Abstract
When visual stimuli (letters, words or pictures of objects) are presented sequentially at high rates (8–12 items/sec.), observers have difficulty in detecting and reporting both occurrences of a repeated item: this is Repetition Blindness. Repetition blindness also occurs for sequences of pictures of non-objects (Arnell & Jolicoeur, 1997). Their non-objects were line drawn figures similar in complexity to drawings of real objects. Our experiments investigated the effects of repetition of various types of novel objects, and whether the representations bound to episodic memory tokens that yield repetition blindness are viewpoint-dependent or whether they are object-centred. Subjects were shown three types of three-picture sequences: familiar things, non-existent chimerical combinations of features of familiar things or non-objects like those used by Arnell and Jolicoeur (1997). Stimuli occurred in a vertical or in a rotated form. Recognition performance was superior for familiar stimuli and chimeras, but repetition blindness was observed for all three types of stimuli, both when the repeated stimuli were the same and when they differed in orientation. In further experiments, we presented simple coloured drawings of 3-dimensional novel objects, and manipulated rate of presentation and rotation in depth. Repetition blindness occurred only at the higher rate (105ms/item), and was found even for stimuli differing in orientation. We conclude that object-centred representations are bound to episodic memory tokens, and that these are constructed prior to object recognition. These results are contrasted with those found with written materials, and implications for explanations of repetition blindness are considered.