This article was downloaded by: [Carnegie Mellon University] On: 13 January 2015, At: 12:12 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Journal of General Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vgen20

The Role of Sensory Attributes as Retrieval Cues Or as Interference in Retrieval of Information from Episodic Memory Ann Robins Krane

a

a

Mount Saint Vincent University Published online: 06 Jul 2010.

To cite this article: Ann Robins Krane (1979) The Role of Sensory Attributes as Retrieval Cues Or as Interference in Retrieval of Information from Episodic Memory, The Journal of General Psychology, 101:2, 227-247, DOI: 10.1080/00221309.1979.9920077 To link to this article: http://dx.doi.org/10.1080/00221309.1979.9920077

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or

indirectly in connection with, in relation to or arising out of the use of the Content.

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

The Journal of Ceneml Psychology, 1979, 101, 227-247

T H E ROLE O F SENSORY ATTRIBUTES AS RETRIEVAL CUES OR AS INTERFERENCE I N RETRIEVAL O F INFORMATION FROM EPISODIC MEMORY*

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

Mount Saint Vincent University

ANN ROBINSKRANE'

SUMMARY The sensory attributes of to-be-remembered (TBR) events and of distractor stimuli were systematically manipulated (both within and between modalities) in two Brown-Peterson tasks [Experiments 1 (N = 40 male and female university students or faculty) and 2 (N = 22 male and female undergraduates)] and in a delayed recognition task (Experiment 1). Performance on these recall and recognition tasks (in which the semantic attributes of the TBR events were relatively controlled) was consistent with episodic theory. Memory of an event was reasonably completely described in terms of (a) the degree of overlap between the perceptible properties of the retrieval environment and of the memory trace of the event and ( b ) the perceptible similarities between the memory trace of the event and the traces of information encoded temporally adjacent to it. These sensory encoding specificity effects (as well as other research reviewed) were interpreted as demonstrations that a semantic interpretation of the encoding specificity principle cannot account for all encoding specificity phenomena. A.

INTRODUCTION

Tulving's (34, 36) episodic memory theory asserts that when an S is presented with a to-be-remembered (TBR) word or list of words, his experience with that event results in the creation of a unique memory trace in the

* Received in the Editorial Office, Provincetown, Massachusetts, on October 16, 1978. Copyright, 1979, by The Journal Press. I Experiment I is based on portions of a dissertation submitted to the Psychology Department at the University of Rochester, Rochester, New York, in partial fulfillment of the requirements for the Doctoral degree. The author would like to thank the members of her thesis committee, Ralph Norman Haber, James R. Ison, and the chairman, Patricia Siple, for their helpful advice throughout that project. She also thanks Richard V. Krane; the research could not have been completed without his advice and support. Experiment 2 was supported by a grant from the Research Committee of Mount Saint Vincent University to the author. Send reprint requests to the author at the address shown at the end of this article. 227

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

228

JOURNAL OF GENERAL PSYCHOLOGY

episodic memory system. The memory trace is assumed to be unique because it represents the S’s encounter with a particular item in a particular context a t a particular time. Although the linguistic knowledge an S possesses about an item may influence the nature of the trace, the episodic memory trace is conceived as being retained independently of any permanent knowledge used in setting up the trace: i.e., independently of the representation of the item in semantic memory. Thus, it is assumed that the representation of an event in episodic memory can be reasonably completely described in terms of (a) the perceptible, semantic, and autobiographic properties of the event as the S experienced it, and (b) the temporal-spatial relation of the event to other experienced events. T o retrieve a specific event from episodic memory it is necessary to be able to identify the location (temporal date) of the event in memory. In other words, it is necessary to differentiate a specific T B R event from similar temporally adjacent encodings. Cues that assist in accessing an episode are those that accurately specify its location. The cues, also, must have been stored a t that location. Thus, “what is stored is determined by what is perceived and how it is encoded, and what is stored determines what retrieval cues are effective in providing access to what is stored” (37, p. 353). According to Tulving (34) this statement, called the encoding specificity principle, describes all known phenomena of episodic memory. Included among these phenomena are all the traditional paradigms of verbal learning: all short-term memory and long-term memory recognition or recall tasks that involve verbal, imaginal, or pictorial events. 1.

