Interfering with remembering and knowing: Effects of divided attention at retrieval

Available online at www.sciencedirect.com Acta Psychologica 127 (2008) 211 221 www.elsevier.com/locate/actpsy Interfering with remembering and knowing: Effects of divided attention at retrieval Erin I. Skinner *, Myra A. Fernandes Department of Psychology, 200 University Ave. W., University of Waterloo, Waterloo, Ontario, Canada N2L 3G1 Received 8 November 2006; received in revised form 17 April 2007; accepted 9 May 2007 Available online 27 June 2007 Abstract A Remember Know paradigm was used to examine the cognitive resource requirements of recollection and familiarity memory processes at retrieval. Younger and older adults studied a list of words, and in a later auditory recognition test indicated whether each word was Remembered, Known, or New. Retrieval was performed under full or divided attention (DA) conditions, with either a digit task to numbers, or an animacy task to words, presented visually. Younger and older adults showed an increase in false Remember responses during both DA conditions, indicating a general effect of attention on illusory recollection. Both age groups also showed decreased accuracy in Know responses, but only during the word-based DA condition, indicating a material-specific effect on familiarity. Aging was associated with decreased accuracy in Remember, but not Know, responses, and with increased latency in distracting task responses under DA conditions. Results suggest that avoiding false recollective responses during retrieval requires attentional resources, whereas accurate familiarity responses require the reactivation of content-specific representations. Ó 2007 Elsevier B.V. All rights reserved. PsycINFO: 2343 Keywords: Memory; Divided attention; Recollection; Familiarity; Age 1. Introduction According to dual process models, there are two ways in which information can be recognized, referred to as Recollection and Familiarity (Gardiner, 1988; Jacoby, 1991; Mandler, 1980). Recollection refers to the effortful retrieval of detailed (e.g., contextual) information about individual personal episodes, whereas familiarity is thought of as an unspecific sense of having previously encountered a given event. The literature describes recollection as a more controlled and analytic process than familiarity (Jacoby, 1991; Kelley & Jacoby, 1998). For example, shallower levels of processing (Gregg & Gardiner, 1994; Rajaram, 1993) and divided attention at study (Gardiner, * Corresponding author. Tel.: +1 519 888 4567x37776; fax: +1 519 746 8631. E-mail address: eiskinne@watarts.uwaterloo.ca (E.I. Skinner). Gregg, Mashru, & Thaman, 2001; Gardiner & Parkin, 1990; Yonelinas, 2001) decrease recollection to a greater extent than familiarity. These findings have led researchers to conceptualize recollection as a more attention-demanding process than familiarity. In contrast, familiarity is generally described as an increase in an item s processing fluency (Johnston, Dark, & Jacoby, 1985; Kelley & Jacoby, 1998), or quantitative memory strength (Yonelinas, 1994). For example, changing the perceptual characteristics of word stimuli at test decreases familiarity-based processing while leaving recollection unaffected (Rajaram, 1993; Rajaram & Geraci, 2000). In the present study we used the divided attention (DA) technique to further characterize the cognitive resources required to perform recollection- and familiarity-based memory recognition. We tested the hypothesis that recollection requires attentional resources during retrieval, whereas familiarity depends on resources involved in 0001-6918/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.actpsy.2007.05.001

212 E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 accessing content-specific memory representations. We examined how performance on a Remember Know memory test 1 was affected, in younger and older adults, by two different DA conditions at retrieval. While previous work has considered the role of attention during encoding, no study to date has examined how divided attention at retrieval affects recollection and familiarity in a Remember Know paradigm. The DA technique has been widely used to measure the resources and component-processes required to perform a cognitive task. With respect to memory, previous research shows that when attention is divided during encoding, subsequent memory performance is largely disrupted. Dividing attention during retrieval, in contrast, has only a small, or non-significant effect on memory (Baddeley, Lewis, Eldridge, & Thompson, 1984; Craik, Govoni, Naveh-Benjamin, & Anderson, 1996), leading some researchers to suggest that encoding is a more attention-demanding process than retrieval (Naveh-Benjamin, Craik, Guez, & Dori, 1998). Nonetheless, other work has shown that, under certain circumstances, DA at retrieval can produce large disruptions to memory. These exceptions shed light on the components of memory retrieval that can be affected by DA conditions during retrieval. The first exception is when the memory test requires controlled processing at retrieval, such as organizational or source memory judgments. Large deficits in memory from DA are observed when the test requires recall from categorized word lists (Moscovitch, 1994; Park, Smith, Dudley, & Lafronza, 1989), list discriminations (Jacoby, 1991), or involves release from proactive inhibition (Moscovitch, 1989, 1994). These studies indicate that attentional resources are required during retrieval when the memory test can benefit from organizational strategies, or involves re-creation of the contextual information tied to item memory. Hicks and Marsh (2000) similarly argued that retrieval will be disrupted by DA only if control processes are used during retrieval. They found that memory for words read intact during encoding were unaffected by DA at retrieval, but the hit rate for words studied as anagrams decreased under DA conditions. The authors argued that the words read intact were more shallowly processed, and consequently, retrieved by relatively automatic familiarity processes unaffected by DA at retrieval. In contrast, words generated from anagrams were believed to be more deeply processed at encoding and subsequently retrieved using more recollective-based recognition processes, making the 1 The Remember Know paradigm was developed in order to study recollective and familiarity-based memory processes (Tulving, 1983, 1985). During a recognition memory test, participants are asked to make a Remember (R) response if they recollect specific information about the item from the study phase, a Know (K) response if the item is familiar in the absence of a specific recollection of the study episode, or a New (N) response if the item is not believed to be from the study list. In general, R responses are believed to reflect recollective memory processes, whereas K responses align more with familiarity-based recognition processes (Yonelinas, 2001; Yonelinas & Jacoby, 1995). retrieval of these items more susceptible to the effects of DA. However, since the authors did not directly test recollection and familiarity during their memory test, it is difficult to determine whether the effect of DA on words generated from anagrams occurred because of a change in recollection per se, or because of a change in some other retrieval mechanism, such as strategy-use. In addition, Hicks and Marsh found that false alarm rate increased under DA conditions, raising the possibility that DA may affect control processes used to monitor the validity of the contents of memory. It is possible that the increase in false alarms occurred