Modality and Illusions of Recognition 1 Running head: MODALITY AND ILLUSIONS OF RECOGNITION The Effect of Study Modality on Illusions of Recognition in Young and Older Adults

According to the dual-process model of recognition memory, recognition is based in part upon attributions about the relative ease of processing a stimulus. Ease of processing can be used as a heuristic for deciding whether an item is familiar. Experimental manipulations of processing fluency in a recognition task can cause a memory illusion by making an item seem more familiar. The memory illusion in this experiment is an increased probability of calling a word "old" for both studied words and words new at test following a manipulation of processing fluency. Perceptual and conceptual elements of familiarity were manipulated in this experiment to create a memory illusion in 24 young and 24 older adults. Recognition judgments and naming latencies were collected from each participant. It was found that younger adults were more susceptible to an illusion derived from manipulations of perceptual processing, while older adults were more affected by manipulating conceptual processing. The findings from the present experiment are in support of the hypothesis that reliance upon conceptual and perceptual processes in recognition memory changes throughout the lifespan. Modality and Illusions of Recognition 3 The Effect of Study Modality on Illusions of Recognition in Young and Older Adults Perceptual illusions have a long history of study in the fields of psychology, philosophy, and physics. Illusions are experimentally interesting because they offer insight into the opaque processes involved in perception and cognition by showing situations in which the underlying mechanisms can produce erroneous perceptions (Coren, 1994). In the same way that the study of perceptual illusions has added to our understanding of perceptual processes, the study of memory illusions and other errors of memory can be used to add to our overall understanding of human memory (Roediger, 1996). The motivating question behind the research reported here was whether there are underlying differences in the mechanisms underlying recognition memory between younger and older adults. To address this question, a memory illusion paradigm was used to manipulate conceptual and perceptual aspects of the recognition process. Differential sensitivity to these manipulations would be suggestive that the basis of recognition memory changes throughout the lifespan. To better understand how memory illusions can be used to investigate this issue, research relevant to recognition memory, aging, and memory illusions is first discussed. Recognition Memory The attributional model of recognition memory is derived from the attributional model of emotion (Schacter & Singer, 1962) in which emotional states are determined by a person’s interpretation of his physiological arousal in the context of some environmental situation. According to the attributional model of memory, feelings of familiarity are based on attributions about how easily an item is processed (Jacoby, Kelley, & Dywan, 1989). This ease of processing is referred to as fluency and can be used as a heuristic in guiding recognition decisions (Whittlesea, 1993). An item that is readily processed is more likely to feel familiar, and Modality and Illusions of Recognition 4 therefore, has a greater probability of being called "old" in a recognition task. Besides being based on feelings of familiarity, item recognition can also occur from deliberate recollection of information from long-term memory (Atkinson & Juola, 1974; Jacoby & Dallas, 1981; Mandler, 1980; Mandler & Boeck, 1974). The two of these processes form the dual-process model of recognition memory. According to this model, recognition can be based on directed memory retrieval, unconscious inferences of familiarity based upon the ease of processing an item, or from both of these processes working together (Holender, 1986, Whittlesea, Jacoby, & Girard, 1990). The dual-process model can be seen as consisting of both an intentional component (deliberate recollection) and an automatic component (familiarity). Intentional forms of memory can be tested directly by such tasks as cued recall and free recall that make explicit reference to a person’s previous experience with an item. Direct tests are a measure of explicit memory. Indirect tests such as word fragment completion, word stem completion, and perceptual identification do not rely upon a person’s deliberate recollection of a previous experience with an item, but rather are seen as a retention of information without a person’s conscious awareness (Richardson-Klavehn & Bjork, 1988; Roediger, 1990; Schacter, 1987). These tasks are indirect because participants are not instructed to rely upon studied items to complete the memory test (Blaxton, 1989). Indirect tests are sometimes treated as synonymous with implicit memory (Richardson-Klavehn & Bjork, 1988). The explicit memory system is subserved by structures in the limbic system, basal forebrain, thalamus, and sensory areas of the neocortex (Kolb & Wishaw, 1996). Implicit memory functions are dependent upon structures such as the basal ganglia for implicit motor learning (Heindel, Salmon, Shults, Walicke & Butters, 1989; Knowlton, Squire, Paulsen, & Swerdlow, 1996) and the primary visual areas of the occipital Modality and Illusions of Recognition 5 lobes for visual word priming (Gabrieli, Fleischman, Keane, Reminger, & Morrell, 1995). There is strong evidence to suggest that the implicit and explicit memory systems are functionally separable as can be seen from both physiological and behavioral studies. For example, Gabrieli, Fleischman, Keane, Reminger and Morrell (1995) described a patient, known as MS, who has a right occipital lobe lesion and shows intact conceptual priming and explicit memory for words, but shows a deficit in visual repetition priming for words. In contrast, amnesic patients show a profound deficit in explicit memory, but are generally not impaired on measures of implicit memory. In addition, patients diagnosed with probable dementia of the Alzheimer's type show impairments in conceptual priming, but not perceptual priming. This pattern of spared and impaired functioning across different neurological conditions shows a double dissociation between perceptual implicit memory for words compared to conceptual implicit memory and explicit memory for words. One way to examine whether a memory process is perceptually or conceptually based is by using tasks that are sensitive to a specific type of processing. The idea that retention is a function of encoding and retrieval is known as transfer-appropriate processing (Blaxton, 1989; Bransford, Franks, Morris, & Stein, 1979). Perceptually based tasks such as graphemic recall and perceptual identification derive a benefit from congruency between item presentation at study and test. These tasks are therefore said to be data-driven and to rely upon the physical characteristics of an item. Tasks that rely upon the meaning of an item are considered to be conceptually driven. Conceptually driven tasks such as free recall, semantic cued recall and tests of general knowledge obtain benefits from item generation and levels-of-processing manipulations at study. Conceptually driven tasks are also thought not to derive a benefit from item congruency between study and test such as sharing the same font or being presented in the Modality and Illusions of Recognition 6 same size (Blaxton, 1989). Differences between perceptual and conceptual tasks have also been supported by positron emission tomography studies that show that the two processes rely on separate neurological areas (Blaxton, Bookheimer, Zeffiro, Figlozzi, Gaillard & Theodore, 1996). It is overly optimistic to say that conceptual and perceptual tasks only measure their intended constructs, however. In a recent meta-analysis of perceptual and conceptual manipulations of implicit memory, Brown and Mitchell (1994) found that contrary to the claim that there is no effect of conceptual manipulations on indirect perceptual tests of memory, there was evidence of increased perceptual priming following conceptual manipulations at study. This effect may be in part due to contamination of indirect tests with explicit memory, but Brown and Mitchell (1994) argue that test contamination cannot account for the entire effect. Aging Older adults have difficulty with tests that rely upon deliberate recollection, or explicit memory, such as cued recall (Craik, 1994) and deliberate recall (Craik & McDowd, 1987), but perform nearly as well as younger adults on indirect tests of memory (La Voie & Light, 1994). Performance decline on direct memory tests can be 50% or more (Graf, 1990). Older adults have more difficulty than younger adults with tasks such as elaborative encoding (Grady et al., 1995), and remembering the context an item was studied in when tested directly (Light, La Voie, Valencia-Laver, Alberston Owens, & Mead, 1992; Spencer & Raz, 1995). However, indirect tests of memory are relatively