A Semantic Interpretation of Encoding Specificity

The history of the episodic view together with its modus operandi (the encoding specificity principle) has been closely linked to two phenomena: the failure to recognize recallable words when the semantic context of such “forgotten” words has been changed between the recognition and recall attempts (e.g., 35, 37, 38, 39, 40, 41) and the failure of a close semantic associate of a target word to cue recall of the target when the target word has been paired at input with a weak semantic associate (e.g., 32, 37). These two phenomena have been the subjects of considerable debate (e.g., 12, 15, 2 2 , 23, 28, 29, 30). The most widely cited critics, Reder, Anderson, and Bjork (28), offered a semantic interpretation of the encoding specificity principle. They suggested that studying a list of words involves associating information about the occurrence of the word in a list to the representation of the word in semantic memory: such occurrence tags must be accessible a t the time a recall or a recognition attempt is made in order for that attempt to be

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

ANN ROBINS KRANE

229

successful. Most words have several meanings or semantic senses and consequently several representations in semantic memory. Now if the meaning of a word is altered by changing the semantic context of the word between study and test (or between a recall test and a recognition test), then a second semantic memory location will be accessed by a retrieval attempt. No occurrence tags will be found at this new location, and consequently the retrieval attempt fails. In other words, semantic context effects are produced by “the same mechanism that produces lexical disambiguation in connected discourse,” and these phenomena are “a manifestation of an extremely significant aspect of man’s linguistic ability” (1, p. 411). Moreover, as pointed out by Light, Kimble, and Pellegrino (12), demonstrations of semantic encoding specificity may not be sufficient to justify the postulation of a memory system for the temporal and perceptual characteristics of autobiographic experiences that is unique from a system of semantic-conceptual information. 2.

The Case Against the Semantic Interpretation

A number of recognition studies challenge (to some extent) the semantic interpretation of encoding specificity. In the continuous recognition procedure or the probe recognition procedure, using a different type of print for presentation and test (7) or changing the sex of the speaker between occasions (4) or switching from an auditory to a visual presentation format (3, 8, 9) leads to poor recognition performance in comparison with performance when a TBR word reoccurs in the same sensory form. These studies show that the sensory context in which TBR words are presented Gust like the semantic context of these words) will influence the subsequent recognition of the words. A number of theorists (e.g., 1, 14, 26, 27) have stressed that the occurrence (perceptual) encodings and the semantic-conceptual encodings of a TBR item contain different types of information about the TBR item and are involved in separate and independent stages (or mechanisms) of information processing. Thus, two apparently similar phenomena (sensory context effects and semantic context effects) must be related to two different processing stages which presumably use two distinct processing mechanisms. On the other hand, sensory and semantic context effects are equally predicted by the encoding specificity principle and thus the episodic memory theory. The case against the semantic interpretation of encoding specificity would be stronger if we could demonstrate that sensory cues in the absence of nominal copy (semantic) cues could facilitate memory performance: i.e., that

