Koriat, A. (2000). Control processes in remembering. In E. Tulving, & F.I.M. Craik (Eds.), The Oxford handbook of memory (pp. 333-346). New York: Oxford University Press

Control Processes in Remembering

ASHER KORIAT

This chapter examines some of the processes ognize it among distractors. The discrepancy that take place in attempting to probe memory between the availability of information and its for a needed piece of information. In present- accessibility to consciousness testifies for the day conceptualization memory processes are critical role of retrieval processes—the ability divided into three phases: encoding, storage, to conjure up stored information. and retrieval. Strangely enough, the distinc- Two observations illustrate this discrep- tion between storage and retrieval, so basic in ancy. Tulving (1967) had subjects study a list current memory theorizing, is relatively new of words, and their recall was tested three (see Baddeley, 1997; Roediger & Guynn, 1996). times in succession. Only about 50% of the In classical, S-R learning theory, based primar- words were recalled on all three tests. For ex- ily on animal research, storage, and retrieval ample, a subject might recall words on the sec- were lumped together, and even the distinc- ond test that he failed to recall on the first test. tion between encoding and storage was not ex- A second observation comes from a study plicitly made. It was the extension of S-R prin- (Williams & Hollan, 1981) in which subjects ciples to human verbal behavior that has spent one hour every day trying to recall the brought to the fore the importance of distin- names of people with whom they had gradua- guishing among encoding, storage, and re- ted from high school 4-19 years earlier. Sub- trieval. jects were found to recall new names after as much as 10 hours in the experiment, spread over two weeks! Clearly these names must Availability and Accessibility have been available even in the first hour of the experiment. The critical role of memory retrieval becomes These observations raise several questions: apparent when we realize that the amount of What prevented all the items ultimately re- information stored in our memory exceeds by called to surface right from the beginning of far the amount of information that we can re- testing? What allowed them to become acces- trieve from it. In terms of the terminology in- sible later on? And more generally, what is troduced by Tulving and Pearlstone (1966), the process by which people search for and re- much more information is available in mem- collect stored information from long-term ory than is accessible at any moment. Thus, memory? although we may momentarily fail to retrieve In attempting to address these questions, the name of an acquaintance, we may still be we should note that although most of our able to recall it on some later occasion or rec- knowledge about memory processes derives

333 334 MEMORY IN LIFE

from the use of relatively simple laboratory way to the sought-after target. At other times tasks, control processes in remembering are they may be recognized as misleading "inter- particularly transparent in the retrieval of in- lopers" and effort will be exerted to oppose formation under naturalistic conditions, par- their interfering influence (Jones, 1989). ticularly when retrieval is laborious and pro- This interplay between top-down and bot- longed. Therefore, this chapter will emphasize tom-up processes derives from the fact that studies concerned with this type of retrieval there is hardly ever a cognitive algorithm that despite the fact they do not always attain the can safely lead from a retrieval description methodological rigor characteristic of labora- (e.g., "what is the name of that street") to its tory experimentation. resolution—the memory target (e.g., "Hilda"). Therefore, while retrieval may begin with a controlled, goal-oriented search that takes off Retrieval as Problem Solving from the retrieval description, it must also be receptive to activations and associations that Even a cursory examination of everyday epi- emerge during the search. In his analysis of sodes of recollection suggests that there is problem solving, Duncker (1945) discussed much more to remembering than simply fetch- this interplay in terms of the idea that the so- ing out a solicited piece of information from lution to a problem involves the matching of some storage place. Take, for example, the fol- suggestions from above with suggestions com- lowing episode recounted by Nickerson (1981): ing from below. In fact, his definition of a problem aptly describes the situation in which Consider the following effort to recall the retrieval is called for: "A problem arises when name of a street that is located a few blocks a living creature has a goal but does not know from where I live. The name would not how that goal is to be reached" (p. 1). come to mind, but I did know it to be the The problem-solving character of remem- name of a friend. The name Elliott sug- bering can be seen in many everyday memory gested itself, but did not seem to be correct. tasks that tax memory, such as answering I thought the name I was looking for was a questions about course material or retrieving first name, and although Elliott can be ei- episodes from the distant past (see Burgess & ther a first or last name, it is in fact the last Shallice, 1996; Reiser, Black, & Abelson, name of a friend of mine. I also was fairly 1985). Williams and Hollan (1981), in fact, sure the sought-for name was the first name proposed a jigsaw puzzle metaphor of remem- of a female, and the Elliot in mind was a bering, in which one begins with starter pieces male . . . As the search continued, the name (specified by the memory query), and by fo- Cellier surfaced, the last name of a close cusing on some section of the puzzle, searches friend of Elliott's, who was also a friend of for a piece that may fit, and then verifies that mine. Next came Emil, the first name of it indeed does.The problem-solving character Cellier; then Hilda, wife of Emil. Hilda was of remembering makes extensive use of work- immediately recognized as the name of the ing memory (see e.g., Conway, 1992) in con- street, (p. 79) trolling the operation of several processes, coordinating between them, holding the out- Similar experiences of wandering through come in a temporary store, and orchestrating the paths of one's memory in search for some the entire process. longed-for record are not uncommon, and Different components of the retrieval pro- many have been reported in some detail in the cess are suggested by experimental studies as literature. These reports suggest that in real- well as naturalistic studies in which subjects life situations, retrieval often involves a com- were asked to think aloud in the course of re- plex interplay between two types of processes membering. They are discussed in what fol- (see Jacoby, 1991; Moscovitch, 1989), a con- lows. trolled, strategic process that guides retrieval, coordinating between different operations di- rected toward the recovery of the elusive The Importance of the memory target, and the automatic, involuntary Instigating Conditions emergence into consciousness of ideas and as- sociations throughout the search. Sometimes In everyday life we are generally unaware of the controlled process will seize onto these using our memory. When walking in a famil- ideas and use them as stepping stones on the iar neighborhood, we