False From Semantic Associates are Reduced by Item-Method Directed Instructions JANE STOUT

SARAH TAUBER In item-method directed forgetting studies, participants are shown a series of words and given a corresponding instruction to either remember or forget each word. This DANIEL P. C ORTS * method has consistently lead to superior for to-be-remembered (TBR) words relative to to-be-forgotten (TBF) words. The present experiment examined the effects Augustana College of item-method directed forgetting in the context of the Deese, Roediger, and McDermott (DRM) paradigm. Participants were shown lists of words that con- tained both semantically related (DRM) words and unrelated words. False mem- ory for a strong semantic associate occurred significantly more often when they were told to remember DRM words than when they were told to forget DRM words. The results support the notion that directed forgetting effects using this procedure are due to differential and rehearsal of remember and forget items.

irected Forgetting studies typically involve explicit nition test (Basden, Basden, & Gargano, 1993; Block, directions to forget or remember each item in 1971; Timmins, 1974). This suggests that the previ- Da series of words. Two procedures referred to ously studied TBF items are inhibited during retrieval as list-method and item-method directed forgetting have but a recognition test can release this inhibition. been used to measure the effect of these instructions Bjork (1970, 1972) argued that differential on memory. The list-method involves an instruction to rehearsal accounts for directed forgetting in the item- forget an already studied list of words and remember a method through two procedures: (a) the segregation second word list whereas the item-method utilizes a of TBR items from TBF items and then, (b) selective word-by-word memory instruction. Research using these rehearsal for TBR items rather than TBF items due procedures has shown that participants consistently to a word-by-word memory instruction. Past research significantly more words that they were trying to has supported this notion by demonstrating superior remember when compared to words that they had been recall and recognition for TBR items when compared instructed to forget (for reviews, see Bjork, Bjork, & to that of TBF items (e.g., Davis & Okada, 1971; Anderson, 1998; Johnson, 1994). MacLeod, 1975). In essence, TBF words are less acces- Researchers have attempted to explain these sible by all measures. This trend suggests that item- directed forgetting effects by mental processes referred method directed forgetting effects may be due to poor to as retrieval inhibition in the list-method and differ- encoding rather than actually forgetting TBF items. ential rehearsal in the item-method. Bjork et al. (1998) Research has examined factors that can influence posit that retrieval inhibition is the inability to inten- participants’ ability to demonstrate item-method tionally retrieve the memories that are otherwise avail- directed forgetting effects including depth of pro- able to influence behavior. Research has shown that cessing and memory cue position (Horton & Petruk, participants demonstrate superior recall for items on 1980; Wetzel, 1975), memory cue delay time (Bjork the to-be-remembered (TBR) list in comparison to & Geiselman, 1978; Wetzel & Hunt, 1977), and cate- items on the to-be-forgotten (TBF) list, but this effect no longer exists when participants are give a recog- * Faculty supervisor

Winter 2005 ¨ PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH 127 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).

