摘要
人們往往無法偵測環境中的重大改變，此現象稱為改變盲。決策失敗為改變盲的成因之一。然而，改變偵測的決策機制尚待釐清。因此，本論文主要探討改變偵測的決策歷程。根據過去研究，改變前、後物體的相似性會影響改變偵測的決策判斷。當改變前、後物體越相似時，偵測困難度上升，則偵測表現下降。本論文探討不同類別刺激材料之間的相似性（如：人臉（第二章）、物體（第三章）、與Gabor（第四章）），如何影響改變偵測。實驗一-三顯示，改變前、後臉的相似性會調節人臉的偵測優勢。實驗四-六測試物體間知覺和語意的相似性，在不同的刺激呈現時間下，如何影響訊息累積速度。利用Diffusion model，發現兩種層次的相似性都會使得決策訊息的累積速度變慢，但不影響非決策的時間。實驗七-九利用systematic factorial technology，進一步探討刺激中不同特徵的處理如何影響決策歷程。結果發現，當不同特徵改變量不等時，受試者會使用不同的決策策略（實驗七）；當不同特徵改變量相等時，受試者會傾向使用平行自我終止式的處理方式來偵測改變（實驗八）。當畫面中有兩個物體時，受試者採用序列自我終止式的處理方式來偵測改變（實驗九）。決策歷程會隨著情境不同而有所改變。本論文探討改變偵測中決策歷程的各個不同面向，研究結果更進一步釐清決策歷程在改變偵測扮演的角色。
並列摘要
People often fail to detect a large change in the visual environment. This phenomenon, change blindness, has attracted researchers’ interests. The decision failure has been proposed as one of the causes of change blindness. Yet, the processes that underlie the decision mechanism are still unclear. As similarity between the pre- and post-change stimuli can influence the decision, I tested how similarity between faces (Chapter 2), objects (Chapter 3), and Gabor patches (Chapter 4) influences the decision. Previous results have shown that similarity costs detection performance (similarity cost), and the results from Experiments 1 - 3 showed that the similarity cost can override the face-detection advantage. Experiments 4 - 6 tested how different levels of similarity (visual and conceptual similarity) influences the information accumulation process when the exposure duration was 2,000, 1,000, or 180 ms. Ratcliff’s diffusion model was used to analyze the information accumulation process. Results showed that both visual and conceptual similarity decreased the rate of information accumulation. The rate of information accumulation also decreased as a function of the exposure duration, suggesting that participants spend more time on accumulating evidence for a decision when the memory strength of the pre-change objects is weak. In Experiments 7 - 9, I used the systematic factorial technology to investigate the processing of each feature underlying change detection. Results showed that participants used different decision strategies when the detection threshold of each type of change was unequal (Experiment 7). When the detection threshold is controlled, participants conducted a parallel process with a self-terminating rule in detecting changes (Experiment 8). They used a serial processing with a self-terminating rule to detect changes when two objects were presented (Experiment 9). Results suggested that participants are able to change the decision strategy depending on the context. Results from this thesis furthers the understanding of the different aspects of the decision mechanism, including the decision difficulty, the rate of information accumulation, the response criterion, the process architecture, the stopping rule and the process capacity, in the context of change detection. Results may shed light on the comparison and decision processes involved in change detection.