The relationship between a free-recall measure and intelligence and memory abilities was investigated in a sample of 72 college students. After the subjects responded to a battery of eight intelligence and memory-marker tests, they were randomly assigned to two groups; one recalling immediately upon presentation of 30 pictorial stimuli, and the other after a delay of 30 seconds. Striking differences were obtained in the correlational patterns involving recall performance and scores on intelligence and memory tests, both between experimental groups and stages of acquisition. Intelligence variables were found to be good predictors of recall performance at later stages of acquisition under delayed recall, while under immediate recall, memory variables predicted recall performance best at early stages of acquisition. Results are discussed within the framework of cumulative learning models, and implications for the analysis of ability-learning interactions and the specification of learning components are presented.

Transler ellects in the A-B, A-C paradigmThere is increasing interest in the effects of task variables in determining the direction and magnitude of transfer and retroaction. For example, List-I response meaningfulness is directly related to the magnitude of the negative transfer in the A-B, A~ paradigm (Goulet, 1965;Jung, 1963). Furthermore, retroactive inhibition is substantially reduced in the same paradigm when the form cIass of the responses (e.g., letters, adjectives) is changed from List I to List 2 (Postman, Keppei, & Stark, 1965).The concern with the identification of task variables affecting transfer and retroaction has been paralleled by an increasing emphasis on identifying the basic unit of interference. In this framework, two hypotheses have been suggested. The first has been deemed specific interferenee by Ceraso (1964) where the unit of transfer is the specific stimulus-response pair. Thus, the unit of interference for a specific A-C pair on List 2 would be the A-B pair on List I which shared the same stimulus. The second hypothesis, deemed general interference by Ceraso, implies that the unit of interference on List 2 is the entire first list. Thus, learning of any A-C pair on List 2 would be subject to interference from all A-B pairs rather than from the single pair which shared the same stimulus on List I.The concept of general interference finds indirect and direct support from a number of sources. For example, Goulet (1967) and Goulet, MeHzer, & O'Shaunessy (1967) have found that negative transfer in the A-B, A-C paradigm is unaffected by partial reinforcement procedures implemented on List I which assured differential degrees of learning for individual sets of pairs. In other ~ords, the magnitude of negative transfer on List 2 was independent of the degree of learning of specific pairs within the list. Similarly, Battig (1966) has found that the magnitude of negative transfer for A-C pairs varied direct1y with the number of C-D (mixed-list) pairs in the list, transfer for the A-C pairs shifting from negative to positive as the number of C-D pairs was increased from two to six.Other evidence for general interference factors in retroactive inhibition is aviülable. Ceraso (1964) has found that unlearning effects generalize to pairs which have an A-B, C-D relationship on List I and List 2 as long as some of the pairs in the mixed list had an A-B, A-C relationship on List I and List 2.The present experiment was concerned with the effects of two variables on transfer. The first variable was level of List I partial reinforcement which was manipulated in terms of percentage of occurrence of response members (% ORM). This technique involves deleting the response terms on a fixed percentage of the learning trials. This procedure effectively IS8 retards both response and associative learning in direct proportion to the percentage of trials that the responses are deleted. ORM was varied at three levels in a mixed list, 10000, 50%, and 0%. In contrast to the 100%-and 50%-0RM pairs, no associative lear...