ナビゲーションバイアスの危険地帯：バイアスカウンター策としての線形逐次マスキング（LSU-E）の（非）効率性の評価

Linear Sequential Unmasking (LSU-E) is a method of managing the flow of information so as to improve decision making and minimize cognitive bias in expert domains. This article examines the specific factors determining the (in)efficiency of LSU-E. Central to this evaluation is the "bias danger zone," where decisions involve human judgement, complex data near decision thresholds, and strong directional bias. In these situations, LSU-E reduces bias by prioritizing information based on its objectivity, relevance, and biasability. However, research indicates that the sequencing of information, a core component of LSU-E, is not always effective. LSU-E's (in)efficiency depends on whether the experts perceive specific information as exceptionally strong or weak. Overpowering information can dominate a decision regardless of its position in the sequence, whereas weak information may be disregarded even if presented early. As part of a broader "Context Management Toolbox," LSU-E must be deployed proportionately and with an understanding of these boundary conditions. Recognizing these factors is critical for forensic science to successfully navigate contextual contamination through appropriate and effective countermeasures.