must be constantly con- CONTROL PROCESSES IN REMEMBERING 335 suiting our memory of the geographic layout tating retrieval. In fact, in naturalistic situa- to find our way, but we are hardly aware of tions, the process of remembering has more in doing so. We may become aware of using our common with cued-recall than with free-recall memory when we lose our orientation, or testing. Such is not the case in many institu- when someone we meet asks us for directions. tionalized memory testing situations (e.g., It is in those cases that we deliberately probe achievement tests, memory experiments), as our memory for the needed information. when a person is required to answer a variety It is important to distinguish between two of general-information questions (e.g., Kel- types of situations that motivate remembering. ley & Lindsay, 1993; Koriat, 1995) that have In the first, retrieval is prompted by a specific nothing to do with the immediate goals and question posed by an external agent". Typical circumstances. This contrast helps bring to the examples are a course examination, a job in- fore two related principles of memory. The terview, or a police interrogation. Importantly, first is that retrieval depends critically on the this type of situation is characteristic of practi- presence of retrieval cues, and the second is cally all memory studies because in these that retrieval success varies depending on the studies the memory questions are presented match between the study and test situations. and formulated by the experimenter. When re- membering occurs in response to externally presented specific questions, it tends to be de- liberate and relatively focused and restricted The Contribution (see Bekerian & Dritschel, 1992). Retrieval Cues The second type, which is more typical of naturalistic situations, is when memory que- Tulving (1983) has promoted the notion that ries are generated spontaneously by the per- memory is a joint product of the stored mem- son himself, or triggered by accidental en- ory traces and the cues that are present when counters or task demands. A simple example retrieval is called for. This view implies that was given by Mandler (1980): given the same conditions of study, retrieval success can vary greatly depending on the Consider seeing a man on a bus whom you conditions of testing. For example, Tulving are sure that you have seen before; you and Pearlstone (1966) observed that memory "know" him in that sense. Such a recogni- for a list of words was considerably better un- tion is usually followed by a search process der cued-recall testing, where the cues were asking, in effect, Where could I know him the semantic categories to which the words from? Who is he? The search process gener- belonged, than under free recall testing. This ates likely contexts (Do I know him from result supports the importance of the distinc- work, is he a movie star, a TV commenta- tion discussed earlier between availability and tor, the milkman?) Eventually, the search accessibility, and indicates that information may end with the insight, That's the that is not immediately accessible can become butcher from the supermarket! (pp. 252— accessible when proper cues are present. 253) In the experiment just described, the cues were presented experimentally. In real-life sit- This type of situation has hardly any analogue uations the conditions that instigate retrieval in memory experiments. When retrieval is ini- normally provide many useful cues. In exter- tiated by the person himself, often in response nally posed queries, the leading cues can be to task demands or social interaction, the "re- found in the query itself. Even when these trieval description" (Norman & Bobrow, 1979) cues are not sufficient to directly trigger the tends to be loose and ill defined. In other target item, they help delimit the memory re- cases still, retrieval may be spontaneous, trig- gions in which that item is likely to be found. gered automatically by some external cues Retrieval cues are particularly critical for (see Salaman, 1982). Unfortunately, because of prospective memory—that is, memory to per- the nature of experimental investigation, we form intended acts in the future, such as know very little about self-generated queries showing up for an appointment or taking a and about automatic retrieval. medicine (Brandimonte & Ellis, 1996). What is Because in everyday life we typically try to most important in prospective memory is not retrieve information from memory when we the retrieval of the specific content of the to- need it, the immediate, instigating conditions be-performed act, but the retrieval of the in- may play a critical role in guiding and facili- tention to perform it at the appropriate time (Einstein & McDaniel, 1996). To do so, one 336 MEMORY IN LIFE must rely on cues that are available in the en- matic process in which the cue directly vironment to trigger the intention, devise facilitates the emergence into memory of the one's own aids ("the knot in the handker- solicited target (see Neely, 1976). The con- chief," a timer), or rely on internal cues. Bur- trolled exploitation of cues is particularly gess and Shallice (1996) note that the context transparent when retrieval is difficult or pro- setting that occurs when a person "sets up" an longed. As we shall see below, much of the intention to perform a future action (e.g., to process of remembering appears then to in- mail a letter) involves a recollection of the volve the deliberate use of cues to probe one's context of retrieval (e.g., imagining the jour- memory for additional cues that may bring ney home, the mailbox), so that the potential one closer to the desired target information. cues for retrieving the intention are stored The starting point, of course, is the cues pres- with the intention itself. ent in the situation that drives retrieval. These Often a failure to perform the planned act can be exploited to delimit a context for fur- stems from the failure to identify the "cue" as ther search (Koriat & Lieblich, 1974; Nor- a cue for retrieving the intention. Sometimes man & Bobrow, 1979). external cues operate retrospectively, remind- In parallel, cues may aid retrieval through ing one of a scheduled intention that was automatic activations (see Collins & Loftus, missed (e.g., a smell of a burnt cake). Note that 1975; Nelson et al, 1998) emanating from the some of the cues, like the knot in the handker- cognitive context of retrieval, or from the in- chief, carry little information about the con- formation already recovered. We may try in tent of the act to be performed. Their function vain to recall the name of a person, but the is to induce the person to retrieve the inten- name may suddenly pop up the moment we tion to perform some act. see that person. In fact, this is the way in Cues differ considerably in their effective- which many "reminders" operate in everyday ness in prompting retrieval. In the case of pro- life: Passing by a drug store reminds us of a spective memory, the most effective cues are prescription that we have to take. Schank those that are more distinctive and less famil- (1982), who considered reminding to be a cru- iar (McDaniel & Einstein, 1993). In retrospec- cial aspect of human memory, offered an ex- tive memory, in contrast, the effective cues, of tensive analysis of the type of reminding pro- course, are those that relate to the content of cesses that occur in everyday life. the solicited