A consideration of the object-quality discrimination task as a dependent variable.

Abstract

Consistency of individual performances is an important requirement for most dependent variables. If one measures performance on a relatively unreliable test, the probability of obtaining significant results attributable to the effects of an independent variable is poor. Lack of internal consistency in the dependent variable leads, in general, to an underestimate of the action of an independent variable. In the investigation of learning, however, a constant level of performance scores is neither expected nor desired. By definition, changes in performance are expected between repeated tests or trials. When a learning task is employed as a measure, the more sensitive task is the one on which 5s' ranks remain relatively constant from trial to trial or from one stage of learning to another. A task on which ,5*3 vary widely in rank position from one stage of learning to another will be an insensitive indicator of the action of an independent variable.Carr (2, 3) has emphasized some of the practical and theoretical hazards that often occur in interpreting learning task reliabilities; however, if precautions are taken, the reliability figures undoubtedly contribute materially to the use of the task in question. In addition, the availability of reliability coefficients makes it possible to assess the effects of minor changes in the apparatus. For example, both Leeper (8) and Ruch (10) have demonstrated the increased reliability of animal mazes when retrace doors are added. McNemar (9, p. 122) has discussed the problem of the meaningfulness of reliability determination for learning tasks.The aim of the present investigation was to determine the reliability characteristics of the object-quality discrimination task in order to improve the usefulness of the test as a dependent variable.

Social interactions of rhesus monkeys: II. Effects of social interaction on the learning of discrimination tasks.

“…The learning of objectquality discrimination by inexperienced Ss was not altered by the testing in pairs on alternate clays. The rise in performance for both groups was consistent with the data collected from other groups of monkeys learning this task (12). It was found, however, that the testing of animals in pairs did significantly improve the learning of the oddity discrimination task.…”

Progressive discrimination learning in platyrrhine monkeys.

“…The success with which this measure differentiates the species is doubtless due to many factors, including a test reliability higher than that for many other measures (14). Perhaps one reason for this high reliability is the fact that the performance measure used above is a composite one involving several factors including acquisition, extinction, and transfer of relevant and irrelevant habits (5,16).…”

“…multiple-discrimination task is eventually performed by the rhesus monkey and the reliability of the resultant performance measures (14), caused us to investigate the behavior of New World monkeys on this task. Inasmuch as the cebus monkeys seem to surpass the rhesus on instrumentation problems (Biercns de Hann, 1; Kliiver, 10; Warden, Koch, and Fjeld, 19; and Harlow,6), we wondered whether their superiority would again be reasserted in the discrimination series.…”

Prompted vs. trial-and-error 3-trial object discrimination learning by monkeys

“…However, theUterature reveals a disconcerting lack of consensus regarding the absolute level of performance around this apparently practical lower limit of 3 tip. With monkeys, for example, both Miller et al (1955), using 3 tip, and Davis (1953), using 4 tip, obtained poorer performance than that reported by Levine et aI, and Hayes (1953) has even reported no interproblem improvement in one chimpanzee trained under 2 tip.…”