FALSE MEMORIES FROM SEMANTIC ASSOCIATES ¨ Stout, Tauber, and Corts gorical relatedness of words (Wilson, Kipp, & Chapman, uli, the latter refers to the meaning or general theme 2003; Horton & Petruk, 1980). The results of such of incoming stimuli. In the context of the DRM pro- inquiries indicate that manipulating the amount of cedure, false memory for critical lures is based on a and/or type of encoding for TBF words can diminish “gist” representation as a result of the weakening of directed forgetting effects. The fact that more encod- “verbatim” memory for presented words. Brainerd, ing and deeper processing reduce directed forgetting Reyna, Reese, and Kail (2001) proposed an extension effects is consistent with the differential rehearsal to fuzzy trace theory in which two processes called explanation. direct access and reconstruction are involved during retrieval. The direct access process corresponds with False Memories and the DRM Paradigm verbatim traces in which information is read directly Past research has also focused on the effects of from memory. The reconstruction process usually the list-method on false memory creation using the occurs after direct access and is slower and less accu- Deese-Roediger-McDermott (DRM) procedure (Cokely, rate. This is because participants attempt to construct 2003; Kimball & Bjork, 2002; Seamon, Chun, Shulman, themes or meanings from the studied ideas, which Toner, & Caglar, 2002). In order to study false mem- results in memory errors. ory occurrence in , Roediger and McDermott (1995) used a series of word lists each DRM and Directed Forgetting containing fifteen semantically related words. In this Several recent attempts have been made to under- procedure, known as the DRM procedure, words are stand how intentional forgetting might affect false all semantically associated with what is called the crit- recollection from DRM lists. On the surface, it would ical lure. This refers to the word that is most strongly seem that being instructed to forget a list of words associated with the words on the list but is absent from would lead to fewer true and false memories. However, the actual list. For example, items such as bed, snooze, two studies using list-method directed forgetting found doze, rest, etc., are all strongly associated with the crit- just the opposite. Kimball and Bjork (2002) reported ical lure sleep. When participants are tested for mem- that false recall of a critical lure actually increased ory of the DRM lists, a significant proportion under directed forgetting, even as correct memories consistently and falsely recalls and recognizes the crit- decreased. Seamon et al. (2002) reported similar ical lure, thus constituting a false memory. For instance, results, and viewed the outcome as support for a fuzzy- Roediger and McDermott (1995, Experiment 1) found trace approach to the DRM procedure. Both reported that during recognition tests, participants were con- that when inhibiting TBF items, the critical lure fident that they had been shown the critical lure over becomes more available to memory. In contrast, Cokely half of the time (.58), which is comparable to their (2003) attributed a significant portion of the increase confidence in the actual studied words (.75). in false recall to the specific method of presentation One theory that attempts to explain this effect is rather than retrieval inhibition alone. In his experi- the Implicit Activation Hypothesis. Proposed by ments, study time for each word was reduced, and as Underwood (1965), this theory indicates that seman- a result, both true recall of TBF items and false recall tically-related words are activated during the encoding of semantic associates decreased in this study. Thus, process. Nonstudied critical words are activated because Cokely indicated that in some circumstances, directed they have the highest associates and are therefore forgetting instructions could reduce false memories as likely to be falsely remembered later. As an extension well as veridical ones. to this idea, Roediger McDermott, and Robinson While DRM effects interact with presentation for- (1998) argued that an activation/monitoring error mat in list-method directed forgetting, there is little evi- can result in this type of false memory. That is, when dence as to how DRM effects might or might not occur completing a memory test, participants sometimes when cues are presented individually. Because the are unable to distinguish an activated semantic asso- processes for directed forgetting vary according to ciate from previously studied words. Thus, Roediger method, the present research incorporated the DRM et al. extended Underwood’s hypothesis to include procedure into item-method directed forgetting in an active memory search at retrieval, which may con- order to observe the effects of differential rehearsal on fuse internal and external activation for critical lures false memory creation. Because the item-method and lead to . focuses mainly on encoding processes that lead to typ- An alternative explanation lies in Brainerd and ical directed forgetting effects, we were particularly Renya’s (1998) fuzzy trace theory. It states that memory interested in the effects of minimal encoding on the is based on both verbatim and gist memory. While the creation of false memories for semantically related former refers to detailed perceptual memory for stim- words. If the DRM procedure relies on the activation

128 PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH ¨ Winter 2005 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).

