Psychological timing of multi-alternative decision: Modelling study

s / Neuroscience Research 71S (2011) e108–e415 e273 P3-n09 Comparison between temporal discounting of reward value in schedule task with and without decisionmaking by monkeys Tsuyoshi Setogawa , Takashi Mizuhiki, Kiyonori Inaba, Munetaka Shidara Grad. Sch. of Comprehensive Human Sci., Univ. of Tsukuba, Tsukuba, Ibaraki, Japan We previously reported that the monkeys showed better behavioral performances in the decision-making schedule task than the tasks without decision-making. This phenomenon is called “self-choice effect”. In this study, we analyzed error rates of three kinds of reward schedule task: the computer assigned reward schedule task with and without matched probability (RSm and RSe, respectively), and the decision-making schedule task. In RSe, the monkey had to complete the schedule composed of 1, 2, 3 or 4 trials of visual discriminations to earn 1, 2, 3 or 4 drops of liquid reward. The number of trials and reward amount were determined by the computer with equal probability. In the decision-making schedule task, the monkey was required to choose one of two visual targets. The brightness and the length of the targets were proportional to the amount of liquid reward (1–4 drops) and the required number of visual-discrimination trials (1–4 trials) to be performed, respectively. After one target was chosen, the corresponding schedule was presented. In RSm, the probability of the schedule length and the amount of reward were set to be that measured in the chosen schedules in the decision-making schedule task performed in the prior probe period. We fitted the error rate of the reward schedule part in three tasks by two types of temporal discounting model of reward value. The exponential and hyperbolic discounting models were used: E = exp(kD)/aR and E = (1 + kD)/aR where E: error rate, D: number of trials, R: reward amount, k: discounting factor, a: constant. The monkey’s error rate was better fit by the exponential discounting model. The discounting factor, k, and the constant, a, showed the highest values in the decision-making schedule task across three tasks. These results suggest that the monkeys discount the reward value more in the decision-making schedule task while increase in the constant value reflects the rise in reward value, which contributes to the lower error rate. Research fund: KAKENHI (17022052, 22300138), Japan. doi:10.1016/j.neures.2011.07.1192 P3-n10 Psychological timing of multi-alternative decision: Modelling study Hiroshi Okamoto 1,2 , Tomoki Fukai 1 1 RIKEN Brain Science Institute, Wako, Japan 2 Research & Technology Group, Fuji Xerox Co., Ltd The timing of decision between multiple alternatives is known to be characterized by Hick’s law, which states that the response time (the amount of time taken for choosing one alternative) is a log-linear function of the number of alternatives. Although Hick’s law is widely used as a design principle in human–computer interaction design, surprisingly little is established on the neural basis of this empirical law. Here we propose a cortical-circuit model that reproduces Hick’s law. The model assumes multiple pools of neurons corresponding to multiple alternatives. Simulation analysis revealed a characteristic time course of neuronal activation. At first the activity of one pool, defined by the neuronal firing rate averaged over this pool, is slowly ramped up, while those of the other pools gradually decrease. Then at a certain time the former sharply increases to the maximum, while the latter ones abruptly drop to the minimum. We regard the alternative corresponding to the maximally activated pool as the chosen alternative and define the response time by the timing of the sharp onset of this maximal activation. We found that if the values of the parameters in the model are properly adjusted the response time averaged across trials increases log-linearly with the number of alternatives. The model also gives testable prediction to psychophysical experiment: When the timing of multi-alternative decision follows Hick’s law, the coefficient of variation of the response time (the ratio of the standard deviation to the mean) is kept constant with the number of alternatives. Preliminary data from a recent psychophysical experiment supports this prediction (see Fig. 6 of (1)). Research fund: KAKENHI (20500279).