selectively for items that made use of recollective-based processing (anagram word recognition); however, since both generate and read items were tested in the same recognition test, this hypothesis could not be specifically examined. Gardiner, Gregg, and Karayianni (2006) also considered how conscious awareness, or available resources, at encoding and retrieval affect recollection and familiarity. They examined how Remember and Know responses were influenced by perceptual effects of study test congruence when attention was divided during encoding, and whether these were influenced by a speeded responding manipulation (intended to reduce available conscious resources) during retrieval. They found that the perceptual effects in remembering and knowing depended more on available conscious resources at encoding than retrieval. However, other studies suggest that conscious resources are also critical for recollection at retrieval, as recollective responding declines under speeded response conditions, though familiarity is unaffected (Benjamin & Craik, 2001; Toth, 1996; Yonelinas & Jacoby, 1994, 1995). These later findings correspond with those from process-dissociation studies, which show that DA at retrieval affects recollection more than familiarity (Dodson & Johnson, 1996; Gruppuso, Lindsay, & Kelley, 1997; Jacoby, 1991; Mulligan & Hirshman, 1997). In the current study, we wished to determine whether DA at retrieval would disrupt recollective, and not familiarity, processing measured by a Remember Know paradigm, and we performed separate analyses on hits, false alarms, and corrected recognition to examine the effects of DA at retrieval on both veridical and false recognition. The second exception in which DA at retrieval disrupts memory performance is when the memory and distracting task use similar material. Fernandes and Moscovitch (2000, 2002, 2003) found that recall of a list of unrelated words was disrupted when participants concurrently performed a word-based distracting task, but not when they performed a digit- or picture-based distracting task. They suggest that the locus of memory interference in their studies lies in reactivation of content representations of the item memory, rather than competition for general attentional resources. In the present study, we tested whether this materialspecific interference effect would act selectively on familiarity-based processing. It has been suggested that familiarity involves responding to the overall similarity of an item

E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 213 across study and test condition, in a global matching process (Gillund & Shiffrin, 1984; Hintzman & Curran, 1994), and work has shown that knowing is more sensitive to variables that affect the perceptual processing fluency of items, than is remembering. For example, Rajaram (1993) showed that presentation of a test word, preceded by an identical masked test prime, affected knowing and not remembering, and Dewhurst and Hitch (1997) found that performing an auditory lexical decision task between the study and test phase of an auditory recognition memory task had a detrimental effect on accuracy of knowing, when lures in the recognition test were created from the non-words in the preceding lexical decision task. These studies suggest that knowing is particularly sensitive to manipulations that interfere with re-creation of the representational content of memory. This led us to predict that manipulations of attention at retrieval could disrupt familiarity, but only when the distracter task uses similar material as the memory task, or when there is competition for a common representational system during dual-task conditions. The cognitive requirements of recollection and familiarity have also been examined through work with older adults. It has been suggested that older adults have less effective attentional processes, which results in poorer detection, discrimination, and selection of task-relevant stimuli, and increased susceptibility to interference (Hasher & Zacks, 1988; McDowd & Shaw, 2000). Psychologists often characterize the aging process as a loss of available general attentional resources as a consequence of changes in brain structures, especially in the frontal lobes, with advancing age. Older adults often show both a decrease in the amount of recollective memories they experience and a tendency to report higher false recollective memories (Bunce, 2003; Perfect & Dasgupta, 1997; Prull, Dawes, Martin, Rosenberg, & Light, 2006). In contrast, familiarity shows either no or a less pronounced decrease with advancing age (Norman & Schacter, 1997; Perfect, Williams, & Anderton-Brown, 1995). As recollection is believed to be a more attention-demanding process than familiarity, the decline in recollection noted with advancing age has been linked to this population s reduced levels of efficient processing on attention-demanding tasks (Parkin & Walker, 1992; Davidson & Glisky, 2002). In comparison, familiarity is believed to involve more representational aspects of cognitive function, which are relatively preserved with aging (Park, 2000; Salthouse, 1991; Wingfield & Stine- Morrow, 2000). In the current study we examined how reduced levels of attentional resources, brought on by an experimental manipulation (DA), and by a participant variable (aging), affect recollection and familiarity. We also tested whether the material-specific interference effect from DA at retrieval, reported by Fernandes and Moscovitch (2000, 2002, 2003), would act selectively on familiarity-based processing. Younger and older adults performed a Remember Know memory test under full attention (FA) or two DA conditions, with either a number- or word-based distracter task. In line with previous work (Fernandes & Moscovitch, 2000, 2002), we expected overall recognition performance to be disrupted when retrieval was performed concurrently with a word-based distracting task (DA-word) but not with a digit-based distracting task (DA-digit), and that the magnitude of this disruption would be similar in younger and older adults (Fernandes & Moscovitch, 2003). Also, in line with previous work (Anderson, Craik, & Naveh-Benjamin, 1998; Naveh-Benjamin et al., 1998), we expected that costs to distracter task performance, during both DA conditions, would be greater in older, as compared to younger, adults. To the extent that recollection requires attentional resources, R responses should be reduced during either of our DA conditions, as both distracting tasks would reduce the amount of available attentional resources to devote to the task. We refer to this as a general effect of DA on recollection. In addition, since older adults have fewer available attentional resources than younger adults, this disruption should be greater in older adults, as their limited resources are further reduced by the DA conditions. We also tested the hypothesis that familiarity would be particularly disrupted in the DA-word condition, as this distracting task would interfere with re-creation of the representational content of memory (Fernandes & Moscovitch, 2000, 2002, 2003), critical to the familiarity process. We refer to this as a material-specific effect of DA on familiarity. Since representational aspects of cognitive function are preserved with aging, the magnitude of disruption to familiarity should be similar in younger and older adults. 2. Method 2.1. Participants Sixty people took part in the experiment. Thirty healthy undergraduate students from the University of Waterloo received course credit and 30 older adults recruited from the Waterloo Research Aging Pool (WRAP) at the University of Waterloo received token monetary remuneration for participating in the study. The WRAP pool is a database of healthy seniors in the Kitchener Waterloo area recruited by means of newspaper ads, flyers in community centers, and through local television segments featuring research at the University of Waterloo. The mean age was 20.20 (SD = 2.19) for the younger adults and 72.60 (SD = 7.05) for the older adults. All participants were fluent English speakers, and had normal or corrected-to-normal hearing and vision. The mean number of years of education was 15.30 (SD = 2.04) and 14.65 (SD = 2.14) for the younger and older adult groups, respectively, which did not significantly differ. The National Adult Reading Test Revised (NART-R) was also administered to allow an estimate of Full Scale IQ (FSIQ), based on number of errors in pronunciation during vocabulary reading (Blair & Spreen, 1989; Nelson, 1982). Younger and older adults had mean FSIQ estimates of 108.52 (SD = 6.85) and 114.80

