Researchers have shown that a simple reading-vocabulary can be trained by repeatedly pairing picture and word ant that fading the picture ovaer trials increases the efficiency of training. An attention hypothesis attributes the effectiveness of fading to an increase in attention to the word caused by the fading out of the picture. In an alternative hypothesis, stimulus change, it is proposed that the mechanism of fading derives from the increased salience of the changing stimuli, so that fading in the word would be as effective as fading out the picture. These competing hypotheses were tested by training 48 nonreading retarded persons (in four equated groups) under the following conditions : fade picture out, fade word in, double fade (pictures fades out and words fades in simultaneously), and standard (picture and word paired without fading). Tests of vocabulary-learning showed that the fade picture out and double-fade conditions, which were equally ineffective. These findings were inconsistent with the stimulus-change hypothesis, but consistent with a two-look attentional model, based on the Zeaman and House attention theory (1963).