item. Research examining the ef- We know a great deal about automatic acti- fectiveness of extra-list words in prompting vations from experiments on verbal priming the recall of studied words (Nelson, McKin- (e.g., Collins & Loftus, 1975; Neely & Keefe, ney, Gee, & Janczura, 1998) indicates that re- 1989) and implicit memory (see Roediger & trieval success varies considerably with a McDermott, 1994). For example, subjects re- large number of associative properties of the trieve faster the answer to a general-informa- cue and of the target. For example, the larger tion question if that answer has been pre- the number of words that a cue word elicits in sented earlier in the context of an unrelated word association norms, the lower its effec- task (Kelley & Lindsay, 1993). In a word-asso- tiveness in facilitating the retrieval of a stud- ciation task subjects are more likely to re- ied word. spond with a word recently studied (e.g., The most effective cues for retrieving an Koriat & Feuerstein, 1976). While such activa- event are personal cues associated with the tions generally aid retrieval, they may give encoding of that event, because these cues get rise to inadvertent plagiarism (e.g., Marsh, to be integrated into the memory trace of the Landau, & Hicks, 1997) or produce false mem- event (e.g., Mantyla, 1986). Methods of mem- ories (e.g., Roediger & McDermott, 1995; see ory improvement make use of this principle, Koriat, Goldsmith, & Pansky, in press). by having people utilize some system of cues Automatic activations may bring to mind during encoding, which they can latter use to entire scenes or events, as when a certain aid retrieval. smell or music conjure up emotionally laden old memories (see Conway, 1992; Salaman, 1982). Such activations might also underlie How Do Cues Assist Retrieval? the recurrence of "flashbacks" memories in people suffering from posttraumatic stress, Cues may aid retrieval either through a con- when memories of the traumatic event intrude trolled process in which the person deliber- into the person's consciousness against his or ately makes use of them, or through an auto- her will. CONTROL PROCESSES IN REMEMBERING 337

The Encoding-Specificity ple, Godden and Baddeley (1975) studied the memory of divers when they learned a list of Principle words either on land or underwater and were later tested in the same or in the opposite en- The importance of the instigating conditions vironment. Recall was better when learning for retrieval is also stressed by the encoding- and recall took place in the same environment specificity principle (Tulving & Thomson, than in different environments. Subjects have 1973), which states that a cue presented dur- also been found to recall a larger number, of ing testing will be effective in aiding retrieval words when they were tested in the same to the extent that it has been encoded together room in which they had studied than when with the solicited memory target at study. they were tested in a different room (Smith, Thus, a critical condition for effective retrieval Glenberg, & Bjork, 1978). It would seem that is the extent to which the processing that oc- context-dependent effects are more likely to curs during retrieval reinstates the processing be found when the environmental contexts that took place during encoding. differ substantially, and when subjects delib- Tulving and Thomson provided some coun- erately associate the studied material with fea- terintuitive results supporting the encoding tures of the study environment. These effects specificity principle. For example, the word are obtained only for recall, not for recogni- hot is more strongly associated with cold than tion (Eicii, 1965), suggesting uiai oontexLual the word ground, and indeed, in a free-recall reinstatement specifically facilitates informa- task, hot is more effective than ground for tion retrieval. prompting the recall of cold. Nevertheless A study by Smith (1979) suggests that men- ground will actually act as a better cue for re- tal reinstatement of the learning environment calling cold if cold has been originally en- may be almost as beneficial for retrieval as ac- coded in the context of ground. What is more, tual, physical reinstatement. This idea has when cold is encoded in the context of been incorporated in the cognitive interview ground, subjects who are successful in recall- (Fisher & Geiselman, 1992), designed to en- ing cold in response to hot sometimes fail to hance witness recollection: prior to answering recognize it as one that has appeared in the specific questions about a past event, wit- study list! This is like the failure to recognize nesses are instructed to mentally recreate the an acquaintance when we accidentally run contextual state that existed at the time of the into him in a very different context from that original event. in which we are accustomed to see him. Evidence for state dependency comes from Similar results consistent with the encod- findings indicating that memory performance ing specificity hypothesis have been reported is best when the same internal state is main- by Morris, Bransford, and Franks (1977). Al- tained across the learning and testing phases. though memory for words is superior when What subjects learned when drunk they re- during encoding subjects attend to the mean- membered better when drunk than when so- ing of the words rather than to their sound, the ber, and vice versa (Goodwin, Powell, Bremer, reverse pattern was found when during re- Hoine, & Stern, 1969). A similar pattern was trieval subjects were induced to attend to the observed in a study on the effects of marijuana phonemic properties of the words. on free recall (Eich, Weingartner, Stillman, & Gillin, 1975). Evidence for state-dependent re- trieval is more clearly observed for free recall Reinstating the Conditions of than for recognition or cued recall (see Eich, Learning during Testing 1980). Several studies suggest that retrieval is also mood dependent: memory performance is Retrieval is also affected by the extent to better when people's moods during study and which the testing conditions reinstate the test match than when they do not (Eich & Met- overall conditions of study with regard to the calfe, 1989). external stimulus conditions (context), the in- We now proceed to examine some of the ternal state of the person (state), or the per- processes that occur during retrieval. As son's emotional feelings (mood). noted, these processes are not easy to trace ex- There is evidence that retrieval is context cept when retrieval is effortful and extended dependent—that is, that memory is best when over some period of time. Before focusing on testing occurs in the same physical environ- effortful retrieval, however, we shall discuss ment in which learning took place. For exam- briefly effortless retrieval. 338 MEMORY IN LIFE