214 E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 (SD = 6.54), respectively, which differed significantly, t(58) = 3.62, p <.05. The Trail-Making test was also administered to establish an estimate of executive functioning (Reitan & Wolfson, 1985), believed to be reduced in the older adult group. The mean time to complete Trails A was 19.50 s (SD = 5.14) and 33.19 s (SD = 8.86) in younger and older adults, respectively, which differed significantly, t(58) = 7.32, p <.001. Time to complete Trails B was 33.43 s (SD = 10.75) and 67.79 s (SD = 20.29) for younger and older adults, which also differed significantly, t(58) = 8.20, p <.001. These means are in the normal range for each age group (Spreen & Strauss, 1998), and are in line with the expected deficit in executive functioning in older adults. Thus, we consider our sample of older adults to be representative of the general aging population. In addition, older adults were administered the Mini-Mental State Exam (MMSE; Folstein, Folstein, & McHugh, 1975) to screen for gross neurological conditions. All had MMSE scores of greater than 27/30 (M = 28.73, SD = 1.26), indicating that they were free from major cognitive and neurological impairments. 2.2. Materials Word stimuli for the recognition and word-based distracting tasks were medium to high frequency words chosen from Celex, a lexical database available on CD-ROM (Baayen, Piepenbrock, & Gulikers, 1995). Words were spoken by author MF in a sound proof booth and recorded into a.wav file using SoundDesigner II software (Palo Alto, California). For the memory task, three study lists were created by randomly choosing 50 words for each list, from a pool of 470 unrelated common nouns. Recognition test lists were then created by randomly choosing 30 words from the study list and an additional 20 words (to serve as lures) from the pool. The three study and corresponding test lists were equated on letter length (M = 5.69, SD = 1.67), utterance length (M =.89, SD =.19), and word frequency (M = 67.44, SD = 113.04), and were counter-balanced across conditions and participants. A 50- and 100-word list was created for the word-based distracter in the single- and DA-word conditions, respectively (see Procedure). Lists consisted of words with a mean of six letters (SD = 1.69), representing living (e.g., kitten) and man-made (e.g., pencil) objects. Another 50-word list was created for use in the auditory continuous reaction time (CRT) task (see Procedure). The three word lists were equated on word frequency (M = 36.31, SD = 63.51), and each list was created to contain half living and half manmade objects. Fifty two-digit numbers for the digit-based distracter task in the single-digit condition and 100 two-digit numbers for the DA-digit condition (see Procedure) were used. Each two-digit number was presented flanked by two X s on either side (e.g., XX47XX), so that the visual display consisted of six items, as in the word-based distracting task. Participants were required to add the two digits together, and make a key-press if the two numbers added to a number greater than 10. The test lists were made such that half of the digit combinations required a response. 2 An additional 50-item list was created for use during the CRT task. 2.3. Procedure Stimulus presentation and response recording were controlled by an IBM PC, using E-prime v.1.1 software (Psychology Software Tools Inc., Pittsburg, PA). Participants were tested individually, and completed the experiment in approximately one hour. All participants began by performing the NART-revised and Trail Making Test, and older adults were also administered the MMSE. Participants were then given a short practice block (10 20 items per condition) of all experimental tasks. During the practice session, participants were instructed with respect to how to properly report Remember, Know, and New responses. 3 Following practice, the five task conditions were administered, with presentation order counter-balanced across participants according to a Latin-square design. For the digit- and word-based distracting tasks, participants indicated, via a key-press, every time the two digits added to a number greater than 10, or a word representing a man-made object, appeared on the computer screen, depending on the distracting task. Participants performed 2 We conducted an earlier experiment in which the two distracting tasks were (1) the same word-based distracter task described in this study, and (2) odd-digit identification of numbers, with 30 younger and 30 older adults. The pattern of performance was identical to that reported in the current study. We chose to report results using the digit-addition distracter task because this is a more complex task than the odd-digit-identification one, requiring a transformation of the items prior to responding. We reasoned that such a task is more analogous to the word-based distracter task in which participants need to first read the word, then access its semantic meaning, prior to making a response. 3 Participants were told that they would hear some new and some old words, and were asked to make one of these three responses. Participants were instructed to say N for New if they believed that the word was not from the study list. If they thought the word was from the study list, however, they had two options, R or K. They were told to report R for Remember if the word was old and they could recall specific details associating that word with the study episode. They were given examples of such details: they may remember how the word sounded, the temporal order, or an image, thought, or feeling they had associated with the word during study. These contextual details meant they had a specific recollection of that word. If however, the word was old but did not have these specific associations with the study episode, they were asked to say K for Know. They were also given the example of a K memory to clarify this type of memory response: meeting someone on the street that they knew they had met before, but not being able to determine the specific instance in which they had met them. They were told that this same kind of experience can happen for the words they may know that they had heard the word previously, but not recall the specific instance in which they had heard it. Participants were then asked if they understood the distinction between Remember and Know responses and, after the practice session, participants were asked to give the details of the memory accompanying a Remember and Know response, in order to ensure that they understood the difference between the responses, and were not responding on the basis of response confidence.