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

230

JOURNAL OF GENERAL PSYCHOLOGY

sensory information can effectively cue recall. A number of studies involving the immediate recall of a word list have shown that recall is enhanced by supplying extralist cues that share sensory features with list items. Homonyms ( 1 1) and rhymes (19, 20) have been used as successful retrieval aids. Moreover, Nelson and Brooks (19) have found that in some instances rhymes (sensory cues) are better extralist retrieval aids than are synonyms (semantic cues). The superiority of rhymes to synonyms as retrieval cues in this study was consistent with the encoding specificity principle. With the short presentation times used (two sec per item), sensory encoding probably had been completed, but such short presentation times probably were not sufficient for elaborative semantic encoding. Possibly, rhyme attributes (but not synonym attributes) were encoded and stored, and thus rhymes (but not synonyms) were effective retrieval cues. A subsequent study (2 1) showed that increasing presentation time or having the cue available with the target during the presentation interval (and thus priming either semantic or sensory encoding) increased the cueing effectiveness of synonyms but not of rhymes. More important, perhaps, is the fact that even when a list was primed for deep semantic encoding, rhymes (sensory attributes) remained effective extralist retrieval aids. Recall cued by rhymes was better than uncued recall. The studies just reviewed may not provide an adequate test of the value of sensory attributes as cues for recall. Possibly processing rhymes or homonyms requires the active involvement of semantic memory. If so (with the use of the terminology of tagging theory), rhymes would not be simply occurrence markers. Instead they would be semantic-conceptual codes rather than perceptual codes. A study by R. Krane and L. Hatton (10) attempts to deal more directly with the issue of sensory attribute cueing. A modified Brown-Peterson paradigm was used in which the first word of a T B R list of five words was presented as a retrieval cue following a 30-sec rehearsal-preventing arithmetic task. The study involved a 2 X 2 X 3 factorial design in which the target list and the arithmetic task were presented in either the auditory or visual modality, and the retrieval cue was presented in either the auditory or visual modality or simply was not presented. The results of that study accord well with episodic memory theory. First, when a target list was presented in the auditory modality, an auditory intralist retrieval cue resulted in better recall than did a visual intralist retrieval cue, and when a target list was presented in the visual modality, then a visual intralist retrieval cue resulted in better recall than did an auditory intralist cue. Second, the occurrence of a distractor task in the same modality as that in which the retrieval cue was to be presented exerted a strong interference effect, and thereby reduced the magnitude of the observed encoding

A N N ROBINS KFUNE

23 1

specificity effects. Although, these data support episodic theory, it is worth noting that the retrieval cue used in this study was the first word in the five-word target list: i.e., the cue contained semantic as well as sensory information.

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

3.

Objectives and Hypotheses

The objective of the present research was to investigate some boundary conditions for sensory encoding specificity phenomena. In particular, can a sensory cue (in the absence of any nominal copy cues) enhance memory performance? And, can sensory context effects be demonstrated for finegrain (within-modality) shifts, as well as for shifts from the auditory to the visual modality or vice versa? Four sensory attributes (red and black visual stimuli and male- or female-voiced auditory stimuli) were used to investigate sensory encoding specificity in a Brown-Peterson recall task (the Recall Phase) and in a subsequent recognition memory task (the Recognition Phase). Briefly, the Recall Phase involved bisensory, dichotic presentation of TBR events following the procedures of Murdock (18). With the use of this procedure, the similarity of the sensory attributes of TBR words, of stimuli presented during the distractor task, and of extralist retrieval cues were systematically varied (within a sensory modality and between modalities) in an attempt to evaluate (a) whether extralist sensory attributes, like intralist (partial nominal copy) attributes (lo), can be used to facilitate recall and (6) whether short-term recall performance varies as a function of the similarity between the sensory properties of TBR items and the sensory properties of temporally adjacent events (distractor stimuli). After a number of short-term recall trials; a subsequent recognition test was administered in which the sensory attributes of the test items again were varied in an attempt to evaluate (a) whether delayed recognition performance varies as a function of the similarity of the presentation forms of the items at test with their form on initial presentation and (6) whether delayed recognition performance varies as a function of the relation of sensory attributes of recognition items to the sensory attributes of events encoded temporally adjacent to the memories of the TBR events (i.e., the sensory attributes of the distractor stimuli during a Brown-Peterson trial). B.

EXPERIMENT 1 1.