well preserved in older adults (Graf, 1990; Light & Singh, 1987; Light, Singh, & Capps, 1986). Examples of indirect tests of memory which show little age differences are word stem completion in which exposure to a word prior to testing increases the likelihood of using that word to complete a word stem (Light & Singh, 1987), and word fragment Modality and Illusions of Recognition 7 completion in which previous exposure to an item increases the probability of using that item to complete the fragment (Light, Singh, & Capps, 1986). One methodology used to examine whether automatic processes remain intact throughout aging is the process dissociation procedure (Jacoby, 1991), which is a way to estimate the relative contributions made by automatic and intentional memory processes to a task. The process dissociation procedure consists of two tests: one where familiarity and deliberate recollection both facilitate performance (inclusion), and one where familiarity is put into opposition with deliberate recollection (exclusion). By algebraic manipulation, estimates of the relative contributions familiarity and recollection make to a task can be obtained. When older adults are compared to younger adults using the opposition procedure on tasks such as cued recall (Jacoby, Toth, & Yonelinas, 1993), two-item forced choice, or a false-fame paradigm (Jennings & Jacoby, 1993), it is consistently found that older adults show a deficit in consciously controlled memory processes, but that automatic processes are unaffected (Light & Albertson, 1989). Furthermore, younger adults under divided attention at study perform comparably to older adults in that their ability to deliberately recollect an item is severely affected, but judgments based upon automatic processes are not affected by attention allocation at study (Jennings & Jacoby, 1993). These results are consistent with findings that performance on indirect tests of memory do not decline in aging, but that direct tests of memory do show substantial decline (Graf, 1990; La Voie & Light, 1994, Light & Singh, 1987). The process dissociation procedure has been controversial, however, because of its theoretical assumption that automatic and intentional aspects of memory are independent of one another (Curran & Hintzman, 1995). If the two processes are redundant, then interpretation of the estimates made by the procedure is difficult because of the overlap between familiarity and recollection. Modality and Illusions of Recognition 8 Although the exact relationship between familiarity and deliberate recollection has not been satisfactorily clarified, opposition tests still are able to provide strong supporting evidence for two processes working in recognition. Furthermore, the use of opposition logic in testing can show clear age differences that might not be apparent when using traditional tests of recognition memory (Jennings & Jacoby, 1997). The difference between younger and older adults on recognition tests is often small and non-significant (Craik & McDowd, 1987). The lack of consistent age differences, however, might be due to insufficient statistical power to detect a difference (La Voie & Light, 1994). Thus, just because age differences are not reliably found does not mean they do not exist. La Voie and Light (1994) found in a meta-analysis of age differences on direct and indirect measures of memory that the standardized effect size for recognition was .5, which by the conventions established by Cohen (1992) is a moderate effect. In terms of the dual-process model, the better performance of older adults on recognition tasks compared to a task such as free recall can be accounted for by the joint contributions that automatic and intentional processes make to a recognition decision. Older adults may be able to perform nearly as well as younger adults on recognition tasks because they can rely upon fluency attributions when making a recognition decision. Free recall offers no contextual support and does not invite the use of a fluency heuristic. The amount of contextual support a task affords has been argued to be a primary determinant of memory differences seen in older adults (Craik et al., 1987; Spencer & Raz, 1995, but see Light & Albertson, 1989). Age differences between automatic and intentional memory processes have also been studied by asking participants to distinguish between whether they Remember (R) or Know (K) that an item appeared previously (Gardiner, 1988; Tulving, 1985). Remember judgments are Modality and Illusions of Recognition 9 thought to rely upon contextual information, and are hypothesized to be a measure of deliberate recollection. Know judgments differ from R judgments in that K judgments are presumed to be based upon automatic processes associated with familiarity. The distinction between R and K responses has been suggested because of differential performance on recognition test manipulations when asking individuals to indicate whether their decision was based upon remembering or knowing that an item had been previously studied. R judgments are sensitive to levels-of-processing manipulations, generation effects, and divided attention at test (Gardiner, 1988; Jacoby, 1991), but K judgments are resistant to these study and test manipulations. K judgments are affected by manipulations such as whether the to-be-remembered item is a word or a non-word (Gardiner, 1988). These differential test results suggest that R judgments are an intentional aspect of memory, and that K judgments are an automatic aspect of memory. The dissociations in performance between R and K judgments are comparable to dissociations found between implicit and explicit memory tests. When examining R and K judgments in young and older adults, Parkin and Walter (1992) found significant age differences in how recognition decisions were made. Younger adults made more R responses indicating better deliberate recollection ability. Older adults made more K judgments than younger adults, and had lower overall R response rates. The decline in R responses for older adults may be due to an inability to access specific temporal and contextual information associated with the learning of an item (c.f., Knowlton & Squire, 1995). It should be noted, however, that the pattern of increased K responses from older adults is not always found. Perfect, Williams, and Anderton-Brown (1995) reported both increased K responses (Experiments 1 and 2B), and decreased K responses (Experiment 2A). No age difference in K responses has been reported by Mantyla (1993). Light, Prull, La Voie, and Healy (in press), Modality and Illusions of Recognition 10 however, performed a meta-analysis of age differences in K judgments and found that overall there is evidence of a slight increase in K responses for older adults. The dissociation between R and K responses in aging is in accord with the process dissociation literature, and further support the proposition that automatic processes such as assessments of familiarity or knowing remain relatively intact in older adults, while tasks that require deliberate retrieval decline with age (Spencer & Raz, 1995). Memory illusions False recognition can be induced by making an item seem more familiar by manipulating how the item is processed prior to a recognition decision (Jacoby & Dallas, 1981; Prull, 1997; Whittlesea, 1993; Whittlesea et al., 1990). If an item is readily processed and a deliberate memory search cannot confidently produce a previous occurrence of the item, then according to the attributional framework of memory, the recognition judgment is likely to be based on attributions about the ease of perceptually processing the stimulus (Johnston, Dark, & Jacoby, 1985). Thus, attributions towards the perceptual fluency of an item work in conjunction with attempts to deliberately recollect an item in a recognition task. The use of a fluency judgments is not restricted merely to assessments of perceptual processing, however. The ease of processing a word's meaning is referred to as conceptual fluency and is also a basis of familiarity. Manipulating the conceptual processing of a word prior to a recognition decision can also create memory illusions (Luo, 1993). For example, Whittlesea (1993) found that created a memory illusion by having a target word appear after a predictive sentence frame prior to the recognition decision. In Whittlesea (1993, Experiment 2), participants saw sentence frames that were either predictive or nonpredictive of the concluding target word. The predictive nature of the sentence frame served to automatically activate the target word in memory, so that when the target was Modality and Illusions of Recognition 11 presented the item was already available and processed more fluently. This ease of processing led to a misattribution about the item's conceptual fluency and resulted in a spurious sense of familiarity. In fact, manipulations of conceptual fluency have been shown to create larger memory illusions than manipulations of perceptual fluency (Whittlesea, 1993). In support of this finding, Wagner, Gabrieli, and Verfaellie (1997) found that conceptually