关键词: 罪犯, 爱荷华赌博任务, 前景效用学习模型, 决策功能缺陷

Abstract: Affective decision making is an individual ability to make choice under uncertainty. Criminal behavior is a result of criminals’ decision-making under uncertainty. Criminals are to be fettered and thrown into prison because they have made wrong decisions. Are there any defects of different types of criminals’ decision-making function? If there are, what caused their defects? And are these reasons different or the same?
The Iowa Gambling Task (IGT) which developed by Bechara et al. (1994) has been widely used to evaluate decision-making ability under uncertainty. In this study, 222 male criminals and 32 controlled normal males were recruited. All of them were asked to fill out a demographical form and then complete IGT. A 10 (block)×9 (participant type: 8 types of criminal group and 1 control group) experimental design was adopted. The independent variables were patricipant type and block, dependent variable were the proportion of advantage decks [(deck3+deck4)/100] and each deck. The Prospect Utility Learning model (PUL) (Ahn et al., 2008) was applied to analysis participants’ choices behavior. PUL model contained four parameters: loss aversion (λ), shape of function (α), learning update rate (A), and choice consistency (c).
A repeated ANOVA was applied to the data analysis. (1) Along with time passing, control group chose more from advantage deck 3 and 4, but 8 types of criminals preferred disadvantage deck 2. There was no significant difference in disadvantage deck 1 between criminals and control group. (2) Comparing to the control group, violent offenders and mafia-like criminals’ loss aversion parameter and shape of function parameter were significant lower, which indicated that they were not sensitive to reward and punishment. Their learning update rate was higher than control and other types of criminals which indicated that they were fast discount the utility of past events. And their choice consistency was significant higher than other types of participant indicated that they did not form right utility of each deck. Drug abstainers, drug criminal, thief and robbery were lower in the loss aversion parameter, which indicated that they were insensitive to punishment; there higher choice consistency indicated that they did not form right utility of each deck. Sex offender’s choice consistency parameter was higher than control, but their loss aversion, shape of function, and learning update rate parameters were normal. So, their high choice consistency may indicate that their behavioral reversal function were defected. Economic criminal’s choice consistency was very low may indicated that they were more cautious than other types of criminal.
In conclusion, the present research suggested that different types of criminal have decision-making functional deficits, and the reasons caused these deficits were different.

Key words: criminal, Iowa Gambling Task, prospect utility learning model, decision-making defects