Bernard Jean-Baptiste, Castet Eric
Laboratoire de Psychologie Cognitive, Centre National de la Recherche Scientifique, UMR 7290, France; Aix Marseille Université, France; Brain and Language Research Institute, France.
Laboratoire de Psychologie Cognitive, Centre National de la Recherche Scientifique, UMR 7290, France; Aix Marseille Université, France.
Vision Res. 2019 Feb;155:44-61. doi: 10.1016/j.visres.2018.12.006. Epub 2019 Jan 15.
Letters and words across the visual field can be difficult to identify due to limiting visual factors such as acuity, crowding and position uncertainty. Here, we show that when human readers identify words presented at foveal and para-foveal locations, they act like theoretical observers making optimal use of letter identity and letter position information independently extracted from each letter after an unavoidable and non-optimal letter recognition guess. The novelty of our approach is that we carefully considered foveal and parafoveal letter identity and position uncertainties by measuring crowded letter recognition performance in five subjects without any word context influence. Based on these behavioral measures, lexical access was simulated for each subject by an observer making optimal use of each subject's uncertainties. This free-parameter model was able to predict individual behavioral recognition rates of words presented at different positions across the visual field. Importantly, the model was also able to predict individual mislocation and identity letter errors made during behavioral word recognition. These results reinforce the view that human readers recognize foveal and parafoveal words by parts (the word letters) in a first stage, independently of word context. They also suggest a second step where letter identity and position uncertainties are generated based on letter first guesses and positions. During the third lexical access stage, identity and position uncertainties from each letter look remarkably combined together through an optimal word recognition decision process.
由于诸如视敏度、拥挤效应和位置不确定性等视觉限制因素，识别视野中的字母和单词可能会很困难。在这里，我们表明，当人类读者识别出现在中央凹和中央凹旁位置的单词时，他们的行为就像理论观察者一样，在经过不可避免且非最优的字母识别猜测后，独立地最优利用从每个字母中提取的字母身份和字母位置信息。我们方法的新颖之处在于，我们通过测量五个受试者在没有任何单词上下文影响的情况下拥挤字母的识别性能，仔细考虑了中央凹和中央凹旁字母身份和位置的不确定性。基于这些行为测量，由一个观察者对每个受试者的不确定性进行最优利用，为每个受试者模拟词汇通达过程。这个自由参数模型能够预测出现在视野中不同位置的单词的个体行为识别率。重要的是，该模型还能够预测在行为单词识别过程中出现的个体字母错置和身份错误。这些结果强化了这样一种观点，即人类读者在第一阶段通过部分（单词字母）识别中央凹和中央凹旁的单词，而与单词上下文无关。它们还表明了第二步，即基于字母的首次猜测和位置产生字母身份和位置的不确定性。在第三个词汇通达阶段，通过最优的单词识别决策过程，每个字母的身份和位置不确定性显著地结合在一起。