E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 215 each task alone (single-digit or single-word conditions) and with the memory task (DA-digit or DA-word conditions), with items presented at a rate of one every 2 s for younger, and one every 2.5 s for older adults. 4 During the study phase for each memory condition, participants were told to listen to a list of words and to try to memorize these for a later memory test; hence encoding was always performed with full attention. Study lists were presented auditorily via computer speakers at a rate of one word every two seconds, and the volume of word presentation was adjusted during the practice phase of the experiment such that participant could hear the words clearly without straining their hearing. After study, participants counted backwards by threes from a number presented visually on the computer screen for 30 s, in order to eliminate recency effects (as in Craik et al., 1996). During the retrieval phase, participants heard words auditorily, and were asked to make one of three verbal responses: (1) say R for Remember, (2) say K for Know, or (3) say N for New. Responses were recorded manually by the experimenter. Younger adults were given 4 s and older adults 5 s to make a response to each word in the recognition test. 4 For the DA conditions, two distracting task items were presented for each word presented in the recognition task, with onset of the first distracting task item simultaneous with that for the recognition task word, thus requiring participants to make two digit or animacy decisions for every memory response. The importance of placing 50% of their effort on the recognition task and 50% of their effort on the distracting task was emphasized prior to each DA condition. 2.4. Comparing difficulty of the distracting tasks 4 Since age-related slowing of processing speed might affect participant s ability to respond to the memory and distracting tasks, we adjusted the presentation rates for the auditory recognition and distracting task in younger and older adults based on pilot testing in three younger and four older adults. Pilot data from two older and three younger subjects showed that when participants were required to make a recognition decision to words heard every 4 s, and distracting task items every 2 s, overall recognition accuracy under FA conditions was 42% in older adults, as compared to 60% in the younger adults. We tested an additional two older adult pilot participants, who heard a word every 5 s, and distracting task items presented every 2.5 s, and found that accuracy in the FA condition increased to 57%, roughly equivalent to the younger adults, thus in our study we used these longer presentation durations for the older adult group. If the digit and word distracting tasks differ with respect to resource requirements, this could contribute to any differences observed in memory interference from each DA condition. In order to compare resource demands of each distracter task, we had participants perform the wordand digit-based task concurrently with an auditory continuous reaction time (CRT) task (as in Fernandes & Moscovitch, 2000). Participants were instructed to identify computer-generated tones either alone (single-task condition) or in combination with the digit- or word-based distracting task (dual-task conditions) for 100 s, with task order counterbalanced. For both distracting tasks, items were presented once every 2 s. This procedure was conducted only in the young adult sample, as seniors have extreme difficulty discriminating tones due to normal sensory loss with increasing age (Ostoff, McDonald, Schneider, & Alain, 2003). The response time (RT) and number of correct responses on the auditory CRT task were recorded and analyzed as a means of gauging how demanding each distracting task was, with longer RTs indicating greater resource demands. 3. Results 3.1. Memory task performance Our results are separated into three sections. We analyzed overall recognition, Remember, and Know responses separately. Since there is evidence that hits and false alarms may show different effects from DA and age (Hicks & Marsh, 2000; Perfect et al., 1995), we analyzed hit rate (out of 30), false alarm rate (out of 20), and recognition accuracy (hit rate false alarm rate) for each response separately. Data were always analyzed in a three (attention) 2 (age group) 5 (task order) analysis of variance (ANOVA), with the first variable being within participants and the other variables being between participant manipulations. Means for each response, condition, and age group are presented in Table 1. 3.1.1. Overall recognition The pattern of results was similar for overall hits, overall false alarms, and overall accuracy, thus only statistics relating to the last measure are reported. There was a main effect of attention, F(2, 100) = 6.93, MSE =.18, g 2 =.12, p <.005, with poorer overall recognition in the DA-word than in both the FA, F(1, 50) = 11.91, MSE =.70, g 2 =.19, p =.001, and DA-digit conditions, F(1, 50) = 4.08, MSE =.24, g 2 =.08, p <.05. Recognition in the FA and DA-digit conditions did not differ significantly, F(1,50) = 3.32. There was a main effect of age group, F(1, 50) = 13.44, MSE =.39, g 2 =.21, p <.001, with older showing poorer accuracy than younger adults. There was no attention age group interaction. 3.1.2. Remember responses Analysis of the R hit rate and Recollection accuracy showed no effects of attention or age group, and no interaction. The R false alarm rate analysis showed a main effect of age group, F(1, 50) = 18.62, MSE =.22, g 2 =.27, p <.001, with older adults producing more R false alarms than younger adults. There was also a main effect of attention, F(2, 100) = 5.48, MSE =.05, g 2 =.16, p <.02, with fewer R false alarms in the FA than in the DA-digit, F(1, 50) = 5.13, MSE =.07, g 2 =.09, p <.05, and DAword conditions, F(1,50) = 8.54, MSE =.20, g 2 =.14,