Method

a. Subjects. The S s , 14 male and 26 female volunteers, were either faculty or students of Mount Saint Vincent University, Halifax, Nova

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

232

JOURNAL OF GENERAL PSYCHOLOGY

Scotia. All Ss had normal or corrected vision, were not color-blind, and were not impaired in hearing. 6 . Apparatus and stimulus materials. The 256 T B R words, as well as the 128 recognition task distractors, were common, three- to five-letter English words obtained from Thorndike-Lorge's (33) high-frequency list (rated AA to 20). All homonyms were excluded, and care was taken to avoid any obvious semantic or imageable associations among the four TBR words appearing together on a Brown-Peterson trial. For the interference task in the Recall Phase, the stimuli were the letters A, B, C presented in pairs. Ss were required to shadow the letter pairs: i.e., to respond to each stimulus by depressing a particular Lafayette response key labelled with the appropriate letter. Performance on this task was monitored on a bank of three Lafayette (model 5804) cumulative data recorders. All auditory stimuli were recorded with the use of a four-channel Sony tape recorder. During the Recall Phase stimuli were presented dichotically. If the E could detect any asynchrony in the onset of the male and female words of a recall trial, that trial was re-recorded. Four auditow stimulus tapes (including practice, recall, and recognition trials) were prepared to permit the four stimulus presentation forms (male voice, female voice, red type, and black type) to be counterbalanced across four tapes. One half of the Ss that heard each tape had male-voiced words directed to the left ear and female-voiced words directed to the right, while the other Ss had the female-voiced words directed to the left ear and the male-voiced words to the right. Similarly, eight sets of visual materials were prepared to permit the counterbalancing of presentation forms and presentation locations. A11 stimuli to be presented visually were typed on geofilm (Hughes-Owens) and then mounted for display on an El-Mo side projection screen by means of a Kodak 850 carousel projector. The cycling of the projector and the exposure time of each slide were controlled by a sequence of tones recorded, with the use of the four-channel Sony tape recorder, on the right-rear channel of four-channel magnetic tapes. Each tone triggered a voice-activated relay (Lafayette model 18010) that opened a Gerbrands tachistoscopic shutter and thus displayed each slide for 400 msec. The off-pulse from a Gerbrands timer controlling the shutter triggered a Hunter timer that in turn cycled the projector. Synchrony of the auditory and visual presentations was dependent upon the use of the same timing sequence (recorded on the leftrear channel of each four-channel tape) in the production of the other three channels (the visual program channel, the male channel, and the female channel). Ss questioned about the degree of perceptual simultaneity

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

ANN ROBINS KRANE

233

achieved reported that synchronization was fairly good and that there definitely was no consistent asynchrony. c. Design. On each of 64 Brown-Peterson trials, four TBR words were presented; two of the words were presented dichotically (one in the male voice and one in the female voice) and two were presented visually (one in lowercase black type and one in uppercase red type). Immediately following the presentation of the four TBR words on each trial, the Ss were engaged in the rehearsal-preventing shadowing task that might serve as a source of differential interference for the four different TBR words. All stimulus pairs of the interference task for one trial were presented in a single presentation form: in the male voice or the female voice or in uppercase red type or lowercase black type. After five seconds of shadowing, the retrieval cue “recall” was presented in one of the four possible presentation forms. The cue signalled the S to attempt to recall all four TBR words. Thus the 64 Brown-Peterson trials involved 256 TBR words (4 words x 64 trials). These 256 words belonged to one of 64 treatment conditions (obtained by factorially varying the four word forms, the four interference task forms, and the four retrieval cue forms). Moreover, four different treatment conditions were presented on a single trial. To illustrate, a trial involving an interference task presented in black lowercase type and a black retrieval cue would include the four treatment conditions: black TBR word, black interference task, black retrieval cue; red TBR word, black interference task, black retrieval cue; male TBR word, black interference task, black retrieval cue; and female TBR word, black interference task, black retrieval cue. Each of 64 treatment conditions occurred four times in the 64 trials, once in each of four blocks of 16 trials. A block of 16 trials consisted of one trial involving each of the 16 possible interference task-retrieval cue combinations (4 interference task forms x 4 retrieval cue forms), and in each of the four blocks of STM trials, the 16 interference task-retrieval cue combinations were presented in a different random order. The delayed recognition and judgment test began shortly following the last recall trial and involved 384 trials; 256 words from the recall task (old words), and 128 similar distractor words (new words) were used. The 384 words were presented one at a time in one of the four presentation forms, such that one quarter of all words were presented in each form. Moreover, the 256 old words could be classified into 256 conditions based upon the four recognition task forms and the three, four-level factors involved in the recall task (TBR word form, interference task form, and retrieval cue form). The variable of retrieval cue form, however, was not immediately relevant to the