Effortless Retrieval Reder (1987) proposed that the familiarity of a question also affects the general strategy Not every memory search is laborious; in of answering that question. When the familiar- many cases the solicited information will ity of a question is low, subjects would tend come to mind immediately (see Nickerson, to resort to a plausibility strategy, inferring the 1981). The retrieval of well-practiced, well-re- answer from a variety of cues, rather than to a hearsed information often has the character of direct retrieval strategy. She noted, however, habits or stimulus-response associations. that even when familiarity is high subjects What makes retrieval automatic? First, may still respond on the basis of the gist of there is the case of incidental remembering. the question. For example, when asked "How Automatic activations may increase the acces- many animals of each kind did Moses take on sibility of a memory entry to the extent that it the Ark?" many subjects reply "two," even can emerge into consciousness spontaneously. though they know that it was Noah who did Such activations have been studied exten- so (Erikson & Mattson, 1981). sively in connection with implicit memory. Second, as far as intentional retrieval is con- cerned, it would seem that practice retrieving Specifying the Initial Context an item from memory is what makes retrieval of Search of that item more automatic (see Bjork & Bjork, 1992). Thus, although we know the numbers When preliminary FOK is high, the next step from 1 to 10 "by heart," their retrieval is more is to determine the initial context in which the effortful when we have to list them from 10 to search has to be conducted. This decision is 1. Counting forward is a much more practiced suggested by the verbal protocols of subjects habit than counting backwards. produced in the course of remembering (e.g., Let us now focus on effortful retrieval. Burgess & Shallice, 1996; Norman & Bobrow, When memory retrieval is effortful and pro- 1979; Reiser et al., 1985; Williams & Hollan, longed, complex regulatory processes of moni- 1981). This is analogous to that of choosing toring and control operate in guiding the under what heading to look for a certain topic search. in a book, or under what directory a computer file is likely to be stored. For example, in at- tempting to answer questions about course material, students often begin by deliberating Preliminary Monitoring and whether the question is "from the textbook" Choice of Mode of Attack or "from the lecture." Or they may attempt to specify more precisely the relevant chapter in When we are presented with a memory ques- the textbook. When the question requires the tion we do not immediately proceed to answer retrieval of some autobiographical detail, the it, unless the answer pops instantaneously person may start by recovering a scene or an into our head. Rather, in many cases a prelimi- episode in which that detail is likely to be nary monitoring stage exists in which we found. Duncker (1945) referred to this process make a rough assessment about the availabil- as one in which "4:he jacket is sewn to the but- ity of the answer in memory and the effort ton" (p. 83). needed to access it. Memory questions differ considerably in The initial feeling of knowing (FOK) associ- the extent to which they delimit, explicitly or ated with a question is apparently based on a implicitly, an effective search domain. Con- process that monitors the overall familiarity of sider the type of questions that specify a par- the question (Nhouyvanisvong & Reder, 1998; ticular memory entry to be retrieved—for Schwartz & Metcalfe, 1992) and the extent to example, a word or a name. This type of ques- which it brings some fragmentary clues to tions, designated "memory pointers" by Koriat mind (Koriat, 1993, 1995). If the question and Lieblich (1977), have been extensively leaves us completely blank, chances are that used in studies of the tip-of-the-tongue (TOT) we would not initiate a deliberate search for and FOK states in which the person initially the answer. Note, however, that a controlled fails to retrieve the solicited target from mem- decision to interrupt search apparently does ory (Brown & McNeill, 1966; Nelson & Narens, not prevent automatic activations that may ul- 1990). Such pointers differ widely in the ex- timately lead to the solicited target (see Ko- tent to which they offer a plan for search. For riat & Lieblich, 1977). example, a question such as "what biblical CONTROL PROCESSES IN REMEMBERING 339 character allegedly lived 969 years?" is likely considered partial retrieval to be the central to activate a more useful search domain ("bib- principle that constraints and determines the lical characters") than the comparable ques- shape of the reconstructive retrieval process. tion "Which person allegedly lived 969 Some of the characteristics of partial re- years?" The delimitation of a search domain is trieval are revealed in studies of the TOT state useful not only for memory but also for meta- (see Brown, 1991). People in a TOT state are rnemory: it contributes to the accuracy of the able to provide correct guesses about the num- initial FOK associated with the memory ber of syllables in the word they are grappling pointer (Koriat & Lieblich, 1977). The most ef- for, some of its letters, and the location of pri- fective pointers are those that cast the specifi- mary stress (Brown & McNeill, 1966). Subjects cation of the solicited target in a format that can also access semantic and associative as- simultaneously constitutes an effective plan pects of the elusive word, such as whether it for search. There are different strategies for has a good or bad connotation (Koriat, 1993; specifying an initial context for search. For ex- Schacter & Worling, 1985). ample, one of the strategies used for retrieval The partial information initially retrieved of the names of old classmates (Williams & provides a lead to a deliberate search, but it Santos-Williams, 1980) was the location strat- also affects the search through its automatic, egy: the subject searches a mental map where implicit influence. This is suggested by find- target items are likely to be recalled. Reiser, ings such as that of Durso and Shore (1991): Black, and Kalamarides (1986) also identified subjects can distinguish between correct and several strategies in the retrieval of specific incorrect uses of rare English words even autobiographical events, which they classified when they classify them as nonwords, sug- into those involving finding a context, and gesting that available partial clues can implic- those involving searching within a context. itly affect choice of response. Also, in Koriat's Impairments in the ability to generate a fo- study (1993), when subjects responded with cused contextual description of the sought an incorrect word in cued recall, that word after information has been seen to underlie tended to have the same connotative meaning some of the memory errors encountered as the correct word that they failed to retrieve. among brain damaged patients (Schacter, Nor- There are indications that during forgetting man, & Koutstaal, 1998). the more specific aspects of the encoded infor- mation are lost before the general attributes (Ceraso, 1987). This implies that generic infor- mation is accessible long after the more de- Access to Partial Information tailed, item-specific information has ceased to and Zooming in on a Memory be accessible. Thus, studies of the long-term Target retention of course material suggest that mem- ory for higher level, superordinate information Many observations suggest that retrieval is not declines less rapidly than memory for specific an all-or-none matter. When we fail to retrieve details (e.g., Cohen, Stanhope, & Conway, a word or a name, we may still be able to ac- 1992). Similarly, categorical or gist informa- cess some of its fragments or attributes. These tion is lost less rapidly than item information can sometimes provide the initial lead for re- (e.g., Dorfman & Mandler, 1994; see also trieval. The