【正文】 ountant’s biases. This study looks at one source of bias suggested in the escalation of mitment literature: involvement in the original decision to invest in the asset. The escalation of mitment literature demonstrates that when an accountant is responsible for an investment decision, subsequent judgments made by the accountant concerning the asset may be affected. For example, the accountant will increase his or her mitment to a failing investment when given the chance (see, for example, Schulz and Cheng, 2020). Similarly, an accountant’s asset impairment judgments may be affected when he or she is involved in the investment decision. The accountability literature suggests that, under certain conditions, accountability may attenuate bias(Lerner and Tetlock, 1999). If an asset impairment decision has a material impact on the financial statements, an accountant’s decision that an asset is (is not) impaired likely will be questioned by an external auditor. The accountant must justify the decision, which is one form of accountability pressure (DeZoort, Harrison, amp。s operations and scan the environment to determine if any indicators of potential asset impairments exist. SFAS No. 144 provides examples of impairment indicators (see Table 1)。 Ross 1987 for a review), Brockner (1992) notes that there is much theoretical controversy over the explanation of escalation. Brockner argues that selfjustification theory provides an important, but only partial explanation of the escalation behavior. He notes that to explain escalation more pletely, other theoretical perspectives (., expectancy theory and prospect theory) must be considered. One theory which may offer insight into behaviors such as escalation of mitment is catastrophe theory. Catastrophe theory helps explain why smooth or continuous changes in inputs to a system canresult in discontinuous responses by the system (Zeema n, 1977). For example, economic indicators may change in a smooth, somewhat linear manner over time, but stock market indices may exhibit sudden jumps or discontinuities in response to economic news. Catastrophe theory provides a basis for studying such occurrences because it helps model psychological and other phenomena where changes in system responses (., behavior) are not linear In nature. Flay (1978) proposed a series applications of catastrophe theory to attitudes and social behavior. Flay based many of the applications on one particular model of catastrophe, the cusp catastrophe. As explained by Flay, the model is a threedimensional representation of how the behavior of a system can change. The top surface of the model (labeled the “Behavior or Response Surface”) represents the system’s response to changes in inputs known as normal factors. The back edge of the response surface is smooth and illustrates a continuous, linear change in behavior in response to changes in the normal example, as the normal factor changes (represented by movements to the left or right in the model),the behavior of the system changes in linear fashion (represented by movements up or down along the vertical axis). In contrast, the front edge of the surface has a fold or inaccessible region, which illustrates a discontinuous, abrupt change in behavior in response to smooth changes in the normal factor. The factor which determines where on the response surface the system will operate is known as the splitting factor. Changes in the level of the splitting factor are represented by movements from the back of the model to the front and vice versa. Thus, as the level of the splitting factor increases (represented by movement toward the front of the model), the system’s responses will bee more abrupt and discontinuous. By incorporating normal factors and splitting factors, the cusp catastrophe model can be used to explain changes in attitudes and behavior that are primarily linear in nature as well as changes that exhibit varying degrees of hystersis (., stickiness). For example, a normal control variable for an investment behavior might be the favorability of information about the project or investment. As the information about the project bees more unfavorable (., negative consequences are being incurred), one would expect the mitment of resources to the project (., the behavior) to decrease linearly in response to the greater unfavorability of the information. In this happens, changes in behavior are dependent on the magnitude of the normal factor, and those changes occur in a continuous fashion. However, observed behaviors do not always conform to this expectation. According to the cusp catastrophe model, the reason why such changes i n behavior are not always observed is that splitting factors can cause a bifurcation (or discontinuity) in behavior. An example of a splitting factor for an investment decision might be the degree of responsibility felt by the individual for any negative consequences resulting from past investment decisions (as in the escalation of mitment phenomenon). As the magnitude of the splitting factor increases (., as the degree of personal responsibility increases), hystersis in the behavior results. This hystersis causes the individual to remain mitted to the previously chosen course of action. Only when unfavorable information is consistently received and, therefore, bees overwhelming, will the individual change his or her mitment to that action. Furthermore, when change does occur, the change may be catastrophic with a shift to a contrasting position. In terms of the model, the change occurs abruptly because hystersis creates a region of inaccessibility in the behavior. Thus, catastrophe theory would predict that, even in the face of negative consequences, an individual with a high degree of personal responsibility for the original investment will