FALSE MEMORIES FROM SEMANTIC ASSOCIATES ¨ Stout, Tauber, and Corts

TABLE 1 a different set of 10 unrelated words. Again, the DRM words were TBF in one presentation and TBR in DRM Lists and Critical Lures another. The DRM lists were drawn from lists pub- lished by the 15-item lists developed by Roediger and Sleep Chair McDermott (1995) which reliably produce false mem- Bed Rest Table Sit ories. The five words from each DRM list with the weakest associations to the critical lure were excluded Awake Tired Legs Seat (see Table 1). Dream Wake Desk Recliner Each presentation consisted of two instruction Snooze Blanket Wood Cushion slides followed by 20 slides that each contained one of Doze Slumber Couch Sofa the words. Each of the 20 words was followed by either a red slide (the “forget” cue) or a green slide (the “remember” cue) and word-cue slides were placed in the of semantic meaning for its effect to take place, word- slide show in a randomized order. Word slides and color by-word intentional forgetting instructions should slides were each displayed for one second apiece, based greatly limit this activation. That is, if participants do on McDermott and Watson’s (2001) assertion that false in fact selectively rehearse TBR words but ignore TBF memory output is higher for encoding times that are words, then only the meaning of the TBR words should shorter than 2 seconds per word without substantial be activated and the critical lure should be subject to detriment in veridical recall. This was also the presen- retrieval and recognition. However, if TBF words are tation time for Cokely’s (2003) study of the DRM and simply inhibited—which seems to be the case in at list-method. Once the learning phase of the slide show least two studies, (Kimball & Bjork, 2002; Seamon et was initiated, it took 40 seconds to complete. al., 2002)—participants should falsely recognize the crit- Participants were given one sheet of paper on ical lure though they will be less likely to retrieve it. which they were to record their responses during each In conjunction with Bjork’s (1970, 1972) con- recall test. They were also given two folded pieces of ception of differential rehearsal in the item-method, paper that contained a recognition test for each pre- we hypothesized that participants would selectively sented list, which were labeled so that the experi- rehearse TBR words but not TBF words. We expected menter could instruct participants to open the that this type of encoding would result in the activa- appropriate test at the appropriate time. Each recog- tion of the critical lure when semantically related nition test consisted of the 20 already presented words, words were in the TBR condition, which would be the critical lure that corresponded to the DRM list most noticeable in recognition. The opposite should presented, and 19 distracter words for a total of 40 be true for semantically related words in the TBF con- words per recognition test. The unrelated and dis- dition because a lack of elaborative rehearsal should tracter words were common nouns that were four to make the critical lure less subject to activation and seven letters in length used by Tulving, Schacter, and subsequently, recognition. Stark (1982) and MacLeod (1989).

Method Design Participants Four conditions were created for the experiment Participants were 53 traditional college-aged stu- and all participants were exposed to two conditions: dents (35 women and 18 men, 18-22 years old) at Condition A always included the first DRM list and Augustana College. Students were enrolled in psy- Condition B always included the second DRM list. chology courses and participated for course credit. Within Condition A, the DRM list could be either TBR (Remember-A) or TBF (Forget-A). Similarly, Condition Materials B also included one version with the DRM list TBF Four PowerPoint slide show presentations were (Forget-B) or TBR (Remember-B). Thus, a total of created for the experiment. Two presentations con- four presentations were created so that the first set of tained 10 DRM words with sleep as the critical lure and participants completed the Remember-A and Forget- both included the same set of 10 unrelated words (i.e., B conditions, and the second set of participants com- words within this list were neither semantically nor pleted the Remember-B and Forget-A conditions. phonologically related to each other); in one pres- entation the DRM words were TBR, in the other they Procedure were TBF. The other two presentations contained 10 Stimuli were projected onto a large screen at the DRM words with chair as the critical lure, along with front of the testing room to one group of 24 partici-

Winter 2005 ¨ PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH 129 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).