based familiarity processes are more important to the recognition process than are perceptually based familiarity processes. Luo (1993) has also found that memory illusions can be generated by manipulating an item's conceptual processing. In Luo’s Experiment 3 (1993), prior to the recognition decision, participants typed the target word backwards, primarily a perceptual task, or completed a wordconstruction manipulation that entailed focusing on the meaning of the target word. Luo found that both of these manipulations led to a strong memory illusion again indicating that the fluency heuristic can be based upon both perceptual and non-perceptual mechanisms. Along with Lou's (1993) finding of separate perceptual and conceptual processes involved in recognition, recent neuroimaging studies have found that the perceptually and conceptually based familiarity systems are functionally distinct (Gabrieli, Desmond, Demb, Wagner, Stone, Vaidya and Clover, 1996, but see Brown & Mitchell, 1994). In normal participants, it would be expected that both perceptual and conceptual processes could have some role in a recognition test, but whether there is a change in the relative involvement of each process in normal aging is still an open question. If older adults have more difficulty with deliberate recollection compared to younger adults, and are therefore more likely to rely upon automatic processes in recognition, we would predict that older adults should be more susceptible to memory illusions than younger adults. Research on memory illusions indeed Modality and Illusions of Recognition 12 does show that older adults are more prone to memory distortions based upon source memory errors (Craik, Morris, Morris, & Loewen, 1990; Schacter, Kazniak, Kihlstrom, & Valdiserri, 1991), in recognizing names (Jacoby & Dywan, 1991), faces (Bartlett, Strater, & Fulton, 1991), and photographs (Schacter, Koustaal, Johnson, Gross, & Angell, 1997). Older adults are also more likely to make processing misattributions about the ease of retrieving an item as can be seen by false-fame effects (Dywan & Jacoby, 1990), and to be more susceptible to a gist-based memory illusion than younger adults (Koustaal & Schacter, 1997). There is also evidence that the increased susceptibility to memory illusions may not be the same for paradigms that manipulate perceptual and conceptual processing. For instance, when manipulating perceptual fluency using a revelation procedure in which target items were shown with decreasing levels of masking (Prull, 1997, Experiment 1; Watkins & Peynircioglu, 1990), older adults did not show a significant memory illusion. However, Prull (1997, Experiment 2) performed a follow-up experiment using Whittlesea's (1993) paradigm of manipulating conceptual fluency by placing the target after a predictive or nonpredictive sentence frame. A predictive sentence frame activates a target word’s meaning prior to its display resulting in more fluent processing of the item, thus making the item seem more familiar. Older adults showed a significantly larger memory illusion than younger adults in this task. The memory illusion found for studied and new items in young adults was 14% and 17% respectively, while the older adults showed an illusion for studied and new items of 29% and 14%. A problem with this procedure, however, was that the conceptual and perceptual aspects of processing were confounded because target words appeared visually at both study and test. Although the conceptually based memory illusion in Prull Experiment 2 (1997) was larger than the perceptually based illusion in Prull Modality and Illusions of Recognition 13 Experiment 1 (1997), it is possible that this was due to perceptual and conceptual fluency having an additive effect and exaggerating the magnitude of the illusion. The memory illusion studied in the present experiment was based upon incomplete sentence frames that either predicted or did not predict the concluding target word (Whittlesea, 1993). A predictive sentence facilitates processing of the concluding word because presumably the word has been activated in memory prior to its actual presentation. An example of a predictive sentence frame is, “The hills are alive with the sound of”, which would subsequently be completed by the word, “music.” Because the participant knows what the concluding word is going to be before it appears on the screen, the increased fluency of processing the item can cause a misattribution of item familiarity. This misattribution would thus increase the likelihood of saying that the item was previously studied. A nonpredictive variant for the target word “music” would be, “the man was annoyed by the.” In this sentence, the concluding word is not predicted by the sentence frame, and therefore processing of the word does not begin until the item is presented. To isolate the relative contributions of perceptual and conceptual processes in the memory illusion found by Prull (Experiment 2, 1997), a modality manipulation during study was used. Items at study were presented either auditorily or visually. The reasoning behind this manipulation is that an item that has been heard at study and is then seen at test will not receive a benefit from modality specific perceptual processes (Blaxton, 1989). An item seen at study will receive a benefit from both conceptual processing and perceptual processing of the word at test. If the magnitude of the memory illusion is greater for words seen at test, this would suggest that both perceptual and conceptual fluency were contributing to the recognition decision. The memory illusion in this experiment is an increased probability of responding “old” to words occurring after predictive sentence frames. To the extent that the memory illusion is Modality and Illusions of Recognition 14 based upon the ease of processing an item, then there should be an increased proportion of “old” judgments for words appearing after predictive sentence frames regardless of whether the words were previously studied or not. Older adults, because of impairments with deliberate recollection, are expected to rely more upon intact automatic processes when making a recognition decision. Therefore, it is expected that older adults will be more susceptible to fluency manipulations than younger adults and will show a larger memory illusion. To the extent that the memory illusion is based upon misattributions about the ease of processing an item, then the illusion should be greater for words seen at study than for words heard at study because of the additive effect perceptual and conceptual fluency will have in the recognition decision. To the extent that perceptual congruency between study and test facilitates item processing, there should be a decrease in naming latencies as seen in a repetition priming effect for items studied visually. Naming latencies are also expected to decrease for items in predictive sentence frames due to a processing benefit from semantic context (Fischler, 1977). Method Participants Twenty-four young adults (12 men, 12 women) and 24 healthy older adults (11 men, 13 women) participated in this experiment. Twenty-one of the young adults were students enrolled in the summer session of the Claremont Colleges and lived on campus. These participants were recruited by randomly calling campus dorm rooms and asking if the occupant would be interested in participating in a psychology experiment. The three remaining younger participants were from the local Claremont community and had indicated to the experimenter that they would be interested in participating in psychological research. The young adults had a mean age of 21.1 years (range = 18-29). The older adults lived independently in the local Claremont Modality and Illusions of Recognition 15 community and had an average age of 72.0 years (range = 63-80). Demographic measures indicated that the two groups were similar in digit span capacity. Neither the forward (M young = 7.46, M older = 7.42), t(46) = .11, nor backward (M young = M old = 5.71), t(46) = 0, digit span measures differed significantly between age groups. The two groups also performed identically in terms of mean word production on a verbal fluency test (M = 18.3) in which participants were to name as many words they could think of that began with the letter 'F' in one minute (Borkowski, Benton & Spreen, 1967). Older adults had somewhat lower subjective ratings of health (M = 7.33) than younger adults (M = 8.25) as measured on a 10-point Likert scale with higher scores meaning better health, t(46) = 1.97, p = .06. The older adults also had more years of formal education than the younger adults (M young = 15.21, M old = 17.81), t(46) = -3.78, p < .01, and higher vocabulary levels as measured by a modified 25-item version of the Nelson-Denny vocabulary test (Nelson & Denny, 1960), (M young = 16.71, M old = 21.13), t(46) = -4.29, p < .01. Data from an additional four young adults (4 women) and 12 older adults (1 man, 11 women) were collected but excluded from the final analysis. One younger adult had learned English after the age of seven, and three