The generation effect with homographs: Evidence for postgeneration processing

The generation effect is the phenomenon in which words are remembered better when generated than when read. These experiments test the possibility that at least one consequence of generating is enhanced semantic processing. Homographs were used as targets, presented with rhymes in Experiment 1 so as not to bias meaning, and with synonyms in Experiment 2 to bias one meaning of each homograph. In beth experiments, extralist synonym cues were provided at recall. In Experiment 1 a generation effect was obtained when the retrieval cues biased the dominant meaning of the homograph (determined from free association norms), whereas in Experiment 2 a generation effect was found when the retrieval cues biased the same meaning that was biased during study. In neither experiment was a generation effect obtained with retrieval cues that biased the other meaning of each homograph. These results indicate that the generation effect is dependent upon the compatibility of the semantic processing conducted at study and test. Since it is impossible to process the meaning of a homograph when generating it from a rhyme cue, the meaning of the homograph could only have been processed after the word had been generated. The finding in Experiment 1 that a generation effect was obtained with rhymes when semantic retrieval cues were provided demonstrates that the enhancement properties associated with generation are not restricted to the information used to guide the generation process. This finding also indicates that one locus of the generation effect is in the processing that occurs after the word has been generated.     

Recency effects in direct and indirect memory tasks

In three experiments, subjects learned two lists under incidental conditions and were then given either a part-word or a word (extralist associate) cue. Each cue was related to one word in each list. Half the subjects were given production instructions (an indirect memory test), and half were given cued recall instructions (a direct memory test). When the interval between List 2 and the test was shortened, recency effects were found for part-word cues for both cued recall and production instructions. Little or no recency effects were found with word cues. These results are incompatible with a simple distinction between the types of memory trace or information that are tapped by direct as opposed to indirect memory tasks. Possible causes for the recency effect and for the difference between word and part-word cues are discussed.     

Semantic specificity in cued recall

In three cued recall experiments, extralist retrieval cues that were congruous in meaning to an encoded target pair of words produced better recall of the targets than wh6n the cuetarget relation was incongruous. However, this result, which differs from that of other experiments, depended in some cases on scoring recall of a target pair when either member of the pair was recalled. It is argued that (1) pairs of words are typically stored as higher order units, (2) the best test procedure is to request recall of both members of the pair when an extralist cue is presented, and (3) semantic features provide an important dimension in the mnemonic representation of word events.     

Tests of the generality of the principle of encoding specificity

Two empirical tests of the principle of encoding specificity are reported. In Experiment I, the normative strength of the cues presented on the input and on the recall trial was varied factorially. To lessen the emphasis on strictly associative learning, only half the items were cued in each phase of the study-recall cycle. Recall was higher when the cues remained the same than when they changed. However, regardless of the condition of input cuing, strong output cues were substantially more effective than weak ones. In Experiment II, the to-be-remembered words were shown in the presence of weak cues on the input trial. Recognition in the context of strong extralist cues was compared with recall to the original input cues. On the test of cued recognition, the target words were either generated by the subjects as free associates or presented to them as items on a test constructed by the experimenter. Contrary to previous findings, recall was not found to be superior to recognition. The phenomena of cue-dependent forgetting that have been interpreted as evidence for the principle of encoding specificity appear to have limited generality.     

Negative Cueing Effects with Weak and Strong Intralist Cues

In three experiments, we studied cueing effects of relational and itemspecific information after enacted and non-enacted encoding of short sentences (e.g. lift the pen, fold the paper). In Experiment 1, all subjects were instructed at encoding to remember only the nouns of these sentences; half of the subjects were informed about the categorical nature of the nouns, whereas the other half were not. At retrieval, all subjects were given a free recall test and a cued recall test with the verb of each sentence as the cue. In Experiment 2, all subjects were instructed at encoding to remember the whole sentence; as in Experiment 1, half of the subjects were informed about the categorical nature of the nouns and half were not. At test, all subjects were given two cued recall tests, one categorical cue for each noun in the first test and one verb cue and one categorical cue for each noun in the second test. In Experiment 3, at encoding, all subjects were informed about the categorical nature of nouns and were instructed to remember the whole sentence. In this experiment, the actions were performed with imaginary objects; free recall and cued recall tests were given to different subjects. In all three experiments, there was a negative effect of intralist cueing with verbs. This finding is at odds with the Encoding Specificity Principle, which assumes facilitation of cueing at retrieval if the cues were encoded together with the to-be-remembered information at encoding. Also, the effect of intralist cueing was different after encoding with enactment than after encoding without enactment; this difference holds true for enactment with real objects but not for enactment with imaginary objects. Enactment increased both the relational and the item-specific cueing efficiency. The results are discussed in terms of encoding interference between cues and targets and between item-specific processing and relational processing. Enacted encoding is conceived as integrating episodic information both with respect to item specificity and relational aspects of the information.     