FALSE MEMORIES FROM SEMANTIC ASSOCIATES ¨ Stout, Tauber, and Corts

TABLE 2 Mean (and Standard Deviation) for the Number of Words Correctly Remembered for a Within-Subject Design (N = 53)

Recall Test Recognition Test Remember Forget Remember Forget 8.00 (.19) DRM 2.28 (.21) Unrelated 9.25 (.15) DRM 6.81 (.32) Unrelated 6.97 (.19) Unrelated 2.38 (.25) DRM 9.00 (.18) Unrelated 5.55 (.33) DRM

pants and one group of 29 participants. In addition ations presented in Table 2). A significant main effect to presenting the instructions for each task on the for Memory Instruction revealed a typical directed first two slides, the experimenter read the instructions forgetting effect across memory tests, F(1, 52) = 408.74, aloud. Participants were told that they would see a p < .01, η2=.88, regardless of whether participants were series of words followed by a slide that was either red instructed to remember DRM words or the unrelated or green. They were told that a red slide indicated words. The main effect for Memory Test revealed a that they should forget the previous word while a green greater number of words recognized than recalled, slide indicated that they should remember the previ- F(1, 52) = 285.12, p < .01, η2=.85. Finally, there was a ous word. main effect for DRM Condition showing that more After the last color slide was presented in the first words were recalled when the DRM list was TBR, memory task, participants were immediately instructed F(1, 52) = 23.93, p < .01, η2=.32. The DRM list to write down all of the words that they could remem- presumably aided correct recall by serving as a sort of ber, even if they remembered some of the words they semantic/organizational . were supposed to forget. They were allowed one minute Of greater interest was the number of critical lures to complete this task. Immediately after the that were falsely remembered in each memory condi- task, participants were given a recognition test with tion. We hypothesized that participants would show a no time limit. The time allotment for free recall and greater amount of false memories when they were recognition were taken from Wetzel’s (1975) meth- instructed to remember DRM words relative to when ods and, in the present study, all participants had time they were instructed to forget DRM words. To measure to record the words that were readily available to mem- this, participants completed two tests (recall and then ory. Participants then underwent a filler task in which recognition) for each condition. Each false memory they were given one minute to mentally complete a that was reported on a recall test was also reported on simple maze. Following the filler, the second-directed the recognition test. The recall tests had expected forgetting task was performed in the exact same man- frequencies of less than five, the minimum expected ner as the first. The entire experiment took less than frequency for a chi-square analysis. The observed criti- 10 minutes to complete. cal lures were approximately evenly distributed across DRM Remember (four critical lures reported) and DRM Results Forget conditions (three critical lures). However, false Each participant’s test form was scored by count- memories were not evenly distributed on recognition ing the number of correctly recalled or recognized tests χ2 (1, N = 44) = 8.20, p = .01. Critical lures were rec- words in each condition as well as whether the critical ognized much more often in the DRM Remember con- lure was recalled or recognized. Only words that had ditions (32 critical lures falsely recognized) than in the been correctly remembered were used in the analy- DRM Forget condition (12 critical lures). sis. All analyses were within-subject designs using an alpha of .05. Discussion We first conducted a 2 (Memory Instruction: In the present study, we observed typical directed Remember/Forget) X 2 (Memory Test: forgetting effects for the item-method using seman- Recall/Recognition) X 2 (DRM Condition: tically related words. Even when TBF words were all Remember/Forget) full factorial analysis of variance semantically related, they were still remembered at a (ANOVA) with the mean correctly remembered words significantly lower rate than TBR words that were unre- as the dependent variable (means and standard devi- lated in meaning. False memory for the critical lure was

130 PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH ¨ Winter 2005 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).

