The effects of types of supplementary information, example deviation, redundancy, and experience on categorization in four category sorting task

Abstract

The present study investigated categorization of stimuli from four populations, or schema families, as a function of five types of Supplementary Information, two levels of Example Deviation, two levels of Redundancy, and four Blocks of 40 Trials. Past research, finding contradictory effects of Supplementary Information, had studied one type of Supplementary Information in difficult tasks with short acquisition periods. In addition to the major variables, task difficulty, amount of acquisition trials, and time limits during the task were studied in the present research. Results showed that categorization was significantly better as Blocks of Trials increased, at 70% Redundancy than at 50% Redundancy, at low Example Deviation than at high Example Deviation. The generality of the above findings is limited by the significant interactions between all the variables. The effects of Supplementary Information depended upon the type of Supplementary Information, the level of Redundancy, the amount of Example Deviation, and the Block of Trials. The Correction condition was detrimental in comparison with the Noncorrection condition at 70% Redundancy, but the two types did not differ significantly at 50% Redundancy. At 70% Redundancy with low Example Deviation, differences in categorization for groups receiving the Examples, the Examples + Noncorrection, and the Examples + Correction conditions did not differ significantly and performance was nearly perfect. At all other combinations of Redundancy and Example Deviation, i.e., 70% Redundancy, high Example Deviation; 50% Redundancy, low Example Deviation; and 50% Redundancy, high Example Deviation, both the Examples + Noncorrection and the Examples + Correction conditions were significantly better than the Examples condition, but the two conditions did not differ significantly from each other. The results support the following conclusions. Task difficulty or short acquisition periods precluded assessment of effects of Supplementary Information in past research. The detrimental effects of a Correction condition were attributed to the identification of deviant stimuli as correct members of categories by E. Such identification might have induced Ss to discard potentially useful hypotheses, increased memory requirements, confused Ss as to category label. The detrimental effects of a Correction condition were reduced by addition of the Examples condition Which probably helped in rule abstraction, reduced memory requirements, and provided stable guides to categories. In summary, the three contradictory effects for Supplementary Information reported in past literature were found in the present study, and were shown to depend upon the type of Supplementary Information, the amount of Redundancy, the amount of Example Deviation, and the number of acquisition trials.