utilization of such partial clues is Brainerd & Reyna, 1993): a person might re- nicely illustrated by the example presented at member that there was a bird on the list with- the beginning of this chapter, of searching for out recalling which bird it was. Such superor- the name of a street. The partial clues avail- dinate information may help define an initial able to Nickerson appeared to shape the entire domain for search. remembering process. The remembering pro- The TOT state discloses some further fea- cess sometimes looks as if the rememberer tures of retrieval. It has been proposed that a grasps the thread provided by the initial clues cursory analysis of a memory pointer activates and follows their course as they gradually un- a relatively broad region of memory that in- fold. cludes the target proper but also other entries Williams and Hollan (1981), discussing the that satisfy the retrieval description only processes involved in recalling the names of grossly (Koriat & Lieblich, 1977). The activa- classmates, argued that a great deal of the pro- tions emanating from the neighboring memory cess can be seen as a reconstruction from a va- entries exert two conflicting effects: they inter- riety of bits and pieces of information. They fere with accessing the correct target but at the 340 MEMORY IN LIFE

same time enhance the subjective feeling that ample, Williams and Hollan (1981) noted that the target is about to emerge into conscious- in attempting to retrieve the names of high ness (Koriat, 1998; Schwartz & Smith, 1997). school classmates, subjects produced an enor- Several researchers have proposed that the mous amount of information that was inciden- difficult retrieval that is characteristic of the tal to the task of recalling the names, includ- TOT state results precisely from the interfer- ing details about the school, about where ing effect of neighboring targets, and that these people lived, and so forth. Examination of this compelling but wrong candidates must be first information suggests that its main function suppressed before the correct target can be re- was to probe one's memory for additional trieved (Jones, 1989). clues that can better specify a new context for Indeed, there is evidence suggesting that search. They proposed that remembering con- when a target item is retrieved from memory, sists of a series of "kernel retrieval" processes, neighboring targets are concurrently inhibited each including three stages: a context is re- (see Dagenbach & Carr, 1994). Dagenbach, trieved (e.g., the volley tennis group), a search Carr, and Barnhardt (1990) found that the fail- is conducted within that context, and the in- ure to retrieve the meaning of a word results formation recovered is verified. in inhibitory priming of semantically related Reiser and his associates (Reiser et al., words. Anderson, Bjork, and Bjork (1994) ob- 1985, 1986), who studied recall of autobio- served that practice retrieving a target item graphical episodes, also emphasized that one from memory renders related items less acces- memory retrieval can be undertaken in order sible. The more likely were these items to in- to provide cues for a subsequent retrieval. Ac- terfere with the retrieval of the target item, the cording to their context-plus-index model, more they suffered from practice retrieving it. specific personal episodes are recalled by first In sum, when people fail to retrieve a mem- recovering the general context in which they ory target, they may access partial clues about were likely to have been encoded, and then it, and these can help in guiding the retrieval specifying the features that uniquely distin- process. Because retrieval of the general attri- guish these experiences from others in that butes of items precedes the recovery of more context. They proposed that scripts (e.g., specific features, retrieval sometimes looks "eating in restaurants," see Schank, 1982) typ- like an attempt to close in on the target ically serve as convenient retrieval contexts. through a progressive narrowing of its descrip- Burgess and Shallice (1996), too, noted that tion (Kolodner, 1983; Koriat & Lieblich, 1974). subjects did not always retrieve the target During this process competing memory candi- memory record directly, but sometimes recov- dates are suppressed to allow zooming in on ered a useful cue first. Thus, it was not un- the target. common for subjects to answer the question "what was the weather like yesterday morn- ing" by crying to remember first what they Probing One's Memory were wearing. during Retrieval Similar processes seem to take place in re- trieving information from semantic memory. How is partial information utilized in the One of the best-studied tasks in memory re- course of remembering? Several observations search is that of retrieving the members of nat- highlight the importance of self-cueing during ural categories such as vegetables, furniture, retrieval—that is, of cognitive operations and the like (Raaijmakers & Shiffrin, 1981). A whose immediate aim is the recovery of fur- study by Walker and Kintsch (1985) suggests ther cues that can lead to more refined cues, that even in this task retrieval relies on the re- and ultimately to the target itself. covery of contexts in which a search is con- How are additional clues recovered? Sev- ducted. Verbal protocols suggested a series of eral studies concur in identifying a recursive two-stage cycles: generating a context in pattern that occurs in the course of arduous which category members are likely to be remembering: a memory environment is speci- found, and then using that context as a re- fied in which a search is to be conducted, that trieval cue to produce the category members environment is searched for additional clues, themselves. Importantly, most of the contexts and the information retrieved is evaluated. generated were episodic rather than abstract- This cycle is repeated, gradually refining the semantic (e.g., in searching for automobiles, description of the information to be searched, one may picture a parking lot, the cars in front until the search closes in on the target. For ex- of the dorm, etc.). CONTROL PROCESSES IN REMEMBERING 341

The Strategic Regulation role of "suggestions from below" has also been of Memory Retrieval acknowledged. Often spurious activations lead the search astray. Williams and Santos- Williams (1980) noted that subjects sometimes In the previous sections we discussed some of abandon one strategy in response to the re- the recurrent processes in prolonged retrieval. trieval of information that appears to be par- Examination of the overall regulation of re- ticularly useful in the context of a different membering, however, brings to the fore a vari- strategy. Furthermore, verbal protocols dis- ety of operations that are not specific to mem- close moments in which the person seems to ory, but fall within the domain of higher deliberately relinquish strategic control alto- order, executive-supervisory functions. These gether, adopting a passive-receptive attitude. operations are involved in the overall regula- Nickerson (1981) noted that in retrieving tion of the remembering process, and are so words from lists, subjects often begin with a intertwined with lower level memory pro- passive attitude, and then switch to an active, cesses that it is difficult to understand remem- systematic search when the passive approach bering without considering their indispens- no longer yields a