from prior levels of operations that the term qualitative should be introduced is a question that is closely related to the topic of this paper. It is, though, not crucial to the general case which will be made about learning abilities and their differences between species as revealed by empirical research.
The question is, how can we study indices of mediating mechanisms, cognition, and their possible relationship to what we call intelligence at the human level through comparative psychological studies with the diverse forms which comprise the order Primates?
Harlow ( 1949) defined the phenomenon known as learning set -- the formation of a facility to learn new problems of a given class as a function of experience in attempting to learn other problems from the same class. In its ultimate form, a learning set allows for one-trial learning of two-choice visual discrimination problems, with only the outcome of the choice on the first trial being necessary for the subject to perform errorlessly on subsequent trials, to know which object or stimulus must be selected subsequently if reward is to be obtained.
Learning set ability is, in part, a function of opportunities to learn problems of a given class, where each problem is presented for too few trials for mastery before the next is encountered. It also is a function of how well developed the brain of the learner is; at least this holds true, generally speaking, for species within the order Primates. (The fact that some birds are found to have facile learning set acquisition skills does not in any way conflict with the fact that there is an orderly relationship between brain development within the order Primates and ability to form a learning set. Rather, it is taken to mean that outside the order Primates there are life forms that have specialized learning skills that resemble at least basic learning set formation.) In Harlow's original view, it was with the formation of a learning set that the learner became freed of stimulus- response bondage to become a proficient, one-trial, insightful learner.
As stated, learning set has been found to be related, in general, to the development of the brain, and this makes learning set appear at first blush to be of obvious value for broad use as a comparative psychological tool. However, the diverse and ill defined differences in species' receptor- mechanisms and in their perceptual and motivational systems lead to questions regarding the wisdom of direct comparisons between species' learning set skills when their absolute performance levels are used as data.
In an attempt to improve Harlow's learning set measurement so as to achieve a defensible assay of primates' abilities to learn-how-to-learn, a variant of the discrimination-reversal learning set was developed to obtain what is called a Transfer Index ( Rumbaugh, 1970; Rumbaugh & Gill, 1973).
Discrimination-reversal tasks differ from the basic learning set tasks in that at some point in training the initial cue values are reversed for the remainder of the problem so that the initially correct stimulus (S+) becomes incorrect (S-) and vice versa. The basic tenet of the Transfer Index is that the trials after reversal serve as a test of transfer of