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

234

J O U R N A L O F GENERALPSYCHOLOGY

hypotheses under investigation for recognition performance. Across conditions involving the cue factor, there were four occurrences of each of 64 relevant treatment conditions (obtained by factorially varying recognition form, recall form, and interference task form). Words were presented in a random order on the delayed recognition task with the restrictions that old words that had occurred in the first, second, third, and fourth block of 16 recall trials occurred in the first, second, third, and fourth quarters, respectively, of the recognition test and that distractor words were equally distributed among the four quarters of the recognition task. d . Procedures. The Ss, tested individually, were told that they were participating in a study of human memory and were given a sheet of written instructions to read. The instructions described the bisensory, dichotic presentation of T B R words, outlined the sequence of events on a single Brown-Peterson trial, and informed Ss that in response to the retrieval cue they were to attempt to recall aloud the TBR words from that trial. Finally, Ss were encouraged to attend to the cue “recall,” since this signal might facilitate retrieval of the TBR words. Ss then attempted 10 practice trials to ensure that all instructions were understood and to gain accuracy on the interference tasks. Some Ss found the task to be quite difficult; for these Ss the practice set was repeated until each S gained confidence and proficiency with the procedures. Then the 64 recall trials were presented. Immediately following the last recall trial, the Ss were given a sheet of instructions for the delayed recognition test. In the instructions Ss were informed that the recognition test involved 384 words-256 “old” words and 128 “new” words and that the S’s task now was to rate each word on a six-point confidence scale from “certain word is old” to “certain word is new.” A copy of this rating scale was placed where the S could see it at all times. After any questions that an S had were answered, the 384 recognition trials were initiated. The E recorded all oral responses. The entire data collection process required about 90 minutes per S . 2.

Results and Discussion

a. Brown-Peterson rccnll phase. The 64 recall trials involved 256 T B R words (four words per trial). The 256 TBR words belonged to one of 64 treatment conditions (obtained by factorially varying the four T B R word forms, the four interference task forms, and the four retrieval cue forms). Thus, four treatment conditions occurred on each trial, and all 64 conditions occurred once in each of four blocks of 16 trials. To evaluate the effects of TBR word form (TBR Form), interference task form (IT Form), and re-

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

ANN ROBINS KRANE

235

trieval cue form (Cue Form) on recall, an analysis of variance was performed on the numbers of words recalled over the four occurrences of each of the 64 treatment conditions. In other words, the dependent variable in the Brown-Peterson phase of the study was defined by the number of words (0-4) correctly recalled by each S over the four occurrences of a treatment condition. The phrase, “number of words recalled,” will refer to this variable. These data were analyzed by an analysis of variance involving three, fourlevel, within-Ss factors (4TBR Forms x 4 I T Forms X 4 Cue Forms). This analysis resulted in only two statistically significant effects (a = .Ol):TBR Form x Cue Form [F (9,351)= 5.601 and TBR Form X IT Form [F (9,351)= 27.241.Both of these interactions are relevant to the evaluation of the applicability of episodic memory theory to recall performance. This evaluation involves comparisons among “theoretically relevant” linear combinations of the 64 treatment combinations which were required to counterbalance completely for presentation forms across the three, four-level factors of TBR Form, I T Form, and Cue Form. That is, the hypotheses tested were formulated [as linear combinations following the procedures of Hays (5)]in relation to the degree of similarity between the presentation forms of two variables rather than in terms of specific presentation form combinations. For example, the TBR Form x Cue Form interaction was expected to reflect the encoding specificity principle. Recall that each Brown-Peterson trial involved the presentation of four words in the four different presentation forms, and that in the 64 trials the retrieval cue “recall” was presented in black type on 16 trials, in red type on 16 trials, in a male voice on 16 trials, and in a female voice on 16 trials. Thus, in accord with the presumed characteristics of episodic memory and the encoding specificity principle one might expect that in collapsing over the trials that involved a particular form of retrieval cue (e.g., retrieval cue in black type), the probability of recalling the TBR words that had been presented in the same presentation form as the retrieval cue (e.g., black TBR words) would be greater than it would be for words that had been presented in a presentation form that was similar (in the sense that the same modality was involved) to the form of the retrieval cue (e.g., red TBR words); and that the probability of recalling words that had been presented in a presentation form that was similar to that of the retrieval cue would be greater than it would be for words that had occurred in a presentation form that was diflerent (in the sense that a different modality was involved) from that of the retrieval cue (e.g., male-voiced TBR words and female-voiced TBR words). The expressions TBR X Cue, Same and TBR X Cue, Similar will refer, respectively, to the linear combinations of

