The mirror effect in recognition memory

The mirror effect in recognition memory refers to the fact that, with several different classes of stimuli, performance on new items from each class mirrors (is correlated with) performance on the corresponding classes of old items. Classes of stimuli that are accurately recognized as old when old are also accurately recognized as new when new; those that are poorly recognized as old when old are also poorly recognized as new when new. The statement above is shown not to be a tautology. A survey demonstrates that the effect holds for several types of variables (ways to classify stimuli)—word frequency, concreteness, meaningfulness, and others. The survey includes a total of 80 findings. The theoretical implications of the effect are considered.     

The mirror effect in recognition memory: data and theory

The mirror effect is a regularity in recognition memory that requires reexamination of current views of memory. Five experiments that further support and extended the generality of the mirror effect are reported. The first two experiments vary word frequency. The third and fourth vary both word frequency and concreteness. The fifth experiment varies word frequency, concreteness, and the subject's operations on the words. The experiments furnish data on the stability of the effect, its relation to response times, its extension to multiple mirror effects, and its extension beyond stimulus variables to operation variables. A theory of the effect and predictions that derive from the theory are presented.     

The variance theory of the mirror effect in recognition memory

The mirror effect refers to a rather general empirical finding showing that, for two classes of stimuli, the class with the higher hit rates also has a lower false alarm rate. In this article, a parsimonious theory is proposed to account for the mirror effect regarding, specifically, high- and low-frequency items and the associated receiver-operating curves. The theory is implemented in a recurrent network in which one layer represents items and the other represents contexts. It is shown that the frequency mirror effect is found in this simple network if the decision is based on counting the number of active nodes in such a way that performance is optimal or near optimal. The optimal performance requires that the number of active nodes is low, only nodes active in the encoded representation are counted, the activation threshold is set between the old and the new distributions, and normalization is based on the variance of the input. Owing to the interference caused by encoding the to-be-recognized item in several preexperimental contexts, the variance of the input to the context layer is greater for highthan for low-frequency items, which yields lower hit rates and higher false alarm rates for high- than for low-frequency items. Although initially the theory was proposed to account for the mirror effect with respect to word frequency, subsequent simulations have shown that the theory also accounts for strength-based mirror effects within a list and between lists. In this case, consistent with experimental data, the variance theory suggests that focusing attention to the more difficult class within a list affects the hit rate, but not the false alarm rate and not the standard deviations of the underlying density, leading to no mirror effect.     

Global subjective memorability and the strength-based mirror effect in recognition memory

Between-list manipulations of memory strength through repetition commonly generate a mirror effect, with more hits and fewer false alarms for strengthened items. However, this pattern is rarely seen with within-list manipulations of strength. In three experiments, we investigated the conditions under which a within-list mirror effect of strength (items presented once or thrice) is observed. In Experiments 1 and 2, we indirectly manipulated the overall subjective memorability of the studied lists by varying the proportion of nonwords. A within-list mirror effect was observed only in Experiment 2, in which a higher proportion of nonwords was presented in the study list. In Experiment 3, the presentation duration for each item (0.5 vs. 3 sec) was manipulated between groups with the purpose of affecting subjective memorability. A within-list mirror effect was observed only for the short presentation durations. Thus, across three experiments, we found the within-list mirror effect only under conditions of poor overall subjective memorability. We propose that when the overall subjective memorability is low, people switch their response strategy on an item-by-item basis and that this generates the observed mirror effect.     

Encoding, repetition, and the mirror effect in recognition memory: Symmetry in motion

Attention/likelihood theory has been used to explain the mirror effect in recognition memory. The theory also predicts that any manipulation that affects the recognition of old items will also affect recognition of the new items—more specifically, that all the underlying distributions will move and that they will move symmetrically on the decision axis. In five experiments, we tested this prediction. The first two experiments used encoding tasks during study to change recognition performance for high- and low-frequency words. The results show symmetrical dispersion of the underlying distributions. The final three experiments used repetition to increase recognition performance. Repetition produced a symmetrical pattern of movement that was different from that produced by encoding task. This pattern is, however, also covered by attention/likelihood theory. A further extension of the theory was used to predict response times.     

