question the Reality of the Bay Area dot-com mania. And my mother would periodically ask if I was thinking about grad school, as if I was failing to reach my natural spawning grounds.
I actually was, occasionally, oscillating between AI research and wanting to muck with weird quantum engineering and wanting to go into economics and test Jane Jacobs's wackier ideas about how economies work. But it never felt real.
Then came 2001, and the great dot-com bust. I didn't end up applying to grad school as a refuge from the real world -- it's a lousy refuge which is a year away, and at any rate I know too many grad students who lacked full enthusiasm for their projects and slipped into hell. But I think the enforced idleness did get me thinking about what I wanted to do, right when I'd had a year to recharge from working full time, and if I'd had an interesting job it'd probably have distracted me from all grad school thoughts for at least another year. As it was, I decided I did want to do more than write web applications, and reading Steven Pinker's books pushed me over the edge: cognitive science (slicked up AI) was it. I wanted to figure out how I thought, and how to make machines think.
So I've applied to Indiana University, MIT, and UCSD. IU's accepted me. IU has Douglas Hofstadter, and it'd be very cool if I ended up working with him; modelling concepts is much of exactly what I want to do. His lab creates programs which can perceive analogies in various microdomains, and in one case dabble in creativity (gridfonts) and in another case self-awareness (Metacat). Self-awareness is something I want to explore, and also, now that we have decent models of concepts, I want to explore ways for a program to find and create new concepts. That's real machine learning -- not just dynamic adaptation of values, or finding regularities, but being able to label the regularity and use it in mental combinations just like built-in concepts.
Another interest is to make a program which can actually hold a conversation. Possibly in a limited version of English, or even a more artificial language, and probably in a 'toy domain' at first, but nonetheless able to have a real back and forth of ideas, and to learn from what the human says. I have no idea what the state of the art here is.
I'm going to Indiana this fall; got my apartment and funding and everything. I'm going through the CS department, since cog sci doesn't offer its own programs; I hope they don't ask me to TA some advanced course whose material I won't know. I wasn't really a CS major...
Refinement of my research interests: what happens when we understand a sentence? What changes are made to concepts and their relations when we read "Jill threw Spot the ball"? How are concepts conveyed through speech? How is ambiguity in sentences resolved?
What are the different ways sentences are understood? We can make syntactic sense of semantically void sentences, such as Chomsky's famous "Green ideas sleep furiously." We can make semantic sense of syntactically poor phrases: "Me heap strong." "Me elephant strong." The latter has two possible meanings -- 'I am as strong as an elephant', which may seem more likely if it follows my first phrase, or 'My elephant is strong'. But even with that remaining ambiguity there has been substantial communication: the uncertainty as to what the speaker might have in mind has been reduced from well-nigh infinite before the phrase was uttered to a single bit -- is he talking about himself or his elephant?
How does that semantic-based comprehension work? Is it even possible without a syntactic backdrop, in which we expand "me elephant strong" to possible correct sentences and choose the appropriate one based on context?
Behold, my interests. My thesis project is yet to be clear.