FALSE MEMORIES FROM SEMANTIC ASSOCIATES ¨ Stout, Tauber, and Corts

rare in the case of free recall, but false recognition of forgetting effects and a reduction in false recall from the critical lure was significantly higher when the DRM lists. Although that study used the list method, semantically related words were in the TBR condition Cokely argued that his results were due to a different when compared to the TBF condition. In summary, level of encoding, which was manipulated through the results are consistent with our hypothesis that item- shorter list presentation times. Thus, the present by-item directed forgetting instructions can reduce research seems to be more in line with Cokely’s the DRM effect. research in which the DRM effect is determined largely One factor to consider is that the hypothesis was by the strength or level of encoding. only supported by recognition tests; participants freely Our research suggests that false memories occur recalled the critical lure at the same, low rate in the most reliably when the meaning of an interrelated set DRM Forget conditions. This pattern of results dur- of information is activated. If one is left with an unor- ing retrieval is consistent with Bjork’s assertion that ganized, relatively meaningless amount of informa- encoding occurs at different levels in the Remember tion, this activation does not take place. The and Forget conditions when using the item-method implications for this type of finding are that strongly (1970, 1972). What is more important is that the seman- related information may facilitate memory fabrica- tic activation also seems to be weaker in Forget con- tion but only if this related information is effectively ditions. The implicit activation hypothesis argues that encoded. This suggests that false memories may not the semantic meaning of the DRM words must be acti- only be dependent on one’s intention to remember or vated in order for the false memory effect to occur. forget but may be largely determined by their quality Fuzzy trace theory says that the meanings of the DRM of encoding. words are integrated into a unitary “gist” meaning so Future research would benefit from investigating that when verbatim memory is exhausted, gist mem- factors that affect differential encoding, leading TBF ory will promote memory for the critical lure. In both information is more susceptible to being stored along explanations, the meaning of the semantically related with the TBR information. Also, it would be benefi- words must be activated to some extent in order for the cial to observe the effects of directed forgetting on activation of the strongest semantic associate to occur. other DRM lists to determine the generalizability of In applying this shared idea to the results of our study, our findings within this type of research. This knowl- it seems that the critical lure generally was not acti- edge could shed light on the limits of set differentia- vated when participants were instructed to forget DRM tion capability and it would also elaborate on the effects words. Had the meaning of the TBF DRM words been of intentionally forgetting instructions on the DRM activated and simply inhibited during retrieval, the false memory effect. critical lure should have been recognized more fre- quently or perhaps at the same frequency as that of References the DRM Remember conditions. Since this was not Basden, B. H., Basden, D. R., & Gargano, G. J. (1993). Directed the case, it is assumed that TBF words were not well- forgetting in implicit and tests: A comparison encoded in the first place, which further supports of methods. Journal of Experimental Psychology: Learning, Memory, & Cognition, 19, 603-616. Bjork’s (1970, 1972) notion of differential rehearsal Bjork, E. L., Bjork, R. A., & Anderson, M. C. (1998). Varieties of as an explanation for directed forgetting effects for goal-directed forgetting. In J. M. Holding & C. M. Macleod item-method studies. (Eds.), Intentional forgetting: Interdisciplinary approaches (pp. 103- 137). Mahwah, NJ: Lawrence Erlbaum Associates Inc. It is also important to examine the results in the Bjork, R. A. (1970). Positive forgetting: The non-interference of light of other directed forgetting research which has items intentionally forgotten. Journal of Verbal Learning & Verbal shown the opposite effects of the DRM. Both Kimball Behavior, 9, 255-268. Bjork, R. A. (1972). Theoretical implications of directed forget- and Bjork (2002) and Seamon et al. (2002) report ting. In A. W. Melton & E. Martin (Eds.), Coding processes in that false memories occurred more frequently under human memory. Washington, D.C.: V. H. Winston. directed forgetting conditions than when presented Bjork, R. A., & Geiselman, R. E. (1978). Constituent processes in the with typical “remember” instructions. The most appar- differentiation of items in memory. Journal of Experimental Psychology: Human Learning & Memory, 4, 347-361. ent difference between prior research and the cur- Block, R. A. (1971). Effects of instructions to forget in short-term rent project is the method of presenting forgetting memory. Journal of Experimental Psychology, 89, 1-9. instructions (list method vs. item method). The dif- Brainerd, C. J., & Renya, V. F. (1998). When things that were never experienced are easier to “remember” than things that were. ference in DRM results should not be surprising Psychological Science, 9, 255-268. because list-method is believed to involve inhibitory Brainerd, C. J., Reyna, V. F., Reese, H. W., & Kail, R. (2001). Fuzzy- processes rather than differential rehearsal and encod- trace-theory: Dual processes in memory, reasoning, and cogni- tive neuroscience. In H. W. Reese, & R. Kail (Eds.), Advances in ing (Bjork, 1970, 1972). Conversely, Cokely (2003) child development and behavior (pp. 41-100). San Diego, CA, US: reported the only prior study to find both directed Academic Press.

