重驗思覺失調症患者於愛荷華賭局作業之決策行為—以期望值和輸贏頻率為觀點

諸多研究顯示思覺失調症(schizophrenia, Sz)患者有執行功能逐漸退化的現象，此類患者多用威斯康辛卡片分類測驗(Wisconsin Card Sorting Test, WCST)觀察此功能之缺損。然近來之研究指出此類病患亦有決策功能之缺陷，故許多研究採用愛荷華賭局作業(Iowa Gambling Task, IGT)觀察其在動態不確定環境下之決策模式。本研究回顧過去比較Sz組和控制組(control, Ctrl)於IGT表現之38篇文獻。若依據原始版IGT期望值觀點之分析方式，可觀察到其中以Beninger等人(2003)為代表之24篇(63%)文獻報告Sz組和Ctrl組存在顯著差異，而以Wilder等人(1998)為代表之14篇(37%)文獻並未發現兩組有顯著差異。顯然文獻間存在不一致之結論，為解決上述之矛盾現象，本研究開展兩部分之研究： (1) 研究一整理並分析38篇文獻之IGT分數，並觀察期望值觀點、輸贏頻率觀點等其他變項之影響力，及兩觀點（既期望值觀點指標與輸贏頻率（奇異B牌現象））之相互關係，以此釐清文獻不一致之現象。 (2) 研究二以實徵資料再驗IGT各變項能否有效區辨Sz組和Ctrl組之決策模式，並探討決策功能、執行功能與臨床症狀之相關。本研究一納入38篇文獻，選取其中呈現四疊牌選牌數之18篇文獻，進行IGT分數之分析。研究二納入經精神專科醫生評估篩選61名Sz個案（思覺失調症或情感性思覺失調症）和62位Ctrl參與者，每位參與者均接受IGT和WCST電腦版作業，Sz組還需進行活性與負性症狀評量表(Positive and Negative Syndrome Scale, PANSS)及個人及社會功能量表(The Personal and Social Performance Scale, PSP)之評估。研究一顯示期望值觀點等諸多指標可區辨兩組決策能力之差異，但文獻間Sz組之選牌模式一致且呈現顯著的奇異B牌現象；Ctrl組選牌模式變異較大，亦呈現顯著的奇異B牌現象。顯然單純只觀察期望值之指標會掩蓋奇異B牌現象。研究二顯示兩組在期望值觀點、輸贏頻率觀點之相關指標均無顯著差異，故IGT大部分指標皆不具區辨Sz組與Ctrl組決策能力之效力。Sz和Ctrl兩組卻仍具有明顯的奇異B牌現象。研究二之Sz選牌模式與文獻一致而Ctrl之選牌模式仍存在變異。值得注意的是研究二觀察到IGT之B、D牌及期望值觀點指標與WCST部分指標具相關性；而IGT各指標與PANSS、PSP均無顯著相關，本文此部分之觀察與所回顧之大部分文獻一致。研究一發現Sz和Ctrl之選牌模式受輸贏頻率影響，且文獻間結果不一致可能來自於各文獻Ctrl組IGT分數之變異而非Sz組；研究二顯示各指標均不足以區辨兩組之決策模式但兩組選牌模式均受輸贏頻率影響，另IGT（決策）與WCST（執行功能）兩測驗所測量之心智功能存在部分相關。綜合兩研究之結果，輸贏頻率為主導兩族群IGT表現之主要變項，且只觀察期望值之效果易忽略輸贏頻率之效果。IGT之大部分行為指標不具備區辨Sz組與Ctrl組決策模式之效力，但值得注意的是臨床版IGT(Bechara, 2007, 2016)仍宣告其能評估Sz之決策能力，此工具之臨床應用實有待修正。

關鍵字

思覺失調症；決策；愛荷華賭局作業；期望值；輸贏頻率；奇異B牌現象

並列摘要

Many studies have shown that patients with schizophrenia (Sz) have gradual degeneration of executive function revealed by the Wisconsin Card Sorting Test (WCST). Recent studies have pointed out that schizophrenia may also exist decision-making dysfunction. And most studies apply the Iowa Gambling Task (IGT) to assess decision-making function under dynamic and uncertain conditions. Our study reviewed 38 relevent studies focusing on comparing IGT performance bewtween Sz and control(Ctrl). Based on the analysis of net score defined by the original IGT, there is a significant difference between Sz and Ctrl in 24 articles (63%) represented by Beninger et al. (2003), While 14 articles (37%) represented by Wilder et al. (1998) indicated no significant difference between two groups. Obviously, this revealed inconsisitent results among the literature of IGT performanc on schizophrenia. In order to elucidate the contradictory phenomenon, our research will carry out two studies: (1) Study one organized and analysed the IGT indicators of 38 articles, and observed the influence of different variables such as expected value and gain-loss frequency, and explored the relationship between these two views (expected value indicator and gain-loss frequent indicator (prominent deck B phenomenon)), so as to clarify the inconsistency of the literature. (2) Study two reexamined the validity of IGT indices between Sz and Ctrl in empirical experiments, and explored the correlations among decision-making behaviors, executive function and clinical symptoms. Study one included 38 articles and selected 18 articles which demostrated the tables or charts of four-deck choices in IGT. Study two recruited 61 schizophrenia patients (schizophrenia or schizoaffective disorder) diagnosed by psychiaists and 62 Ctrl participants. Each participant received computerized IGT and WCST, and Sz group additionally complete the Positive and Negative Syndrome Scale (PANSS) and the Personal and Social Performance Scale (PSP) evaluated by researchers. Study one indicated that expected value index and others can distinguish the difference between two groups. The IGT selection pattern of Sz in 18 literatures is relatively consistent and illustrated a robust prominent deck B phenomenon. However, the Ctrl’s selection pattern in 18 literatures is unstable, but still demonstrated prominent deck B phenomenon. Apparently, dependently obsevering the effect of expected value index might conceal the prominent deck B phenomenon. Study two reveals that no significant difference between two groups in terms of expectancy and the gain-loss frequency indicators. Therefore, most IGT indicators can’t distinguish the IGT performance between two groups. IGT performance of Sz and Ctrl both highlight with prominent deck B phenomenon. Additionally, it is notable that deck B and D of IGT and expected value index are correlated with partial indicators of WCST. Conversely, it is no significant correlation between IGT and PANSS and PSP, which is consistent with most of the literatures reviewed before. Study one finds gain-loss frequency impose a strong effect on selection pattern of Sz and Ctrl, and the inconsistency of IGT performance in the literature may results from the variation of the Ctrl rather than the Sz. Study two implies each index can’t distinguish the IGT patterns in two groups, but also shows IGT patterns largely influented by gain-loss frequency. Partial correlation exists between the IGT (decision-making) and WCST (executive function). According to the results of the two studies, gain-loss frequency exerts a dominant impact on IGT performance in two populations. Most behavioral indicators of IGT are not valid enough to discriminate the decision-making function between Sz and Ctrl. Yet, it is worth noting that the clinical version of IGT (Bechara, 2007 and 2016) still declares its reliable and valid to evaluate Sz, the clinical application of this assessment need to be revised urgently.