Recall is a type of memory (definition 1), brain process (definition 2), and test situation (definition 3). It is important to distinguish between the three. Recall as a memory faculty is the phenomenon that most people have in mind when they refer to 'remembrance' or 'recollection'. It is 'memory par excellence' (Bergson 1908). Some authors, especially in the older literature, equate recall with "retrieval. This should be avoided, because retrieval is a universal process, which must occur to actualize any type of learned information, even the simplest ones such as "habituation and "sensitiza-tion, whereas recall corresponds to retrieval only under the conditions specified in definition 1. Recall as a process is inferred to underlie the recall faculty. Recall as a test involves the generation by the "subject of the internal representation of a fact or event according to instructions, in the absence of the corresponding online sensory information, for example, reproduce in mind a remote autobiographical episode while sitting in the laboratory. But recall could also be involved in tasks that are not intentionally meant to tap into it. For example, in a "recognition test, a subject instructed to identify an on-line sensory "stimulus probably uses recall of off-line information to evaluate the meaning of this target.
A few additional words on variants of recall. When recall is tested in the absence of intentional "cues, it is called 'free recall'. When cues are provided, it is called 'cued recall' or 'prompted recall'. Each of these terms can be used, as far as it is remembered that there is probably no recall without cues—only that in prompted recall these cues are provided by the experimenter, whereas in free recall they are generated by the subject. When recall is initiated by instructions or by focused intention, it is focused or intentional recall, whereas if it is the spin-off of a stream of "associations, it is incidental recall. When the recall yields what is judged to be the complete target, it is complete or total recall. But in many cases recall is only partial. Two examples of partial recall are provided by the 'feeling of knowing' (FOK) and the 'tip of the tongue' (TOT) phenomena. In FOK the subject judges that the target is in memory despite a failure to retrieve it at present time (Hart 1965); in TOT, the subject judges as if verbal information resides on the tip of the tongue ("metaphor) despite the failure to express it (Brown and McNeill 1966; "metamemory).1 An attempt to recall an item in memory may at first prove futile but later succeed without further learning. This is called 'reminiscence', i.e. the recall of previously unrecalled items (Ballard 1913; Payne 1987). Under certain conditions, the overall recall performance of the subject on a certain type of test and material may improve with repeated recall trials; this is termed 'hyperamnesia' (Erdelyi and Becker 1974; Payne 1987).
Starting from the early days of experimental psychology (e.g. MacDougall 1904; Hollingworth 1913), many attempts have been made to explain how recall works, and what distinguishes it from recognition. It should be said at the outset that this focus on the distinction of recall from recognition may be a bit misleading, because it suggests that the two phenomena or processes are decisively different, which may not be true. In a nutshell, there are two main types of recall 'theories', or "models. One type refers to recall and recognition as lying on a continuum of retrieval (Tulving 1976). The main difference between recall and recognition, according to this view, is the nature and availability of retrieval cues: supplied in recognition, self-generated in recall (see above). The other type of models distinguishes basic differences between recognition and recall. These models propose that recall operates in two *phases, or stages (Bahrick 1970; Kintsch 1970). The first phase is the search, generation, or retrieval phase. The second phase is the identification, or decision, or recognition phase, once the target has been retrieved.2
Two points about the two-stage models of recall are noteworthy. One, recognition also involves retrieval. Therefore, distinguishing one phase of recall as the retrieval phase, as opposed to recognition, disregards the universality of retrieval as the process necessary to actualize stored information. Second, it is tempting to assume that if recognition is a subprocess in recall, recognition will be easier than recall. Intuitively it makes sense: how easy it is to recognize a face, how difficult is it to recall the name that belongs to that face (McWeeny et al. 1987). Yet it can be shown that under certain conditions items are recalled but not recognized (Bahrick and Bahrick 1964; Tulving and Thomson 1973). For example, consider the following experiment (Tulving and Thomson 1973): subjects were presented with an input list of