This item is only available as the following downloads:
SHOP TIL YOU DROP: A WORKING MEMORY TRAINING PROGR AM FOR OLDER ADULTS TO IMPROVE MEMORY IN A GROCERY SHOPPING TRANSFER TASK BY MELANIE BAUER A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Dr. Heidi Harley Sarasota, Florida May, 2010
ii Acknowledgments I would like to thank my thesis sponsor Dr. Heidi Harley for her guidance throughout the thesis process, especially for her c ritical eye when it came to making revisions to this final thesis. I was both motivat ed by her clear direction and comforted by her positive attitude. I would also like to tha nk Dr. Michelle Barton and Dr. Charlene Callahan for being on my committee. All three have been great influences throughout my academic career in my ways of thinking and writi ng as a psychological researcher, and helped me to be accepted to a wonderful graduat e program for this coming school year. I want to thank my older adults who endured the ri gorous memory training program. It was a long road but their perseverance is admirable. Their participation taught me valuable lessons about the research proce ss, and I will greatly benefit in the future from my experiences with them. I would also like to thank Barbara Celnar, Erin McLeod, and others at the Senior Friendship Center who allowed me to use their facilities to conduct the training as well as Marci a Ryan from the Jefferson Center who helped me to gather participants, and Jennifer LeLa urin for pointing me there. Finally, I want to thank my friends Corrie Ethered ge for our thesis-writing sessions, and Upom Malik for helping me to find tha t article that was eluding me.
iii Table of Contents ACKNOWLEDGEMENTS ii TABLE OF CONTENTS iii LIST OF FIGURES AND TABLES v ABSTRACT viI INTRODUCTION 1 Working Memory and Age-Related Decline 1 Cognitively Demanding Daily Activities 3 Activity inventories 4 Single-domain skills 14 Cognitive Training: Participants with Cognitive Im pairments 18 ADHD in children 18 Dementia in older adults 23 Cognitive Training: Healthy Participants 26 Mnemonic strategies as software 27 Working memory as hardware 30 Present Study 41 METHOD 45 Participants 45 Working Memory Tasks 47 Grocery Shopping Task 49 Questionnaires 51 Procedure 52
iv RESULTS 57 Training Tasks: Working Memory 57 Transfer Tasks: Grocery Shopping and Daily Functio ning 59 DISCUSSION 61 Effects of Training within the Trained Domain 6 2 Effects of Training on Two Far-Transfer Tasks 6 7 Future Research 73 Conclusion 74 REFERENCES 76 FIGURES 81 TABLES 91
v List of Figures and Tables FIGURES 81 Figure 1: Model of Working Memory and Long-Term Me morys Relations to Training and Transfer Tasks 81 Figure 2: Alphabet Span Stimulus and Participant R esponse Sequence 82 Figure 3: Operation Span Stimulus and Participant Response Sequence 83 Figure 4: Reading Span Stimulus and Participant Re sponse Sequence 84 Figure 5: Layout of Grocery Store with Food Items from Preand Post-Tests 85 Figure 6: Timeline of Study 86 Figure 7: Participant-by-Participant Performance o n WM Span Tasks Preand Post-Test 87 Figure 8: Performance of Participants 3B and 31B o n Training and Transfer Tasks 88 Figure 9: Participant-by-Participant Performance o n the Grocery Shopping Task Preand Post-Test 89 Figure 10: Performance of Participants 21A and 30B on Training and Transfer Tasks 90 TABLES 91 Table 1: Participant Characteristics and MANOVA An alysis 91 Table 2: Performance on Training and Transfer Task s by Group 92 Table 3: Grocery Shopping Lists for Preand PostTest 93 Table 4: Daily Activities Questionnaire Items 9 4
vi Table 5: Daily Functioning Questionnaire Items 95 Table 6: Effect of Strategies on Performance on Tr aining and Transfer Task Means 96 Table 7: Effect of Shopping Methods on Performance on Shopping Task Means 97 Table 8: Effect of Gender on Performance on Shoppi ng Task Means 98
vii SHOP TIL YOU DROP: A WORKING MEMORY TRAINING PROGR AM FOR OLDER ADULTS TO IMPROVE MEMORY IN A GROCERY SHOPPING TRANSFER TASK Melanie Bauer New College of Florida, 2010 ABSTRACT Working memory (WM) is a foundational component of cognition. As such, benefits achieved through training in WM have the p otential to transfer and improve other areas of cognition, such as long-term memory (LTM). WM functioning decreases slowly and steadily after the age of 20 years and i mpedes complex tasks and daily functioning as one ages. The present study added t o previous research on WM training by training healthy older adults on 3 WM span measu res: alphabet span, operation span, and reading span. Training occurred in a group set ting over five 1.5-hour sessions, during which the difficulty level of each WM span t ask was progressively increased. Improvement from preto post-test was measured not only in the trained domain on the 3 WM tasks, but also in an untrained domain on a groc ery shopping transfer task. The grocery shopping task occurred in a grocery store a nd entailed a short study period for a list of 15 items followed by participant retrieval of those items from the store shelves. This task can be seen as tapping both LTM and WM sk ills. Overall, participants did not improve from preto post-test on either the traine d WM tasks or transfer grocery
viii shopping task. Reasons for a lack of training and transfer effects include various methodological issues with the training program, su ch as lack of adaptivity and rigor, and a group setting. However, there were a few example s of success in both the training and transfer tasks that are cited. Future research sho uld consider the weaknesses of the present study as well as include real-life transfer tasks. ______________________________________ Dr. Heidi Harley Division of Social Sciences
1 Working Memory and Age-Related Decline Working memory (WM) is one of the cognitive primi tives, along with inhibition and processing speed, thought to influen ce all other areas of cognition (Borella, Carretti, & De Beni, 2008). There exists only a li mited amount of cognitive resources available to WM, which are used to store information while simultaneously processing incoming or recently accessed information in servic e of other high-order cognitive tasks (Baddeley & Hitch, 1974 as cited in Borella et al., 2008). This information is limited in capacity and duration in WM, with a small amount of information being available mostly for just a short period of time (Olesen, Westerberg & Klingberg, 2004). To investigate this multi-tasking area of cognition, typical WM me asures (e.g., complex span tasks) require participants to store some piece of informa tion while completing a concurrent attention-demanding task. For example, participant s may be required to mentally solve a series of simple mathematical equations while remem bering a word that is paired with each one (Turner & Engle, 1989). In this way, part icipants must divide their WM resources between processing the math equations and storing the paired words. WM does not function at a consistent level through out ones life but rather travels a fairly clear trajectory of decline. To examine i ts changes across the lifespan, it is useful to categorize WM into one of the two components of intelligent thought, one representing accumulated knowledge of the world and the other representing the ability to use that knowledge flexibly and adaptively (Park e t al., 2002, p.132). WM is a component of this latter area of cognition, the fle xible, adaptive use of knowledge through cognitive control. Put differently, this d ichotomy entails the former stated representational knowledge, or crystallized cognit ion, and the latter stated cognitive
2 control, or fluid cognition. Crystallized cognit ion increases during childhood, slows but still increases throughout adulthood, and remai ns stable in old age (Craik & Bialystok, 2006). Fluid cognition, on the other ha nd, increases in power, speed, and complexity from infancy to young adulthood, but de clines through old age (Craik & Bialystok, 2006, p. 132). The findings of Park et al. (2002) support this assertion of differential trajectories of functioning for fluid and crystallized cognition into old age. The researchers conducted a cross-sectional study w ith participants across the lifespan, measuring their levels of WM, processing speed, lon g-term memory (LTM), short-term memory, and verbal knowledge. They found that age was negatively correlated with all cognitive measures, except with the measures of ver bal knowledge (crystallized cognition) that showed an increase with increasing age. Crystallized cognition aside, the decline in the other areas of cognition seems to be steady from age 20 to 80 and beyond, with the same magnitude of decline from age 20-30 y ears and 70-80 years (Park et al., 2002). However, the proportional loss becomes grea ter in later years, with small decreases in fluid cognition producing great effect s. One reaches a point when the absolute loss experienced begins to impede demandin g tasks (likely between 40 and 50 years) and then has implications for daily function ing (between 80 and 90; Park et al., 2002). Decline in fluid cognition has been attributed to deterioration of the frontal regions of the brain (Raz, 2000 as cited in Craik & Bialystok, 2006). These frontal regions are important for planning, decision making and other executive control functions including WM (Craik & Bialystok, 2006). As a fundamental component of cognition, this deterioration in WM has implication s for other aspects of a persons