216 E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 Table 1 Mean Remember and Know performance measures for younger and older adults (standard deviation in parentheses) Measure Response and condition Response Full attention Divided attention digit Divided attention word Younger Older Younger Older Younger Older Overall accuracy (hit rate false alarm rate).62 (.19).46 (.24) 59 (.24).41 (.19).49 (.29) 36 (.21) Hit rate Remember.34 (.22).38 (.29).35 (.27).42 (.33).35 (.27).41 (.32) Know.37 (.20).32 (.29).38 (.22).31 (.28).31 (.21).27 (.28) False alarm rate Remember.00 (.01).11 (.13).02 (.03).16 (.18).06 (.14).17 (.19) Know.08 (.09).14 (.13).12 (.10).16 (.17).11 (.09).15 (.16) Accuracy (hit rate false alarm rate) Remember.34 (.22).27 (.27).33 (.27) 26 (.26) 29 (.30).24 (.27) Know.28 (.20).18 (.29).27 (.21).16 (.17).20 (.18).12 (.21) Independence model measures Recollection.34 (.22).27 (.27).33 (.27).26 (.26).29 (.30).24 (.27) Familiarity.41 (.27).17 (.35).41 (.28).19 (.20).30 (.27).12 (.24) p =.005. R false alarms in the DA-digit and DA-word conditions did not differ, F(1, 50) = 1.81 (see Table 1). 3.1.3. Know responses The analysis of K accuracy showed a main effect of attention, F(2,100) = 5.64, MSE =.08, g 2 =.10, p =.005, with poorer performance in the DA-word than in the FA, F(1,50) = 10.15, MSE =.32, g 2 =.17, p =.002, and DA-digit conditions, F(1,50) = 4.75, MSE =.15, g 2 =.09, p <.05, though the latter two conditions did not differ significantly. There was no effect of age group, nor an age group attention interaction. The analysis for K hit rate and independence models of familiarity showed similar results, except that the latter showed an additional main effect of age group, F(1, 50) = 13.62, MSE =.59, g 2 =.21, p =.001. Analysis of the K false alarm rate showed no effect of age group or attention, and no interactions. We remind the reader that the same pattern of interference reported above was found in an initial experiment 2 using the word-based and a digit-identification distracter task, showing a general effect of attention on recollection and a material-specific effect on familiarity, demonstrating the replicability of our results. 3.2. Distracter task performance Accuracy and response time (RT) for correct responses in each distracter task were analyzed in a two (distracter task) 2 (attention) 2 (age group) 5 (task order) ANOVA. The accuracy analysis showed a main effect of distracter task, F(1,50) = 20.82, MSE =.23, g 2 =.29, p <.001, a main effect of attention, F(1,50) = 113.37, MSE = 1.52, g 2 =.69, p <.001, a distracter task age group interaction, F(1,50) = 8.71, MSE =.10, g 2 =.25, p =.005, and a distracter task attention interaction, F(1, 50) = 13.55, MSE =.13, g 2 =.21, p =.001 (see Table 2 for means). However, this was all accompanied by a three way distracter task attention age group interaction, F(1, 50) = 4.5, MSE =.44, g 2 =.08, p <.05. Planned comparisons showed that although accuracy was lower in the word-based than digit-based distracter task under dualtask conditions, t(59) = 2.43, accuracy on the distracter tasks did not differ under single-task conditions, t(58) = 1.49. Planned comparisons also showed that there were no age differences in accuracy for the single-digit, t(58) =.40, single-word, t(58) = 1.19, or dual-word, t(58) =.57, conditions, but older adults showed significantly lower accuracy during the dual-digit condition than younger adults, t(58) = 3.24. The RT analysis showed a main effect of age group, F(1,50) = 400.12, MSE = 30447997.44, g 2 =.89, p <.001, a main effect of attention, F(1, 50) = 1264.40, MSE = 36122280.55, g 2 =.96, p <.001, and an age group attention interaction, F(1,50) = 668.78, MSE = 19106041.19, g 2 =.93, p <.001, such that older adults had slower RTs than younger adults, and this difference was greater during dual-task than single-task conditions (see Table 2). Importantly, we did not find a main effect of distracter task, nor any interactions with this factor. 3.3. Distracting task performed concurrently with the auditory CRT task In order to examine the resources necessary to perform the distracting tasks, we examined CRT task performance alone, and in combination with each of the distracting tasks, in our young adult sample. The mean RT to identify correct tones and the number of tones correctly identified was analyzed using a three (attention) 2 (task order) ANOVA. Data from one participant were lost due to computer failure. Mean RT for the single CRT, dual-digit CRT, and dual-word CRT conditions was 724 (SD = 139), 1110 (SD = 220), and 1140 (SD = 260), respectively. There was a main effect of attention, F(2,54) = 73.48, MSE = 1,545,808, g 2 =.73, p <.001, with faster RTs in the single CRT compared to the dual-digit, F(1,27) = 109.38, MSE = 4265987.85, g 2 =.80, p <.001, and dual-word CRT conditions, F(1,27) = 101.43, MSE = 4981167.95, g 2 =.79, p <.001. Importantly, RT

E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 217 Table 2 Mean accuracy and response time (in ms) for each distracter task, under single- and dual-task conditions for younger and older adults (standard deviation in parentheses) Measure Condition Single-digit DA-digit Single-word DA-word Younger Older Younger Older Younger Older Younger Older Accuracy.96 (.04).96 (.04).90 (.08).79 (.17).93 (.07).96 (.08).73 (.16).75 (.16) Reaction time 652 (106) 834 (106) 863 (116) 2172 (309) 716 (93) 829 (136) 928 (104) 2172 (316) in the dual-digit CRT and dual-word CRT conditions did not differ. The mean number of tones correctly identified in the single CRT, dual-digit CRT, and dual-word CRT conditions was 99 (SD = 21), 59 (SD = 16), and 57 (SD = 18) tones, respectively. There was a main effect of attention, F(2, 54) = 181.15, MSE = 90.48, g 2 =.87, p <.001, with a greater number of tones identified in the single CRT than dual-digit F(1, 27) = 199.17, MSE = 200.08, g 2 =.88, p <.001 and dual-word CRT conditions, F(1, 27) = 268.87, MSE = 229.93, g 2 =.91, p <.001. Notably, the means in the two dual-task conditions did not differ significantly, F(1,27) = 2.15. 