An evaluation of exemplar-based models of generalization was provided for illdefined categories in a category abstraction paradigm. Subjects initially classified 35 high-level distortions into three categories, defined by 5, 10, and 20 different patterns, followed by a transfer test administered immediately and after 1 wk. The transfer patterns included old, new, prototype, and unrelated exemplars, of which the new patterns were at one of five levels of similarity to a particular training (old) stimulus. In both experiments, increases in category size and oldnew similarity facilitated transfer performance. However, the effectiveness of old-new similarity was strongly attenuated by increases in category size and delay of the transfer test. It was concluded that examplar-based generalization may be effective only under conditions of minimal category experience and immediacy of test; with continued category experience, performance on the prototype determines classification accuracy.

Ease of prototype abstraction and recognition of new instances belonging to the prototype was facilitated by increasing the number of instances sorted together during original learning. The 5s sorted distorted dot patterns into groups of 3, 6, and 9 instances, each group containing distortions generated from a single prototype. Following the sorting task, half of the 5s were tested immediately on their ability to correctly classify old and new patterns as well as the prototype; another group of 5s was tested 4 days later. It was found that (a) correct classification of both the prototype and new instances increased as a function of the number of old instances sorted together in the original learning task; and (b) old instances exhibited some forgetting over the delay, but neither the prototype nor new instances did. It was concluded that the abstraction of a prototype undergoes repeated change as a function of the number of instances which define it, and that the ability to correctly recognize new exemplars of a concept is dependent upon the number of instances as well.A number of recent experiments have attempted to understand how information is abstracted from a set of instances that are classified together (Franks & Bransford, 1971;Posner & Keele, 1968Reed, 1972;Strange, Keeney, Kessel, & Jenkins, 1970). Unlike more typical concept formation tasks (e.g., Bourne, 1966), which require the discovery of verbalizable rules along well-defined dimensions, Posner and Keele (1968, 1970) have investigated the abstraction process with stimuli which are not easily verbalizable and whose dimensions are poorly understood. The basic paradigm requires that an initial "prototype" be determined (by randomly assigning 9 dots to a 30 X 30 matrix), and that instances of that prototype be derived (usually by distortion of the prototype, in which each dot is moved according to some statistical decision rule). A number of prototypes are formed in a like manner, each with its own set of instances. In the first part of the experiment, 5s are shown only the instances of the prototype in a random order, and learning is complete when the instances of each prototype are classified together. In the second part, Ss are tested 1 Requests for reprints should be sent to Donald

No abstract

Three experiments investigated the identification or localization of a letter that was displaced from the fixation point by 1 degree-3 degrees. The subject's task was to identify a fixated letter and identify (Experiment 1) or localize (Experiments 2 and 3) the displaced letter. On uncued trials, the displaced letter could appear at any of eight locations on any of three rings surrounding the fixated letter; on cued trials, the ring containing the displaced letter was specified. The results indicated that cuing improved subjects' identification and localization of the displaced letter. Invalid cuing (Experiment 3) produced costs comparable in magnitude to the benefits. The distance of the target from the cued ring determined cost, but costs were unaffected by the appearance of a target within the presumed beam of attention. It was proposed that attention should be viewed as a general, rather than feature-specific, resource that can be voluntarily allocated to multiple regions of the visual field.

No abstract