Winter 2005 ¨ PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH 131 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).

FALSE MEMORIES FROM SEMANTIC ASSOCIATES ¨ Stout, Tauber, and Corts

Cokely, E. T. (2003). Inhibiting false memories: Influences of encoding Roediger, H. L., III, McDermott, K. B., & Robinson, K. J. (1998). and intention. Unpublished master’s thesis, Florida State University, The role of associative processes in creating false memories. In Tallahasee, Florida. Retrieved October 24, 2004, from M. A. Conway, S. E. Gathercole, & C. Cornoldi (Eds.), Theories http://etd.lib.fsu.edu/theses/available/etd11242003210832/ of Memory, Vol. II (pp. 187-245). Hove, UK: Psychology Press. unrestricted/CokelyThesis.pdf Seamon, J. G., Chun, R. L., Shulman, E. P., Toner, S. K., & Caglar Davis, J. C., & Okada, R. (1971). Recognition and recall of positively S. (2002). False memories are hard to inhibit: Differential effects forgotten items. Journal of Experimental Psychology, 89, 181-186. of directed forgetting on accurate and false recall in the DRM Horton, K. D., & Petruk, R. (1980). Set differentiation and depth procedure. Memory, 10, 225-237. of processing in the directed forgetting paradigm. Journal of Timmins, W. K. (1974). Varying processing time in directed for- Experimental Psychology: Human Learning & Memory, 6, 599-610. getting. Journal of Verbal Learning & Verbal Behavior, 13, 539-544. Johnson, H. M. (1994). Processes of successful intentional forget- Tulving, E., Schacter, D. L., & Stark, H. A. (1982). effects ting. Psychological Bulletin, 116, 274-292. in word-fragment completion are independent of recognition Kimball, D. R., & Bjork, R. A. (2002). Influences of intentional and memory. Journal of Experimental Psychology: Learning, Memory, & unintentional forgetting on false memories. Journal of Experimental Cognition, 8, 336-342. Psychology: General, 131, 116-130. Underwood, B. J. (1965). False recognition produced by implicit ver- McDermott, K. B., & Watson, J. M. (2001). The rise and fall of false bal responses. Journal of Experimental Psychology, 70, 122-129. recall: The impact of presentation duration. Journal of Memory Wetzel, C. D. (1975). Effect of orienting tasks and cue timing on & , 45, 160-176. the free call of remember- and forget-cued words. Journal of MacLeod, C. M. (1975). Long-term recognition and recall follow- Experimental Psychology: Human Learning & Memory, 5, 556-566. ing directed forgettting. Journal of Experimental Psychology: Human Wetzel, D. C., & Hunt, R. E. (1977). Cue delay and the role of Learning & Memory, 1, 271-279. rehearsal in directed forgetting. Journal of Experimental Psychology: MacLeod, C. M. (1989). Directed forgetting affects both direct and Human Learning & Memory, 3, 233-245. indirect tests of memory. Journal of Experimental Psychology: Wilson, S. P., Kipp, K., & Chapman, K. (2003). Limits of the retrieval- Learning, Memory, & Cognition, 15, 13-21. inhibition construct: List segregation in directed forgetting. Roediger, H. L., III, & McDermott, K. B. (1995). Creating false mem- Journal of General Psychology, 130, 341-358. ories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory & Cognition, 21, 803-814.

132 PSI CHI JOURNAL OF UNDERGRADUATE RESEARCH ¨ Winter 2005 Copyright 2005 by Psi Chi, The National Honor Society in Psychology (Vol. 10, No. 4, 127–132 / ISSN 1089-4136).