Part 1
Short-Term or Primary Memory
The idea that episodic memory can be divided into two main compartments or processes has appeared in various forms in human experimental psychology over the last 150 years. The basic distinction is between memory for material presented and recalled recently, and material acquired some time ago, with the two stores or systems referred to typically as short-term memory (STM) and long-term memory (LTM). William James (1890) was one of the first writers to describe the two systems explicitly; he distinguished "primary memory" from "memory proper," where primary memory referred to information just experienced and still maintained in conscious awareness, as opposed to stored information that must be retrieved in order to be re-experienced. The notion of two different systems emerged again strongly with the advent of information-processing models of attention and perception. For example, Broadbent's (1958) model had a pre-attentive "s-system" and a limited capacity "p-system." Incoming material could be held briefly in the s-system before selection by the attentional filter mechanism, then identified and experienced consciously in the p-system, and if necessary recycled through the s-system. This last procedure enabled the participant to rehearse the material until a response was required.
Research on STM was further boosted by articles written by John Brown (1958) in England and by Lloyd and Margaret Jean Peterson (1959) in the US. Both studies showed that retention of small amounts of material fell off drastically after only a few seconds of interfering activity. For example the Petersons found that memory for three-letter stimuli fell to 10 percent after 18 seconds of counting backwards. Murdock (1961) followed these demonstrations by showing that three words were forgotten at the same rate as three letters, and therefore concluded that the appropriate unit of storage in STM is "the chunk"—any meaningfully integrated unit of information. The growing enthusiasm for a separate short-term store was dampened, however, by Melton (1963) who argued that the hallmarks of memory in general were (a) learning to increase retention and (b) interference to reduce retention. In his 1963 study Melton demonstrated that both learning and interference occurred in STM paradigms, and therefore argued that since both short-term and long-term memory obeyed the same laws there was no case for postulating two distinct storage systems.
The counterargument was produced independently by Peterson (1966) and by Waugh and Norman (1965). These authors pointed out that the term "short-term memory" had been used both to describe a postulated mechanism and more generally to refer to any situation in which material was retained for a short time. They argued that the two usages were not necessarily synonymous, and specifically that retention over short intervals could involve a contribution from LTM as well as from a separate STM. Waugh and Norman further suggested that to reduce confusion around the term "short-term memory," the short-term component should be referred to by James's label of "primary memory" and LTM be re-labeled "secondary memory." The hunt was therefore on in the late 1960s and early 1970s to show that primary memory (PM) and secondary memory (SM) did indeed have different characteristics, and these are spelled out in one of the reproduced articles (Craik, 1971). Briefly, PM has a strictly limited capacity, verbal material is held in an acoustic/articulatory code, and information is forgotten primarily by being knocked out by subsequent interfering material, although passive decay may also play some part. These characteristics contrast with SM, which has an essentially unlimited capacity, with a variety of codes depending partly on perceptual aspects of stimuli and partly on the meaningful implications of these stimuli. Several factors contribute to forgetting from SM, including proactive and retroactive interference as well as forgetting owing to retrieval failure although the information may still be potentially available.
I became interested in the PM/SM distinction in the late 1960s while serving on the faculty at Birkbeck College, University of London. One contribution (Craik, 1968) showed that when the PM and SM components in the free recall of word lists were separated, older adults retrieved fewer words than younger adults in the SM portion, but there were no age differences in the PM portion. I also showed that as list length increased, overall recall increased, but that this increase was attributable entirely to retrieval from SM; the PM component remained constant. A second contribution stemmed from an interest in the Atkinson and Shiffrin (1968) model of memory—the dominant model of the day. One of their ideas was that incoming material was held in a short-term store by virtue of a process called the rehearsal buffer operating within the structure of the store, and that information was transferred to LTM as a result of this rehearsal and as a function of the time each item was rehearsed. When participants are given a list of words to recall in any order (free recall) they typically recall the last three or four items first of all (those still in PM) and then some further words from the beginning and middle of the list. This means that the last one or two items are rehearsed least, since they are presented and then recalled in quick succession. In turn, this should mean that terminal words in a list are least well represented in LTM (or SM) despite being best retrieved in immediate recall. I demonstrated this effect by unexpectedly asking participants to recall as many words as they could from all lists in a second recall session ("final free recall") after all ten lists had been presented and recalled and, as predicted from the Atkinson and Shiffrin model, this resulted in a "negative recency effect," that is, poorest recall of the last few words in the original lists (Craik, 1970).
A second, unexpected, finding from the same study was that words recalled latest in immediate recall (and thus presumably entirely from SM) were recalled best in final recall (Craik, 1970, Fig. 3). I speculatively attributed this finding to the idea that initial retrieval with great difficulty or effort is associated with particularly good long-term retention. We followed this idea up in a later study (Gardiner, Craik, & Bleasdale, 1973) in which participants were given definitions and asked to retrieve the relevant words. We measured retrieval latency and found that longer latencies in initial recall were associated with better recall in an unexpected final recall session. However, this benefit was restricted to words that the participant reported "knowing and almost retrieving" for some seconds before actually recalling them. It therefore seems that the benefit is attributable to some degree of semantic processing of the word's features during the initial recall phase.
