• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>IEEE Symposium Series on Computational Intelligence >Letter position encoding in a neural framework
【24h】

Letter position encoding in a neural framework

机译：神经框架中的字母位置编码

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Visual word recognition requires the identification of a word's component letters as well as their position within a word. The position of letters, once thought to be coded to the precise location in a word, is now understood to be more flexible. Indeed, a wealth of experimental evidence supports a view of visual word recognition that is forgiving to local changes in the position of letters. We present a model of letter position encoding, as an extension of the Overlap model, that treats letters as distributions along a normalized retinotopic space. Additionally, letters that do not belong in a target word, or are distant from the expected location, are modeled to inhibit the activation of the target word. This method has not yet been explored in the letter position encoding literature, and can help increase the biological plausibility of these models. We show that model estimates fit well with a database of priming studies used to investigate the effects of letter position manipulations on participants' reaction time and accuracy.
机译：视觉单词识别需要识别单词的组成字母以及它们在单词中的位置。字母的位置曾经被认为是编码到单词中的精确位置，现在被认为更加灵活。确实，大量的实验证据支持了视觉单词识别的观点，这种观点原谅了字母位置的局部变化。我们提出了一种字母位置编码模型，作为Overlap模型的扩展，该模型将字母视为沿标准化视网膜视位空间的分布。此外，对不属于目标单词或远离预期位置的字母进行建模以禁止激活目标单词。该方法尚未在字母位置编码文献中进行探索，并且可以帮助增加这些模型的生物学可信度。我们表明，模型估计值与用于研究字母位置操纵对参与者的反应时间和准确性的影响的启动研究数据库非常吻合。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号