Effects of encoding instructions and retrieval cuing on recall in Korsakoff patients

The relationship between the depth of encoding a word and its subsequent recall, either cued or noncued, was investigated in this study. In Experiment 1, Korsakoff subjects and alcoholic controls were shown a categorized word list under one of three different encoding instructions: (1) nonsemantic, that is, detecting the presence or absence of the letter “e” in each word, (2)semantic, that is, assigning words to their correct taxonomic category, and (3) no encoding instructions. Semantic encoding instructions resulted in higher recall for both diagnostic groups than the other instructions. In Experiment 2, subjects were again assigned to one of the three encoding instructions as in Experiment 1, but all groups received cues (category labels) at the time of recall. Cuing increased recall for all but the group receiving instructions to encode nonsemantically. Experiment 3 was a replication of the previous experiments. The results indicated that Korsakoff subjects were capable of encoding semantically without specific instructions to do so but were impaired in the ability to generate retrieval cues at the time of recall.     

Aging and the stability of activation and sampling in cued recall

Age effects in cued recall were investigated as a function of activation and sampling of preexisting associates of the test cue. Young adults, community-dwelling elderly, and elderly patients studied lists of unrelated words and were tested with extralist cues. Preexperimental strength between test cues and studied words was manipulated to discern differences in activation, and normative size of the set of associates was manipulated to discern differences in sampling. Test delay and prior testing were also manipulated in Experiment 1. Although large age effects were found with phonemic and taxonomic test cues, young and older subjects showed comparable effects of strength and set size, suggesting that age effects were not due to activation and sampling differences. Test delay and prior testing also had comparable effects. Implications for age effects in episodic cued recall are discussed.     

Effects of retention interval and modality on sensory and semantic trace information

Subjects studied a long list of individual words that were presented either visually or auditorily. Recall was tested immediately or after a filled delay by using either word endings or taxonomic categories as extralist retrieval cues. Two interactions were of particular interest. First, word ending cues were just as effective as taxonomic cues on the immediate test. On the delayed test, however, ending cues were less effective. This result suggests that sensory information encoded about a word decays at a faster rate than semantic information. Second, although modality had no observable influence on the taxonomic cues, word ending cues were more effective when all items were shown visually than when they were presented auditorily. Taken together, these findings indicate that the visual features of words are encoded at study and that this information can be accessed during test if it is recapitulated by the retrieval cue shortly after acquisition.     

Release from generation failure: The role of study list structure

Three experiments, using the original encoding-specificity paradigm, investigated the role of study list structure in producing Higham and Tam's (2005) generation failure effect. Generation failure occurs when cued recall performance for strong, extralist cues is worse than target production in a control group that is given no study list but is instead required merely to generate responses to the same test cues. In the present study, generation failure was replicated in Experiment 1, and Experiment 2 demonstrated that strong, extralist cues were more likely to elicit targets in pure generation groups when participants had studied a list of strong associates than when they had studied a list of weak ones. In Experiment 3, participants were released from generation failure when a study list of moderate associates was used and the cue-to-target associative strength was equated between the reinstated- and extralist-cue conditions. Together, these results suggest that generation failure is partly attributable to participants' searching inappropriate domains that, though consistent with the study list structure, are unlikely to contain targets.     

