測謊光電容積頻道（PPG）之效益──以比對問題技術為例

The Effectiveness of Polygraph Photo-Plethysmograph Channel: Examining Subjects Tested with the Comparison Question Technique

本研究於傳統測謊頻道上，增添光電容積頻道（Photo-Plethysmography, PPG），記錄「周圍血管收縮」變化，檢驗新頻道能否提升鑑判率來評估其效益。研究以刑事警察局111～112年間，Lafayette-LX6電腦化測謊儀所蒐集之受試者生理訊號作為素材，結果發現，最高鑑判率的量分-頻道組合為「4頻道，三分位法」，達71.8%，惟與其他組合未達顯著差異。後續與測謊結果及司法採用結果相比，亦無證據主張增添頻道得使效能大幅提升。相對來說，國外成果不免有高估效益之議，過分淡化其他因子對鑑判率所造成的影響。
另觀察頻道間表現，膚電反應顯具主導地位，呼吸與心脈則分別在鑑別度指數及臨界比分析上取得高評價。光電容積對比下難謂出色。另研究在頻道間一致性檢驗上，發現膚電、心脈及光電容積相對於呼吸，儼然為2大族群，惟不應逕推導呼吸頻道價值較低或區辨差等結論。至於評分者間信度達.90以上，顯見在特徵點及標準明確下，即能確保評分者間信度。此外，性別及年齡確實造成頻道量分差異，但僅少數頻道受影響的情況下，無須過分推論此結果。
最後，本文保守以為新式頻道可達增加生理資訊的基本要求，但堅持開發更多生理訊號，強化電腦測量輔助工具應用，始能有效降低「無法鑑判」，文末揭示研究限制及實務具體建議。

This study incorporated the Photo-Plethysmography (PPG) channel into traditional polygraph channels to record changes in peripheral vascular constriction, aiming to assess its potential in enhancing diagnostic accuracy. The physiological signals of examinees, collected using the Lafayette-LX6 computerized polygraph between 2022 and 2023 by the Criminal Investigation Bureau, served as the study materials. The findings indicated that the highest classification accuracy was achieved with a“4 channels, three-position scale scoring system”, reaching 71.8%, though statistical significance was not attained. Subsequent comparisons with polygraph results and judicial decisions provided no substantial evidence supporting a significant improvement in diagnostic performance. These results suggest that prior international studies may have overestimated the efficacy of the PPG channel while downplaying the influence of other contributing factors on classification accuracy.
Further examination of individual channel performance revealed that the electrodermal activity (EDA) played a dominant role, while respiratory and cardiovascular responses received high evaluations in item discrimination index and critical ratio analysis. In contrast, the PPG channel demonstrated relatively limited effectiveness. Moreover, consistency testing indicated that EDA, cardiovascular responses, and PPG formed a distinct cluster compared to respiration; however, this should not be interpreted as evidence of the latter’s lower value or weaker discriminatory capability. Inter-rater reliability was found to exceed .90, suggesting that when deceptive physiological features and scoring criteria are well-defined, consistency among raters can be ensured. Moreover, while gender and age differences contributed to variations in the distribution of channel scores, the limited number of affected channels suggests that these findings should not be over-interpreted.
In conclusion, this study conservatively posits that the inclusion of the PPG channel meets the fundamental requirement of expanding physiological information. However, to effectively reduce cases of“inconclusive results,”greater emphasis should be placed on signal development and the enhancement of computer-assisted measurement tools. The study concludes by outlining its limitations and providing practical recommendations for future research and applications.