236

JOURNAL OF GENERAL PSYCHOLOGY

TABLE I h I E A S S r M B E R OF \VORDS RECAI.I,ED" A S .4 FL1SCTlOS O F T H E SISIILARITY Jlale-male Female-female Mean

I .85 I .85 2.17 2.05 1 .98

1.73 1.74

Similar Blac k-red Rerl-black Male-female Female-male Mean

I .68 1.72 I .83 1.82 I.i6

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

Form

Different Black-male Black-female Red-male Red-female Male-black Male-red Female-black Female-red Mean

I .82 1.90 I .03

1.85 1.99

2.16 2.20 2.2i 2.02

1.47

1.56 1.62 1.61 1 .. 1i

1.72 1.82 I .6i 1.93

2.10 I .98 2.0.1 2.50 2.47 2.48 2.4i 2.24

'' Based on the four occurrenccs of a treatment condition.

means that were formed by combining over those treatment conditions in which the words recalled had been presented in the same form as, or in a form similar to, the form of the retrieval cue. The expression T B R x Cue, Different will refer to the linear combination of means that was formed by combining over those conditions in which the words recalled had been presented in a form different from that of the retrieval cue. The linear combinations of experimental conditions relevant to an interpretation of the T B R Form x Cue Form interaction are presented in Table 1. It is apparent from Table 1 that the obtained outcomes were not in full accord with the expected outcomes. A planned comparison of the mean number of words recalled under condition T B R x Cue, Same = 1.98) and condition T B R x Cue, Similar (TI = 1.76) indicated that the recall of words that had been presented in the same form as the retrieval cue was significantly better than the recall of words that had been presented in a form similar to that of tht: retrieval cue [!(351) = 2.364, p < .01, one-tailed. (All preplanned statistical

(x

A N N ROBINS KRANE

comparisons-i.e., evaluations of episodic theory-will be reported as onetailed values.) However, the finding that the mean number of words recalled under condition TBR x Cue, Different = 2.02) was significantly higher than the number recalled under condition TBR X Cue, Similar [t(351) = 3.08, p < .01] and the finding that recall under condition TBR X Cue, Same did not differ reliably from recall under condition T B R Cue, Different [t C 11, were findings that were not anticipated. The obtained pattern of results suggests that the word “recall” did not serve as an effective retrieval cue, but instead interfered with recall. It also appears that this increment in interference was greater for words that had been presented in a form similar to that of the retrieval cue (i.e., in the same modality, but in another form) than it was for words that had been presented in a form that was either very different from (i.e., in a different modality) or identical with that of the retrieval cue. One thing is certain: the data from the TBR Form x Cue Form interaction of this study do not reflect the encoding specificity principle. Perhaps sensory attributes can affect recall only when they accompany some portion of the semantic encoding of a memory episode [cf Krane and Hatton (lo)]. On the other hand, the interference effects of the postlist cue in this study may reflect a peculiarity of the rather complex bisensory, dichotic presentation procedure used in these two studies. Data supporting the latter hypothesis are provided by Experiment 2 . The analysis of the TBR Form x IT Form interaction was predicted to show that recall of those words that had been presented in a particular form would decrease as the degree of similarity between the form of the TBR words and the form of the IT increased. The expressions TBR X IT, Same; TBR x IT, Similar; and TBR X IT Different will refer, respectively, to the linear combinations of means that were formed by combining over those experimental conditions in which the words recalled had been presented in a form that was the same as, similar to, or different from, the form of the stimuli used in the IT. It was predicted that recall of those words that had been presented in the same form as that in which the IT had been presented (i.e., the mean number of words recalled under condition TBR x IT, Same) would be poorer than the recall of words that had been presented in another form but in the same modality as that in which the IT had been presented (i.e., condition TBR X IT, Similar). Recall of the words that had been presented in a form similar to that of the IT was expected, in turn, to be poorer than the recall of words that had been presented in a modality that was different from that in which the IT had been presented (i.e., condition