younger adults had microphone error rates (see below) in excess of 16%. For the older adults, ten had microphone error rates greater than 16%, one was over 80 years old, and one had naming latencies greater than 2.5 SD from the group mean in more than half of the naming latency trials. Design and Materials The experimental design was a 2 (age: young, older) x 2 (study modality: auditory, visual) x 2 (sentence frame: predictive, nonpredictive) x 2 (study status: old, new). The first factor was between-subjects, while the remaining three were within-subjects. The experimental Modality and Illusions of Recognition 16 stimuli were taken from Prull (1997). Fifty-six critical words were assigned to eight groups, allowing for counterbalancing of the within-subjects factors. For the 56 critical words, the mean number of syllables was 1.86 (range = 1-3), the mean number of letters per word was 6.05 (range = 5-9) and the mean Francis-Kucera (1982) word frequency was 35.36 (range = 1-187). Three one-way analyses of variance (ANOVA) indicated the counterbalancing groups were comparable in terms of number of syllables, length in letters, and Francis-Kucera frequency, all Fs ≤ 1. The predictive and nonpredictive sentence frames used as test stimuli had previously been found to effectively elicit memory illusions (Prull, 1997). An example of a predictive sentence frame is “She passed the salt and”, which would be concluded with the critical stimulus, “pepper”. A nonpredictive sentence frame such as, “She accidentally dropped the”, is one in which many different words could logically conclude the sentence. In this case, “pepper” would be the critical stimulus. The sentence frames were crossed with the critical stimuli so that for a given word, half of the participants would read the predictive sentence frame, while the other half would read the nonpredictive sentence frame. The study status of the word—whether it was heard, seen, or not presented—was also counterbalanced with the sentence frames. In addition, the critical stimuli were arranged such that no test condition was repeated more than three times consecutively, and that the critical stimuli appeared in a different testing condition for each counterbalancing block. Sound files for auditory word presentations were created using Soundblaster, Soundedit, a Macintosh computer, and an IBM computer. Each sound file was constructed with 30 ms of silence at the beginning in order to eliminate extraneous start-up clicks from the auditory file. The words were pronounced in a normal, conversational tone by a female speaker. After word pronunciation, silence was added to each audio file to fix file length at 1 s. Modality and Illusions of Recognition 17 Procedure The procedure was based upon Whittlesea (1993, Experiment 2) and Prull (1997, Experiment 2). Participants were asked to sign an informed consent form before the experiment proceeded. Because participants would be both hearing and seeing words at study, it was necessary to control for differential hearing ability between the groups. Participants adjusted the volume of the auditory presentation prior to the beginning of the experiment proper. A calibration program was designed to play a sound file that said, “check one, check two...” for three repetitions. Participants told the experimenter if the volume was too loud or too soft and adjusted it accordingly. Before moving on to the next stage, participants indicated that they could understand the words, and that the words were being presented at a comfortable sound level. At the beginning of the study phase, instructions were presented to the participant on the computer monitor. Participants were told that they would see and hear a number of words, and were asked to remember them for a later memory test. The nature of the test was not specified. After participants indicated that they had finished reading the instructions, the experimenter repeated the instructions to make sure that participants understood the task. The study list consisted of five primacy buffers, five recency buffers, 14 visually presented critical stimuli, and 14 auditorily presented critical stimuli. Half of the buffers were presented visually, and half were presented auditorily. The modality of the buffers was randomly determined and held constant for all individuals. The interstimulus interval was 1,500 ms. That is, visually presented words were displayed on the monitor for 1,000 ms followed by a blank screen of 500 ms duration. For auditory presentation, the sound file of 1,000 ms was played followed by a 500 ms pause. The screen was blank during auditory presentations. Study list presentation lasted 57 s. Modality and Illusions of Recognition 18 After the study session, participants were given a filler word-finding task that lasted for seven minutes. Pilot data for this experiment indicated that recognition scores were at ceiling when participants took the recognition test immediately after study list presentation. The filler task was thus introduced to reduce recognition rates. The memory test consisted of 10 practice trials and 56 test trials presented on a computer. Each trial began with a focal cue (++++) before stimulus presentation. Participants were instructed to read the incomplete sentence frames aloud as quickly as they could, and that they should not try to guess what the concluding word was going to be. The sentence frames were on the screen for 2,000 ms. Following the sentence presentation, the screen went blank for 500 ms after which the critical stimulus to be named appeared. Participants were instructed to pronounce the word loudly and clearly into the microphone positioned in front of them. The microphone recorded response times beginning from critical stimulus presentation to the onset of word naming. Naming a word triggered the microphone, which caused a screen to appear that prompted participants to indicate whether they had previously studied the word. Participants initiated each trial by pressing the space bar when they were ready to proceed. Immediately after the test, participants were questioned as to whether they were aware of the nature of the task. Previous research by Whittlesea, Jacoby, and Girard (1990, but see Prull, 1997) suggests that awareness of the test manipulation can attenuate or reverse the effect of memory illusions. To assess awareness of the task, participants were first asked how they made their recognition decisions. Next, they were asked if they felt as if some of the test words seemed easier to say into the microphone than others. Participants who responded “yes” to this question were asked to give a percentage of the time that the words seemed easier to say, as well as if they thought this influenced their judgment. After this, participants were asked if they Modality and Illusions of Recognition 19 thought that some of the sentences predicted the final word. Finally, participants were asked if they thought that predictive sentences might have influenced their recognition judgments. To conclude the experiment, participants completed a demographic questionnaire. Participants were also given a digit span test, a verbal fluency test in which they were to name as many words as they could beginning with the letter F in one minute, and a 25-item modified version of the Nelson-Denny vocabulary test (Nelson & Denny, 1960). Participants were thanked and paid $12.00 at the end of a 1-1.5 hour session. Results The presence of a memory illusion in this experiment would be seen as a main effect of increased "old" responses for words following predictive sentence frames compared to words following nonpredictive sentence frames regardless of an item’s study status. Study status refers to whether the item was presented at study either auditorily, visually, or not at all. This analysis is based upon recognition judgments and a subsequent analysis of participants who were aware of the test manipulation. The recognition data were analyzed using repeated-measures ANOVA. Recognition scores were computed as the percentage of "old" responses in each condition. d′ and B′′D measures were calculated from the hit and false alarm rates of the recognition responses (Donaldson, 1992; Hochhaus, 1977) and were also submitted to ANOVA. For all conclusions drawn from inferential statistics reported in this paper, p < .05. To simplify data analysis, the original design of eight word groups was collapsed into six groups. New words did not have a modality condition, and were therefore combined into two groups of new words that followed either predictive or nonpredictive sentences. This change affected the overall number of trials for targets that were new at test. The reaction time means for old words were based upon a maximum of seven items, while the means for new words were Modality and Illusions of Recognition 20 based upon a maximum of 14 items per condition. Because some naming latency trials were removed due to the microphone not triggering and other difficulties, the n used to pool reaction times to accomplish the test correction varied for each participant. The new standard deviations