The Effect of Presence and Type of Encoding Cue on

Memory

By: Erica Starr

AbstractO Two experiments conducted to provide continued

evidence for the encoding specificity principle and the levels of processing theory as they relate to memory.

O Whether or not the presence of cues during the encoding process would affect recall for target words, and if so, which types of encoding cues would provide the greatest performance on subsequent retrieval tests.

O Experiment 1: Encoding specificity + cued recallO Experiment 2: Levels of processing + recognition

Important Definitions

O Encoding specificity refers to the idea that successful recall of a previous event depends on the interaction between encoding and retrieval cues.

O The levels of processing theory suggests that information is remembered as a function of how deeply it is processed. Thus, information processed using semantic processing tends to lead to deeper processing than rhyme processing.

Previous ResearchO Common conditions used to test encoding specificity:

O Rhyme-matched condition

O Semantic-matched condition

O Rhyme-semantic mismatched condition

O Semantic-rhyme mismatched condition

O Performance was best when encoding conditions matched retrieval conditions, and was also better for the semantic matched condition compared to the rhyme matched condition.

O For individuals who make great use of the relationship between the encoding and retrieval cues, mismatched conditions are particularly harmful to recall performance.

O However, the encoding specificity principle is often misrepresented to suggest that an increase in the encoding-retrieval match always leads to an increase in recall or recognition performance…

Memory as Discrimination

O Retrieval is not just about picking the correct target, but also about rejecting targets that are incorrect.

O Poirier, Nairne, Morin, Zimmermann, Koutmeridou, & Fowler (2012) challenged the encoding-retrieval match principle

OIncreasing the encoding-retrieval match principle can actually lead to no change, an increase, or a decrease in retrieval performance.

Cue Overload HypothesisO As the number of individual items in memory associated

with a specific cue increases, the effectiveness of that cue decreases.

O The cue overload hypothesis is consistent with a belief that being provided with too many cues can serve as a source of interference for memory.

O This suggests that what determines the probability that a target word will be recalled is the degree to which a single cue is uniquely associated with the target word (Poirier et al., 2012).

Learning to LearnO Can learners make adaptive and qualitative changes in the

way they learn material after experiencing the demands of a subsequent test given in a specific format (Finley & Benjamin, 2011)?

O The test-expectancy paradigm compares performance on a particular test format by participants led to expect that test format versus the performance of individuals led to expect a different format (Finley & Benjamin, 2011). Their study used word pairs and both cued and free recall tests.

O What about more realistic test formats?

Reading Span TasksO Some examples of stimuli from Loaiza et al.’s (2011) task

include:

O The brother of one of your parents is an UNCLE.

O A word made up of five letters is UNCLE.

O The brother of one of your parents is a LETTER.

O Deeper levels of processing led to better immediate recall performance than shallow processing for this task, as well as led to better performance on a delayed recall test.

Reversal of the Levels of Processing Effect

O On study trials, participants saw a question which related to either a semantic or rhyming encoding process before hearing a word.

O They responded yes or no to the question. O They received either the standard recognition test or a

rhyme recognition test, which involved deciding if the test word rhymed with a word presented during study.

O Rhyme recognition was better for the phonological than semantic condition, at least for yes items.

O Recollection depends on the nature of the retrieval task.

Aims of the Current ResearchO Replicate results that show support for both the theories of

encoding specificity and levels of processing.

O Experiment 1 Matched and mismatched encoding and retrieval conditions

O Matched Better recall on cued recall test

O Experiment 2 Different levels of encoding tasksO Semantic Better recognition on recognition test

Experiment 1O Purpose: To determine whether one’s ability to recall a

target word is affected by the presence of a cue at encoding, and if so, how performance is affected when the cue presented at encoding matches or does not match the cue present at retrieval.

O Two phases, Three Cuing Conditions No Cue, Cue A, Cue B

O Responses recorded demonstrated the percentage of times that the target words were recalled correctly for the three cue conditions.

Predictions

O Participants would be more accurate in recalling the target words when the cue at retrieval matched the cue at encoding than when there was a mismatch.

O In relation to the encoding specificity principle, the conditions of No Encoding Cue-No Retrieval Cue and Encoding Cue-Same Retrieval Cue should warrant equal recall performance.

O However, without the context of a meaningful cue, it was predicted that the No Encoding Cue-No Retrieval Cue condition would elicit suboptimal performance compared to that of the Encoding Cue –Same Retrieval cue condition.