satisfactory return (see also able role. Unfortunately, little systematic work Walker & Kintsch, 1985). Koriat and Melkman exists regarding the overall regulation of the (1987) observed a similar pattern, but also remembering process. An act of remembering often has the char- diverted, the retrieval of words from a list be- acter of a goal-oriented process that is con- comes less controlled, moving along associa- trolled and guided by an overriding program tive links between the words rather than along (see the example from Nickerson, cited ear- conceptual-logical relations. lier). This supervisory program is responsible for choosing a starting point, recruiting a strat- egy, monitoring its execution, and changing it when it proves unsuccessful. Thus, remember- The Strategic Regulation of ing may involve a complex set of problem- Memory Reporting solving routines that are interlaced with processes concerned with memory retrieval We turn now to the final stages of the process, proper (Burgess & Shallice, 1996). those involved either in selecting an answer Discussing the global structure of a remem- and providing it, or in reporting "I don't bering act, Norman and Bobrow (1979; see know." also Morton, Hammersley, & Bekerian, 1985) Consider first cases in which the person identified three general stages: specification of fails to provide any answer. A fast "don't the information needed (and hence of the veri- know" response may be issued based on the fication criteria), a matching process in which assessment that the needed information is un- memory records are accessed and compared available in memory. Examination of the la- with the target description, and evaluation of tency of "don't know" responses suggests that suitability of the recovered records (see also such responses are not simply a result of scan- Conway, 1992). Each of these stages calls for a ning one's memory and failing to find the ap- variety of monitoring and control operations. propriate target, but actually depend on rather Burgess and Shallice (1996) reached similar complex processes (see Glucksberg & McClos- conclusions regarding the broad structure of key, 1981; Klin, Guzman, & Levine, 1997). the prototypical retrieval process. However, Even after initiating a memory search, a their analysis suggested a more complex or- person may abort the process if the search fails ganization in which layers of control lie be- to produce the solicited target. It should be tween general problem solving and specific noted that when the preliminary FOK associ- memory retrieval, with monitoring running ated with a question is high, subjects spend parallel with the different stages of the pro- more time searching for the target before giv- cess. An important feature of their scheme is ing up than when initial FOK is low (Nelson & that the supervisory processes of monitoring Narens, 1990). The decision to continue and control are assumed to run parallel with searching reflects the operation of two con- the different stages of remembering rather flicting tendencies: the reward for finding the than being confined to any one stage. correct answer, and the cost for spending time While there is agreement regarding the con- searching (Barnes, Nelson, Dunlosky, Maz- trolled, goal-oriented nature of retrieval, the zoni, & Narens, 1999). 342 MEMORY IN LIFE

What are the processes underlying the out- wrong. This was true for both recall and recog- putting of an answer? All of the retrieval models nition memory testing. How do people regu- reviewed earlier incorporate verification pro- late their memory accuracy? According to a cesses in which the memory records recovered model proposed by Koriat and Goldsmith are evaluated for suitability. These processes oc- (1996), when recounting past events, people cur throughout the retrieval process, often lead- monitor the likelihood that each item of infor- ing to self-corrections. Burgess and Shallice mation that comes to mind is correct. They (1996] stressed the frequent occurrence of "er- then apply a control threshold to the monitor- rors" in verbal protocols, which would nor- ing output for the item with the highest sub- mally not show up in the final response re- jective probability of being correct. The item ported because they are corrected or edited out. will be reported if its assessed probability pas- They proposed that these might provide the key ses threshold, and will be withheld otherwise. for explaining the occurrence of confabulations The setting of the control threshold depends among patients with frontal lobe damage (see on the relative utility of providing complete also Moscovitch, 1989). According to them, versus accurate information: the stronger the memory errors are a standard part of the normal motivation for accuracy, the more selective memory retrieval so that mechanisms that guard people are in their reporting, and hence the against them must exist. higher the level of memory accuracy attained. Williams and Hollan's study (1981) of the Several experiments provided support for the memory for classmates also revealed a consid- model, revealing the manner in which moni- erable number of "fabrications" (about a third toring and control processes mediate between of the total number of names reported), some memory retrieval on the one hand and mem- of which were later corrected by the subject. ory performance on the other. A quantity-ac- Walker and Kintsch (1985) examined a differ- curacy tradeoff was observed: subjects could ent kind of error: reporting a category member achieve a higher level of memory accuracy that has already been reported. Whereas sub- only by withholding a larger number of cor- jects who were asked to think aloud produced rect answers as well. The degree of tradeoff, i large number of such "errors," control sub- however, varied strongly with monitoring ef- jects produced practically none, suggesting fectiveness—that is, the ability to distinguish that they were able to edit out retrieved but between correct and incorrect answers. Effec- already reported items. tive monitoring allowed a person to achieve a Editing processes are particularly impor- higher level of memory accuracy at a smaller tant when the accuracy of what one reports is cost in quantity. at stake. A person on a witness stand, for ex- Another means by which subjects can regu- ample, must be concerned not only with late the accuracy of their memory reports is by "telling the whole truth" but also with controlling the "grain" or level of generality of telling "nothing but the truth." In order to the reported information (see Goldsmith & Ko- meet both requirements, an eyewitness must riat, 1999). Rather than withhold an answer monitor the correctness of the information entirely, a person may choose a level of gener- that comes to mind and weigh the costs of ality at which he is less likely to be wrong providing a piece if information that may be (e.g., "in the early afternoon" rather than "at 2 incorrect against he costs of withholding a P.M."). The choice of grain size is guided by correct piece of information. the attempt to compromise between the ten- Koriat and Goldsmith (1994, 1996) exam- dency to be accurate and the tendency to be ined the strategic regulation of memory informative (Goldsmith & Koriat, 1999; Ya- reporting within the traditional, item-based niv & Foster, 1995). memory assessment framework. Memory In sum, in this chapter we focused on con- quantity was defined as the likelihood of trol processes in remembering. Some of these remembering an input item, whereas memory processes undoubtedly take place in simple, accuracy was defined as the likelihood that a laboratory contexts, but most of them are more reported item is correct. Their results clearly apparent in the everyday use of mem- suggested that the option f free report—that ory retrieval, particularly when retrieval is is, the option to decide which items to more laborious. A greater effort is being made volunteer and which to with-old, allows in recent years to bring some of these pro- subjects to enhance their memory accuracy at cesses under systematic experimental investi- the expense of memory quantity by gation. screening out items that are likely to be CONTROL PROCESSES IN REMEMBERING 343

