在目标导向的阅读理解过程中,人类的注意力依赖于任务优化.
Jiajie Zou1,2, Yuran Zhang1, Jialu Li3
1Key Laboratory for Biomedical Engineering of Ministry of Education, College of Biomedical Engineering and Instrument Sciences, Zhejiang University, Hangzhou, China.
eLife
|November 30, 2023
概括
计算模型预测了目标导向阅读期间的注意力. 优化阅读任务的深度神经网络 (DNN) 解释文字阅读时间,反映人类对文本特征和问题相关性的注意力模式.
科学领域:
- 认知科学 认知科学
- 计算神经科学是一种神经科学.
- 人工智能的人工智能
背景情况:
- 了解复杂的,以目标为导向的任务中的注意力分配是认知科学中的一个重大挑战.
- 阅读以回答问题是一种常见的现实世界任务,需要集中注意力.
研究的目的:
- 调查能够解释目标导向阅读期间注意力分布的计算模型.
- 确定任务优化如何影响阅读中的注意力分配.
主要方法:
- 使用基于变压器的深度神经网络 (DNN),优化了阅读理解任务.
- 相关联的DNN注意力权重与通过眼睛跟踪获得的人类文字阅读时间.
- 对阅读理解和词预测任务进行训练的DNN进行了比较.
主要成果:
- 通过对同一阅读任务训练的DNN的注意力权重,准确地预测了每个单词的阅读时间.
- 眼睛追踪显示,在人类中,对文本特征 (第一次通过) 和问题的相关性 (重读) 给予了明显的关注.
- 浅层DNN层通过文本特征调节注意力,而深层则通过问题相关性调节注意力.
- 经过训练的DNN用于单词预测,而不是阅读理解,预测了没有问题存在的阅读时间.
结论:
- 基于变压器的DNN为目标定向阅读中的注意力分布提供了可行的计算计算计算.
- 任务优化显著调节了注意力分配的方式,无论是在DNN还是人类读者中.
- 这些发现提供了关于低级别文本特征与高级别任务目标之间的相互作用的见解.
相关概念视频
01:30Information Processing Approach
42
The information-processing theory of cognitive development centers on fundamental mental processes, including attention, memory, and problem-solving skills. Researchers in this field examine how cognitive abilities, such as working memory, evolve and influence children's overall development. Studies indicate that children with stronger working memory tend to excel in reading comprehension, math, and problem-solving compared to peers with less efficient memory skills. Low working memory is...
42
01:19High-Level and Low-Level Awareness
273
Controlled processes in human consciousness represent high-alert mental states where individuals deliberately focus their attention on achieving specific goals. Controlled processes can be seen in situations like mastering new technology, where a person might become so absorbed that they ignore surrounding distractions. Such processes involve selective attention, requiring one to concentrate on particular elements of experience while disregarding others. These are governed by executive...
273