3 cognition. Since there is a steady deterioration i n most areas of cognition with age decline in speed of processing, LTM, and short-term memory in addition to WM (Park et al., 2002)it is possible that there is a relation between these concurrent deteriorations. If so, then training in one of them may affect all of them. This suggestion of training one area to improve another is called transfer. Ther e has been much research in the area of training and transfer in aging research, but the ma jority of the results has not evidenced much transfer. Training effects have mostly been m odest or nonexistent, and the ability for these training benefits to transfer to new cont exts is rare, especially in older adults (Persson & Reuter-Lorenz, 2008). However, there are clear cases of successful train ing programs for people across the lifespan that compel continued investigation in the area. Both through examination of a variety of cognitively demanding activities in wh ich people participate in their everyday lives as well as laboratory-based training interven tions designed to target specific skills, research adds to the knowledge of which activities may serve as protective factors against or aid to reverse cognitive decline. These data su ggest avenues for training programs that may be effective in addressing age-related decline. The present study focused on this latter area of research to target WM in a lab-based WM training program. Moreover, given the fundamental value of WM for other areas o f cognition, transfer to LTM task in a real-life context (the grocery store) was also ex amined. Cognitively Demanding Daily Activities There is no consensus about the precise factors, ph ysiological and/or environmental, that buffer against cognitive declin e. Why one person ages successfully while another ages less successfully is still large ly unknown. Therefore, it is important to
4 consider all aspects of a persons life to gain clu es about which activities may be useful buffers. Research in this area of cognitively dema nding daily activities is diverse, and its results are similarly mixed. In this section two a reas of research are cited: activity inventories in which participants report on their g eneral cognitive activities as well as social and physical ones, and single-domain skills in which participants performance in a specific activity is examined in relation to cognit ion. In general, engagement in cognitively demanding daily activities seems to be related to greater performance on various cognitive measures, and participation in a single skill has less transfer potential. It is important to interpret these data carefully a s the direction of causation is not certain: greater participation in cognitively demanding acti vities may lead to higher cognitive scores, or higher cognitive scores (and a more acti ve mind) may permit participation in more cognitively demanding activities. Nonetheless the research examines ways older adults stay active and provides information about t he types of people and lifestyles that are related to higher cognitive performance. Activity inventories. A cognitively-engaging lifestyle has been suggeste d as one means of providing a buffer against cognitive decli ne. Studies have examined leisure activities ranging from physical and social activit ies to more explicitly cognitively demanding ones. Most find only modest domain-speci fic positive effects, with cognitive activities being the only ones to transfer to lab -based cognitive measures. Notably, many of these effects were mediated by WM measures. In some studies, lifespan differences in activity participation and effects w ere also addressed. The demand on cognition that a given activity affo rds may vary greatly between individuals. Therefore, it is useful to inquire di rectly about the ways in which people
5 engage with their environment. Schooler and Mulatu (2001) sought to analyze peoples perceptions of the cognitive demands of complex lei sure activities and how they affected cognitive level. Participants ( N = 635) were part of a longitudinal study and were couples interviewed in 1974 and again at a 20-year follow-u p (range of 41 to 88 years at followup). A complex activity was defined as requiring thought and independent judgment. The degree of participation in complex leisure acti vities was measured as follows: number of books read (within past 6 months), number of magazines read regularly and their intellectual complexity; frequency of visits to museums, concerts, and plays; and number and hours spent engaged in special interests and hobbies. The degree of complexity of these activities was determined along two dimensions: interaction with things and with data. Things were defined as equ ipment or products and were rated on a Likert-style complexity scale ranging from mere handling to as complex as setting up machines and equipment. Data were defined as numbers/words and ideas/concepts and were rated from simply reading instructions t o synthesizing. The overall complexity of the activity was also rated as not a t all complex to setting up of a complex system of analysis, synthesis, or both. C ognitive level was measured along two dimensions: intellectual flexibility and standard c ognitive functioning. Intellectual flexibility (fluid cognition) was ones ability to handle the complexity and cognitive demands of a complex situation (based on several pe rformance-based measures). General cognitive functioning (crystallized cogni tion) included many objective measures such as memory recall and a vocabulary tes t, as well as three more subjective measures assessed by an interviewer (interviewers assessment of the participants intelligence, the adequacy of the participants res ponse to a hypothetical question
6 requiring consideration of potential costs and bene fits, and the adequacy of providing reasons in support of an argument). The measure of intellectual flexibility was very strongly related to the more standard measures of c ognitive functioning, and the two were used together as the measure of cognitive level. However, this intellectual flexibility measure was not found to be significantly related t o age. This result is unlike other measures of fluid cognition in which these skills t end to decrease with age (Park et al., 2002). Overall, as age increased, cognitive level (compos ite score of intellectual flexibility and cognitive functioning) decreased. However, participation in more complex leisure activities was positively correlated with i ncreased cognitive level. Relatedly, participation in less complex leisure activities wa s correlated with decreased cognitive level. Also, having a higher cognitive level was p ositively correlated with greater participation in more complex leisure activities. This finding is in line with previous studies that have suggested people of higher cognit ive level participate in more cognitively demanding activities until decline begi ns to occur (e.g., Hultsch, Hertzog, Small, & Dixon, 1999). These results can also prov ide support for the rough-hewn theory of cognitive functioning: since complex env ironments reward cognitive effort, individuals in such environments should be motivate d to develop their intellectual capacities to generalize the resulting cognitive pr ocesses to other situations (Schooler & Mulatu, 2001, p. 492). Considering the findings of this study, participants may be challenged by their complex activities and so devel op their own cognitive skills to continue to succeed.
7 Extending the research of the above study on leisu re activities, Schooler, Mulatu, and Oates (1999) explored the effects of occupation al activities on cognition. Participants ( N = 233) were part of the same longitudinal study ci ted above but were examined in reference to their work-related activit ies as opposed to leisure ones. The complexity of their work was determined along three dimensions: things and data as in the previous study, with the addition of people . People was assessed based on required interactions with human beings and was rat ed from less complex (serving) to more complex (mentoring). The overall complexity of work was also rated and the same scales as the previous study were used for thi s and the other dimensions (things and data), with complexity for each dimension ran ging from less to more complex. Cognitive level was also measured along the same two dimensions: intellectua l flexibility and standard cognitive functioning. As was previously found, more complex work activit y was related to higher cognitive functioning, and lower cognitive function ing was related to less complex work. However, unlike the previous study, participants of increased age seemed to receive a greater cognitive benefit from participating in mor e complex work activities than did younger. Therefore, more complex work was related to greater increases in older participants cognitive level than younger particip ants. Due to the differential results found between these two studies, it seems there is a marked difference between occupational complexity and leisure-time complexity These findings can again be explained using the rough-hewn theory of cognitiv e functioning in which participants must maintain their cognitive skills in order to su cceed at complex work. Older workers seem to benefit more from this experience with grea ter increases in cognitive functioning
8 as related to complex work than younger workers. T he differential results for the two age groups may be a difference in ability and motivatio n. Older workers may have to work harder than younger workers at more complex activit ies, and so receive greater resultant gains, while for younger workers the complex activi ties are not as cognitively demanding and so do not challenge them and provide the same o pportunity for cognitive gains. Unlike the previous studies findings that partici pation in cognitively demanding activities buffers against cognitive decline, Salth ouse, Berish, and Miles (2002) did not find protective effects based on cognitive activiti es. Participants ( N = 204, aged 20 to 91 years) were measured on their cognitive levels (mea sures of spatial ability, reasoning, episodic memory, and vocabulary) and weekly partici pation frequency for 22 various cognitive activities (e.g., television watching, ch ess playing, and shopping). While the cognitive activities were similar to those of previ ous studies, in this study they were accompanied by ratings of participants perceptions of how cognitively demanding each activity was. This measure was included because wh ile a 20-year-old may perceive grocery shopping as not especially cognitively dema nding, a 91-year-old may perceive it to be very cognitively demanding. Additionally, the tendency to participate in cognitively demanding activities (need for cognition) was als o measured. This measure was included to distinguish between individuals seekin g, and quality of experience in, activities in that individuals high in need for co gnition tend to have active, exploring minds, and, through their senses and intellect, the y reach and draw out information from their environment (Cacioppo, Petty, Feinstein, & J arvis, 1996 as cited in Salthouse, Berish & Miles, 2002, p. 550).
