D'Angiulli Amedeo, Reeves Adam
Northeastern University, Boston, Massachusetts, USA.
Mem Cognit. 2002 Dec;30(8):1179-88. doi: 10.3758/bf03213401.
In three experiments, participants pressed a reaction time key and rated vividness after reading descriptions of common objects and imagining them as accurately as possible within a visual display subtending 1.2 degrees or 10.8 degrees. Display size had small effects on vividness and on image latency. Latency was much faster (approximately 2.5 sec) for vivid images than for nonvivid ones (approximately 7.5 sec), regardless of display size and initial size at which the images were generated (natural image size). Our findings characterize images as (1) detailed, as indicated by ratings of detail and by lack of category membership effects; (2) as occurring at many natural image sizes, so that time-consuming size adjustments are typically not needed; and (3) generated not concurrently with but after lexical access, as shown by latencies for the same image evoked by synonyms like pail and bucket. We conclude that image vividness and latency reflect the activation level of the visual imagery system, implying that image generation in everyday life mostly involves retrieving already-computed representations episodically stored in memory, which seldom require extensive re-elaboration such as size adjustment.
在三项实验中，参与者在阅读常见物体的描述并在一个张角为1.2度或10.8度的视觉显示器内尽可能准确地想象这些物体之后，按下反应时间键并对生动程度进行评分。显示器大小对生动程度和图像延迟的影响较小。无论显示器大小以及生成图像时的初始大小（自然图像大小）如何，生动图像的延迟都比不生动图像的延迟快得多（约2.5秒），不生动图像的延迟约为7.5秒。我们的研究结果表明，图像具有以下特点：（1）细节丰富，这由细节评分以及缺乏类别归属效应所表明；（2）在许多自然图像大小下都会出现，因此通常不需要进行耗时的大小调整；（3）不是与词汇提取同时生成，而是在词汇提取之后生成，正如由“桶”（pail）和“水桶”（bucket）等同义词唤起的同一图像的延迟所显示的那样。我们得出结论，图像的生动程度和延迟反映了视觉表象系统的激活水平，这意味着日常生活中的图像生成大多涉及从记忆中情景性地检索已经计算好的表征，这些表征很少需要诸如大小调整之类的广泛重新加工。
