目擊證人辨認臉孔之眼動分析

Abstract

本研究旨在探討信心程度與背景脈絡對目擊證人辨認臉孔的影響，及目擊證人在學習和辨認階段的凝視位置，因此設計兩個實驗。實驗一以73名大學生為對象，隨機分派到簡單或複雜背景脈絡組，其中在學習階段和辨識階段的臉孔圖像各8和48張，並以臉孔辨識問卷測得信心程度。結果發現目擊證人的信心程度與辨別力沒有關聯，顯示華人在這部分的表現與西方無異；另外，在複雜背景脈絡下的目擊證人辨別力，相較簡單背景脈絡組，前者表現較佳，顯示複雜脈絡能提供目擊證人有意義的提取線索。 實驗二主要探討高、低辨別力目擊者的臉孔辨認表現是否與個體本身的空間工作記憶能力有關，亦或是初始編碼時，目擊證人對於各臉孔特徵區域選擇性注意力的個別差異所引起，由此進而了解目擊證人辨認能力表現良窳的因素。在實驗二以61名大學生為對象，取其前後各1/3高低辨別力組（高辨別力組21人，低辨別力組18人），在學習階段與辨認階段的臉孔圖片各使用8和36張，以眼動儀進行分析，並以空間短期記憶作業測得每個人的空間工作記憶能力。結果發現高、低辨別力目擊證人的辨別力表現與空間工作記憶能力無關，用眼動儀分析其選擇的臉孔特徵確實是存有差異，在辨認階段，低辨別力組比高辨別力組花較多時間凝視嘴巴特徵，高辨別力則比低辨別力組花較多凝視時間在外部特徵上，然因不同階段設定反應時間的因素，因此在學習階段與辨認階段的選擇特徵略有不同，這部分可能應該加以排除階段任務的干擾再進一步探討。This study aims to examine the effect on face recognition of eyewitness. Experiment 1 investigated the correlation between confidence and the accuracy of face recognition, and the difference between the recognition accuracy in simple and complex background condition. 73 National Taiwan University of Science and Technology undergraduate participates in this experiment were randomly assigned to two background conditions. Testing comprised of a learning session, in which participants viewed 8 faces for 15 seconds each, and a recognition section, in which participants viewed 48 faces (including 8 faces in learning session) for 20 seconds each. The result indicates that, on an individual, confidence doesn’t correlate to recognition accuracy, which is consistent with past studies. The recognition accuracy of eyewitness in simple background is significantly different from the accuracy in complex background.. Experiment 2 examines the correlation between the spatial working memory, the accuracy of face recognition, and the difference of the selective attention to the different facial features. 61 National Taiwan Normal University undergraduate and graduate participates in this experiment, viewed 8 faces for 15 seconds each in learning phase and then were tested those faces from 36 faces in test phase. According to participants’ performance on face recognition, they were divided into two groups: high discrimination-abilitied group (21 people) and low discrimination-abilitied group (18 people). All participants’ eye movements and judgments were recorded by Eye Tracker. The result shows that, The difference of discrimination ability and spatial working memory doesn’t correlate to the recognition accuracy. Comparing to the high discrimination-abilitied group, low discrimination-abilitied group spent more fixation time on invalid facial features (e.g. mouth and ear), it means that the selective attention of eyewitness changes by different facial features.