pairs of words, one a target word, such as baby, the other a weak semantic associate that was intended to be used as a "cue for the target word, e.g. grasp. Afterwards, the subjects were presented with strong semantic associates of the target words, e.g. infant, and asked to generate a list of associated words. The subjects were then asked to mark all the words in the generated list that they recognized as having occurred in the input list. Finally, the subjects were presented with the original input list cues and asked to recall the target words. Under these conditions, the prompted recall was superior to the recognition of words from the input list. This seems incompatible with recall being a generation-plus-recognition process. The recognition failure was explained by the authors by invoking the 'encoding-specificity principle'. This principle ("acquisition, "retrieval) states that memory performance is best when the cues present at retrieval match those present in acquisition. The encoding of the target words in the input list was in the presence of the input list cues, and influenced by the subjects' expectation that they will be tested with these cues. In the later part of the experiment, which involved the recognition task, the retrieval attempt was made using different cues and in a different "context. The encoding specificity principle predicts that this will hamper the retrieval, hence recognition, of the target words. Another interpretation of the data is that the recognition failure resulted from the formation of two different traces in the different parts of the experiment, rather than from differences in the retrieval efficacy of the same trace (Baddeley 1982). The two different traces could have resulted from the interaction of each target word with two different semantic contexts, shaped by the weak and the strong semantic associates, respectively. If correct, this means that the 'item to be recognized' was not actually identical to the 'item to be recalled', rendering the comparison of the efficacy of recognition and recall in this experiment questionable.
Can animals recall? We find it easy to understand recall because we frequently recall and because we can ask other individuals to declare their own recall experience. But how do we know whether a dog is musing spontaneously, or at least is cued to muse, about things past? In referring to recall, there is an implicit assumption that it does not refer to very short-lived memory, say of a few seconds only. If this assumption is abandoned, than trace conditioning ("classical conditioning) and "working memory involve cued recall functions, and clearly, they can be performed by nonhuman species. Furthermore, animals do seem capable of high-order stimulus-stimulus associations (Dickinson 1980; Mackintosh 1983), so why not consider these as resulting in stimulus-independent retrieval, hence recall? Nonhuman species might also be capable of "conscious awareness of the recall act (Clark and Squire 1998, "declarative memory). A reasonable conclusion is that recall is a spectrum of functions, differing in their dependence on available cues, and their explicitness, and other species are capable of rudimentary forms of recall (see also discussion in "episodic memory).
Which brain circuits subserve recall? The analysis of defective recall in "amnesics indicates that, similarly to recognition, recall depends during the acquisition and "consolidation phases of the memory on the integrity of the "hippocampal formation and other mediotemporal structures (Haist et al. 1992). "Functional neuroimaging of normal subjects confirms that the hippocampus is indeed involved in explicit recall (e.g. Maguire et al. 1997). With time, however, the memory to be recalled becomes practically independent of the hippocampal formation, and remains in the "cerebral cortex (e.g. Teng and Squire 1999). There is evidence that recalling a target in its absence activates areas in the sensory cortex that were originally involved in "perception of the target (e.g. Kosslyn et al. 1995; Zatorre et al. 1996; Wheeler et al. 2000). And finally, to recall a target, monitor the operation and verify it, the prefrontal cortex is needed (Jetter et al. 1986; Buckner and Koustaal 1998; Fletcher et al. 1998; Tomita et al. 1999; Maril et al. 2001). This brings us back to the issue as to whether nonhuman species can recall: those species that have a reasonably developed prefrontal cortex, such as other primates, can probably recall, but less efficiently than Homo sapiens, who happens to have a more sophisticated prefrontal cortex.
Selected associations: Amnesia, Confabulation, Recognition, Retrieval, State-dependent learning
1FOK and TOT apply to "recognition as well. 2The notion of recall involves search and identification can be traced to much earlier literature: it is hinted already in Augustine (400; "classic).
Was this article helpful?