Abstract: This paper describes the concept of computer-aided experimental human geography, which has been proposed as a methodology in geography. Experimental human geography is defined as a geographical methodology to reproduce human geographical phenomenon experimentally by means of computer sciences and GIS technologies. Its academic function is to study human-environment relationship under explicit geographic structure, implicit characteristics of human behavior, and explicit and implicit environment information. Accordingly, the paper has discussed the scope, principles and technical roadmaps of experimental human geography. In comparison to experimental economics, the paper has found that, in principle, many methods and rules of experimental economics are still effective in experimental human geography, but experimental human geography is more concerned about approaching natural processes simulation of geographic environment. Therefore, certain pre-designed models and special algorithms are necessary for experimental human geography. The experimental human geography rules include "value-induced" principle, "response-game" principle, "experimental control" principle, "statistics-based conclusion" principle, "geographical approximation" principle, and "experimental packaging" principle. Using the methodology of computer-aided experimental human geography, this paper studies the urban residential location choices, with two study objectives: (1) When the layout of urban infrastructure changes, what factors affect the dependence of the subjects on the city center, the reasons for destroying this dependence. (2) Whether the two different urban spatial structures, Burgess Model case and Hoyt Model case, would affect differently when the subjects were asked to choose a residential location. Based on the experiment objectives, the author made five experiment assumptions: urban space layout, residential location, residential location price, job location, incentive mechanisms. And the results are as follows: (1) Differences in urban residential space structures didn't have significant impact on residential location choices of the subjects. (2) In the case of the two typical urban spatial structures, the subjects were more concerned about traffic condition and school district around a residential location when asked to choose. (3) The price factor was not an obvious impact on the residential location choices, which might be the reason that not too many subjects accumulate in a particular area. This case study has shown that the thought and means of computer-aided experimental human geography are feasible. In other words, it is feasible to study the characteristics of human behaviors, such as decision-making behavior, in a controlled experimental environment with clear and realistic geographical structures and designed experimental contents based on GIS and computer technology. In short, this paper tries to develop a basic scientific framework for computer-aided experimental human geography, and much more remains to be done in the future.

Key words: experimental human geography, geo-information science, methodology, residential location choice