9 Increased age was related to higher ratings of cog nitive demand for television and lower ratings for teaching/attending class and play ing chess. However, this difference may partially be explained by the shift in the natu re of activities as people age. For example, younger people (less than 50 years old) ar e more likely to attend academic, and cognitively demanding classes than are older people (over 50 years old). The average level of cognitive demand was calculated for each a ctivity, and older people tended to participate less in activities with higher cognitiv e demands. This may be because greater cognitive decline, which is closely tied to increas ed age, may lead people to participate less in cognitively demanding activities. In refer ence to cognitive levels, only the tests of vocabulary were related positively with age and par ticipation in cognitive activities. It is not surprising that increased age saw an increase i n vocabulary (crystallized knowledge) as this has been shown previously (Park et al., 200 2). Interestingly, participants who had higher levels of cognitive stimulation (high partic ipation frequency and high ratings of cognitive demand) had greater vocabulary knowledge with increased age. In this way, cognitive stimulation seems to be related to at lea st one measure of cognition, vocabulary. Overall, measures of fluid cognition ( spatial ability and reasoning) and episodic memory were not found to be related to cog nitive aging. Also, participants need for cognition declined with age, and cogniti ve stimulation did not buffer against age-related decline in cognitive functioning. One additional comparison was made between the par ticipants level of education, which can be considered another measure of cognitive performance, and cognitive stimulation and level. Unlike the sample as a whole, for participants with a relatively low level of education, higher scores on the measures for fluid cognition and
10 episodic memory were significantly related to highe r cognitive stimulation. Perhaps greater stimulation has a greater impact on those p eople with lower cognitive performance (as indicated by lower level of educati on). A final consideration suggested by the researchers is in reference to the cross-sec tional nature of the study, as opposed to longitudinal. Because data were collected only at one point, the lack of relation found between the daily activities and measures of cognit ion may be due to a characteristic of cognitive functioning in which the average or cumul ative level of cognition accounts for group differences, as opposed to an instantaneous c ognitive level as measured in this study. That is, the rate of decline would likely b e more variable between a group of young adults and old adults than a measure of absol ute cognitive level. Relatedly, the rate of decline in ones cognition may be more grea tly affected by participation in cognitively demanding activities than ones absolut e cognitive level, and this effect may be age-specific. Therefore, a person high in cogni tion who had experienced cognitive decline and another person who was initially low in cognition and had experienced no cognitive decline would appear the same on an insta ntaneous measure of cognition. This cross-sectional design may partially explain the la ck of relation found in this study as opposed to the positive correlations found between cognitive activity and cognitive measures in the above longitudinal studies by Schoo ler and colleagues (1999, 2001). Hultsch et al. (1999) examined data from a longitu dinal study in which several aspects of engaging lifestyles were analyzed as pot ential predictors for an extensive array of cognitive measures. Participants ( N = 250, 55 to 86 years of age at start) were tested on these measures (listed below) of lifestyle pract ices and cognition 3 times over 6 years. The measures of cognition included WM (e.g., solvin g math problems while holding one
11 number in memory from each for later recall), fact recall (general knowledge measure), word recall (free recall of word lists), story reca ll (gist recall of stories), vocabulary test, verbal fluency (recall of words with similar and op posite meanings), reading comprehension (of short passages), comprehension s peed (i.e., reading speed), and semantic speed (lexical decision task, determined w hether a string of letters or string of words were plausible as a word or sentence, respect ively). The measures of lifestyle included the frequency over the last year with whic h participants engaged in everyday activities (4 physical, 6 self-maintenance, 7 socia l, and 12 hobby/home activities) as well as 8 passive information processing activities such as radio-listening, and 27 novel information processing activities such as learning a language or playing bridge. Health was measured based on self-reported instances of ch ronic illness, number of illness episodes, instrumental health that prevented activi ty, overall self-rated health, and medication use. Personality was also measured alon g 5 personality traits (neuroticism, extraversion, openness to experience, agreeableness and conscientiousness), and was both stable across the 6 years and not especially v ariable throughout the group. There was an expected decline with age in all cogn itive measures, except the vocabulary and story recall tests, as well as less participation over time in all activities, especially those requiring novel information proces sing. However, those participants who engaged in cognitively demanding activities (no vel information processing activities) experienced less cognitive decline on n early all measures of cognitive change, though only modestly. These cognitive effects were all mediated by WM, such that increased age and decreased activity were related t o lower WM scores and also lower scores on nearly all other measures of cognition.
12 Importantly, no relationship was found between soc ial and physical activity and cognitive change. Given this finding, it might be that changes in cognition as one ages more strongly affect the cognitive activities people choose versus the social and phy sical ones. However, an alternative explanation suggeste d previously is that participants are intellectually active until their cognition begins to decline with increased age and deters participation in these cognitively demanding activi ties. This explanation gains support when the younger and older groups are compared. Th e younger group participated in more physical activities, self-maintenance activiti es, hobbies, and novel information processing than the older group. Additionally, ove r the 6 years of the study, there were significant decreases in physical activities, selfmaintenance activities, hobbies, and novel information processing activities for the group ove rall. These findings are important as they support the idea that participation in more co gnitively demanding tasks, such as novel information processing, as well as other acti vities decreases as age increases. When this is compared to the overall groups signif icant decline in health over 6 years with heightened chronic illness and medication usag e along with lowered instrumental and self-rated health, it becomes clearer, not surp risingly, that with increased age comes decreased activity engagement and overall poorer he alth. Still a third possibility is that the benefits from engagement in cognitive activitie s tend to be within the cognitive domain, while benefits from physical and social act ivities do not transfer to cognition. A final study, conducted by Wilson et al. (2002), used a daily activities inventory similar to the above studies to investigate the rel ation of cognitively-stimulating activities to the incidence of Alzheimers disease. Participa nts were 801 older Catholic priests, nuns, and brothers diagnosed as not having any form of dementia at the studys start.
13 Twenty cognitive tests were used to assess WM, epis odic memory, semantic memory, perceptual speed, and visuospatial ability. The fr equency of participation in cognitive leisure activities (those activities requiring inf ormation processing) was assessed and included the following activities: watching televis ion, listening to the radio, reading the newspaper, reading magazines, reading books, playin g games (e.g., cards, checkers, crosswords), and going to museums. Physical activi ty was also assessed and included such activities as exercise, gardening, and swimmin g. After about 4.5 years, 111 participants had develo ped Alzheimers ( M = 81 years, as compared to those who did not develop Alzheimer s, M = 74 years). Lower cognitive activity participation was related to increased age and lower education, though only weakly. A 1-point increase in cognitive activity s core (range of 1.57 to 4.71) was associated with a significant 33% reduction in the incidence of Alzheimers. Also, a 1point increase in cognitive activity was associated with a significant reduced decline in global cognition (by 47%), WM (by 60%), and percept ual speed (by 30%). To eliminate the possible explanation that some participants wer e in the early stages of Alzheimers at the studys beginning, episodic memory (noted to be affected in the very early stages of Alzheimers) and existence of the ApoE4 allele (an established risk factor for Alzheimers) were measured, but neither were signif icantly different between the groups. Also, physical activity was not related to reduced decline in any measure of cognitive functioning. In sum, greater participation in dail y cognitively demanding activities was related to a lower level of cognitive decline and r eduction in ones risk for Alzheimers disease.