MethodO Participants Eighteen undergraduate college students

enrolled in Dr. Weingartner’s Cognitive Psychology Seminar at Hofstra University.

O Materials & Design Fully within participants

O Loosely based on Thomson and Tulving’s 1970 study

O Participants were given written instructions regarding both phases of the experiment prior to the first trial, and the entire experiment took approximately 30 minutes to complete.

Phase 1O Participants saw a list of 45 word pairs

O Each pair shown on the computer screen for 3 seconds

O Word pairs appeared like cue-TARGET

O No Cue/Cue A/Cue B + TARGET WORD

O The designations of Cue A and Cue B were only used to indicate whether participants would see the same cue at encoding as at retrieval. There were no systematic differences between Cue A and Cue B.

Phase 2O 45 trials

O Participants were asked to produce the target word by filling in missing letters when shown either No Cue, Cue A, or Cue B.

O Participants responded using library classroom computers using the Wadsworth Coglab Online Laboratory 2.0 website.

ProcedureO In Phase 1, participants started a trial by pressing the

space bar.

O They then saw a list of 45 word pairs, each for three seconds.

O Ex: cup-DESK or ????-DESK

O No additional responses were required in Phase 1.

Phase 2 ResponsesO Participants were asked to recall the words that were

shown in uppercase in Phase 1.

O Sometimes a cue was given, in which participants either saw Cue A or Cue B. Sometimes no cue was given.

O This could either match or not match the cue given at encoding.

O If participants received a cue, it appeared on the computer screen as cup-D—K.

O Participants used the keyboard to type in the two missing letters that would complete the target word.

O If they could not remember the target, they were

instructed to type in any two letters.

O If no cue was given, the pair would appear something like ????-D—K, and participants still attempted to fill in the missing letters.

O After recording a response, participants pressed the space bar until they advanced through all 45 trials in Phase 2.

Results

OA 3 by 3 within-participants ANOVA with an alpha level of .05 was used to analyze the effect of the presence of cues as well as the effect of matched and unmatched encoding and retrieval cue conditions on recall.

O Retrieval cue did not have a main effect on mean recall, F (2, 34) = 0.04, p > .05.

O The results did reveal a main effect of encoding cue, F (2, 34) = 80.69, p < .05, such that the marginal means for the No Encoding Cue, Encoding Cue A, and Encoding Cue B conditions were M = 29.72, M = 67.41, and M = 70.93 respectively.

O This main effect was qualified by a significant interaction between encoding cue and retrieval cue, F (4, 68) = 5.94, p < .05, and a Bonferroni adjustment was completed prior to performing all t-tests, for which the alpha level used was .01.

O Recall performance for the matched condition of Encoding Cue A-Retrieval Cue A was significantly greater than that of the Encoding Cue A-Retrieval Cue B condition, t (17) = 3.56, p < .01.

O Recall performance for the matched condition of Encoding Cue B-Retrieval Cue B was also significantly greater than that of the Encoding Cue B-Retrieval Cue A condition, t (17) = 3.22, p < .01.

O Mean percent recall for the conditions of Encoding Cue A-Retrieval Cue A and Encoding Cue B-Retrieval Cue B was not statistically different, t (17) = -0.45, p > .01.

O There were also no statistical differences between mean percent recall for the Encoding Cue B-Retrieval Cue A condition compared to that of the Encoding Cue A-Retrieval Cue B condition, t (17) = 1.10, p > .01.

O In addition, mean recall performance for the No Encoding Cue-No Retrieval Cue condition was significantly lower than that for the Encoding Cue A-Retrieval Cue A condition, t (17) = -7.58, p < .01.

DiscussionO Prediction 1 stated that percent recall would be higher

when a cue at retrieval matched a cue at encoding than when the cues did not match.

O As expected, for conditions Encoding Cue A-Retrieval Cue A and Encoding Cue B-Retrieval Cue B, there were no significant differences between the groups, as both these conditions matched encoding cue to retrieval cue.

O There were also no significant differences between the groups who were oppositely mismatched.

O Encoding Cue A-Retrieval Cue A > Encoding Cue A-Retrieval Cue B supporting the encoding specificity principle

O Encoding Cue B-Retrieval Cue B > Encoding Cue B and Retrieval Cue A

O Encoding Cue A-Retrieval Cue A = Encoding Cue B-Retrieval Cue B

O Encoding Cue A-Retrieval A > No Encoding Cue-No Retrieval Cue

Experiment 2O Purpose: Determine whether or not the type of encoding

cue matters as it relates to how well words can be recalled at retrieval.

