再探網路成癮在動態決策作業之表現：以原始版及逆轉版愛荷華賭局作業為例

愛荷華賭局作業(Iowa Gambling Task, IGT)為一模擬真實生活決策之作業，已成為決策行為之臨床評估工具(Bechara, 2007, 2016)，並宣告可用於評估物質成癮及嗜賭症。此工具在臨床與風險決策研究逐漸扮演重要之角色。部分網路（遊戲）疾患(Internet [Gaming] Disorder, IGD)之研究認為其應與物質成癮及嗜賭症有相似之行為特質。本研究回顧十篇採用IGT於比較IGD與控制組之研究。結果顯示，其中三篇IGD表現較差，有兩篇IGD表現較好，另外有五篇兩組沒有差異。文獻之間顯然產生不一致之結果。本研究認為其問題可能發生在三個向度上：作業、變項、方法。(1)「作業向度」：以IGT為基礎並增加逆轉版IGT交互驗證期望值是否可為區辨指標，並以兩倍的作業長度，觀察決策行為之變化；(2)「變項向度」：過去研究大多以期望值變項為推論之基礎，而新近研究逐漸注意輸贏頻率變項之影響，故本研究同時評估兩變項之作用；(3)「方法向度」：過去研究大多使用期望值淨分數指標，但新近研究皆建議應採用原始四疊牌之分數，以避免淨分數掩蓋單疊牌之區辨效力。本研究兩實驗皆採組間設計，實驗組為排除菸癮及其他成癮疾患之網路成癮者，實驗一採用IGT，網癮組24位與控制組35位；實驗二採用逆轉版IGT(簡稱 rIGT)，網癮組24位與控制組38位。以電腦版賭局作業進行施測及資料分析(Matlab 2015a)。施測程序為，徵求參與者知情同意，再進行兩階段電腦版作業，最後填寫陳氏網路成癮量表，並依照切截分數進行分組（網癮組 vs. 控制組）。 IGT及rIGT之結果顯示，兩作業皆無法有效區辨網癮組與控制組之決策行為差異，僅於IGT之D牌和rIGT之rC牌看到兩組選擇次數些微的差異。兩組在兩實驗中皆顯示「奇異B牌」與「鏡像奇異B牌」現象。在IGT中，「輸贏頻率」較「期望值」有相對高的解釋力。總結來說，IGT及rIGT可能不是評估網癮者決策行為之適當工具。本研究之結果隱含兩可能之解釋：(1)IGT系列作業無法測得網癮者之決策行為特質；(2)網路成癮之行為仍伴隨著資訊科技發達的現代生活環境進行共同演化，其是否會成為行為疾患仍有待觀察。

關鍵字

網路遊戲疾患；網路成癮；愛荷華賭局作業；逆轉版愛荷華賭局作業；期望值；輸贏頻率；奇異B牌現象；鏡像奇異B牌現象

並列摘要

The Iowa Gambling Task (IGT) is a task that simulates real-life decisions. In recent decades, (Bechara, 2007, 2016) the IGT has also been developed as a clinical assessment tool, such that the choice behavior of substance addiction and gambling disorders can be evaluated using the clinical version of the IGT. Relatedly, the IGT has played an important role in both clinical research and risk decision research. Some studies of Internet gaming disorder (IGD) and Internet addiction (IA) have posited that individuals with IGD might share some behavioral characteristics with individuals suffering from substance addiction and gambling disorders. The present study reviewed 10 IGT-based studies of IGD (including some investigating IA) and found that they reported relatively inconsistent results with respect to the mean net IGT score of those with IGD versus control subjects (specifically, IGD > control [2 studies]; IGD < control [3 studies]; IGD = control [5 studies]). This study proposed that these incongruent results in the IGT-based IGD literature might be explained by 3 dimensions, specifically, the task, variable, and method dimensions. (1) Task dimension: this study utilized the standard IGT and the reverse version of the IGT (rIGT) and doubled the number of trials in both tasks in order to integrally observe the changes in choice behavior and test the validity of the net score based on the expected value (EV). (2) Variable dimension: the interpretations of IGT choice patterns in IGT-based IGD studies have typically been primarily based on the EV assumption. However, more and more IGT studies have suggested the influence of a new variable known as gain-loss frequency. In this study, we evaluated and analyzed both of these variables concurrently. (3) Method dimension: the 10 IGT-based IGD studies reviewed by this study mainly adopted the EV net score as the index of choice behavior, but some recent IGT-based studies have suggested that the mean selection number of each deck would be helpful data for observing some effective indices during IGT performance. Both of the experiments performed in this study utilized a between-group design, where the experimental group consisted of individuals with Internet addiction (IA) and no tobacco and other substance addiction. For Experiment 1, a total of 24 participants with IA and 35 control subjects played the IGT computer game, whereas for Experiment 2, a total of 24 participants with IA and 38 control subjects played the rIGT computer game. The computer versions of the IGT and rIGT were programmed with Matlab 2015a. Each participant first signed an informed consent form, then played the computer game over 2 sessions, and then completed the Chen Internet Addiction Scale (CIAS). The categorization of participants as those with IA or as control participants was based on the CIAS score, with a score of 67 serving as the cut-off between the two groups. The two experiments and the subsequent analysis revealed that most indices of the IGT and rIGT were unable to distinguish between the choice behavior of the participants with IA and the controls. More specifically, there were significant differences between two groups only on deck D of the IGT and deck rC of the rIGT. Notably, both groups exhibited the “prominent deck B” (PDB) and “mirrored PDB” phenomenon on the IGT and rIGT. Compared to the EV, the gain-loss frequency variable had more explanatory power in the IGT. Overall, these results suggest that although the IGT has long been used as a clinical neuropsychological assessment tool, it may not be relevant for assessing IA-related choice behavior. It is worth noting that the present study implied two possible inferences: (1) that the IGT serial tasks are invalid with respect to evaluating Internet gaming behavior and (2) that Internet gaming behavior may change in a manner coherent with the development of modern information-technology. As such, there is still some room for discussion of the question of whether Internet gaming behavior should be categorized as a behavior disorder.

並列關鍵字

Internet gaming disorder ； Internet addiction ； Iowa gambling task ； reverse version of Iowa gambling task ； expected value ； gain-loss frequency ； prominent deck B phenomenon ； mirrored prominent deck B phenomenon