Employing computer technology to assess visual attention in young children and adolescents with severe mental retardation.

The intent of this investigation was to establish a valid and sensitive computer measurement technique for educational assessment applications. An integral part of the investigation was to establish the similarities and differences in how prior reinforcement histories of individual stimuli affect attention to compound visual cues for young children of normal development versus adolescents with severe mental retardation, both groups having comparable mental age. A series of identical conflict compound discrimination tasks was presented to the two groups. In addition, generalization effects were investigated for both groups by presenting compounds containing some or all novel cues. The similarities and differences in performance for young children of normal development and adolescents with severe mental retardation were analyzed using multiple testing procedures. In addition to assessing stimulus control by presenting stimulus components separately following acquisition of compound discriminations, response topographies of the compound stimuli were recorded with a touch screen attached to a computer monitor screen. This study demonstrated that overselective attention did not occur only for students with severe mental retardation but also for young children of normal development if multiple tests were employed. A difference was found, however, between the two populations in the efficiency with which they shifted attention among elements of complex stimuli depending on prior conditioning histories. Presentation of compounds whose components had conflicting reinforcement histories was found to be a more sensitive assessment technique than presentation of compounds containing some or all novel components for distinguishing between the two groups. The use of multiple testing procedures was critical in preventing false conclusions from altered test performances arising from reinforcement contingencies in effect during the test.