(x

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

237

238

J O U R N A L O F GENERAL PSYCHOLOGY

T B R x IT, Different). The data presented in Table 1 appear to be in accord with the predictions. The mean number of words recalled under TBR x IT, Different (% = 2.24) was significantly higher than the mean number of words recalled under T B R X IT, Similar [% = 1.67, L(3Sl) = S . 8 9 , p < .01, one-tailed. The mean number of words recalled under T B R x I T , Similar was greater than that under TBR x IT, Same = 1.62). However, these means were not reliably different (t < 1). It should be noted that unlike the Krane and Hatton study (lo), in the present study no interference task by retrieval cue interaction was found. The absence of this interaction was neither surprising nor contradictory, since the extralist cue used here was not an effective retrieval cue, while the intralist cue of Krane and Hatton was. The effects of the interference tasks in both studies would seem to provide strong support for episodic memory theory. In both studies the degree of similarity in the perceptible properties of the TBR events and the perceptible properties of the temporally adjacent experiences (the stimuli of the interference task) influenced subsequent access to the T B R events. These data are difficult to reconcile with the semantic interpretation of encoding specificity. Within the tagging theory framework there is no apparent reason why the perceptible properties of experiences extraneous to the semantic context of a T B R event should influence subsequent access to the T B R event. 6. Recognition-memory phase. To evaluate the effectiveness of sensory attribute retrieval cues (in this case, the presentation forms of the old words) in aiding delayed recognition of old words, analyses of variance were performed on two different measures of performance-confidence ratings (based upon a six-point rating scale) and numbers of hits (defined as a confidence rating of “old,” that is, a rating of one, two, or three). The fundamental unit of each analysis was based upon the four recognition trials involving one of the 64 treatment conditions obtained by factorially combining the four recognition task presentation forms, the four recall task presentation forms, and the four recall task IT presentation forms. In other words, the confidence ratings analysis of variance was based on the mean confidence rating for each S over the four occurrences of each of the 64 treatment conditions. Both analyses involved three, four-level, within3 factors (four recognition task forms, four recall task forms, and four recall task IT forms). Because the analysis of variance that was performed on the confidence ratings resulted in a pattern of outcomes that was essentially the same as the pattern of outcomes that resulted from the analysis of variance of the numbers of hits in each condition only the results of the latter analysis