The third article reproduced in this section (Craik & Watkins, 1973) takes the argument a stage further. The basic demonstration is that prolonged rehearsal of words in an initial recall session does not guarantee their boosted recall in a final recall session; it depends on the qualitative nature of the rehearsal. In the Craik and Watkins study we induced participants to rehearse the last four words in half of the lists "by rote," that is simply maintain words for 20 seconds in the articulatory loop (Baddeley & Hitch, 1974) without attempting any semantic or associative processing. The result was that performance in a final free recall test (after all lists had been presented and recalled) was not enhanced in any way for the final four words in the "rehearsed" lists despite the fact that such words had been rehearsed between eight and twelve times relative to the two to six rehearsals carried out on comparable words in "non-rehearsed" lists. The conclusion was that maintenance rehearsal simply maintains items in an accessible state but does little or nothing for longer-term retention; for the latter to be enhanced the person must carry out "elaborative" or semantic rehearsal. Clearly this conclusion fits well with the levels of processing notions described in the next section. The result also meant that the Atkinson and Shiffrin (1968) model must be modified to take different qualitative types of processing into account.
My co-author Michael Watkins worked with me as a graduate student at Birkbeck College and later emigrated with his wife Olga to North America where he worked with Endel Tulving at Yale and Toronto, and was then a faculty member at Princeton and at Rice University in Houston. He is now retired and living in Europe. Mike was always tremendously creative and very involved in ideas and experiments; on one occasion I received a Christmas card from him with conventional greetings on one page and some excited notes on a proposed experiment on the reverse. Truly a dedicated man!
The results of these and other contemporary experiments left me in no doubt that the STM/LTM distinction in some form is a real one—a position I still maintain today. Most of the further work in this general area has followed the very influential lead of Baddeley and Hitch (1974) and focused on the concept of working memory (WM). In my opinion, the concepts of primary memory (PM) and WM are clearly related in that both refer to material held in conscious awareness. They differ, however, in that whereas information "in PM" is simply presented, maintained, and given back as a response in unchanged form (as in looking up a telephone number), the essence of WM is that the information in mind is processed, elaborated, or transformed in various ways. I therefore think of PM and WM as lying on a continuum depending on the amount of work that has to be done on the maintained material. Some evidence in favor of this view comes from studies of aging. There are essentially no age-related differences in PM, but age-related decrements do occur in WM paradigms to the extent that processing manipulations are required. That is, the problem for the older adult lies in the manipulations and transformations that must be carried out, not in the simple processes of maintenance (Craik & Rabinowitz, 1984).
As a final word, how should we think of PM—as a structural store, some kind of transient trace, or as a set of processes? I strongly endorse the last possibility and suggest briefly in the Craik (1971) review that PM may be equated with attention paid to one or several features of the information held in mind. These ideas of PM as a process were spelled out more fully in a further review chapter (Craik & Levy, 1976) and are very compatible with some current ideas about WM (e.g., Cowan, 1999; Shipstead, Harrison, & Engle, 2015).
References
Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. In K. W. Spence and J. T. Spence (Eds.), The psychology of learning and motivation, Vol. 2, pp. 89–195. New York: Academic Press.
Baddeley, A.D., & Hitch, G. (1974). Working memory. In G. H. Bower (Ed.), The psychology of learning and motivation, Vol. 8, pp. 47–89. New York: Academic Press.
Broadbent, D.E. (1958). Perception and communication. London: Pergamon Press.
Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology, 10, 12–21.
Cowan, N. (1999). An embedded-processes model of working memory. In A. Miyake and P. Shah (Eds.), Models of working memory: Mechanisms of active maintenance and executive control (pp. 62–101). Cambridge: Cambridge University Press.
Craik, F.I.M. (1968). Two components in free recall. Journal of Verbal Learning and Verbal Behavior, 7, 996–1004.
Craik, F.I.M. (1970). The fate of primary memory items in free recall. Journal of Verbal Learning and Verbal Behavior, 9, 143–148.
Craik, F.I.M. (1971). Primary memory. British Medical Bulletin, 27, 232–236.
Craik, F.I.M., & Levy, B.A. (1976). The concept of primary memory. In W.K. Estes (Ed.), Handbook of learning and cognitive processes, Vol. 4 (pp.133–175). Hillsdale, NJ: Erlbaum.
Craik, F.I.M., & Rabinowitz, J.C. (1984). Age differences in the acquisition and use of verbal information: A tutorial review. In H. Bouma and D.G. Bouwhuis (Eds.), Attention and performance X (pp. 471–499). New York: Psychology Press.
Craik, F.I.M., & Watkins, M.J. (1973). The role of rehearsal in short-term memory. Jour...