Computational research on neural mechanism and learning principle for human language acquisition

为了定量地研究大脑的高阶功能，如语言处理，任务的反应时间特征（RT）可以作为一个很好的衡量难度的指标。在本研究中，我们测量了光分辨反应时间任务中依赖于刺激持续时间的RT和错误率。在每次试验中，受试者分别在短时间（150ms）或长时间（1s）内看到两个不同亮度的光块。当刺激持续时间“短”时，参与者的反应更快，辨别能力也比长时间时差。结果表明，视觉皮层的不同感觉反应决定了刺激时间对反应速度和准确度的影响。结果表明，简单的“赢者通吃”型神经网络模型从初级视觉皮层接收瞬态和持续刺激信息，成功地再现了正确反应和错误率的RT分布。此外，从模型网络中获得的时间尖峰序列与猴子在其他视觉决策任务（如运动识别和古怪检测任务）中的前额叶或顶叶区域的神经活动非常相似。对于小刺激发生异步（SOA）对两选择亮度-分辨RT任务的影响，我们测量了对SOA的依赖性，并通过在模型中引入全局抑制剂来防止不必要的感觉信息整合，以及具有前馈相互抑制的神经元链作为两次闪光的异步检测器，成功地再现了对SOA的依赖性。修正后的模型不仅再现了有和没有SOA的两种选择的亮度辨别任务的反应，而且还再现了对大范围刺激亮度的超阈值单刺激检测任务的反应。

To quantitatively investigate the higher-order functions of the brain such as language processing, characteristics of reaction time (RT) of the subjects for tasks can be used as a good measure of difficulty. In this research, RT and error rate that depend on stimulus duration were measured in a luminance-discrimination reaction time task. Two patches of light with different luminance were presented to participants for short (150ms) or long (1s) period on each trial. When the stimulus duration was 'short', the participants responded more rapidly with poorer discrimination performance than they did in the longer duration. The results suggested that different sensory responses in the visual cortices were responsible for the dependence of response speed and accuracy on the stimulus duration during the luminance-discrimination reaction time task. It was shown that the simple winner-take-all-type neural network model receiving transient and sustained stimulus information from the primary visual cortex successfully reproduced RT distributions for correct responses and error rates. Moreover, temporal spike sequences obtained from the model network closely resembled to the neural activity in the monkey prefrontal or parietal area during other visual decision tasks such as motion discrimination and oddball detection tasks.As for the influences from small stimulus onset asynchrony (SOA) in a two-choice luminance-discrimination RT tasks, the dependence on SOAs was measured and successfully reproduced by introducing global inhibitors to prevent unnecessary integration of sensory information, and neuron chains with feedforward mutual inhibition as an asynchronism detector of two flashes, into the model that we proposed. The revised model not only reproduced responses of these two-choice luminance- discrimination tasks both with and without a SOA but also responses in a suprathreshold single-stimulus detection task for a wide range of stimulus luminance.