Automatic attraction of attention to former targets in visual displays of letters

In a classical experiment, Shiffrin and Schneider (1977, Experiment 4d) investigated how prolonged and consistent training in visual search for particular alphanumeric characters (digits among letters) affected performance after a change in the search criterion. After training, the participants were told to search a particular diagonal of a 2 3 2 matrix of characters for particular members of the set of characters (letters) that had been used as distractors during the training. Characters on the other diagonal should be ignored. The results showed that when digits (former targets) appeared on the diagonal to be ignored, detection of simultaneous letter targets on the diagonal to be attended was impeded. Apparently, the former targets ( foils) captured attention automatically, contrary to the will of the participants. A theoretical account of the effects of training described by Shiffrin and Schneider (1977) was outlined by Shiffrin and Dumais (1981) and Shiffrin, Dumais, and Schneider (1981). In this account, attentional weights of targets and distractors diverge during training, because stimuli gain in weight when they are used as targets and lose in weight when used as distractors. Quantitative formulations were later developed by Bundesen (1990) and Shiffrin and Czerwinski (1988). In Bundesen’s (1990, 1998) theory of visual attention (TVA), every stimulus is assigned an attentional weight, which determines the likelihood that the subject will attend the stimulus. The attentional weight, wx, of stimulus x is defined as