4. Discussion The purpose of this study was to examine the resource requirements of recollection and familiarity-based recognition during retrieval. First, we tested the hypothesis that recollection requires attentional resources during retrieval by examining how R responding was affected by manipulations of available attention during retrieval. In addition, since older adults are believed to have fewer available attentional resources than younger adults, we compared the pattern of R responding across age groups, and investigated whether DA in the older adults group had an additive effect on R responses. We found evidence in support of a general effect of DA on recollection; false alarms increased significantly during both digit- and word-based DA conditions (see Fig. 1). While older adults showed a general increase in false alarms compared to younger adults, the interaction with attention condition was not significant. Second, we tested the hypothesis that familiarity would be particularly disrupted in the DA-word condition, as this distracting task would interfere with re-creation of the representational content of memory (Fernandes & Moscovitch, 2000, 2002, 2003) critical to the familiarity process. We found that verbal memory performance suffered when attention was divided during retrieval with a word-, but not a digit-based, distracter task, replicating past research (Fernandes & Moscovitch, 2002, 2003). What is novel in the current data is that we were able to show that this material-specific interference is selective to K responding (see Fig. 2). K hit rate, K accuracy, and independence model measures of familiarity decreased only during the DAword, and not DA-digit condition, as compared to FA. Aging did not interact with this effect. We discuss the implications of these findings in turn. 4.1. General interference effects Younger and older adults made more false R responses during both DA conditions (see Fig. 1). That the magni- Fig. 1. Mean false alarm rate for Remember (R) responses under full attention (FA), divided attention digits (DA-digit), and divided attention word (DA-word) conditions in younger and older adults. Error bars show the standard error of the mean.

218 E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 Fig. 2. Mean recognition accuracy under full attention (FA), divided attention digit (DA-digit) and divided attention word (DA-word) conditions in younger and older adults. Grey bars show Remember (R) accuracy and white bars show Know (K) accuracy, measured as hit rate false alarm rate. Error bars show the standard error of the mean. tude of the increase in R false alarm rate did not differ across DA conditions suggests that manipulations of attention at retrieval have a general effect on recollective-based memory processes. This finding is similar to that reported in Hicks and Marsh (2000), who also found that false responding increased under DA conditions. Importantly, our data suggest that the false memory effects found under DA conditions are selective to R responses. At first glance, it may be surprising that participants are prone to making any false remember responses at all. After all, these memory decisions require that additional context, or details of encoding, be recalled along with the word before participants make such a response. Other research, however, has documented numerous conditions under which false, or illusory, recollective responses can be given. False recollections have been explained in terms of impaired source monitoring (Lane & Zaragoza, 1995), the improper assignment of old contexts to new items (Holmes, Walters, & Rajaram, 1998), an inability to inhibit or control gistrelated processes (Balota, Dolan, & Duchek, 2000), or phantom recollection (Brainerd, Wright, Reyna, & Morjadin, 2001). Regardless of the mechanism by which false recollections are produced, our study suggests that availability of attention during retrieval is critical to avoid such memory errors. The claim that accurate recollection is reliant upon available attentional resources at retrieval is in contrast to Gardiner et al. s (2006) suggestion that reduced resources at retrieval do not affect recollection. The discrepancy in our findings may be due to differences in the manipulation used to limit cognitive resources (speeded responding as compared to divided attention), differences in study and test materials, or because Gardiner and colleagues performed their analyses on overall accuracy, rather than separate hit and false alarm analyses. Nonetheless, our findings support other work suggesting that during retrieval, recollection is a more attention demanding process than familiarity (Jacoby, 1991; Yonelinas & Jacoby, 1994, 1995). If reduced attentional resources at retrieval contribute to false R responding, one would expect that older adults would have more false recollective experiences. Several studies have found that older adults are more inclined to falsely recall and recognize information than younger adults (Norman & Schacter, 1997; Rankin & Kausler, 1979; Smith, 1975), and research suggests that the increased rate of false remembering shown in older adults is a result of age-related changes in the integrity of frontal lobe structures (Butler, McDaniel, Dornburg, Price, & Roediger, 2004; Craik, Morris, Morris, & Loewen, 1990). As expected, we found that older adults made fewer accurate recollective responses than younger adults, as indicated by their elevated R false alarm rate across all experimental conditions. However, older adults did not show elevated levels of false recollections under DA conditions. The lack of interaction between attention and age manipulations suggests that although an age-related loss in attentional resources may contribute to increased levels of false memories in older adults, other factors also contribute to age differences in false recollections. For example, Bunce (2003) found an interaction between false R responding, the amount of cognitive support at encoding, and frontal lobe function in older adults. In that study, older adults reported more false R responses than younger adults when they were given a random, but not categorizable, word list only if they had low frontal lobe functioning. How addi-