Lexical and semantic search in cued recall, fragment completion, perceptual identification, and recognition

Search processes in word-stem cued recall, fragment completion, perceptual identification, and recognition are contrasted. These retention tests involve letters as cues, but the lexical characteristics of these cues vary considerably. In word-stem cued recall, ending letters are presented as recall cues for studied targets (e.g., ONEY as a cue for HONEY). In fragment completion, the test cues consist of letters and spaces (e.g., HO__Y); in perceptual identification, they consist of letter features that survive the mask; and in recognition, they consist of all the letters of the studied word (e.g., HONEY). These differences in retention tests and lexical characteristics were evaluated by manipulating three variables with known effects in cued recall: (a) the presence of study context words emphasizing lexical information, (b) lexical set size corresponding to the number of words that fit the letter cue, and (c) meaning set size corresponding to the number of meaningful associates linked to the studied targets. The results indicated that (a) the presence of study contexts emphasizing lexical information reduced accuracy and response time equally in all tasks, (b) larger lexical set sizes reduced accuracy and response time in all tasks except recognition, and (c) larger meaning set size reduced accuracy in cued recall but not in the other tasks. Lexical search appears to be a significant component process in word-stem cued recall, fragment completion, and identification. Searching through meaning-related concepts encoded during study is a significant component process only in cued recall.     

Primacy and recency in primed free association and associative cued recall

Two experiments compared the serial positions of primed words in an implicit free association test with words recalled in a cued recall test. In both tests, weakly or strongly related word pairs were studied, and the first words of each pair formed the test cues. In the implicit test, weakly related words pairs showed primacy and extended recency effects but strongly related word pairs did not. In the explicit test, both weakly and strongly related word pairs showed primacy and extended recency effects. These functional dissociations between implicit and explicit memory tests indicate that strongly related word pairs are encoded together because they have unitized memory representations that function as integrated units without requiring any additional associative links to be made, but that an additional system or process is required to strengthen weakly related word pairs during encoding. In addition, a further additional system or process is accessed by explicit retrieval.     

How implicitly activated and explicitly acquired knowledge contribute to the effectiveness of retrieval cues

The extralist cued recall task simulates everyday reminding because a memory is encoded on the fly and retrieved later by an unexpected cue. Target words are studied individually, and recall is cued by associatively related words having preexisting forward links to them. In Experiments 1 and 2, forward cue-to-target and backward target-to-cue strengths were varied over an extended range in order to determine how these two sources of strength are related and which source has a greater effect. Forward and backward strengths had additive effects on recall, with forward strength having a consistently larger effect. The PIER2 model accurately predicted these findings, but a plausible generation-recognition version of the model, called PIER.GR, could not. In Experiment 3, forward and backward strengths, level of processing, and study time were varied in order to determine how preexisting lexical knowledge is related to knowledge acquired during the study episode. The main finding indicates that preexisting knowledge and episodic knowledge have additive effects on extralist cued recall. PIER2 can explain these findings because it assumes that these sources of strength contribute independently to recall, whereas the eSAM model cannot explain the findings because it assumes that the sources of strength are multiplicatively related.     

The anatomy of free recall

The relative importance of various covert cues used in free recall is inferred on the basis of conditional probabilities of free recall given success vs failure of cued recall of the same material. Thirty-six names associated with pictures arranged in a 6 by 6 matrix were learned. A free recall test of names was followed by cued recall of names, with pictures, positions, or joint pictures and positions used as cues. Matching tests based on these cues were also administered. The tests were given at one of two stages of training, immediately or after 2 days. Pictorial and position cues are equally well encoded, but pictorial cues are less accessible and, therefore, relatively more useful in cued recall than in free recall. Position and pictorial information related to the names appears to be encoded and forgotten independently, and there is no evidence for summation of subthreshold encoding effect.     

Part-set cuing of false memories

Part-set cuing inhibition describes the common finding that re-presenting items from a word list can reduce subjects’ overall recall performance for studied items. Do part-set cuing effects occur for false memories as well? In the present experiments, subjects studied lists of words drawn from Roediger and McDermott (1995). After studying each list, subjects completed math problems and then recalled the list items either with or without accompanying list cues. In Experiment 1, the recall cues consisted of items drawn randomly from the original list. In Experiment 2, an additional type of cued recall task was added in which the even numbered list items were used as cues. Taken together, these experiments demonstrate robust part-set cuing effects for critical nonpresented items. In addition, they show that whereas recall of critical words is reduced by the presence of cues at test, retrieval cues do not affect critical words and studied words in exactly the same manner.     