O Experiment 2 traded the presence of matched conditions at encoding and retrieval for the presence of supposedly stronger and weaker encoding cues preceding this recognition test.

O Experiment 2 was also presented in two phases.

O Phase 1 Participants saw a target word and a judgment task which represented the level of processing to be used to encode the target word.

O The shallow encoding task comparing a consonant-vowel word structure to the target word.

O The medium encoding task determined if the cue rhymed with the target word

O The deep encoding task determined if the cue was similar in meaning to the target word.

O Phase 2 Participants saw a list of words, half of which had appeared in Phase 1.

O Their task was to determine if the word had appeared in Phase 1 or not.

O Manipulated the level of processing induced by the orienting task in Phase 1 to include shallow, medium, and deep encoding processes

O Experiment 2 sought to measure the proportion of times target words were correctly recognized as being in Phase 1 for each encoding condition.

Prediction

O In support of Loaiza et al.’s (2011) study, which found evidence for the levels of processing theory using a reading span task:

O It was predicted that participants would correctly recognize a greater number of target words which were learned using a deep encoding process compared to those learned using a shallow encoding process.

MethodO Participants Twenty undergraduate college students

enrolled in Dr. Weingartner’s Cognitive Psychology Seminar at Hofstra University. One participant’s data was not included in the analysis of this experiment due to experimenter error.

O Materials & Design Modeled after Craik and Tulving’s 1975 study

O Employed a fully within-participants design and an incidental learning task to test recognition of target words.

O Two phases completed one after the other.

O Participants were given written instructions regarding both phases before the start of the experiment.

O Phase 1 60 judgments made

O The judgments were split into blocks of 20 incidental learning tasks in order to cover the three levels of processing. Judgments involved letters, rhyming words, and synonyms.

O Phase 2 120 judgments made regarding whether words shown had appeared in Phase 1.

O Half of the 120 words shown did appear in Phase 1; the other half were new words.

O Participants knew this would occur as per their given instructions, but new and old words appeared in random order.

O The entire experiment took participants approximately 30 minutes to complete.

Procedure

O Participants recorded their own responses on library classroom computers using the Wadsworth Coglab Online Laboratory 2.0 website.

O In Phase 1, participants pressed the space bar to start each trial. The [ z ] key was pressed to indicate a NO response, and the [ / ] key was pressed to indicate a YES response.

O For each trial in Phase 1, two words (or a word and a word structure) appeared on the screen side by side, separated by a word indicating one of three tasks that would be used to compare the words.

O Phase 1:O dog LETTERS cvc YES (/)O dog RHYME boat NO (z)O angry SYNONYM mad YES (/)

O Phase 2:O Was this word in Phase 1? O YES (/) or NO (z)

O Participants pressed the space bar to advance through all 120 trials of Phase 2.

ResultsO The results from a one-way, within participants ANOVA indicated

that the level of processing used at encoding influenced participants’ mean accuracy on the recognition test, F (2, 36) = 16.34, p < .05.

O Mean correct recognition for those words that appeared in the rhyming task of Phase 1 was 9% greater than the correct number of words recognized from the letters task, t (18) = -2.31, p < .05.

O Mean correct recognition for those words which appeared in the synonym task was 16% greater than those words recognized from the rhyming task, t (18) = -3.73, p < .05.

O Mean correct recognition was also 25% higher for words that appeared in the synonym task compared to the letters task, t (18) = -4.88, p < .05.

DiscussionO By manipulating the level of processing induced by the

orienting task in Phase 1 to include shallow, medium, and deep encoding processes, Experiment 2 measured how often target words were correctly recognized as being in Phase 1.

O The type of encoding cue used does matter for retrieval in that deep, semantic cues used at encoding produced much greater recognition than shallow or medium level processing cues.

O As predicted, a significantly greater number of words shown in Phase 2 were recognized that had been encoded using the synonym task in Phase 1 than those that had been encoded with either the rhyming task or the letters task.

General DiscussionO The present experiments were conducted in response to

equivocal research concerning the encoding specificity principle and the levels of processing theory.

O Overall, the aim of the current studies was to provide evidence in support of the encoding specificity principle using a simple matched/non-matched encoding and retrieval condition paradigm.

O In Experiment 1, when encoding cues were presented, recall was better than when no encoding cue was presented, regardless of whether the condition included no encoding cue and no retrieval cue (a matched condition).