Forgetting in recognition memory with and without recollective experience

Retention interval was manipulated in two recognition-memory experiments in which subjects indicated when recognizing a word whether its recognition was accompanied by some recollective experience ("remember") or whether it was recognized on the basis of familiarity without any recollective experience ("know"). Experiment 1 showed that between 10 min and 1 week, "remember" responses declined sharply from an initially higher level, whereas "know" responses remained relatively unchanged. Experiment 2 showed that between 1 week and 6 months, both kinds of responses declined at a similar, gradual rate and that despite quite low levels of performance after 6 months, both kinds of responses still gave rise to accurate discrimination between target words and lures. These findings are discussed in relationship to current ideas about multiple memory systems and processing accounts of explicit and implicit measures of retention.     

Perception and recognition memory of words and werds: Two-way mirror effects

We examined associative priming of words (e.g., TOAD) and pseudohomophones of those words (e.g., TODE) in lexical decision. In addition to word frequency effects, reliable base-word frequency effects were observed for pseudohomophones: Those based on high-frequency words elicited faster and more accurate correct rejections. Associative priming had disparate effects on high- and low-frequency items. Whereas priming improved performance to high-frequency pseudohomophones, it impaired performance to low-frequency pseudohomophones. The results suggested a resonance process, wherein phonologic identity and semantic priming combine to undermine the veridical perception of infrequent items. We tested this hypothesis in another experiment by administering a surprise recognition memory test after lexical decision. When asked to identify words that were spelled correctly during lexical decision, the participants often misremembered pseudohomophones as correctly spelled items. Patterns of false memory, however, were jointly affected by base-word frequencies and their original responses during lexical decision. Taken together, the results are consistent with resonance accounts of word recognition, wherein bottom-up and top-down information sources coalesce into correct, and sometimes illusory, perception. The results are also consistent with a recent lexical decision model, REM-LD, that emphasizes memory retrieval and top-down matching processes in lexical decision.     

Word frequency and word likeness mirror effects in episodic recognition memory

Estes and Maddox (2002) suggested that the word frequency mirror effect in episodic recognition memory might be due to word likeness rather than to the frequency of experience with a word per se. We examined their suggestion using a factorial manipulation of frequency and neighborhood density, a measure used in lexical memory research to measure orthographic word likeness. For study with no specified task, main effects of density and frequency were in the mirror order, confirming the hypothesized mirror effect of word likeness but not its role in producing the frequency mirror effect. Lexical decision study increased the size of both mirror effects, even though the density manipulation had a negligible effect on lexical decision performance for words. Post hoc analyses showed that neither mirror effect could be explained by differences in lower order measures of word likeness (letter and bigram frequency). The joint orders of frequency and density results were mirrored across new and old conditions in accordance with attention likelihood theory (ALT), but density effects on z-ROC slope suggest that ALT may require extension to accommodate the effect of word likeness on response confidence.     

Reflections of the mirror effect for item and associative recognition

In five experiments, participants studied pairs of words and yes/no recognition memory for both item and associative information was tested. Two stimulus manipulations, nouns versus nonnouns and high versus low word concreteness, produced the mirror effect for both item and associative recognition. The mirror effect was reflected in both measures of accuracy and response latency. A word frequency manipulation, however, produced the mirror effect only for item recognition. Two additional experiments showed that the mirror effect could also be obtained between nouns and nonnouns and between high and low concrete words for associative recognition in a forced-choice recognition procedure. The results extend the generality of the mirror effect to measures of response latency and to associative recognition and also suggest that similar retrieval and decision processes underlie recognition of item and associative information.     