14 Single-domain skills. While the above studies covered a wide range of act ivities, the following studies focused on skill in a single activity. In the previous studies cited, general participation in cognitively demanding acti vities (leisure and occupational; cognitive, social, and physical) was positively cor related with overall cognitive functioning for older adults, but mostly only to a modest degree. This section examines whether intense skill in one area has similar effec ts, though largely the effects were again limited with little transfer. Notably again, WM wa s a mediator of many of the relations found between performance within the given skill an d other cognitive measures. Hambrick and Engle (2002) examined the benefits of knowledge in one domain on memory along a continuum of expertise. Particip ants ( N = 181, 18 to 86 years) with various levels of knowledge and experience with bas eball were asked to recall events that transpired in recordings of fictional baseball game s. Recall of both game-relevant (e.g., number of outs and runs) and game-irrelevant (e.g., number of spectators in attendance) information was tested. General and baseball-speci fic knowledge were measured. The measure of general knowledge included a test of voc abulary and one of cultural knowledge, and the knowledge measure for baseball w as of its rules, regulations, and terminology. WM capacity (e.g., solving math probl ems while remembering target words) and processing speed (e.g., quickly determin ing whether two groups of letters were the same) were also tested with multiple measu res. Not surprisingly, knowledge of baseball (domain-sp ecific knowledge) was the most powerful predictor of memory for the games, ab ove the effects of general knowledge, WM, and age. This effect was greater fo r the game-relevant than for the game-irrelevant information. However, independent of baseball knowledge and age,
15 greater levels of WM were related to greater recall Baseball knowledge did not compensate for lower levels of WM on recall, but di d have a greater effect on improving recall for those who had higher WM levels. General knowledge had no effect on memory performance, though greater WM levels were related to greater baseball and general knowledge. This latter finding may be due to the i dea that WM capacity contributes to knowledge acquisition in general. Finally, increas ed age was related to decreased recall, and high levels of baseball knowledge did not lesse n this effect. This age effect was partially affected by ones processing speed and WM such that decreases in the latter two increased the age-related declines, but did not completely account for this age-related difference. However, increased age witnessed incre ased general and baseball-specific knowledge, which is supported by the hypothesis tha t crystallized knowledge remains stable or increases with age. In sum, domain knowledge enhances memory for domai n-specific information. However, it does not seem to compensate for any ind ividual differences in WM capacity or processing speed, and neither for age-related de clines in WM and memory recall in general. That WM had both beneficial and limiting effects on knowledge supports a less domain-specific and more global effect of WM. The researchers explain these results by supporting the idea of WM as a building block in accruing knowledge and operating on an elemental level. Hambrick, Salthouse, and Mein z (1999) explored skill in another domain: crossword puzzles. Four studies we re utilized to exact these factors, each study adding successively more specialized mea sures and successively more skilled participants. Beyond seeking the cognitive compone nts of crossword skills, potential
16 relations between high crossword puzzle experience and age-related cognitive decline were examined. Approximately 200 people, from 18-8 0 years, participated in each study. In Study 1, participants were of every experience level and were given crosswords of fairly simple skill level. In addition to this test of puzzle proficiency, participants were measured on other components posited to influence s uccessful puzzle completion: general knowledge (vocabulary and general informati on), word retrieval (e.g., write as many words that begin with the letter S), transfo rmation efficiency (transform one word into another by changing only one letter at a time), abstract reasoning (fluid cognition), and perceptual speed. General knowledg e was the strongest predictor of puzzle success, supporting the idea that knowing ge neral information is more useful than having more specific word skills. In Study 2, seve ral measures were added to more appropriately investigate unique crossword puzzle s kills: letter sequence knowledge (the likelihood that certain letter combinations occur o ver others), esoteric vocabulary (unique to crossword puzzles), and popular culture knowledg e. Additionally, participants with greater crossword puzzle experience were recruited and more difficult crossword puzzles were used. Of the new measures, esoteric word know ledge very strongly predicted puzzle success, supporting the idea that crossword puzzlers do possess a specialized vocabulary that facilitates their success in crossw ords, while letter sequence knowledge had an effect to a lesser but significant degree. Due to the inability of the abstract reasoning measure to affect puzzle proficiency in t he previous two studies, Study 3 utilized a measure of analytical reasoning instead of the abstract reasoning measure that had been unrelated to crossword success. This new measure required participants to take several elements into account in order to find a so lution. Further measures of knowledge
17 were also added that inquired about participants s elf-rated interest and knowledge of a variety of subjects to see if greater interest faci litated greater knowledge. However, all these measures failed to predict proficiency in sol ving the crosswords. Finally, in Study 4, participants were given still more difficult cro sswords to solve, considering that success on the more basic previous puzzles did not require the high level of reasoning and knowledge being measured. Additional measures were again added on top of the previous studys to measure analytical reasoning an d knowledge. However, again, on the majority of cognitive measures higher performance w as not related to greater puzzle success. The knowledge measure of advanced vocabul ary was the only measure related to success on the crossword. In sum, knowledge of general information, esoteric words, letter frequency, and advanced vocabulary contributed the most to crosswo rd puzzle success. Interestingly, neither abstract nor analytical reasoning measures seemed to contribute, despite the report of crossword puzzlers that reasoning was an importa nt aspect of success in these puzzles. Fluid and crystallized cognition were examined thro ughout the studies through measures of abstract and analytical reasoning, and general k nowledge and vocabulary respectively. Fluid cognition decreased with age, while crystalli zed cognition remained constant or even increased with age. However, greater proficie ncy in puzzle solving did not buffer against the age-related decrease in the level of fl uid cognition. Participants experienced increases in the practiced skill (crossword-puzzle solving) that enhanced crystallized cognition. In this way, the practiced knowledge sk ill transferred to a similar area of cognition (crystallized cognition), while a dissimi lar area (fluid cognition) was unaffected.