E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 219 tional factors, including executive control and level of frontal functioning, relate to the incidence of false recollective memories will need to be examined in future research. There were differences in demographic variables and executive function across age groups in our study, which may have affected our finding of an age-related increase in false R responding. To examine the effects of education, intelligence, and cognitive flexibility on false R responding, we entered education, FSIQ (derived from the NART), and RT for Trails A and B as covariates in the analyses. All analyses still showed a main effect of age, suggesting that education, intelligence, and cognitive flexibility as assessed by the Trails test cannot account for the increased false R responding observed in older adults. We also found an effect of DA on RT for each distracting task, with increased latency during DA conditions (relative to FA), and that this increase was greater for older adults. This corresponds with past research suggesting that attentional resources are required to establish and maintain a retrieval set during memory retrieval, that distracting task costs provided an index of these resource requirements, and that older adults have a more difficult time establishing and maintaining the retrieval set (Anderson et al., 1998; Whiting & Smith, 1997). To examine whether age differences in distracter task RT are affected by education, intelligence, and cognitive flexibility, we again entered education, FSIQ, and RT for Trails A and B as covariates in the distracter task analyses. All analyses still showed a main effect of age and an age group attention interaction, suggesting that these factors cannot account for the agerelated increase in RT under DA conditions. 4.2. Material-specific effects on memory Recognition memory performance decreased during the DA-word, but not DA-digit task, as compared to FA, and this effect was unaffected by age. Importantly, we found that this material-specific interference effect acted specifically on K measures of hit rate, accuracy, and independence measures of familiarity, and not on R responses, in younger and older adults (see Fig. 2; white bars). Similar results were found in an initial experiment, in which the distracting tasks were the word-based ones used here, and an odd-digit identification task, suggesting that the results are robust. To our knowledge, this is the first time anyone has shown that familiarity-based responding can be disrupted by DA at retrieval. The finding suggests that familiarity-based memory retrieval relies on the reactivation of content representations, as it is disrupted specifically when the distracting task material is similar to that in the memory task. Our claim is supported by other work showing that familiarity, and not recollective-based responding, is affected when processing is disrupted by another variable or task that contains material similar to that in the memory task. For example, when lures on an auditory recognition task are created from the non-words of an auditory lexical decision task that is performed between encoding and retrieval, accuracy of Knowing responses declines (Dewhurst & Hitch, 1997). Our work provides evidence that familiarity-based memory retrieval relies critically on the ability to properly engage in either the perceptual or conceptual processes that re-create the content of the memory. We also examined whether the selective disruption to K responding in the DA-word condition was due to an increased level of difficulty of the word-, compared to digit-based, distracter task, but results argue against this interpretation. On the auditory CRT task, we found that the number of tones identified, and RTs to tones under dual-task conditions, were equivalent when the task was performed concurrently with either the word- or digitbased distracter task. We did not examine performance on the CRT task in older adults, since the auditory sensory loss associated with aging would lead to very poor, and/or variable performance, preventing any meaningful interpretation of differences across conditions. It is thus possible that there was an age task difficulty interaction. Nonetheless, results from the auditory CRT task suggest that our finding of a material-specific effect on K responses cannot be accounted for by differences in level of difficulty of the word- and digit-based distracting task. In line with this, there were no differences in accuracy or RT for these distracting tasks performed under single-task conditions, suggesting the resource demands for these tasks were equivalent. Our study also showed that R hits and R accuracy were unaffected during the DA-word condition (see Fig. 2, grey bars). This suggests that during recollection, the content of memories can be accessed via a different network than during familiarity-based retrieval. Otherwise, R accuracy should have also decreased in the DA-word condition. At this point we can only make speculations as to what the exact mechanisms of this network may be, but since recollection appears to involve additional activation within the frontal lobes (Eldridge, Knowlton, Furmanski, Bookheimer, & Engel, 2000; Henson, Rugg, Shallice, Jospehs, & Dolan, 1999), there may be alternate routes that can access the content of memories, distinct from the network of brain regions recruited during familiarity-based responding. 5. Conclusions Our study shows two novel findings: First, we found that DA at retrieval, regardless of the material in the distracter task, led to an increase in false Remember responding, suggesting that general attentional resources are required to properly search and/or monitor the retrieval of contextual memories. Second, we found a selective decrease in Know responses during a word-based, but not digit-based, DA condition at retrieval, indicating a material-specific effect on familiarity. Aging was associated with an overall increase in false Remember responses, and with increased latency in distracting task responses under DA conditions. Results suggest that avoiding false recollective responses during retrieval requires attention, whereas