Acknowledgments This chapter was pre- lation and the control of recollection. Mem- pared while the author was a visiting profes- ory, 4, 359-411. sor at the Max Planck Institute for Psychologi- Ceraso, J. (1987). On generic recall. In D. S. cal Research, Munich, Germany. The work Gorfein & R. R. Hoffman (Eds.), Memory was supported by the Max-Wertheimer Min- and learning (pp. 319-327). Hillsdale, NJ: erva Center for Cognitive Processes and Hu- Erlbaum. man Performance, University of Haifa, and by Cohen, G., Stanhope, N., & Conway, M. A. the Ebelin and Gerd Bucerius ZEIT Founda- (1992). Age differences in the retention of tion. knowledge by young and elderly students'. British Journal of Developmental Psychol- ogy, 10, 153-164. References Collins, A. M., & Loftus, E. F. (1975). A Anderson, M. C., Bjork, R. A., & Bjork, E. L. spreading-activation theory of semantic (1994). Remembering can cause forgetting: processing. Psychological Review, 82, 407- Retrieval dynamics in long-term memory. 428. Journal of Experimental Psychology: Learn- Conway, M. A. (1992). A structural model of ing, Memory, and Cognition, 20, 1063- autobiographical memory. In M. A. Con- 1087. way, D. C. Rubin, H. Spinnler, & W. A. Wa- Baddeley, A. (1997]. Human memory—theory genaar (Eds,), Theoretical perspectives on and practice. Hove, England: Psychology autobiographical memory (pp. 167-193J. Press. Dordrecht, Netherlands: Kluwer. Barnes, A. E., Nelson, T. O., Dunlosky, J., Maz- Dagenbach, D., & Carr, T. H. (1994). Inhibitory zoni, G., & Narens, L. (1999). An integra- processes in perceptual recognition: Evi- tive system of metamemory components in- dence for a center-surround attentional volved in retrieval. In D. Gopher & A. mechanism. In D. Dagenbach & T. H. Carr Koriat (Eds.), Attention and performance (Eds.), Inhibitory processes in attention, XVII—Cognitive regulation of perfor- memory, and language (pp. 327-358). San mance: Interaction of theory and applica- Diego: Academic Press. tion (pp. 287-313). Cambridge, MA: MIT Dagenbach, D., Carr, T. H., & Barnhardt, T. M. Press. (1990). Inhibitory semantic priming of lexi- Bekerian, D. A., & Dritschel, B. H. (1992). Au- cal decisions due to failure to retrieve tobiographical remembering: An integra- weakly activated codes. Journal of Experi- tive approach. In M. A. Conway, D. C. Ru- mental Psychology: Learning, Memory, and bin, H. Spinnler, & W..A. Wagenaar (Eds.), Cognition, 16, 328-340. Theoretical perspectives on autobiographi- Dorfman, J., & Mandler, G. (1994). Implicit cal memory [pp. 135-150). Dordrecht, and explicit forgetting: When is gist re- Netherlands: Kluwer. membered? Quarterly Journal of Experi- Bjork, R. A., & Bjork, E. L. (1992). A new the- mental Psychology, 47A, 651-672. ory of disuse and an old theory of stimula- Duncker, K. (1945). On problem-solving. Psy- tion fluctuation. In A. F. Healy, S. M. Kos- chological Monographs, 58, 1-113. slyn, & R. M. Shiffrin (Eds.), From learning Durso, F. T., & Shore, W. J. (1991) Partial processes to cognitive processes: Essays in knowledge of word meanings. Journal of honor of William K. Estes (Vol. 2; pp. 35- Experimental Psychology: General, 120, 67). Mahwah, NJ: Erlbaum. 190-202. Brainerd, C. J., & Reyna, V. F. (1993). Memory Eich, J. E. (1980). The cue-dependent nature independence and memory interference in of state-dependent retrieval. Memory & cognitive development. Psychological Re- Cognition, 8, 157-173. view, 100, 42-67. Eich, E. (1985). Context, memory, and inte- Brandimonte, M., & Ellis, J. (1996). Prospec- grated item/context imagery. Journal of Ex- tive memory. Mahwah, NJ: Erlbaum. perimental Psychology: Learning, Memory, Brown, A. S. (1991). A review of the tip-of- the-tongue experience. Psychological Bulle- and Cognition, 11, 764-770. tin, 109, 204-223. Eich, E., & Metcalfe, J. (1989). Mood depen- Brown, R., & McNeill, D. (1966). The "tip of dent memory for internal versus external the tongue" phenomenon. Journal of Ver- events. Journal of Experimental Psychol- bal Learning and Verbal Behavior, 5, 325- ogy: Learning, Memory, and Cognition, 15, 337. 443-455. Burgess, P. W., & Shallice, T. (1996). Confabu- Eich, J. E., Weingartner, H., Stillman, R. C, & 344 MEMORY IN LIFE

Gillin, J. C. (1975). State-dependent accessi- Koriat, A. (1993). How do we know that WE bility of retrieval cues in the retention of a know? The accessibility model of the fee categorized list. Journal of Verbal Learning ing of knowing. Psychological Review, li and Verbal Behavior, 14, 408-417. 609-639. Einstein, G. O., & McDaniel, M. A. (1996). Re- Koriat, A. (1995). Dissociating knowing and trieval processes in prospective memory: the feeling of knowing: Further evidence Theoretical approaches and some empirical for the accessibility model. Journal of Ex findings. In M. Brandimonte, G. O. Ein- perimental Psychology: General, 124; 31: stein, & M. A. McDaniel (Eds.), Prospective 333. memory: Theory and applications (pp. Koriat, A. (1998). Illusions of knowing: The 115-141). Mahwah, NJ: Erlbaum. link between knowledge and metaknowl- Erikson, T. A., & Mattson, M. E. (1981). From edge. In V. Y. Yzerbyt, G. Lories, & B. Da; words to meaning: A semantic illusion. denne (Eds.), Metacognition: Cognitive ai Journal of Verbal Learning and Verbal Be- social dimensions (pp. 16-34). London, havior, 20, 540-552. England: Sage. Fisher, R. P., & Geiselman, R. E. (1992). Mem- Koriat, A., & Feuerstein, N. (1976). The reco ory-enhancing techniques for investigative ery of incidentally acquired information. interviewing: The cognitive interview. Ada Psychologica, 40, 463-474. Springfield, IL: Thomas. Koriat, A., & Goldsmith, M. (1994). Memory Glucksberg, S., & McCloskey, M. (1981). Deci- in naturalistic and laboratory contexts: Di sions about ignorance. Journal of Experi- tinguishing the accuracy-oriented and mental Psychology: Human Learning and quantity-oriented approaches to memory Memory, 7, 311-325. assessment. Journal of Experimental Psy- Godden, D. R., & Baddeley, A. D. (1975). Con- chology: General, 123, 297-315. text-dependent memory in two natural en- Koriat, A., & Goldsmith, M. (1996). Monitor- vironments: On land and underwater. Brit- ing and control processes in the strategic ish Journal of Psychology, 66, 325-331. regulation of memory accuracy. Psycholo^ Goldsmith, M., & Koriat, A. (1999). The strate- cal Review, 103, 490-517. gic regulation of memory reporting: Mecha- Koriat, A., Goldsmith, M., & Pansky, A. (in nisms and performance consequences. In press). Toward a psychology of memory a< D. Gopher & A. Koriat (Eds.), Attention and curacy. Annual Review of Psychology. performance XVII—Cognitive regulation of Koriat, A., & Lieblich, I. (1974). What does a performance: Interaction of theory and person in the "TOT" state know that a pe: application, (pp. 373-400). Cambridge, son in a "don't know" state does not MA: MIT Press. know? Memory & Cognition, 2, 647-655. Goodwin, D. W., Powell, B., Bremer, D., Koriat, A., & Lieblich, I. (1977). A study of Hoine, H., & Stern, J. (1969). Alcohol and memory pointers. Ada Psychologica, 41, recall: State dependent effects in man. Sci- 151-164. Ence, 163, 1358. Koriat, A., & Melkman, R., (1987). Depth of processing and memory organization. Psy- Jacoby, L. L. (1991). A process dissociation chological Research, 49, 173-181. framework: Separating automatic from in- McDaniel, M. A., & Einstein, G. O. (1993). tention uses of memory. Journal of Memory The importance of cue familiarity and cue and Language, 30, 513-541. distinctiveness in prospective memory. Jones, G. V. (1989). Back to Woodworth: Role Memory, 1, 23-41. of interlopers in the tip-of-the-tongue phe- Mandler, G. (1980). Recognizing: The judg- nomenon. Memory & Cognition, 17, 69-76. ment of previous occurrence. Psychologi- Kelley, C. M., & Lindsay, D. S. (1993). Remem- cal Review, 87, 252-271. bering mistaken for knowing: Ease of re- Mantyla, T. (1986). Optimizing cue effective- trieval as a basis for confidence in answers ness: Recall of 500 and 600 incidentally to general knowledge questions. Journal of learned words. Journal of Experimental Memory and Language, 32, 1-24. Psychology: Human Learning and Memory Klin, C. M., Guzman, A. E., & Levine, W. H. 12, 66-71. (1997). Knowing that you don't know: Met- Marsh, R. L., Landau, J. D., & Hicks, J. L. amemory and discourse processing. Jour- (1997). Contributions of inadequate source nal of Experimental Psychology: Learning, monitoring to unconscious plagiarism dur- Memory, and Cognition, 23, 1378-1393. ing idea generation. Journal of Exverimen- Kolodner, J. L. ( 1983). Reconstructive mem- A ______CONTROL PROCESSES IN REMEMBERING 345

Morris, C. D., Bransford, J. D., & Franks, J. J. Reiser, B., Black, J. B., & Abelson, R. P. (1977). Levels of processing versus transfer (1985). Knowledge structures in the organi- appropriate processing. Journal of Verbal zation and retrieval of autobiographical Learning and Verbal Behavior, 16, 519- memories. Cognitive Psychology, 17, 89- 533. 137. Morton, J., Hammersley, R. H., & Bekerian, D. Reiser, B., Black, J. B., & Kalamarides, P. A. (1985). Headed records: A model for (1986). Strategic memory search processes. memory and its failures. Cognition, 20, 1- In D. C. Rubin (Ed.), Autobiographical 23. memory (pp. 100-121). New York: Cam- - Moscovitch, M. (1989). Confabulation and the bridge University Press. frontal lobe system: Strategic vs. associa- Roediger, H. L., & Guynn, J. M. (1996). Re- tive retrieval in neuropsychological theo- trieval Processes. In E. L. Bjork & R. A. ries of memory. In H. L. Roediger & F. I. M. Bjork (Eds.), Handbook of perception and Craik (Eds.), Varieties of memory and con- cognition: Memory (Vol. 10; pp. 197-236). sciousness: Essays in honor of En del Tulv- San Diego: Academic Press. ing (pp. 133-160). Hillsdale, NJ: Erlbaum. Roediger, H. L., & McDermott, K. B. (1994). Neely, J. H. (1976). Semantic priming and re- Implicit memory in normal human sub- trieval from lexical memory: Evidence for jects. In H. Spinnler & F. Boiler (Eds.), facilitatory and inhibitory processes. Mem- Handbook of neuropsychology (Vol. 8; pp. 63-131). Amsterdam: Elsevier. ory & Cognition, 4, 648—654. Neely, J. H., & Keefe, D. E. (1989). Semantic Roediger, H. L., & McDermott, K. B. (1995). context effects on visual word processing: Creating false memories: Remembering A hybrid prospective/retrospective pro- words not presented in lists. Journal of Ex- cessing theory. In G. H. Bower (Ed.), The perimental Psychology: Learning, Memory, psychology of learning and motivation: Ad- and Cognition, 22,803-814. vances in research and theory (Vol. 24; pp. Salaman, E. (1982). A collection of moments. 207-248). New York: Academic Press. In U. Neisser (Ed.), Memory observed: Re- Nelson, D. L., McKinney, V. M., Gee, N. R., & membering in natural contexts (pp. 49- Janczura, G. A. (1998). Interpreting the in- 63). San Francisco: Freeman. fluence of implicitly activated memories Schacter, D. L., & Worling, J. R. (1985). Attri- on recall and recognition. Psychological bute information and the feeling of know- Review, 105, 299-324. ing. Canadian Journal of Psychology, 39, Nelson, T. O., & Narens, L. (1990). Metamem- 467-475. ory: A theoretical framework and new find- Schacter, D. L., Norman, K. A., & Koutstaal, ings. In G. Bower (Ed.), The Psychology of W. (1998). The cognitive neuroscience of learning and motivation: Advances in re- constructive memory. Annual Review of search and theory (Vol. 26; pp. 125-123). Psychology, 49, 289-318. San Diego: Academic Press. Schank, R. C. (1982). Dynamic memory. Cam- Nhouyvanisvong, A., & Reder, L. M. (1998). bridge: Cambridge University Press. Rapid feeling-of-knowing: A strategy selec- Schwartz, B. L., & Metcalfe, J. (1992). Cue fa- tion mechanism. In V. Y. Yzerbyt, G. Lo- miliarity but not target retrievability en- ries, & B. Dardenne (Eds.), Metacognition: hances feeling-of-knowing judgments. Jour- Cognitive and social dimensions (pp. 35- nal of Experimental Psychology: Learning, 52). London, England: Sage. Memory, and Cognition, 18, 1074-1083. Nickerson, R. S. (1981). Motivated retrieval Schwartz, B. L., & Smith, S. M. (1997). The re- from archival memory. In J. H. Flowers trieval of related information influences (Ed.), Nebraska symposium on motivation tip-of-the-tongue states. Journal of Memory (pp. 73-119). Lincoln: University of Ne- and Language, 36, 68-86. braska Press. Smith, S. M. (1979). Remembering in and out Norman, D. A., & Bobrow, D. G. (1979). De- of context. Journal of Experimental Psy- scriptions: An intermediate stage in mem- chology: Human Learning and Memory, 5, ory retrieval. Cognitive Psychology, 11, 460-471. 107-123. Smith, S. M., Glenberg, A. M., & Bjork, R. A. Raaijmakers, J. G. W., & Shiffrin, R. M. (1981). (1978). Environmental context and human Search of associative memory. Psychologi- memory. Memory S- Cognition, 6, 342-353. cal Review, 88, 93-134. Tulving, E. (1967). The effects of presentation Reder, L. M. (1987). Strategy selection in ques- and recall in free recall learning. Journal tion answering. Cognitive Psychology, 19, of Verbal Learning and Verbal Behavior, 6, 90-138. 175-184. 346 MEMORY IN LIFE

Tulving, E. (1983). Elements of episodic mem- Williams, M. D., & Hollan, J. D. (1981). The ory. New York: Oxford University Press. process of retrieval from very long-term Tulving, E., & Pearlstone, Z. (1966). Availabil- memory. Cognitive Science, 5, 87-119. ity versus accessibility of information in Williams, M. D., & Santos-Williams, S. memory for words. Journal of Verbal Learn- (1980). Methods for exploring retrieval pro- ing and Verbal Behavior, 5, 381-391. cesses using verbal protocols. In R. S. Nick- Tulving, E., & Thomson, D. (1973). Encoding erson (Ed.), Attention and Performance specificity and retrieval processes in epi- VIII (pp. 671-689). Hillsdale, NJ: Erlbaum. sodic memory. Psychological Review, 80, Yaniv, I., & Foster, D. P. (1995). Graininess of 352-373. judgment under uncertainty: An accuracy- Walker, W. H., & Kintsch, W. (1985). Auto- informativeness trade-off. Journal of Exper- matic and strategic aspects of knowledge imental Psychology: General, 124, 424— retrieval. Cognitive Science, 9, 261-283. 432.