18 Cognitive Training: Participants with Cognitive Imp airments In an effort to study the power to protect against and reverse cognitive decline, some researchers have sought to train specific cogn itive skills. There are two areas of cognitive training research that have employed quit e different methods to target different populations: those suffering from a particular cogn itive decline, and those not experiencing any medical cognitive decline. The fo rmer is explored in this section in reference to people who have already experienced sp ecifically WM or overall cognitive decline: children with attention deficit hyperactiv ity disorder (ADHD) and older adults with dementia. Both areas of research have as thei r goal to improve the daily functioning of their participants through some type of domain-s pecific training. These studies may be considered somewhat comparable to the single-domain skill studies in that the trained skill is within a limited domain but its implicatio ns/transfer effects to other domains are also examined. They also provide more hope for tra nsfer to everyday life, being more concentrated and intense than everyday cognitive ac tivities explored in the previous section. ADHD in children. ADHD includes impairment in WM, due to a deficit in the frontal lobe, and represents an opportunity to stud y WM training effects on younger participants who are experiencing such impairment ( Klingberg, Forssberg, & Westerberg, 2002). Transfer effects from trained to untrained skills as well as daily functioning measures are examined and provide support for the g reat potential of cognitive training programs. Klingberg et al. (2002) investigated the impact of WM training on children with ADHD, as well as young adults without ADHD or WM im pairment. Participants trained
19 on three WM tasks (visuospatial WM, visuospatial ve rsion of the backwards digit span, and a spatial-verbal WM task) in addition to a reac tion time (RT) task. The strength of this studys training was its intensity and adaptiv e nature: it utilized a computerized WM training program on which participants trained at l east 20 minutes a day, 4-6 days a week, for at least 5 weeks via an adaptive staircase meth od. This adaptive training adjusted automatically, increasing or decreasing the difficu lty level based on individual performance, allowing participants to train right a t their skill levels. Specifically, difficulty level was increased with each trial by a dding additional items to be remembered until a participant missed two trials in a row, at which time the level was decreased by one item. For the visuospatial WM task, a succession of circ les was presented in a 4x4 grid and participants had to remember the sequence of ci rcle placement. For the visuospatial backwards digit span, a series of numbers were read aloud and participants recalled the numbers in reverse order by indicating them on a ke yboard. A letter span task was used as a test of spatial-verbal WM. A series of letter s was read aloud and participants had to remember their order; then a row of lamps was displ ayed and a flashing lamp cued the participants as to which letter they should recall (e.g., if the third lamp was lit, they recalled the third letter in the sequence). Additi onally, given that children with ADHD have impairment in this domain as well, a choice RT task was used in which two grey circles were displayed on a screen and participants were to press one key when one of the circles became green and not to respond when one of the circles became red. To measure training effects, several trained and u ntrained (transfer) tasks were measured both preand post-training. The choice R T task and a variation of the
20 visuospatial WM task included in training were used The variation in the visuospatial WM task was as follows: participants had to press a key whenever they saw a yellow circle (presented in one of two places), in which t he location was cued by a grey circle. First, in a simple RT task, participants were corre ctly cued to the yellow circles location with a grey circle that appeared 1-4 seconds before the yellow one. Then, in a choice RT task, grey circles appeared in both locations even though a yellow circle only appeared subsequently in one. Three transfer tasks were als o included: span board, Stroop task, and Ravens Colored Progressive Matrices. The span board was a visuospatial WM task in which an experimenter pointed to a series of 10 blocks, arranged randomly in front of participants, and participants had to recall the or der in which they were emphasized by pointing to the correct blocks in either the same o rder or the reverse. In the Stroop task (a measure of inhibition), color words were printed in a color different from the color the word indicated (e.g., the word red was printed in a blue color) and participants had to say the color of the ink and not the color word. I n Ravens Colored Progressive Matrices (a measure of nonverbal reasoning developed for chi ldren), participants had to complete a visual pattern by choosing the correct missing segm ent from several answer choices. A measure of motor activity in the form of head movem ent, achieved with the use of an infrared camera that detected movements of a marker placed on the childrens heads, was included as hyperactivity characterizes this disord er as well. In the first experiment, they measured their train ings effect on children ( n = 7, 7 to 15 years old, some of whom took medication for A DHD) as compared to a control ( n = 7, 7 to 15 years old), who trained with the same pr ogram only 10 minutes per day at a constant (not adaptive) low difficulty level. As c ompared to the control, the treatment
21 training group improved significantly from preto post-test in the trained visuospatial WM task and transfer tasks of span board (visuospat ial WM), Stroop task accuracy (inhibition), and Ravens (nonverbal reasoning) as well as reducing the frequency of their head movements. There were no changes in the choice RT task. In a second experiment, young adults without ADHD ( N = 4; 20-29 years old) underwent the same training procedure. They experienced a similar improvement preto post-training on the above measures except that improvement in the Stroop task entailed a decrease in the time needed to complete it, not in the overall accuracy (likely due to ceiling effects as they performed nearly perfectly). However, the improvem ent experienced by young adults was to a much lesser degree than that experienced b y the ADHD children. Therefore, it appears that even people without WM impairment can benefit from such an intense and adaptive training program. Importantly, there was improvement not only in the trained tasks but also in the transfer tasks, which likely use similar brain regions (Olesen et al., 2004) and benefit from activity in those regions. Klingberg et al. (2005) extended the work conducte d above with a larger group of children ( N = 44), all of whom were not taking ADHD medication. The same training tools were used (without the RT task), except that the computer training was not conducted in a lab but rather provided on a CD for participants to take for use at home or in school. The outcomes measures were also reduced to the span board task (visuospatial WM), digit span (verbal WM), Stroop task (inhibitio n), and Ravens (nonverbal reasoning) as well as motor movement as done in the previous study. The timeline of the study was also somewhat different: the median total training time was about 40 minutes, which occurred on at least 20 separate days (within the 5-6 weeks after the initial visit),
22 and the experiment added a third follow-up assessme nt of the outcome measures to test for long-term maintenance 3 months after the end of training. Finally, additional measures of parents and teachers subjective ratin gs of childrens inattention as well as hyperactivity/impulsivity were included. Participants improved more than did a control grou p (who used the same training materials but they were not adaptive and stayed at an easy difficulty level) on all cognitive outcome measures (span board, digit span, Stroop task) as well as parental ratings of ADHD symptoms. Additionally, several me asures maintained their improvement at the 3-month follow-up: span board, d igit span, Stroop task (time required to complete the task, not task accuracy), and paren tal ratings (but not teacher ratings) of ADHD symptoms. However, hyperactivity as measured with head movements did not show any significant improvement, though parents s ubjective ratings of their childrens hyperactivity indicated improvement. Gibson, Gondoli, and Grundy (in press) sought to re plicate the results of the above studies of ADHD children with a closer look a t the mediating effects of fluid cognition (i.e., reasoning ability) to improve ADHD symptoms. Participants were children ( N = 12, 12 to 14 years old) who were being presently medicated for ADHD. They underwent a computerized training program that involved both verbal and spatial WM tasks: verbal tasks to remember phonemes, letter s, and digits; and spatial tasks to remember positions of objects on a 2D or 3D grid. They underwent 1-hour training sessions, 5 days a week, for 25 days. The outcome measures were the same as those used in the previous studies cited (e.g., Klingberg et a l., 2005) and included the span board
23 task, digit span task, and Ravens Standard Progres sive Matrices (the adult version). Parents ratings of their childrens ADHD symptoms were also taken. As compared to pre-training, participants improved equally on both verbal and spatial training tasks as well as on all three cogn itive outcome measures. Also, as assessed with parents reports, ADHD symptoms (both inattention and hyperactivity/impulsivity) dropped significantly, e ven below the clinical range after training. In these ways, this study replicated the findings of Klingberg et al. (2005). The secondary purpose of the study was to investigate t he mediating effects of fluid cognition between improvement in WM and improvement in the in attentive symptoms of ADHD. This was largely supported in the analyses, most st rongly in that spatial WM improvement was significantly related to fluid cogn ition improvement which was related to inattentive symptom improvement. Furthermore, f luid cognition completely accounted for the relation between spatial WM improvements an d inattentive symptom improvements in the statistical model used, though not to a significant degree. This suggests a domain-specific effect as verbal WM impr ovements were not related to fluid cognition improvements, even though there was overa ll equal improvement on both WM tasks. Interestingly, verbal WM improvements were related to increased ADHD symptoms, suggesting that improvement in verbal WM competed negatively with improvement in spatial WM. However, these findings were unexpected by the authors, who advise replication and controls before strong c onclusions should be made. Dementia in older adults. Effective treatment for older adults with dementia may well benefit from a combination of pharmacological strategies and cognitive training (Hofmann, Hock, Kuhler, & Muller-Spahn, 1996). How ever, training programs designed