of each participant’s naming latency trials were obtained by pooling each condition's weighted variance and then dividing by the trials remaining in that condition. This adjustment made the experiment analysis a 2 (age: young, older) x 2 (sentence type: predictive, nonpredictive) x 3 (study modality: audio, visual, new) design. Recognition Recognition decisions were scored as the proportion of "old" responses per condition and are presented in Table 1. A 2 (age group: young, old) x 2 (sentence type: predictive, nonpredictive) x 3 (study status: audio, visual, new) ANOVA with repeated-measures on the last two factors was used to analyze recognition responses. “Yes” responses for words heard at study (.656), words seen at study (.624), and words new at test (.327) significantly differed, F(2, 92) = 88.67, MSE = .035. To find the source of this effect, a set of (nonorthogonal) contrasts was computed. Participants had significantly higher hit rates than false alarm rates indicating that they were remembering words from the study list, F(1, 46) = 130.70, MSE = .020. This was true for both the auditorily studied words, F(1, 46) = 129.98, MSE = .020, and the visually studied words, F(1, 46) = 97.10, MSE = .022. There was no difference in the recognition rates between words seen and words heard at study, F(1, 46) = 2.03, MSE = .012. The sentence manipulation elicited a significant memory illusion. Predictive sentence frames had a higher rate of "old" responses (.583) than nonpredictive sentence frames (.488), F(1, 46) = 20.50, MSE = .032. The young and older adults did not differ significantly in terms of the overall magnitude of the memory illusion, F(1, 46) = .04, MSE = .032. However, there were Modality and Illusions of Recognition 21 some age differences. The young adults showed an illusion of 1.8% for auditory words, 17.8% for visual words, and 10.1% for new words. Younger adults’ illusions for visual, t(23) = 4.31, and new words, t(23) = 2.97, were both significantly greater than zero. Older adults showed an illusion of 12.4% for auditory words, 6.6% for visual words, and 8.4% for new words. Older adults’ auditory illusion, t(23) = 1.99, p = .06, was marginally greater than zero, and their illusion for new words was significantly greater than zero, t(23) = 2.14. The three-way interaction between age group, sentence manipulation, and study status attained marginal significance, F(2, 92) = 2.84, MSE = .026, p = .06, indicating a trend for older and younger adults to differ in illusion size depending upon the study modality of the item. The fact that older adults had similar recognition rates for words presented auditorily at study (.616) and for words seen at study (.604) argues against this effect being due to hearing deficits. Further analysis of the three-way interaction was done by carrying out separate sentence × study status analyses for each group. Modality did not affect the overall recognition rates for older adults, F(1, 23) = .45, MSE = .047, but study modality was a significant factor for younger adults, F(1, 23) = 5.90, MSE = .026, as seen by the greater illusion for visually studied words. This finding indicates that the illusion for younger adults was more influenced by a perceptual mechanism. For younger adults, predictive sentences were not affected by study modality, F(1, 23) = .60, MSE = .03, but nonpredictive sentences were, F(1, 23) = 6.85, MSE = .030. Auditorily studied words had a higher hit rate (.684) than visually studied words (.554) for younger adults in nonpredictive sentence frames. That is, younger adults were less likely to recognize visually studied items in nonpredictive sentence frames compared to studied items presented auditorily. Modality and Illusions of Recognition 22 Analyses of d′ and B′′D Previous work by Luo (1993) and Whittlesea, Jacoby, and Girard (1990) has shown that the occurrence of a memory illusion is associated with a decrease in the signal detection theory measure of bias, β. To determine if the older adults had difficulties with deliberate recollection, and would therefore be more likely to base their recognition judgments upon a fluency heuristic, signal detection measures of discrimination ability and bias were computed. d′ is a standardized measure that can be used in memory tasks to estimate the ability to discriminate between old and new words (Banks, 1970; Swets & Green, 1974). A d′ score can be treated as a standard normal deviation score, and as such, the larger a d′ score the better a person is able to differentiate studied material from unstudied material. B′′D is a nonparametric equivalent to β for situations in which the participant makes multiple “yes/no” judgments (Donaldson, 1992). B′′D ranges from -1 to 1. A value of zero indicates no bias, a negative value means a liberal bias, and a positive value indicates a conservative bias. If the sentence manipulation was biasing participants to be more liberal in their responses, then B′′D would be expected to decrease accordingly. The two modality conditions were crossed with the sentence condition to yield four d′ and B′′D scores. The d′ means and standard deviations are displayed in Table 2. A 2 (age group) x 2 (modality) x 2 (sentence type) ANOVA with repeated-measures on the second and third factors was used in this analysis. d′ differed between groups (M young = 1.40, M older = .85), F(1, 46) = 5.70, MSE = 2.90, but was not affected by the test manipulation or the study modality of an item, all F’s ≤ 2.36. The larger d′ value for younger adults means that they were better able to discriminate old words from new words at test as was expected. The three-way interaction between group, sentence type, and modality attained marginal significance, F(1, 46) = 2.85, MSE Modality and Illusions of Recognition 23 = .530, p = .098, which by visual inspection of Table 2 can be seen to be primarily due to variability within the younger adults when discriminating between items heard or seen at study. The means and standard deviations of the B′′D measure also appear in Table 2. Though the means appear rather different, the age groups did not differ significantly in terms of bias levels (M young = .288, M older = .159), F(1, 46) = .04, MSE = 1.42. Predictive sentences made participants significantly more liberal in their recognition decisions (M = -.082) than nonpredictive sentences (M = .193), F(1, 46) = 10.201, MSE = .36. Study modality interacted with the overall level of bias, F(1, 46) = 4.56, MSE = .09. This interaction was primarily due to nonpredictive sentences. Bias was not affected in predictive sentences by whether the item had been heard or seen, (M auditory = -.071, M visual = -.092), F(1, 46) = .129, MSE = .08, but following nonpredictive sentences, participants were more conservative in their responses to words seen at study (M = .276) compared to words heard at study (M = .111), F(1, 46) = 6.79, MSE = .10. Thus, as expected, being able to rely upon perceptual fluency was affecting recognition decisions. Naming Latencies Reaction times to name target words in the recognition task were collected from each participant and submitted to ANOVA. The presence of a visual repetition priming effect would be seen in shorter naming latencies for studied words compared to words heard at study or words new at test. Before analyzing the reaction time data in the word naming task, microphone misses and outlying scores were removed. The process involved a number of steps determined prior to data collection. First, trials in which the microphone failed to trigger when the participant named the word were eliminated. Second, trials in which the naming latencies were below 300 ms or greater than 2,000 ms were removed. Third, naming latencies greater than or less than 2.5 SD Modality and Illusions of Recognition 24 from the participant’s mean in a condition were removed. Standard deviations were based upon the participant's own data within a condition. Fourth, participants with more than 57% of their data missing in a condition were removed from the analysis. Finally, as discussed earlier, participants who were more than 2.5 SD from the group's mean in more than half of the conditions were replaced. Microphone misses and RT pruning resulted in the loss of 6.59% of the younger adults' data and 6.45% of the older adults’ data. Naming latencies and standard deviations are presented in Table 3. A 2 (age group) x 2 (test condition) x 3 (study status) ANOVA with repeated-measures on the last two factors was used to analyze the data. The younger adults were faster in the naming task (M = 740 ms) than the older adults (M = 829 ms), F(1, 46) = 9.10, MSE = 62,982. There was a marginally significant effect for naming studied targets faster than new targets, F(2, 92) = 2.76, MSE = 9785.16, p = .07, which did not differ between the age groups, F(1, 46) = .768. However, when collapsing the auditory and visual conditions into one variable of old words, there was no indication that the naming latencies differed from new words, F(1, 46) = 2.82, MSE = 2,435. Further analyses of the effect modality had upon naming latencies were performed