幽默是人類高層次的認知能力，相較於幽默指涉的內容，幽默刺激所使用的技巧更能引起幽默的愉悅感受；眼動追蹤技術的特色為客觀直接，可以記錄參與者的認知與情感歷程的內在表徵。本研究透過眼球追蹤技術比較不同笑話技巧（歧義與推論）的眼動差異，由客觀眼動指標與主觀評定，探究笑話理解的認知及情感歷程。以兩個實驗進行研究，採用受試者內設計，共68位參與者。實驗一探討歧義笑話技巧，獨變項為刺激類型（語音歧義笑話、語音歧義非笑話、語意歧義笑話、語意歧義非笑話、語法歧義笑話、語法歧義非笑話），依變項為眼動指標，包括首次凝視時間、總凝視時間、平均凝視次數、回視次數及平均瞳孔大小；以及主觀的理解程度與好笑程度的評定。實驗一結果顯示，參與者閱讀笑話與非笑話的營造句之首次凝視時間並無顯著差異，符合本研究假設。在笑話認知歷程，「語音歧義笑話」相較於「語法歧義笑話」，在總凝視時間最短及回視次數最少。實驗二比較推論笑話與歧義笑話的眼動軌跡，獨變項為刺激類型（語意歧義笑話、語意歧義非笑話、橋界推論笑話、橋界推論非笑話、推敲推論笑話、推敲推論非笑話）。實驗二結果顯示，推論笑話（橋界與推敲）在總凝視時間、回視次數與平均瞳孔大小皆顯著大於「語意歧義笑話」，表示推論笑話的認知理解歷程比語意歧義笑話來得久且較深層處理，其認知涉入較多而產生較高的愉悅感受。綜合兩個實驗，笑話的好笑程度受難度影響，除具備難度適中、可理解的條件，笑話技巧亦是另一重要的關鍵。未來可延伸至認知神經科學領域比較不同笑話類型的認知、情感與笑反應的大腦機制，亦可透過不同笑話技巧進行幽默的訓練課程。

Humor is a high-level cognitive ability in humans. In jokes, it is the humor-generating technique and not the content of the humorous stimulus that generates amusement. Eye-tracking techniques can provide a direct and objective recording of participants' internal representations in various cognitive and affective processes. The present study thus aimed to compare the cognitive and affective processes associated with distinct humor techniques (i.e., ambiguity and inference jokes) by using eye-tracking techniques, with both objective eye movement indices and subjective ratings. Sixty-eight participants participated in two experiments that had within-subjects designs. Experiment 1 investigated various categories of ambiguity jokes by using stimulus categories (phonological jokes, phonological nonjokes, semantic jokes, semantic nonjokes, syntactic jokes, and syntactic nonjokes) as independent variables and both objective eye movement indices (first pass gaze duration, total viewing time, average fixation counts, regression counts, and average pupil size) and subjective ratings (comprehensibility and funniness) as dependent variables. Results supported the hypothesis that first-pass gaze durations for setup lines do not differ significantly between jokes and nonjokes. Additionally, total viewing time and regression counts were shorter and lower, respectively, while reading phonologically ambiguous jokes than for syntactically ambiguous ones. Experiment 2 compared the effects of semantically inferential and semantically ambiguous jokes. The experimental design was the same as in experiment 1, except that the stimulus categories were semantically ambiguous jokes, semantically ambiguous nonjokes, bridging inferential jokes, bridging inferential nonjokes, elaborative inferential jokes, and elaborative inferential nonjokes. Results demonstrated that total viewing time was longer, regression counts were higher, and average pupil size was bigger for inferential jokes (bridging inferential jokes and elaborative inferential jokes) than for semantically ambiguous jokes. It was concluded that the cognitive processing of inferential joke takes more time and is more complicated than that of semantically ambiguous jokes, and this consequently results in deeper cognitive involvement and amusement. Taken together, the results of the present study indicate that the greatest pleasure accompanies this cognitive comprehensibility with the deeper semantic processing but that joke techniques also play a major role. Future research might use neuroimaging techniques to further investigate the neural correlates of cognitive, affective, and laughter processing of various types of joke techniques as well as design humor training courses on joke techniques.