Altering memory through recall: The effects of cue-guided retrieval processing

Three experiments were designed to investigate the influence of initial recall on memory by assessing delayed recall after different immediate cued-recall tests. In all experiments, subjects performed semantic and phonemic encoding tasks on a word list. The subjects then received a cued-recall test that cued the target using the same word as the context word in the encoding task, a test that cued the target with a word from the same level at which the target was encoded, a test that cued the target with a cue from a different level at which the target was encoded, or no immediate-recall test. One day later, the subjects performed a final cued-recall test in which the type of cue (semantic or phonemic) was varied. Consistently, delayed recall was facilitated primarily when the cue on the immediate test was from the same level as the cue on the delayed test. This pattern of facilitation suggests that immediate cued-recall produces an elaboration of an existing memory representation that is closely tied to the type of cue used on the immediate test.     

The effects of natural category size on memory for episodic encodings

Previous findings indicate that natural category size affects cued recall but not recognition performance. Words that define or belong to larger categories are not as likely to be recalled in the presence of an extralist cue. However, category size has no effect on recognition in the presence of the target as the cue. Theoretically, this difference could be due to inherent differences between these tasks, to the use of different types of test cues, or to differences in the nature of the required responses (naming compared with “yes/no”decisions). Three experiments indicated that none of these factors is a sole determinant. Natural category size effects were found in cued-recall and recognition tasks, with extralist and target cues and regardless of the required response. The critical factor is whether the testing conditions require or encourage subjects to search the category defined by the cue. With the initiation of such a search, information represented in semantic memory is likely to influence memory for episodic information.     

Comparing word fragment completion and cued recall with letter cues

Cued recall with word stems as cues and fragment completion rely on different types of letter cues and also differ in the explicit-implicit nature of the retrieval orientation. Despite these differences, variables effective in one task may be effective in the other because both rely on letter cues. Two variables known to affect cued recall were manipulated: Lexical set size (number of words that fit the letter cue) and meaning set size (number of associates generated to the studied words). Across four experiments, subjects in each task were less likely to recover targets from larger lexical sets. However, meaning set size affected cued recall but not fragment completion. These results indicate that fragment completion and letter-cued recall are based on lexical search but that cued recall also involves a semantic search component. Furthermore, type of retrieval cue had a greater effect than type of retrieval orientation.     

Semantic and acoustic processing in free and cued recall by educable mentally retarded adolescents

The abilities of educable mentally retarded adolescents to encode and retrieve words with semantic and acoustic cues were investigated in a free and cued recall task. On each of three trial blocks, seven groups of subjects were presented 20 unrelated stimulus words. Groups received either semantic, acoustic, or no encoding cues along with the stimuli. Free recall was requested from all subjects, followed immediately by a second period of either free recall or cued recall with the semantic or acoustic cues. Semantic cues were most effective when presented both at encoding and retrieval. The subjects were unable to use acoustic information as effective retrieval aids. Results were discussed in terms of encoding dimension dominance and mediational deficiencies.     

The effects of specific and categorical orienting on children's incidental and intentional memory for pictures and words

This study examines the effects of differences in the encoding of specific and categorical information on second graders' (7 years; 7 months), fifth graders' (10; 6), and college adults' cued recall for cue-target picture and word pairs. The cues at retrieval were either same-modal (P-P and W-W) or cross-modal (P-W and W-P) as the cues presented in acquisition, and acquisition encoding was either incidental or intentional and constrained by orienting questions or unconstrained. The most important results were that both picture and word recall varied with the encoding of both specific and categorical information and that children differed from adults in the encoding of both kinds of information in both incidental and intentional encoding conditions. In addition, both children and adults showed congruency effects for pictures and words that were greater for specific than for categorical orienting questions. The results suggest that differences in the encoding of both specific and categorical attribute information contribute to developmental recall differences independently of encoding intent and stimulus modality.     

The generation effect and the modeling of associations in memory

The search of associative memory (SAM) model of Gillund and Shiffrin (1984) was applied to data of two experiments that examined the generation effect (Slamecka & Graf, 1978). Subjects studied a list of related word pairs, in which they either read both words in the pair or generated the righthand response term using the left-hand stimulus term plus the response word fragment as generation cues. Experiment 1 manipulated encoding condition within subjects and used an incidental learning procedure. Experiment 2 manipulated encoding condition between subjects and used an intentional learning procedure. Memory was tested with recognition, cued recall, and free recall. A higher order association model gave a better and more parsimonious fit to the results than did an item-level association model. The relationship between various versions of SAM and current accounts of the generation effect are discussed, particularly the two-factor theory of Hirshman and Bjork (1988).