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

(x

239

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

ANN ROBINS KRANE

will be reviewed. This analysis revealed only one statistically significant effect (a= .01),the Recall Task Form x Recognition Task Form interaction [F (9,351) = 4.881.This interaction was expected to demonstrate encoding specificity. Recognition rates (hits) were expected to increase as the similarity between the presentation form of a word during the Recall Task Phase of the study and the form of that word in the Recognition Task Phase increased. In other words, experimental conditions involving the presentation of words in the Recognitive Phase in the same form as they had been presented in the Recall Phase (Recall x Recognition, Same) were expected to result in a higher mean number of hits than the conditions that involved the presentation of words in the recognition task in the same modality as that in which they had been presented in the recall task, but in another form (Recall X Recognition, Similar); the conditions involving the presentation of words in the recognition task in a different modality from that in which they had been presented in the recall task (Recall x Recognition, Different) wete expected to produce the lowest mean number of hits. The obtained pattern of results was consistent with the pattern that had been predicted. Unfortunately, however, the mean number of hits obtained under condition Recall x Recognition, Same (5? = 2.68)did not differ reliably from the mean number obtained under condition Recall x Recognition, Similar = 2.54), t(35 1)= 1.44,p > .lo],and the mean of condition Recall x Recognition, Similar did not differ significantly from the mean of condition Recall X Recognition, Different = 2.45,t < 1). A significantly higher mean number of hits was obtained under condition Recall x Recognition, Same than was obtained under condition Recall x Recognition, Different [t (351)= 2.35,p < .01]. The mean number of hits variable, however, does not provide the best possible test of encoding specificity, since that measure included all old items from the recognition test, regardless of whether an item had been “recalled” or not recalled (“forgotten”) in the Brown-Peterson test. According to episodic memory theory, in the process of recalling an item, an S automatically recodes the item in terms of the temporal and semantic context of recall (cf. 2, 13). Thus, in the present study “recalled” words should have had a second memorial representation placed in the episodic memory system. Moreover, this second memory, as a consequence of the conditions of the recall attempt, should have been associated with the sensory attributes of the recalling S’s voice. Thus, at a delayed recognition attempt, “recalled” (and hence recoded) words should have been more easily accessed than “forgotten” (and hence not recoded) words, since two memorial representations of each “recalled” word should have been available in episodic memory. In

[(x

(x

240

J O U R N A L O F GENERAL PSYCHOLOGY

TABLE 2 PROPORTIONS O F THE “RECALLED” A N D “FORGOTTEN” ITEMS THAT WERERECOGNIZED AS OLD AS A FUNCTIONOF THE SIMILARITY BETWEENT H E FORMOF A WORDI N THE RECALLPHASE A N D T H E RECOGNITION PHASE

Downloaded by [Carnegie Mellon University] at 12:12 13 January 2015

Items

Same

Recalled

.7s

Forgotten

.60

Degree of similarity Similar

Different

.7.1 .56

.s2

.74

addition, on a subsequent retrieval attempt “recalled” words (unlike “forgotten” words) may not be differentially affected by the degree to which a sensory attribute retrieval cue was similar to the initial presentation form of the word, since the second memorial representation of the recalled word had been associated with another set of sensory attributes. The data represented in Table 2 provide support for these two suggestions. Let us first consider the proportions of “recalled” words (collapsing across the 40 S s) that were identified as old words in the recognition test as a function of the similarity between the recall and the recognition forms of a word. The proportions for the Recall x Recognition, Same (p = .75), Recall x Recognition, Similar (p = .73), and the Recall x Recognition, Different Cp = .74) conditions were ( a ) relatively high and (b) not reliably different (all x’ < 1). The data for “recalled” words supported episodic memory inasmuch as recognition of the recorded words was not differentially affected by sensory attribute retrieval cues associated with their initial (but not their second) registration in the episodic memory system. O n the other hand, the proportions of words that had not been recalled on the Brown-Peterson test (i.e., the “forgotten” words), but that had been recognized as old in the Recognition Phase, did vary as a function of the similarity between their recall and recognition forms, as is also apparent in Table 2 . These data provide strong support for the encoding specificity principle. Statistically, the proportion of “forgotten” items that were recognized under condition Recall x Recognition, Same (p = .60) was greater than the proportion that were recognized under condition Recall x Recognition, Similar (p = .56), x2 ( 1 ) = 3.64 (which approached the critical value of x’ ( 1 ) = 3 . 8 , p < .05), while the proportion of “forgotten” items recognized as old under condition Recall X Recognition, Similar was greater than the proportion recognized under condition Recall x Recognition. Different (p = , 5 2 ) , xz (1) = 5.74,

p

The role of sensory attributes as retrieval cues or as interference in retrieval of information from episodic memory.

This article was downloaded by: [Carnegie Mellon University] On: 13 January 2015, At: 12:12 Publisher: Routledge Informa Ltd Registered in England and...
1MB Sizes 0 Downloads 0 Views