背景和目的：語意與影像訊息是人們認知運作歷程的重要表徵。由於臉孔刺激具有多重語意訊息與影像訊息，因此以臉孔刺激來探討語意訊息與影像訊息的關係。臉孔刺激為一受注視角度影響的刺激，倒立與正立的知覺效果不同，倒立時由於整體組態訊息的喪失，而辨識變差，研究中也將呈現方向視為一重要變項。本研究主要目的在探討語意訊息對臉孔辨識歷程的影響，臉孔影像具有表情、型態的影像訊息，同時臉孔也可引發名字、表情詞、臉孔部件描述等語意訊息。利用促發作業，以文字為促發項，臉孔為目標項，探討(1) 不同層次語意訊息如何影響臉孔知覺，(2) 不同層次語意訊息如何影響倒立臉孔知覺。(3) 隱式促發與顯式促發之差異。方法：以224 個參與者在顯式促發與隱式促發作業的三個實驗中，探討名字、表情與特徵描述等不同語意訊息對臉孔辨識的影響。結果：三個實驗結果發現(1) 倒立臉孔的辨識較正立臉孔困難的現象，符合倒立效果的結果。(2) 人名訊息在顯式促發作業中，除了正立臉孔一致訊息組由於天花板效應，而沒有效果外，一致訊息有助益效果而不一致訊息有減損效果。隱式促發作業，則無效果。(3) 表情訊息在顯式促發作業下，不論訊息是否一致對倒立臉孔的反應正確率皆有助益效果。在隱式促發下，對正立臉孔的反應時間有縮減效果，對倒立臉孔的反應正確率有助益傾向。(4) 特徵訊息在顯式促發作業下，對反應正確率與反應時間皆沒有顯著效果，在隱式促發作業下，對正立臉孔的反應正確率有減損傾向，但對反應時間卻有縮短現象，可能為一種權變現象。結論：不同層次的語意訊息對臉孔辨識歷程的影響不同，名字訊息是對臉孔的類別分辨而言，是一個相當有影響的語意訊息。另外顯式促發與隱式促發的效果在本研究之結果，有相當大的差異，也顯現二種促發作業之機制的差異。

Background and Purpose: Both semantic and imagery information are essential components forprocessing by the human cognitive system. Pertaining to properties of multiple representations of two kindsof stimuli, facial stimuli are valuable for use in investigations of relationships between the two kinds. Theperception of facial stimuli is angle dependent, that is, there are different effects from the perspectives of thesame face being upside-down or right-side up. One of reasons for worsened recognitionwhen people perceivetheir surroundings from a handstand position is the loss of global configuration information. Therefore, thisconsequence was included in the present study as one of independent variables. This investigation mainlyfocused on how semantic manipulations influence facial recognition processes. Primary facial imageryattributes like the expression and configuration, and the semantic information induced by faces, such asnames, expressional words, and the description of facial parts, were all considered in this study. In thisresearch, words were used as the priming items, and selected faces were presented as target objects forprobing the following processes: (1) how different levels of semantic information affect facial perceptions, (2)how that information affects the perception of upside-down faces, and (3) differences of implicit and explicitpriming on these processes. Methods: In total, 224 participants joined the 4 experiments where semanticinformation, including names, facial expressions, and descriptions of facial features, were designed under anarrangement of implicit and explicit tasks. Results: (1) It was found that subjects had greater difficulty identifying upside-down facial exemplars than upright exemplars, in accordancewith the upside-down effectproposed in previous studies. (2) In the explicit priming task, consistency of information on people's namesprovided a beneficial effect on upside-down facial recognition, while inconsistent information had a negativeeffect. However, there was no significant recognition performance with consistent information for uprightfaces due to the ceiling effect. Nevertheless, participants showed no significant differences in the implicitpriming tasks. (3) Facial information offered an enhancing effect on the correct recognition rate in explicitpriming tasks when both consistent and inconsistent information was used. On the contrary, in the implicittasks, the response time decreased for upright exemplars, and there was a trend toward improving theaccuracy of recognition of upside-down faces. (4) For both response rate and response time, information onfacial features showed no impact in the explicit priming tasks, but adversely affected the correct response rateto upright faceswhile speeding up the response time in the implicit priming tasks,whichmight be interpretedas a contingent process. Conclusion: Numerous kinds of semantic information may involve differentprocesses of how faces are appreciated where visual information on a name is relatively effective inrecognizing categories of faces. It was also suggested that manipulations using implicit and explicit primingtasks evidenced rather different mechanisms for the processing of facial recognition.