O Further support was also found for the levels of processing theory examined in Experiment 2, such that cues encoded during an incidental learning task by shallow, medium, or deep processing affected recall for a target word during a recognition test.

O Deep, semantic cues used at encoding produce much greater recall than shallow or medium level processing cues, such as consonant/vowel word structures and rhyming words.

LimitationsO Ideally, participants would have been naïve to the fact that

they would be receiving a memory test following the study period at the end of both Experiment 1 and Experiment 2 and that the judgment tasks in Experiment 2 were designed to control for using different levels of processing.

O Mulligan & Picklesimer (2011) demonstrated a consistent reversal of the levels of processing effect in favor of non-semantic encoding processes such as rhyming. This implication suggests that the results found presently may have been due to a bias toward promoting enhanced recognition for semantically encoded words.

Future ResearchO Move in the direction of determining the effects of the

encoding specificity principle and the levels of processing theory on performance on realistic tests.

O Use different types of study strategies related to learning the same information that will be tested by means of a short answer test given on paper.

O The study strategies will range from passive reading of material in order to learn it, to actively teaching the material to someone else who is not familiar with it. Intermediate strategies will include listening to information related to the topic and discussing the topic with other people who are somewhat familiar with it.

DesignO Between participants design

O Four possible study conditions:

Read, Listen, Discuss, or Teach

O Study sessions for each condition will last 45 minutes.

O Immediately after studying, participants will be presented with short answer questions related to the original information they were presented with.

O They will have 45 minutes to complete the test.

O Following the test, participants will not receive feedback, but will instead complete a distractor task involving simple math problems.

O Participants will return for a second session one week later consisting of another short answer test in order to assess delayed recall for the studied information.

O The procedure will be the same as the first test session, minus the distractor task.

O Performance on both the immediate test and the delayed test for each of the four conditions will be assessed based on accuracy and average time spent answering each question, which will be determined by participants signaling that they have recorded their final answer for the question.

PredictionsO It is predicted that accuracy of test answers will be highest

for the Teach condition, and that the average time spent answering the questions will be the lowest of all the conditions because the study process is very active.

O The lowest accuracy and longest answer times should occur with the Read condition, which employs a passive study strategy.

O The Discuss group should perform almost as accurately as the Teach condition, but it is predicted they will take longer to answer the questions due to interference from having heard others give their ideas about the study topic.

O **All participants will read the story once prior to engaging in the study period.

O Read Condition: Participants will reread the above story during the allotted study period, as many times as time allows.

 O Listen Condition: Participants will listen to a recording of the story

for the remainder of the study period, as many times as time allows.

O Discuss Condition: Participants will be placed in groups of three to discuss the contents of the story. They can refer to the written piece as necessary for the duration of the study period.

O Teach Condition: Participants will teach the story material to one other person and attempt to hold a discussion after teaching for the allotted study period. They can refer to the written piece as necessary.

ReferencesO Finley, J. R. & Benjamin, A. S. (2012). Adaptive and qualitative changes in encoding

strategy with experience: Evidence from the test-expectancy paradigm. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38(3), 632-652.

O Gallo, D. A., Meadow, N. G., Johnson, E.L., & Foster, K. T. (2008). Deep levels of processing elicit a distinctiveness heuristic: Evidence from the criterial recollection task. Journal of Memory and Language, 58, 1095-1111.

O Jou, J. (2010). Can associative information be strategically separated from item information in word-pair recognition? Psychonomic Bulletin & Review, 17(6), 778-783.

O Loaiza, V. M., McCabe, D. P., Youngblood, J. L., Rose, N. S., & Myerson, J. (2011). The influence of levels of processing on recall from working memory and delayed recall tasks. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37(5), 1258-1263.

O Mulligan, N. W. & Picklesimer, M. (2012). Levels of processing and the cue-dependent nature of recollection. Journal of Memory and Language, 66, 79-92.

O Poirier, M, Nairne, J. S., Morin, C., Zimmerman, F. G. S., Koutmeridou, K., & Fowler, J. (2012). Memory as discrimination: A challenge to the encoding-retrieval match principle. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38(1), 16-29.

O Rose, N. S., Myerson, J., Roediger III, H. L., & Hale, S. (2010). Similarities and differences between working memory and long-term memory: Evidence from the levels-of-processing span task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(2), 471-483.

O Unsworth, N., Brewer, G. A., & Spillers, G. J. (2011). Variation in working memory capacity and episodic memory: Examining the importance of encoding specificity. Psychonomic Bulletin & Review, 18, 1113-1118.