Transfer between paired-associate and serial learning

Abstract

Summary It was hypothesized that serial learning takes place by a different process or strategy than PA learning and does not normally consist of the chaining together of S-R connections. It was also hypothesized that specific transfer of S-R pairs from PA to serial learning would be facilitated under conditions that make it relatively easy for the S to carry over his set for PA learning into the serial task as compared with conditions that make it relatively difficult to maintain the PA set. To test this hypothesis two experimental groups learned a PA list followed by a serial list with common S-R elements. A control group learned only the serial list. The two experimental groups, labeled Odd and Even, learned different pairs of items derived from the serial list. The Odd Group learned pairs 1–2, 3–4, 5–6, 7–8; the Even Group learned pairs 2–3, 4–5, 6–7, 8–9. The second task consisted of learning the serial list 1 2 3 4 5 6 7 8 9. The Odd Group, for whom the set for paired-associate responding in serial learning was facilitated by the fact that the first items in the serial list had already been learned as a paired associate, learned the serial list significantly faster than the Control Group. Though the Even Group showed some advantage over the Control Group in the first few trials of serial learning, this advantage disappeared completely by the sixth or seventh trial. This was true even of the adjacent serial items that had been previously learned as paired associates. This result was interpreted as being due to a loss of the PA set and the adoption of a strategy peculiar to serial learning. Other evidence that quite different processes are involved in serial and PA learning was the lack of a significant correlation between individual differences in the two forms of learning. It appears that a serial list is not learned as a chain of S-R connections and that what the S learns in the serial task is somehow quite different from what he learns in the PA task, even though both tasks formally have the same associative elements in common. An hypothesis was suggested to account for these findings, viz., that in PA learning the S-terms serve primarily a cue or stimulus function, while in serial learning the subject integrates a number of responses (supplied by the items in the list) and the items serve primarily as reinforcers without acquiring a cue function, except possibly by incidental learning when the list is overlearned.