Determinants of the word-frequency effect in recognition memory

Possible determinants of the word-frequency effect (WFE), that is, the finding that lowfrequency (LF) words are recognized more accurately than high-frequency (HF) words, are evaluated. Three studies examined the view that, since HF words have more meanings than LF words, it is less likely that the word sense tagged at time of presentation will be accessed at time of test and a correct response will be made. To ensure the same word sense was accessed at time of presentation and time of test in the case of both HF and LF words, sentence contexts used during presentation were combined with cuing procedures at time of test. Recognition performance improved, but the WFE was unaltered even when the identical sentence context was used during presentation and test. A fourth study considered GlarLzer and Bowles’ (19761 suggestion that associates of both HF and LF words tend to be HF words that are (1)likely to be activated and derivatively encoded during presentation and (2) likely to include a number of distractors from the recognition test sufficient to impair performance on HF words. Analysis of associative responses of subjects to HF and LF words, and the errors they made, support strongly such an interference-type interpretation of the WFE.     

The nature of memory processes underlying recognition judgments in the process dissociation procedure

The hypothesis is tested that the memory processes involved in recognition judgments in the process dissociation procedure are the same as those involved in standard source-monitoring tasks. It is shown how source-monitoring response categories can be mapped onto process dissociation response categories. On the basis of this observation, an experiment was conducted in which it was possible to compare, using a multinomial modeling approach, the parameters representing memory processes in the process dissociation procedure with those involved in source monitoring. For the two different encoding conditions realized, the results are compatible with the hypothesis that the same processes are involved in source monitoring and in recognition judgments in the process dissociation procedure. Implications for the interpretation of the model’s parameters are discussed.     

Relational information and the context effect in recognition memory

A role for relational information was examined for the paradigm in which recognition-memory performance on items tested in the same context in which they were studied is compared with performance on items tested in different contexts. Over a series of five experiments, randomly formed pairs were used to manipulate the context of high-frequency English words. Comparisons were made between instructional manipulations designed to influence the use of relational information, and between yes/no, confidence rating (both between- and within-subject), and forced-choice tasks. There was a context effect not due to the use of inappropriate response strategies. However, high-criterion subjects resembled those subjects who were specifically instructed to use relational information, while low-criterion subjects showed little or no context effect. A model specifying the relationship between item and relational information and how relational information influences decisions in recognition-memory paradigms was proposed.     

Lexico-semantic structure and the word-frequency effect in recognition memory

The word-frequency effect (WFE) in recognition memory refers to the finding that more rare words are better recognized than more common words. We demonstrate that a familiarity-discrimination model operating on data from a semantic word-association space yields a robust WFE in data on both hit rates and false-alarm rates. Our modeling results suggest that word frequency is encoded in the semantic structure of language, and that this encoding contributes to the WFE observed in item-recognition experiments.     

Recollection-based recognition eliminates the revelation effect in memory

The revelation effect refers to the tendency to call an item on a recognition test old if it is preceded by a cognitive task that involves the processing of a similar stimulus (Watkins &; Peynircioglu, 1990). It has been proposed that the revelation effect occurs because of an increase in the familiarity of the test items in the revelation condition (Luo, 1993; Westerman &; Greene, 1998). In the present experiments, the revelation effect was investigated in recognition tasks that were not based solely on the familiarity of the test items but,also, on a recall-like retrieval process. A revelation effect was not found on an associative-recognition task or on a plurality recognition task. The results of this study show that the revelation effect does not occur when the contribution of familiarity to recognition decisions is reduced by factors that encourage the recollection of the study episode.     

Forgetting motor programmes: Retrieval dynamics in procedural memory

When motor sequences are stored in memory in a categorised manner, selective retrieval of some sequences can induce forgetting of the non-retrieved sequences. We show that such retrieval-induced forgetting (RIF) occurs not only in cued recall but also in a test assessing memory indirectly by providing novel test cues without involving recall of items. Participants learned several sequential finger movements (SFMs), each consisting of the movement of two fingers of either the left or the right hand. Subsequently, they performed retrieval practice on half of the sequences of one hand. A final task then required participants to enter letter dyads. A subset of these dyads corresponded to the previously learned sequences. RIF was present in the response times during the entering of the dyads. The finding of RIF in the slowed-down execution of motor programmes overlapping with initially trained motor sequences suggests that inhibition resolved interference between procedural representations of the acquired motor sequences of one hand during retrieval practice.     