with pairwise comparisons between words studied auditorily, studied visually, or new at test. Naming latencies for auditory words (M = 773 ms) showed a benefit from previous exposure at study (M = 793), F(1, 46) = 5.13, MSE = 1490, indicating a non-specific priming effect (Light et al., 1992). Visually studied words, however, did not show a repetition priming effect (M = 788), F(1, 46) = .21, MSE = 1868. Naming latencies for auditorily studied words did not differ significantly from seen words, F(1, 46) = 2.09, MSE = 2184. As anticipated, predictive sentences facilitated word naming. Critical stimuli in predictive sentence frames were named faster (M = 743 ms) than critical stimuli in nonpredictive Modality and Illusions of Recognition 25 sentence frames (M = 827 ms), F(1, 46) = 57.55, MSE = 8731, indicating that a semantic context effect was occurring. There was no indication that semantic context interacted with groups, F(1, 46) = 1.72, MSE = 8731. Unlike the recognition data, the Age Group × Sentence Type × Study Status interaction did not indicate a trend towards significance, F(2, 92) = 1.58, MSE = 3905, p < .21. No other effects or interactions approached significance, all Fs ≤ 1. Awareness of Task Because awareness of the experimental manipulation has been shown to negate or reverse the effect of memory illusion manipulations (Whittlesea et al., 1990), all participants were given a post-experiment questionnaire to assess awareness. Participants were considered to be aware of the test manipulation (n young = 10, n older = 6) if they indicated a belief that some sentences predicted the final word, and that this might have influenced their recognition judgments. Means and standard deviations of the recognition decisions for the 14 younger adults and 18 older adults who were unaware of the test manipulation are presented in Table 1. To analyze awareness of the task, the data from aware participants were excluded and a 2 (age group) x 2 (test condition) x 3 (study status) ANOVA with repeated-measures on the second and third factors was used to analyze hits and false alarms in the recognition data. As in the original analysis, there was a significant effect of having previously studied the item when making a recognition decision, (M auditory = .625, M visual = .605, M new = .325), F(2, 60) = 52.96, MSE = .036. The interaction between age and study modality was significant, F(2, 60) = 4.87, MSE = .036. This effect was primarily due to higher false alarm rates for older adults than younger adults (M young = .225, M older = .403). The sentence manipulation elicited a memory illusion of 9.4%, (M predictive = .499, M nonpredictive = .405), F(1, 30) = 7.91, MSE = .038. There was no evidence of the illusion's size differing between age groups, Modality and Illusions of Recognition 26 F(1, 30) = .78, MSE = .038. Inspection of the illusion size for unaware participants in Table 1 shows that the illusion size remained fairly constant when compared to the values from the original analysis, except that younger adults showed a reduction of 7.6% in the illusion size for items presented visually, and a reduction of 6.0% for items new at test. This effect was attributable to an increase in "old" responses for words appearing after predictive sentences as there appears to be no effect for words following nonpredictive sentence frames. The interaction between study modality and the sentence manipulation, F(2, 60) = .04, MSE = .029, failed to reach significance just as when all participants were included in the analysis. Finally, the trend towards a three-way interaction between study modality, the sentence manipulation, and age group found when including all participants did not approach significance in this analysis, F(2, 60) = .81, MSE = .029. In sum, the ANOVA yielded the same basic results as when all including all participants. There was some evidence, however, that awareness of the task might have resulted in an exaggeration of the memory illusion for younger adults. For older adults, awareness of the task did not affect the memory illusion. Discussion Fluency and Memory Illusions In this experiment, the roles of perceptual and conceptual fluency in recognition memory were examined. A memory illusion was created by manipulating the fluency of processing a target item by presenting the item after either predictive or nonpredictive sentence frames. Encoding was manipulated by presenting words either auditorily or visually. Younger adults were more susceptible to a perceptually based memory illusion than older adults. Older adults were less affected by perceptual congruency between study and test, and instead tended to show a larger memory illusion when the judgement was based primarily upon conceptual fluency. The Modality and Illusions of Recognition 27 older adults showed a larger illusion for auditorily studied words (12.4%) than for visually studied words (6.6%) and words new at test (8.4%). Younger adults showed almost no illusion for words studied auditorily (1.8%), a large illusion for visually studied words (17.8%), and an illusion of 10.1% for new words. These results are in accord with Prull (1997, Experiment 1), in which a significant memory illusion based upon a data-driven manipulation of revealing perceptually degraded words at test was found only for younger adults. In Prull (1997, Experiment 2), a manipulation of conceptual fluency led to a greater memory illusion for older adults than for younger adults. The present experiment thus clarifies the results of Prull’s Experiment 2 in which conceptual and perceptual fluency were confounded. In Prull Experiment 2 (1997), words were studied visually meaning that at test both conceptual and perceptual fluency could have been contributing to the memory illusion. In the present experiment, when removing perceptual congruency between study and test, older adults showed a larger memory illusion than younger adults (young = 1.8%, older = 12.4%). From these results, it is clear that older adults are more likely to make memory errors when relying upon conceptually based processes, and that for younger adults this task is primarily perceptual in nature. As predicted from previous research on memory illusions (Luo, 1993), the sentence manipulation biased subjects to call predicted targets “old” as indexed by B′′D. There was no evidence of a between-group difference in bias. This shift in bias towards more liberal responses indicates that the predictive sentences were making items seem familiar (Luo, 1993). Study modality did not affect bias for words appearing after predictive sentence frames. However, from looking at the recognition data, it is clear that only younger adults showed a benefit from perceptual match between study and test. Modality and Illusions of Recognition 28 Older adults had more difficulty with deliberately recollecting studied items than younger adults. The fact that older adults had smaller d′ scores compared to younger adults indicates that the two groups were not equivalent in terms of deliberate recollective ability. A decline in recollective ability predicts that fluency judgments should become more important for older adults, and therefore, they should be more susceptible to fluency manipulations. Given this finding, it was expected that older adults should show a larger memory illusion. The decline in d′ did not necessarily result in a larger memory illusion for older adults, however. The mean illusion size for younger adults (9.9%) was slightly larger than that for older adults (9.1%). The mean illusion size is somewhat misleading though because of the trend towards a three-way interaction between group, test manipulation, and study status. Younger adults showed a memory illusion only when there was a perceptual match between study and test. Older adults showed a memory illusion for both encoding conditions. Evidently, the decrease in ability to deliberately recollect items from the study list left older adults more susceptible to the illusion in general, whereas for younger adults the memory illusion was primarily dependent upon study modality. In examining the overall magnitude of the memory illusion, it is necessary to account for individuals who indicated that they were aware of the test manipulation. If awareness of the task attenuates or reverses the effect, then the overall memory illusion seen in this experiment could have been confounded by combining the aware and unaware groups. This possibility was analyzed by submitting the data of participants who indicated that they did not notice the sentence manipulation to ANOVA. This analysis yielded the same basic pattern of results as when including all participants. However, for younger adults the magnitude of the memory illusion decreased for visually studied items from 17.8% to 10.2%, while the illusion for Modality and Illusions of Recognition 29 auditorily studied words was invariant. If awareness of the test manipulation was causing subjects to respond "old" to any item following a predictive sentence, then this effect should have occurred in all predictive conditions, which it did not. Finding