220 E.I. Skinner, M.A. Fernandes / Acta Psychologica 127 (2008) 211 221 accurate familiarity responses require reactivation of content-specific representations. Acknowledgments This research was supported by a grant from the National Sciences and Engineering Research Council of Canada (NSERC) awarded to M.F., a post-graduate scholarship from NSERC to E.S. References Anderson, N. D., Craik, F. I. M., & Naveh-Benjamin, M. (1998). The attentional demands of encoding and retrieval in younger and older adults: 1. Evidence from divided attention costs. Psychology and Aging, 13, 405 423. Baayen, R. H., Piepenbrock, R., & Gulikers, L. (1995). The CELEX Lexical Database [CD-ROM]. University of Pennsylvania, Philadelphia, PA: Linguistic Data Consortium. Baddeley, A. D., Lewis, V., Eldridge, M., & Thompson, N. (1984). Attention and retrieval from long-term memory. Journal of Experimental Psychology: General, 13, 518 540. Balota, D. A., Dolan, P. O., & Duchek, J. M. (2000). Memory changes in healthy older adults. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 395 410). New York: Oxford University Press. Benjamin, A. S., & Craik, F. I. M. (2001). Parallel effects of aging and time pressure on memory for source: Evidence from the spacing effect. Memory and Cognition, 29, 691 697. Blair, J. R., & Spreen, O. (1989). Predicting premorbid IQ: A revision of the National Adult Reading Test. The Clinical Neuropsychologist, 3, 129 136. Brainerd, C. J., Wright, R., Reyna, V. F., & Morjadin, A. H. (2001). Conjoint recognition and phantom recollection. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27, 307 327. Bunce, D. (2003). Cognitive support at encoding attenuates age differences in recollective experience among adults of lower frontal lobe function. Neuropsychology, 17, 353 361. Butler, K. M., McDaniel, M. A., Dornburg, C. C., Price, A. L., & Roediger, H. L. (2004). Age differences in veridical and false recall are not inevitable: The role of frontal lobe function. Psychonomic Bulletin and Review, 11, 921 925. Craik, F. I. M., Govoni, R., Naveh-Benjamin, M., & Anderson, N. D. (1996). The effects of divided attention on encoding and retrieval processes in human memory. Journal of Experimental Psychology: General, 125, 159 180. Craik, F. I. M., Morris, L. W., Morris, R. G., & Loewen, E. R. (1990). Relations between source amnesia and frontal lobe functioning in older adults. Psychology and Aging, 5, 148 151. Davidson, P. S. R., & Glisky, E. L. (2002). Is flashbulb memory a special instance of source memory? Evidence from older adults. Memory, 10, 99 111. Dewhurst, S. A., & Hitch, G. J. (1997). Illusions of familiarity caused by cohort activation. Psychonomic Bulletin and Review, 4, 566 571. Dodson, C. S., & Johnson, M. K. (1996). Some problems with the processdissociation approach to memory. Journal of Experimental Psychology: General, 125, 181 194. Eldridge, L. L., Knowlton, B. T., Furmanski, C. S., Bookheimer, S. Y., & Engel, S. A. (2000). Remembering episodes: A selective role for the hippocampus during retrieval. Nature Neuroscience, 3, 1149 1152. Fernandes, M. A., & Moscovitch, M. (2000). Divided attention and memory: Evidence of substantial interference effects at retrieval and encoding. Journal of Experimental Psychology: General, 129, 155 176. Fernandes, M. A., & Moscovitch, M. (2002). Factors modulating the effect of divided attention during retrieval of words. Memory and Cognition, 30, 731 744. Fernandes, M. A., & Moscovitch, M. (2003). Interference effects from divided attention during retrieval in younger and older adults. Psychology and Aging, 18, 219 230. Folstein, M. F., Folstein, S. F., & McHugh, P. R. (1975). Mini-Mental State: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189 198. Gardiner, J. M. (1988). Functional-aspects of recollective experience. Memory and Cognition, 16, 309 318. Gardiner, J. M., Gregg, V. H., & Karayianni, I. (2006). Recognition memory and awareness: Occurrence of perceptual effects in remembering or in knowing depends on conscious resources at encoding, but not at retrieval. Memory and Cognition, 34, 227 239. Gardiner, J. M., Gregg, V. H., Mashru, R., & Thaman, M. (2001). Impact of encoding depth on awareness of perceptual effects in recognition memory. Memory and Cognition, 29, 433 440. Gardiner, J. M., & Parkin, A. J. (1990). Attention and recollective experience in recognition memory. Memory and Cognition, 18, 579 583. Gillund, G., & Shiffrin, R. M. (1984). The relationship between recall and recognition in amnesia: Effects of matching recognition between patients with amnesia and controls. Neuropsychology, 15, 444 451. Gregg, V. H., & Gardiner, J. M. (1994). Recognition memory and awareness: A large effect of study modality on know responses following a highly perceptual orienting task. European Journal of Cognitive Psychology, 6, 131 147. Gruppuso, V., Lindsay, D. S., & Kelley, C. M. (1997). The processdissociation procedure and similarity: Defining and estimating recollection and familiarity in recognition memory. Journal of Experimental Psychology, 23, 259 278. Hasher, L., & Zacks, R. T. (1988). Working memory, comprehension, and aging: A review and a new view. In G. H. Bower (Ed.). The psychology of learning and motivation (Vol. 22, pp. 193 225). San Diego, CA: Academic Press. Henson, R. N. A., Rugg, M. D., Shallice, T., Jospehs, O., & Dolan, R. J. (1999). Recollection and familiarity in recognition memory: An eventrelated functional magnetic resonance imaging study. The Journal of Neuroscience, 19, 3962 3972. Hicks, J. L., & Marsh, R. L. (2000). Toward specifying the attentional demands of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1483 1498. Hintzman, D. L., & Curran, R. (1994). Retrieval dynamics of recognition and frequency judgments: Evidence for separate processes of familiarity and recall. Journal of Memory and Language, 33, 1 18. Holmes, J. B., Walters, H. S., & Rajaram, S. (1998). The phenomenology of false memories: Episodic content and confidence. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24, 1026 1040. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513 541. Johnston, W. A., Dark, V. J., & Jacoby, L. L. (1985). Perceptual fluency and recognition judgments. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 3 11. Kelley, C. M., & Jacoby, L. L. (1998). Participative reports and process dissociation: Fluency, knowing, and feeling. Acta Psychologica, 98, 127 140. Lane, S. M., & Zaragoza, M. S. (1995). The recollective experience of cross-modality errors. Memory and Cognition, 23, 607 610. Mandler, G. (1980). Recognizing The judgment of previous occurrence. Psychological Review, 87, 252 271. McDowd, J. M., & Shaw, R. J. (2000). Attention and aging: A functional perspective. In F. I. M. Craik & T. A. Salthouse (Eds.), The handbook of aging and cognition (2nd ed., pp. 221 292). Mahwah, NJ: Laurence Erlbaum Associates. Moscovitch, M. (1989). Confabulation and the frontal system: Strategic versus associative retrieval in neuropsychological theories of memory. In H. L. Roediger, III & F. I. M. Craik (Eds.), Varieties of memory and