24 to target people with dementia have been quite limi ted and their results have been mixed, with some success in transfer beyond the trained ta sks to daily functioning measures. Kawashima et al. (2005) investigated the potential cognitive benefits of solving simple arithmetic problems and reading aloud for el derly people suffering from dementia. In a pre-test, participants were tested on two meas ures of general cognitive ability used to screen individuals for cognitive impairment: the Mi ni-Mental Status Examination (MMSE) and the Frontal Assessment Battery (FAB). T hese standard measures required participants to respond to such items as spell wo rld backwards and in what way are a banana and orange alike?, respectively. Participa nts were then separated into control and training groups (total N = 32, ranging in age from 76 to 96 years). The tr aining group underwent learning tasks in reading and arith metic for 6 months (average attendance of 4-5 days per week) for about twenty m inutes each day. Difficulty ranged from typical learning materials for 4to 10-year-o lds. The reading tasks ranged in complexity from simply reading and writing single s yllables to reading fairy tales aloud, and the arithmetic tasks ranged from counting pract ice to 3-digit division. Prior to beginning the learning tasks, the appropriate degre e of difficulty in each task was determined per participant, though it was not speci fied whether participants were allowed to advance. After 6 months, all participants cognitive levels were again tested. The FAB score of the training group significantly increased over the 6 months, and the MMSE score of the control group significantly decreased over that span. Additionally, the training group obtained higher scores on both cogni tive measures than the control group. These data suggest that repetition of tasks as simp le as arithmetic and reading aloud can
25 aid in the rehabilitation of people suffering from dementia, improving their cognitive performance. Transfer occurred from the trained ta sk to two measures of overall cognitive status (MMSE and FAB). That practice in reading and math would improve seemingly unrelated skills such as identifying qual itative similarities between items suggests that, at least for people already sufferin g from dementia, domain-specific training can have more global effects. Hofmann et al. (1996) also attempted to improve th e cognitive performance of participants with Alzheimers disease but they used a computer-based cognitive training program. Ten participants (49 to 86 years) sufferi ng from mild to moderate Alzheimers disease were trained 3 to 4 times a week for 3 week s using an interactive computer program. In this computer game, photographs of the ir own local and social environments were used and participants had to navigate througho ut their virtual town and perform tasks of social competence, orientation, and memory Participants were tested on several measures of cognitive ability before training, imme diately after training, and finally 3 weeks after the trainings completion. These cogni tive tests included such measures as the MMSE, digit span (test of WM), and immediate an d delayed recall and recognition of words. In addition to these direct measures, parti cipants and their caregivers also completed questionnaires that inquired about their perceived effectiveness of the training on memory, and if they thought the training could t ranslate into the participants real-life experiences. While the patients showed more optimi sm for potential training effects, the caregivers were less positive. Computer-game performance increased as measured by reduced number of mistakes, reduced amount of time needed, and reduce d amount of advice needed to
26 complete the tasks. Taken together, all participan ts tended to experience a reduction in mistakes from the baseline pre-test to both the imm ediate post-test and follow-up 3 weeks later. Though not statistically significant, early -onset Alzheimers participants did experience a fairly large reduction in mistakes in the computer game, but late-onset Alzheimers participants did not. The lack of sign ificant effects may largely be explained by the very small sample size ( n = 5 for each group) of the study. For all partici pants, the time and advice needed were significantly reduced a t both the post-test and follow-up as compared with the pre-test. While there was a tend ency toward greater time and advice needed at the follow-up as compared with the immedi ate post-test, likely due to the cessation of training and a decrease in the trained skills, the reduction from the baseline was still significant. However, there were no tr ansfer effects beyond the computer program improvement to the more standard measures o f cognitive ability (e.g., MMSE or word recall). This is contrary to Kawashima et al. s (2005) findings since the studys trained skills did not transfer outside the trained domain, though small sample sizes and differences in training interaction across particip ants may also be explanations for these findings. Cognitive Training: Healthy Participants This section explores the other area of cognitive training, for participants without cognitive impairment. Recently, there has much res earch on cognitive training interventions for older adults. This research has led to mixed results, with the greatest effectiveness being in programs that adopt a proce ss-specific approach (Buschkuehl et al., 2008). The goal of a process-specific trainin g program is to target a particular cognitive process such as WM and work to increase i ts efficiency, and not to train
27 additional processes such as mnemonic strategies or techniques. Studies on both types of training interventions are discussed below. Mnemonic strategies as software. Mnemonic strategy programs attempt to train a particular memory-enhancing technique that can be applied to a limited number of situations. In this way, mnemonic strategies can b e considered cognitive software that is superimposed on existing cognitive processes and has no enduring effects without their explicit usage. As part of a larger study, Ball et al. (2002) test ed the effectiveness of a mnemonics training intervention for older adults ( N = 2832, aged 65 to 94 years old) who had not experienced any sufficient cognitive declin e as compared to a no-training control group. The training group, which focused on verbal episodic memory, taught participants strategies to remember word lists and item sequence s, text material, and story components. For example, participants were instruc ted on how to group words in a list into meaningful categories or to create visual imag es of the words to enhance recall. The sessions were conducted in a small group setting an d lasted about 1 hour. The 10-session training lasted 5 to 6 weeks: the first 5 sessions consisted of instruction and practice in mnemonic strategies, while the last 5 sessions cons isted of additional practice with no instruction on new strategies. About 10% of each o f the sessions consisted of explicit training of these strategies for daily-life tasks, such as remembering the details of a medication label or the items in a grocery list. I n addition to the 10-session training, a 4session booster training lasting two to three wee ks was offered to a random 60% of participants eleven months after initial training e nded.
28 To measure the effectiveness of the training, all participants were measured specifically on the three cognitive domains targete d in the overall study (memory, reasoning, and speed of processing) and on other as pects of their daily functioning, including performance-based everyday problem-solvin g (e.g., reasoning and identifying information from medication labels) and everyday sp eed of processing (e.g., finding food items in a crowded grocery shelf), as well as selfreported daily functioning (IADL; e.g., meal preparation, managing finances, etc.) and driv ing habits (i.e., driving difficulty and avoidance of specific situations). These effects w ere measured at 1 and 2 years after initial training ended. Overall, participants expe rienced significant improvement from their baseline measures immediately following train ing in their domain-trained skill, though there were some domain differences: 87% of t he speed group, 74% of the reasoning group, and 26% of the memory group experi enced this improvement. Many participants experienced ceiling effects at their b aseline measures, and perhaps this was the cause of the memory groups relatively small im provement. Participants maintained their improvement at each follow-up testing 1 and 2 years after training completion, though an increasing decline at each follow-up afte r the immediate testing occurred. No beneficial effects on either self-reported or perfo rmance-measured real life activities were observed. However, overall declines in the daily f unctioning of participants in the training groups were noted to be less than that of the control group, so it seems that the training had some protective effects. Additionally booster training led to a significant increase over that of the non-booster-trained group on tests of reasoning and speed, but not memory. In sum, participants experienced impro vement in their specific training
29 domains, but this training did not transfer to meas ures of daily functioning and seemed to diminish over time. Willis et al. (2006) extended the above study to in clude additional follow-up measures at 3 and 5 years after training (beyond th e 1and 2-year follow-ups of the above study), as well as a 3-year booster training (beyond the 1-year booster training of the above study). The purpose of the study was to see whether the training effects endured beyond the 2 years measured by Ball et al. (2002). Overall, the immediate improvement experienced in each trained domain that persisted to 2 years in Ball et al. likewise persisted to 5 years in this study. Howev er, while the performance levels were above baseline, they tended to diminish at each suc cessive follow-up as found in the previous study. Also, the 3-year booster training showed a similar beneficial effect over the non-booster trained group on tests of reasoning and speed. An important finding of the study, unlike that of the previous one, was the effect on self-reported daily functioning. At the 5-year follow-up, participants reported less difficulty in performing daily activities themselves. However, this effect was significant only for the reasoning training group. In reference to memory, it seems that training in a mnemonic strategy was not effective in its ability to transfer to common acti vities in daily life (performance-based memory measures within the lab) or overall measures of self-reported daily functioning (IADL). Training improved the trained skill only. A meta-analysis of studies that examined the effec tiveness of mnemonic intervention training on memory was conducted by Ve rhaeghen, Marcoen, and Goossens (1992). Thirty-three studies ( N = 1,539, M = 69 years) were included in the analysis
