Incorporating Bounded Rationality Concept into Route Choice Model for Transportation Network Analysis



Incorporating Bounded Rationality Concept into Route Choice Model for Transportation Network Analysis

Authors

HATO E and ASAKURA Y, Ehime University, Japan

Description

The conventional travel behavior models in network research can be characterized as normative models assuming the rationality in travelers' behavior. Most of these models implicitly provide the following conditions:

Abstract

The conventional travel behavior models in network research can be characterized as normative models assuming the rationality in travelers' behavior. Most of these models implicitly provide the following conditions:

1 ) The Alternative set is given beforehand, and

2) The models are based on a compensatory type of maximum utility theory.

If the alternatives were offered to everyone beforehand, as in the model for selecting means of transportation, there would be no problem for the model in considering alternatives as foresight. In the route choice decision-making, however, the question of how a driver selects one of the innumerable route choices would become very important. In conventional route choice models, the driver's searching process for alternatives has been discarded in most of the behavior models.

There is an occasion where the driver obtains certain information from the given traffic information and 'accordingly selects a new route which had not been one of his/her alternatives in the past. Such a phenomenon is indeed a constitutional matter of leveling the traffic demand by giving the traffic information. In the model assessing the efficacy of given traffic information, the process for generating alternatives needs to be expressly considered, in place of a prepared aggregate of alternatives.

The second proble~m concerns the maximum utility theory. The expected- utility theory was presented by Neumann and Morgenstern (1943) as a basic concept for game theory. A Iogit model and probit model based on this theory now take root as desegregated models. The models based on Multiple Attribute Utility Theory (MAUT) are useful but impractical. This radical hypothesis on behavior, comparing the utility of each alternative, which was calculated by adding all the partial utilities of several attributes in each alternative, seems to be unreasonable, because it requires immense computing performance for the decision maker. A new theory of EBA (Elimination By Aspects) established by Tversky (1972), which eliminates each route that does not fulfill a certain reference value in route selection behavior on real networks with various route alternatives, is considered to be more realistic. Simon (1987) proposed a notion of bounded rationality. The rationality model must be founded on the observation results, which is based on the experiment and corroborative evidence of the decision maker. The bounded rationality theory explores the decision maker's selection processes in order to explain the non-rationality of selection made by the observed drivers. In this study, we propose a route choice model to networks based on the behavior framework within bounded rationality.

The concept of a structural approach (Svenson, 1998) is used to solve two problems in the rational behavior model. In this approach, the heuristics of decision-making can be evidently presented by splitting the rating process of alternatives into several sequences. A prospect theory proposed by Tversky and Kahneman (1979) divides the decision making process into two aspects: "editing" and "evaluating." Hato et al. (1999) have found that the repeatability of driver's route choice behavior increases when expressly considering the information acquiring process and the decision making process respectively.

In this study, we will consider three processes of decision making in selecting routes on networks; that is, network recognition, selection from alternatives, and rating by comparison. In the first process of network recognition, the state of how a driver recognizes the existing road networks will be described by using a mental map model. The second process of selection from alternatives presents a course of behaviors to eliminate impractical routes from the recognized networks based on a certain reference value and to pick out the alternatives that are considered for selection. We will assume the non-compensatory behavior standard; that is, a route is deemed suitable for selection if the route distance satisfies a specific reference value. This will then be modeled. In the third process of rating by comparison, the driver is assumed to compare the utilities among the selected route alternatives and to select a final route of the highest utility. The conventional compensatory behavior standard is assumed in this process. It is weU known by Payne's study (1976) that the strategy suggesting compensatory rules is applied when the number of alternatives is small, while the strategy to promptly reduce the number of target alternatives, based on the limited information retrieval, is applied when the number of such alternatives is large. This model adopts such structure of decision-makings.

Publisher

Association for European Transport