Distinctiveness and the recognition mirror effect: evidence for an item-based criterion placement heuristic

Superior detection and rejection of 1 versus another class of items during recognition is called the mirror effect. Some mirror effects may involve strategic criterion adjustments based on item distinctiveness and its relation to memorability. Three experiments demonstrated mirror effects for known versus unknown scenes and 1 suggested a similar pattern for faces. In opposition to preexperimental familiarity, lures from known and frequently encountered locations were confidently rejected more often than unknown lures. Forgetting and speeding recognition reversed this lure response pattern, suggesting abandonment of strategic adjustment in favor of a single fixed criterion. With sufficient response time and recent encoding, observers demand more evidence for conceptually distinctive items, perhaps because such items typically foster vivid recollection during retrieval.     

A partial matching theory of the mirror effect in immediate probed recognition

A set of experiments on immediate probed recognition of digit triples is reported in which the variables were list length (five, six, seven, or eight triples), the probability that a probe was old (.33, .5, or .67), and whether the digit triples were presented with an auditory component or articulatory suppression. Previous work had suggested that the false alarm (FA) rate in this paradigm was lower when auditory information was available than when it was not; this observation had led to the development of thepartial matching theory of immediate probed recognition, according to which FAs could arise not only as a result of unlucky guesses but also when new probes shared a first digit in common with a partially retained target triple. It was argued that partial memory representations were less likely following auditory presentation than following articulatory suppression. Partial matching theory is contrasted with therational response theory, according to which all FAs are unlucky guesses; partial matching theory gave a better account of the present experimental data than did rational response theory. However, a logical relationship between the two theories was suggested, a consequence of which was that rational response theory could be modified to include partial matching in such a way as to account for mirror effects, not only in unusually difficult immediate probed recognition tasks, but also in the more commonly studied mixed test list paradigm involving words of high or low frequency.     

The long-term recency effect in recognition memory

Three classes of theories explain the recency effect: the modal model, single-store models, and the composite view, which integrates the two positions. None could explain the absence of a long-term recency effect in recognition memory in previous studies. We suggest that prior work did not obtain a recency effect because testing used a multiple-probe rather than a single-probe recognition procedure. Here we tested memory using a single-probe recognition procedure. Experimental conditions included an immediate test, a delayed test after a filled interval, and a continuous-distractor paradigm in which the same filled delay preceded the first word and followed every study word. The long-term recency effect in continuous-distractor recognition was equivalent to the recency effect in immediate recognition. Its absence in the delayed recognition condition demonstrated that it was not attributed to the use of a putative short-term memory store. Single-store models and the composite view can account for this novel finding.     

The mere exposure effect and recognition memory

This investigation explores the relationship between liking ratings and recognition performance for obscure classical and Russian music melodies. Past studies have explored if awareness of stimulus presentation affects the mere exposure effect (MEE) (Bornstein, 1989; Bornstein & D'Agostino, 1992). We investigate if the type of awareness (i.e., "remembering" an actual occurrence of hearing the melody vs. "knowing" that one is familiar with the melody in a more general, less contextualised sense; Tulving, 1985) affects the MEE. In Experiment 1, we administered the liking test and the recognition test within the same test block and found that liking ratings for "remembered" melodies were on average higher than those for melodies that were merely "known". Although the recognition data replicate the findings of Gardiner, Kaminska, Dixon, and Java (1996) whereby remember and know responses to a given stimulus react differently to repetition from one to three trials, liking data did not vary with the degree of exposure. In addition, our adoption of the recognition category "guess" resulted in a pattern of results that is not only different from previous studies but illustrates the importance of judged, instead of true, old/new status in determining liking. The basic findings of Experiment 1 were replicated in Experiment 2 in which the liking test was conducted prior to the recognition test. Implications of these findings for theories of MEE are discussed.