that the exaggerated effect only occurred in the condition where younger adults were already showing the largest illusion again points to the centrality of perceptual fluency in recognition for younger adults. In summary, it appeared that the ability to discriminate between old and new items was not affected by study modality, but that the sentence manipulation did create a significant memory illusion. Although older adults were not as accurate at discriminating between old and new words as the young adults, they did not show a larger overall memory illusion. This effect, as well as the lack of a between-group difference in bias, suggests that there are differences in the familiarity processes between younger and older adults because the magnitude of the illusion tended to vary depending upon study condition. Older adults were apparently more likely to rely upon attributions towards conceptual processing in assessing familiarity because they showed a consistent illusion for both encoding conditions. Younger adults were more likely to rely upon attributions towards perceptual fluency because their illusion appeared only for visually studied items. Although the reason for these qualitative changes in recognition are still unclear, there is reason to believe that the increased tendency to misattribute conceptual fluency in older adults might have a neuropsychological basis situated in the frontal lobes (West, 1996). Naming Latencies From looking at the naming latency data, it was apparent that sentence context facilitated word naming, indicating that the sentences were successfully predicting the target words. An unexpected finding, however, was the lack of a repetition priming effect. Repetition priming, when a second exposure of an item increases identification performance or decreases naming Modality and Illusions of Recognition 30 latencies, is a consistent effect seen in both younger and older adults. One of the most common findings throughout the aging literature is that there are large differences in direct tests of memory, but that indirect tests such as repetition priming show very little decline (La Voie & Light, 1994). The lack of a repetition priming effect may be due to variability in the reaction time data in conjunction with an insufficient number of trials per condition to yield a satisfactorily stable measure of central tendency. Although manipulating familiarity to induce false recognition can successfully account for the memory illusion seen in this experiment, studies of the revelation effect suggest that there might be alternative explanations. Watkins and Peyniricioglu (1990) found that when an item was presented in a perceptually degraded form prior to the recognition decision, there was an increased probability of calling the item "old". Watkins and Peyniricioglu (1990) called this the revelation effect because revealing items prior to a recognition decision induced a memory illusion. The original explanation of this effect was that revealing a word at test caused an increase in either perceptual or conceptual fluency, or was due to the manipulation biasing participants to call the item "old" (Luo, 1993). However, from recent work by Westerman and Greene (1996) it appears that the revelation effect cannot be explained by either fluency misattributions or bias effects. In Westerman and Greene's Experiment 7 (1996), participants were to solve an anagram of a target word prior to making a recognition decision; a manipulation which has been shown to produce a revelation effect. However, instead of presenting the word solved as an anagram as the target, a word unrelated to the anagram was presented as the target item. Although there is no reason why perceptual or conceptual fluency should be influencing familiarity for a word that was not solved prior to the recognition decision, there was still a significant revelation effect. In further studies, Westerman and Greene (1998) found a consistent Modality and Illusions of Recognition 31 revelation effect with manipulations such as time delays prior to the recognition test, being successful at completing a task prior to recognition, and performing non-relevant tasks prior to a recognition decisions such as a memory span task with letters as stimuli. The only condition in which a revelation effect was not found was when a memory-span task using numbers as stimuli was performed prior to recognition. Westerman and Greene (1998) conclude that the willingness to call an item "old" after revelation cannot be explained by either increased familiarity or a biasing effect of the test. Because the only manipulation that did not induce a revelation effect was a digit span task using numbers as stimuli, they argue that any kind of manipulation with words will influence recognition because of a general semantic activation of words throughout memory. As in the study of perceptual illusions, the study of memory illusions can be used to explain underlying processes in memory, not by showing in which situations a memory system works correctly, but by showing in which situations it works incorrectly. Memory illusion paradigms can thus be used to infer underlying processes and systems by showing situations in which people remember incorrectly. In this study, the use of a memory illusion paradigm was successful in showing a qualitative and quantitative difference in the underlying basis of recognition memory in younger and older adults. The present study shows that younger adults' assessments of familiarity are more reliant upon judgments towards perceptual fluency, whereas familiarity for older adults is more influenced by conceptual fluency. Future research may elucidate the differential importance encoding and retrieval strategies have in this change in recognition memory. Future research on familiarity processes in aging should also examine the role frontal functioning has on recognition memory. The frontal lobes are one of the most severely affected neurological areas in natural aging (Scheibel, 1996), and previous research has Modality and Illusions of Recognition 32 shown that frontal deficits are a primary cause of memory illusions (Schacter & Curran, 1995). Therefore, accounting for frontal functioning may be able to account for the differences seen between younger and older adults in this experiment. Finally, future research might reveal whether the memory illusion in this experiment is better explained by fluency misattributions or a global activation model such as that suggested by Westerman and Greene (1998) to account for memory illusions caused by revelation. Modality and Illusions of Recognition 33 ReferencesAtkinson, R. C., & Juola, J. F. (1974). Search and decision processes in recognition memory.In D. H. Krantz, R. C. Atkinson, R. D. Luce, & P. Suppes (Eds.), Contemporary developments in mathematical psychology: Vol. 1. Learning, memory, and thinking. San Francisco: W. H.Freeman.Banks, W. P. (1970). Signal detection theory and memory. Psychological Bulletin, 74, 81-99.Bartlett, J. C., Strater, L., & Fulton, A. (1991). False recency and false fame of faces inyoung adulthood and old age. Memory & Cognition, 19, 177-188. Blaxton, T. A. (1989). Investigating dissociation among memory measures: Support for atransfer-appropriate framework. Journal of Experimental Psychology: Learning, Memory, andCognition, 15, 657-668.Blaxton, T. A., Bookheimer, S. Y., Zeffiro, T. A., Figlozzi, C. M., Gaillard, W. D., &Theodore, W. H. (1996). Functional mapping of human memory using PET: Comparisons ofconceptual and perceptual tasks. Canadian Journal of Experimental Psychology, 50, 42-56.Borkowski, J. G., Benton, A. L., & Spreen, O. (1967). Word fluency and brain damage.Neuropsychologia, 5, 135-140.Bransford, J. D., Franks, J. J., Morris, C. D., & Stein, B. S. (1979). Some general constraintson learning and memory research. In Cermak, L. S. & Craik, F. I. M. (Eds.) Levels of processingin human memory. Hillsdale, NJ: Erlbaum.Brown, A. S. & Mitchell, D. B. (1994). A reevaluation of semantic versus nonsemanticprocessing in implicit memory. Memory & Cognition, 22, 533-541.Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.Coren, S., Ward, L. M., & Enns, J. T. (1994). Sensation and perception. New York: Modality and Illusions of Recognition 34 Harcourt Brace College Publishers.Craik, F. I M. (1994). Memory changes in normal aging. Current Directions in PsychologicalScience, 3, 155-158. Craik, F. I. M., & McDowd, J. M. (1987). Age differences in recall and recognition. Journalof Experimental Psychology: Learning, Memory, and Cognition, 13, 148-151.Craik, F. I. M., Morris, L. W., Morris, R. G., & Loewen, E. R. (1990). Relations betweensource amnesia and frontal functioning in older adults. Psychology and Aging, 5, 148-151.Curran, T. & Hintzman, D. L. (1995). Violations of the independence assumption in process dissociation. Journal of Experimental Psychology, 21, 531-547.Donaldson, W. (1992). Measuring recognition memory. Journal of ExperimentalPsychology: General, 12, 275-277.Dywan, J., & Jacoby, L. (1990). Effects of aging on source monitoring: Differences insusceptibility to false fame. Psychology and Aging, 5, 379-387.Fischler, I. (1977). Semantic facilitation without association in a lexical decision task.Memory & Cognition, 5, 335-339.Francis, W. N., & Kucera, H. (1982). Frequency analysis of English usage: Lexicon andgrammar. Boston, MA: Houghton Mifflin.Gabrieli, J. D. E., Desmond, J. E., Demb, J. B., Wagner, A. D., Stone, M. V., Vaidya, C. J., &Clover, G. H. (1996). Functional magnetic resonance imaging of semantic memory processes inthe frontal lobes. Psychological Science, 7, 278-283.Gabrieli, J. D. E., Fleischman, D. A., Keane, M. M., Reminger, S. L., & Morrell, F. (1995).Double dissociation between memory systems underlying explicit and implicit memory in thehuman brain. Psychological Science, 6, 76-82. Modality and Illusions of Recognition 35 Gardiner, J. M. (1988). Functional aspects of recollective experience. Memory & Cognition,18, 23-30.Grady, C. L., McIntosh, A. R., Horwitz, B., Maisog, J. M., Ungerleider, L. G., Mentis, M. J., Pietrini, P., Schapiro, M. B., & Haxby, J. V. (1995). Age-related reductions in humanrecognition memory due to impaired encoding. Science, 269, 218-221.Graf, P. (1990). Life-span changes in implicit and explicit memory. Bulletin of thePsychonomic Society, 28, 353-358.Heindel, W. C., Salmon, D. P., Shults, C. W., Walicke, P. A., & Butters, N. (1989). Neuropsychological evidence for multiple implicit memory sites: A comparison of Alzheimer's,Huntington's, and Parkinson's disease patients. The Journal of Neuroscience, 9, 582-587.Hochhaus, L. (1977). A table for the calculation of d′ and ß. Psychological Bulletin, 77, 375376.Holender, D. (1986). Semantic activation without conscious identification in dichoticlistening, parafoveal vision, and visual masking: A survey and appraisal. Behavioral and BrainSciences, 9, 1-23.Jacoby, L. L. (1991). A process dissociation framework: Separating automatic fromintentional uses of memory. Journal of Memory and Language, 30, 513-541.Jacoby, L. L., & Dallas, M. (1981). On the relationship between autobiographical memoryand perceptual learning. Journal of Experimental Psychology: General, 110, 306-340.Jacoby, L. L., Kelley, C. M., Dywan, J. (1989). Memory Attributions. In H. L. Roediger III &F. I. M. Craik (Eds.), Varieties of memory and consciousness: Essays in honour of EndelTulving. Hillsdale, NJ: Erlbaum.Jacoby, L. L., Toth, J. P., & Yonelinas, A. P. (1993). Separating conscious and unconscious Modality and Illusions of Recognition 36 influences of memory: Measuring recollection. Journal of Experimental Psychology: General,12, 139-154.Jennings, J. M. & Jacoby, L. L. (1997). An opposition procedure for detecting age-related deficits in recollection: Telling effects of repetition. Psychology and Aging, 12, 352-361.Jennings, J. M., & Jacoby, L. L. (1993). Automatic versus intentional uses of memory:Aging, attention, and control. Psychology and Aging, 8, 283-293.Johnston, W. A., Dark, V. J., & Jacoby, L. L. (1985). Perceptual fluency and recognitionjudgments. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 3-11. Knowlton, B. J., & Squire, L. R. (1995). Remembering and knowing: Two differentexpressions of declarative memory. Journal of Experimental Psychology: Learning, Memory, &Cognition, 21, 699-710.Knowlton, B. J., Squire, L. R., Paulsen, J. S. & Swerdlow, N. R. (1996). Dissociationswithin nondeclarative memory in Huntington's disease. Neuropsychology, 10, 538-548.Kolb, B., & Wishaw, I. Q. (1996). Fundamentals of human neuropsychology (4th ed.). NewYork: Freeman.Koustaal, W. & Schacter, D. L. (1997). Gist-based false recognition of pictures in older andyounger adults. Journal of Memory and Language, 37, 555-583.La Voie, D. L., & Light, L. L (1994). Adult age differences in repetition priming: A meta-analysis. Psychology and Aging, 9, 539-553.Light, L. L., & Albertson, S. A. (1989). Direct and indirect tests of memory for categoryexemplars in young and older adults. Psychology and Aging, 4, 487-492.Light, L. L., La Voie, D., Valencia-Laver, D., Albertson Owens, S. A., & Mead, G. (1992).Direct and indirect measures of memory for modality in young and older adults. Journal of Modality and Illusions of Recognition 37 Experimental Psychology: Learning, Memory, and Cognition, 18, 1284-1297.Light, L. L., Prull, M. P, La Voie, D. J., & Healy, M. R. (in press) Dual process theories ofmemory in old age. In Perfect, T., Maylor, E. (Eds.), Theoretical debate in cognitive aging. Oxford (UK): Oxford University Press.Light, L. L., & Singh, A. (1987). Implicit and explicit memory in young and older adults.Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 531-541.Light, L. L., Singh, A., & Capps, J. L. (1986). Dissociation of memory and awareness inyoung and older adults. Journal of Clinical and Experimental Neuropsychology, 8, 62-74. Luo, C. R. (1993). Enhanced feeling of recognition: Effects of identifying and manipulatingtest items on recognition memory. Journal of Experimental Psychology: Learning, Memory, andCognition, 19, 405-413.Mandler, G. (1980). Recognizing: The judgment of previous occurrence. PsychologicalReview, 87, 252-271.Mandler, G., & Boeck, W. J. (1974). Retrieval processes in recognition. Memory &Cognition, 2, 613-615.Mantyla, T. (1993). Knowing but not remembering: Adult age differences in recollectiveexperience. Memory & Cognition, 21, 379-388.Nelson, M. J., & Denny, E. C. (1960). The Nelson-Denny reading test (revised by James I.Brown). Boston, MA: Houghton Mifflin.Parkin, A. J., & Walter, B. M. (1992). Recollective experience, normal aging, and frontaldysfunction. Psychology and Aging, 7, 290-298.Perfect, T. J., Williams, R. B., & Anderton-Brown, C. (1995). Age differences reported inrecollective experience are due to encoding effects, not response bias. Memory, 3, 169-186. Modality and Illusions of Recognition 38 Prull, M. (1997). Illusions of recognition in young and older adults. Unpublished doctoraldissertation, Claremont Graduate School, California.Richardson-Klavehn, A., & Bjork, R. (1988). Measures of memory. Annual Review of Psychology, 39, 475-543.Roediger, H. L. III. (1996). Memory illusions. Journal of Memory and Language, 35, 76-100.Roediger, H. L. III.(1990). Implicit memory: A commentary. Bulletin of the PsychonomicSociety, 28, 373-380. Schacter, D. L. (1987). Implicit memory: History and current status. Journal ofExperimental Psychology: Learning, Memory, and Cognition, 18, 501-518.Schacter, D. L., & Curran, T. (1995). The cognitive neuroscience of false memories.Psychiatric Annals, 25, 726-730.Schacter, D. L., Kazniak, A. W., Kihlstrom, J. F., & Valdiserri, M. (1991). The relationbetween source memory and aging. Psychology and Aging, 6, 559-568.Schacter, D. L., Koustaal, W., Johnson, M. K., Gross, M. S., & Angell, K. E. (1997). Falserecollection induced by photographs: A comparison of older and younger adults. Psychologyand Aging, 12, 203-215.Schacter, S., & Singer, J. (1962). Cognitive, social and physiological determinants ofemotional states. Psychological Review, 69, 379-399.Scheibel, A. B. (1996). Structural and functional changes in the aging brain. In J. E. Birren,& K. W. Schaie (Eds.), Handbook of the psychology of aging. (4th ed.). San Diego, CA:Academic Press.Spencer, W. D., & Raz, N. (1995) Differential effects of aging on memory for content and Modality and Illusions of Recognition 39 context: A meta-analysis. Psychology and Aging, 10, 527-539.Swets, J. A., & Green, D. M. (1974). Signal detection theory and psychophysics. (2nd ed.).New York: Robert E. Krieger Publishing Co. Tulving, E. (1985). Memory and consciousness. Canadian Psychologist, 26, 1-12.Wagner, A. D., Gabrieli, J. D. E., & Verfaellie, M. (1997). Dissociations between familiarityprocesses in explicit recognition and implicit perceptual memory. Journal ofExperimental Psychology: Learning, Memory, & Cognition, 23, 305-323.Watkins, M. J., & Peynircioglu, Z. F. (1990). The revelation effect: When disguising items induces recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition,20, 671-679.West, R. L. (1996). An application of prefrontal cortex function theory to cognitive aging.Psychological Bulletin, 120, 272-292.Westerman, D. L. & Greene, R. L. (1998). The revelation that the revelation effect is notdue to revelation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24,377-386.Westerman, D. L., & Greene, R. L. (1996). On the generality of the revelation effect.Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, 1147-1153.Whittlesea, B. W. A. (1993). Illusions of familiarity. Journal of Experimental Psychology:Learning, Memory, and Cognition, 19, 1235-1253.Whittlesea, B. W. A., Jacoby, L. L., & Girard, K. (1990). Illusions of immediate memory:Evidence of an attributional basis for feelings of familiarity and perceptual quality. Journal ofMemory and Language, 29, 716-732. Modality and Illusions of Recognition 40 Table 1Mean Recognition Judgments and Standard Deviations (In Parentheses) for All Participants andAware Participants Only as a Function of Age, Test Status, Modality of Study, and Sentence TypeYoungOlder Sentence Type Auditory Visual New Auditory Visual New All participantsPredictive .702