30 whose inclusion was based on the use of healthy par ticipants, training in some mnemonic technique, and memory measurements taken both befor e and after training. Examples of mnemonic techniques included the name-face mnemonic (choosing a unique facial characteristic, finding a word/phrase that is simil ar to the name, and creating a visual image linking the unique facial feature to the word /phrase) and method of loci mnemonic (visualizing a familiar place/route and visually pu tting items/objects needed to be remembered at particular places in the scene). Thr ee groups were compared: training (mnemonic technique training), placebo (training or lecture in a topic unrelated to mnemonic techniques), and control (no training). O lder adult participants received greater benefits due to the mnemonic training than did those of the untrained placebo and control, who did not differ from each other. Train ing effects were limited to tasks in which mnemonic strategies could be utilized (e.g., memories for faces/names but not memory in general), and no particular mnemonic tech nique produced greater benefits. Also, training that took place in a group setting a nd with a relatively short session length (less than 1.5 hours) created the greatest benefits for memory in participants. These findings about group-learning benefits may be due t o positive effects of social comparison, within-group support, or increased moti vation, and shorter session lengths likely reduced fatigue effects. Overall, younger a dult participants experienced greater gains than older adults. In sum, the effects of mn emonic training are quite domainspecific and especially limited for older adults. However, the most successful training programs were those of relatively short length and that occurred in a group setting. Working memory as hardware. WM training, versus domain-specific mnemonic strategy training, takes a process-specifi c approach that facilitates improved
31 performance in a plethora of situations. In this w ay WM can be considered cognitive hardware, (as opposed to the software with whic h mnemonic strategies equip people) a domain-general cognitive process that is trained directly and has enduring effects without the need for explicit implementation. Buschkuehl et al. (2008) sought to study high-func tioning older adults over the age of 80 years in a WM training program. Particip ants ( N = 32) trained on three visual WM tasks and two RT tasks (to help increase speed o f processing, known to decrease with age). They trained 45 minutes in each session twice a week, for 3 months. The training was adaptive like previous studies (e.g., Klingberg, 2005), increasing difficulty with correct participant responses and decreasing w ith incorrect ones. In the first WM task, included primarily to familiarize all partici pants with computer use, a display of four colored squares were shown and the squares dis appeared and reappeared one at a time, after which the participant was cued to repea t the sequence by clicking the appropriate squares. Participants received immedia te feedback on their performance. The second WM task had two parts: 1) participants d ecided if a series of animal pictures (cats or dogs) were presented the right way or upsi de down, and 2) they then had to recall the sequence of animals presented to them by clicki ng the pictures of the two animals in the correct order. The final WM task was a variati on of the second and simply increased the types of animals from 2 to 8 (cat, iguana, dog, rabbit, mouse, toad, butterfly, and bee), but followed the same procedure. The first RT task had participants decide if a presented word was a word (a list of positive aging stereot ypes) or nonword as quickly as possible by pressing particular keys. In the secon d RT task, participants saw a word appear above or below a fixation cross, followed by a distracting series of letters, and had
32 to respond whether the word and distractor were abo ve or below the fixation cross as quickly as possible by pressing particular keys. E ach training session began and ended with a passive activation task in which participant s watched a series of words presented and repeated them aloud. Additionally, at the end of each training session, participants heard a brief presentation about a topic such as me mory or attention systems. Before and after training, as well as at a 1-year follow-up, several transfer tasks were measured: forward and backward digit span (e.g ., Klingberg, 2002), block span, verbal free recall, and visual free recall. In the block span task, participants saw a 4x4 grid and a blue dot appear in a given sequence with in that grid. Participants recalled the sequence by pointing to the different location on t he grid in the correct order. The verbal free recall task presented participants with a shor t prose text read aloud to them. They were instructed to listen carefully and memorize th e text; 30 minutes later they were asked to reproduce the text with as many details as possible. The visual free recall task had two parts: 1) for 3 minutes participants search ed for differences between two near identical pictures presented side-by-side and were told to also look at the picture in a way that they would be able to provide information abou t it later, and 2) after 20 minutes passed participants reported as many items as they could that were present in the previously compared picture. A control group ( n = 19) was also included for comparison that underw ent physical training on an eccentric bicycle ergometer which is a low-metabolic-load exercise bicycle. They followed the same training schedule as the cognitive training group, with warm-up gymnastics and cool-down stretc hing sessions. In previous studies, the cardiovascular component of exercise has led to cognitive improvement (Kramer et
33 al., 2003, as cited in Buschkuehl et al., 2008, p. 747). Therefore, the training bicycle used for this study had low metabolic load and so o nly a minimal cardiovascular component, similar to the low-dose cognitive traini ng control used in Klingberg et al. (2005). The load was manipulated by the participan ts in response to a screen that gave them feedback and had them adjust their performance to the target amount. This selfmonitoring was not expected to have an effect on co gnitive performance as the load level remained constant (not adaptive like the cognitive training) and likely became automatic for the participants. In the cognitive training group, participants impr oved on the trained tasks as well as on the block span (visual WM) and, to a lesser d egree, on the visual free recall (visual episodic memory) transfer tasks, but not the digit span (verbal WM) or verbal free recall (verbal episodic memory) tasks. These findings are interesting in the distinction between near transfer and far transfer. Due to the foc us of the WM training tasks in the visual domain, it is not surprising that the greatest effe cts would be on the visual domain transfer tasks. Additionally, the trained WM task and block span task share the use of a short-term memory store (near transfer), as oppos ed to the LTM store necessary for the visual free recall task (far transfer). The limi ted transfer may also be due to the development of strategies, for which participants r eported usage on the second WM task. The cognitive group also experienced improvement in the second RT task. The control group experienced no changes in the transfer tasks, and by the 1-year follow-up the cognitive training groups training benefits had de creased to the performance level of the control. However, while the training groups trans fer task performance decreased, it was still not significantly different than the post-tes t, so it had not decreased to the pre-test
34 level. Overall, successful near transfer occurred, and the researchers speculated that more efficient training (e.g., prevention of possib le strategy use) might aid in further transfer. Li et al. (2008) had participants train with a mor e challenging computerized spatial WM task: the n -back. In this task participants saw a 3x3 grid in which a sequence of black circles appeared (in any of the squares ex cept the center square) and they had to respond whether each circle appeared in that spot n items (in this case 2 items) earlier. In a more difficult shifting condition, the circles appeared in the same manner but participants had to mentally shift each position on e step clockwise and recall the shifted position. In both conditions participants were enc ouraged to respond as quickly as possible. This training program differed from prev ious ones in that it was not adaptive in nature, but instead had a set increase in difficult y within each training session (from the original condition to the shifting condition). P articipants were younger (20-30 years, n = 11) and older adults (70-80 years, n = 15) who trained for merely 15 minutes per day, for 45 days (almost twice as long as other WM studi es), over the course of about 3 months. They were compared to age-matched, no cont act control groups (younger adults: 21-30 years, n = 27; older adults: 70-80 years, n = 20). Groups were compared on a number of outcome measur es, both near and far transfer tasks. The near transfer tasks were varia tions of the trained n -back but at a greater difficulty level. The first had participan ts perform on a three-back task (respond to each dot in reference to the item 3 trials ago). The others were numerical n -backs (two-back and three-back) in which a series of sing le-digit numbers were presented and participants had to respond if each was the same as the number 2 or 3 items earlier. The
35 far transfer tasks included two complex WM span tas ks (short-term memory requirements) and two non-memory speed tasks. The first complex span task was operation span in which a series of simple mathemat ical equations paired with onesyllable nouns were shown (set sizes varying from 3 to 6 pairs in length). Participants had to decide if the solutions provided for the equ ations were correct, by pressing particular keys, and to remember the paired words i n the presented order. The second complex span task was the rotation span in which le tters were presented in various degrees of rotation (45, 90, 135, 180, 225, 270, or 315 degrees), and participants had to respond if the letter was displayed normally or mir ror-reversed. Additionally, they had to remember the exact letter orientations and recall t hem at the end of the trial in the order they were presented. The final far transfer tasks were two simple decision speed tasks formatted to look like the trained n -back task, but in this case required no WM store a nd could be considered zero-backs. Participants for o ne variation had to respond if a circle presented in the periphery of the 3x3 grid was blac k or white (by pressing the appropriate keys), and for the other variation had to respond i f a number presented in the center of the grid was 1-3 or 4-9 (again by pressing the appropri ate keys). In both tasks, participants were to respond as quickly as possible. Both age groups improved significantly on the trai ned spatial WM task (twoback) from preto post-test, though the maximum pe rformance of the older adults was still lower than younger adults at baseline. There was positive transfer to the spatial three-back and numerical twoand three-backs, and the transfer effect was, surprisingly, not greater in the younger than older adults. Cont rolling for improvement in the speed task, these transfer effects were still beyond a me re increase in speed efficiency. At a 3-
