I've got the final version of my dissertation prospectus, if anyone is interested. How about the abstract?
(Cangelosi and Parisi, 2001) write that, “The study of language origins and evolution is the study of how language emerges from a situation in which there is no language,” and my desire is to create animats that will make this critical jump. My approach will be to investigate the utility of a signaling system to animats organized into tribal groups that compete for scarce resources. Animat performance will be judged based on taking, holding, and defending territory within a two-dimensional continuous world populated with food sources of various values.I'm using the Quake 2 graphics and rendering engine for displaying my virtual world, and I've got screenshots in the prospectus that show how it all looks.Animat power and implementation will be varied so as to explore a wide range of capabilities. For example:
- Do tribes with signaling capability out-compete tribes without?
- What sensory input is required for a signaling system to be of value?
- What facts must be communicated for signals to be of value?
- Can animats develop their own signal system if the initial generation begins with randomly weighted neural nets?
- Is it important for the animats to be differentiated? Does it matter if children, males, females, &c., are functionally distinct?My philosophy is that the design of the experiment should result in animats that are analogous to apes, in that they will have the “physical” tools required for advantageous communication, but will need to learn to use those tools through experience. As described in section 6.1, many low-level language considerations will be ignored so as to focus primarily on how groups (not individuals) can benefit from signaling and cooperation, and on how communication can emerge from suitable building blocks.
My next step is to form my committee (I've got all the profs lined up except 1), and then to do my preliminary examination, where I'll formally present my prospectus. After that, I'll finish the research, write the dissertation, do the thesis defense, and get my degree.
Cool! Getting a proposal completed is a big step forward.
Given your topic, I wonder if the book by Prof. Julian Jaynes, "The Origin of Consciousness in the Breakdown of the Bicameral Mind" might be useful.
Prof. Jaynes, from an anthropological point of view, talks about the relationship between the origin of language, the origin of consciousness, and the origin of civilization. While your work will be considerably more mathematical, it might be interesting to have an anthropological reference thrown in.
It's a good book and well worth the read, IMO, even if it is somewhat off subject.
Yeah, my work does fit in with anthropology, and I've been in discussions with an anthro professor at UCLA. Unfortunately, I can onyl have one out-of-department professor on my committee :/
Much of the work I'm interested in would fall more properly into population genetics, anthropology, linguistics, and psychology, but I went with computer science because I feel that AI is the best tool available for studying any of those other topics.