36 month follow-up these trained and transfer effects were maintained by both age groups, though to a greater extent by the younger adults. There were, however, no transfer effects to the complex span tasks. An important as pect of this training program was that it was not adaptive and so did not adjust to the sk ill level of the individual. For the older adults, the training tasks remained challenging thr oughout the study, but after 1 week the younger adults had reached ceiling performance in a ccuracy. Perhaps if the training had been adaptive, there would have been greater transf er effects for the younger adults. Olesen et al. (2004) trained young adults without WM impairments on similar adaptive WM tasks as above and used similar outcome measures, with the addition of a brain imaging component. In two experiments, young adults (Experiment 1: n = 3, 20-23 years old; Experiment 2: n = 8, Mage = 29.3 years old) trained about 40 minutes each da y, for more than 20 days, and over the course of 5 wee ks. The same cognitive outcome measures were used in both experiments (span board, untrained visuospatial WM, Stroop task, and Ravens). A control group ( n = 11) was included, but in this study did not undergo any form of training. In the first experim ent, participants trained with the same three WM training tasks as Klingberg et al. (2005; visuospatial WM, backwards digit span, and letter span) and were brain scanned using functional magnetic resonance imaging (fMRI) before and after training (the spati al-verbal WM task used during scanning was of a relatively easy difficulty level) to measure activation and interaction of various brain regions. Participants in the trainin g group improved on the span board, Stroop task, and Ravens as compared to the control In reference to the imaging results, there were positive interactions between the prefro ntal and parietal cortices (brain regions
37 associated with WM control and capacity, respective ly), along with an increase in activity in the two regions that corresponded with an increa se in WM capacity. In the second experiment, participants trained on three different WM tasks (Grid, Grid rotation, and 3D Grid) that were all based on the following basic task: a sequence of red circles were presented on a 4x4 grid and partic ipants had to recall the ordered location of the circles. This task was also perfor med while participants were scanned using the fMRI and was harder than that of Experime nt 1. Participants were also asked about mnemonic strategies they used to aid memory f or the test items. Several participants reported using chunking in which the y formed associations between test items, grouping them into units, so they could be r emembered as a whole. However, a trial-by-trial analysis with stimuli varying in the ir ease in chunking found that use of this strategy did not account for improvement on the WM task. Participants were scanned 5 times: once before, three times during, and once af ter training. The same outcomes measures were again used. The results of this expe riment replicated those of Experiment 1: participants in the training group improved on t he span board, Stroop task, and Ravens as compared to the control ( n = 11), as well as experienced an increase in activity and positive interaction between the prefr ontal and parietal cortices. In conclusion, WM performance seems to occur namely in these two interacting regions of the brain; their increased activity may allow for g reater WM capacity, with improvements not only in trained tasks but also in untrained, tr ansfer tasks. Persson and Reuter-Lorenz (2008) conducted another WM training study. They trained young adults ( N = 48, 18-30 years) on three WM tasks in just 8 ses sions, for 40 minutes, over 2 weeks. These WM tasks differed fro m previous studies in that they were
38 manipulated to create high interference situation s, specifically proactive interference. Proactive interference occurs when stimuli from pre vious trials interfere with recall of stimuli from the present trial. The three WM tasks used were: item recognition with letters, item recognition with faces, and a three-b ack task using words. In the item recognition tasks, participants saw four target ite ms (letters or faces) for 1.5 seconds, followed by a 3-second delay, and then a probe item They had to decide if the probe item had been among the target items just presented In the interference conditions, the probes were items from target sets in previous tria ls, which were more difficult for participants to differentiate. In a word version o f the three-back, participants saw a series of 96 words and had to respond if the present word had been presented three words earlier. There were two control groups: one that e xperienced these same WM tasks except without interfering stimuli, and another tha t experienced low difficulty WM tasks again without interfering stimuli. These low-load WM tasks included the following variations of the above tasks: 1) letter/face match ing in which four identical target items were presented followed by a probe that participant s identified as in the previous target set or not, and 2) a one-back in which participants had to decide if the present word matched the previous word. Several transfer tasks were used, each with an int erference and no-interference condition: paired associates (episodic memory), ver b generation (semantic memory), and item recognition (WM; identical in procedure to the trained except that stimulus items were words). In the paired associates task, partic ipants experienced a study phase in which they saw several pairs of words and their ass ociates, each for 3 seconds. Then during the test phase, participants saw the word an d had to generate its previously paired
39 associate. In the interference condition each word was paired with multiple associates (e.g., queen-king, queen-bee, queen-crown ) and participants were instructed to recall the most recent associate, while in the no-interference condition each word was paired with only one associate. In a similar task, the verb ge neration task, nouns were presented and participants had to think of a verb associated with each. In the interference condition participants saw nouns that had several appropriate verb pairs (e.g., ball-throw, kick, bounce ), while the no-interference condition had nouns wi th only one dominant pair (e.g., scissors-cut ). Overall, the interference group performed better t han the control groups only on the interference trials, performing just as well in speed and accuracy on the nointerference trials. The researchers suggest that there is evidence for these results in brain imaging studies that implicate the left inferior fr ontal gyrus in interference resolution. Therefore, shared use of this brain region across t he interference tasks could have helped to produce the transfer results. Adding to research on the nback task, Jaeggi, Buschkuehl, Jonides, and Perrig (2008), in a series of four experiments ( N = 70: Mage = 25.6), used a more demanding n back training task called the dual n -back. Similar to the n -back task used above, participants were required to respond in reference to stimuli n number of items ago (though no interference conditions were included). Specifically, this was a dual task in that participants saw two sequences of stimuli pres ented in synchrony: one sequence was auditory and consisted of a series of letters heard aloud, and the other was visual and consisted of a series of white boxes that moved aro und an invisible 3x3 grid (with the center position omitted). Participants responded a ffirmatively only if both the letter and
40 square position matched the letter and square posit ion n items earlier. As performance improved n increased by one item, and as it worsened n decreased by one item, making the task adaptive to individuals current skill lev els. Participants were young adults trained in this ada ptive training program and compared to age-matched, no-training control groups ( n = 35). The main difference between the four experiments was the number of trai ning sessions between preand posttest measures: participants trained over 8, 12, 17, or 19 sessions (approximately 25 minutes per session). The purpose of the study was to investigate the relation between WM and fluid intelligence. Both cognitive domains may share a common capacity limit in their demand for attention, which is a limited c ognitive resource (Halford, Cowan, & Andrews, 2007 as cited in Jaeggi et al., 2008). To investigate this relation, Jaeggi et al. included a fluid intelligence transfer task (Raven s) measured preand post-training. WM capacity was also measured with two span tasks: digit span (described above in Klingberg et al., 2002) and reading span. In the r eading span task participants saw sentences 13-16 words in length and were required t o read the sentences aloud and recall the terminal words in any order. Overall, participants improved significantly, as c ompared to controls, on trained and most transfer tasks. Those with initially lowe r fluid intelligence scores improved more than those with higher fluid intelligence. Th is improvement was not based purely on WM capacity improvement, as improvement on fluid intelligence went beyond mere improvement on the WM digit span task (the reading span task did not lead to similar transfer improvements). Interestingly, improvement s were dosage-dependent in that the more sessions in which participants trained, th e greater the training (dual n -back,
41 WM) and transfer (Ravens, fluid intelligence) effe cts. To explain these greater transfer effects, the researchers highlighted the adaptive n ature of the training and high attentional control required to succeed on the dual n -back. It was not a task-specific WM capacity that explained the improvement but rather another t ask-general skill. The researchers propose several task-general attentional skills t hat may work together to account for this improvement: multi-task management of attentio n, inhibition of irrelevant information, monitoring of ongoing performance, and updating items in memory. In sum, the results support the claim that WM and flui d intelligence share cognitive resources, which may be their common need for atten tional control. Present Study In an effort to investigate the source of the mode rate and mixed success of various cognitive training programs, especially in referenc e to transfer effects, the present study utilized a similar cognitive training intervention approach. I examined the effectiveness of a WM training program for healthy older adults. WM training programs with healthy older adults have been diverse in their training co ntent, style, intensity, and duration. Some have trained intensely for a short period of t ime (e.g., Persson & Reuter-Lorenz, 2008, trained for 8 sessions over 2 weeks for 45 mi nutes per session), while others have trained more diffusely for a longer period of time (e.g., Li et al., 2008, trained for 45 sessions over 3 months for only 15 minutes per sess ion). Most have focused on training visual WM (e.g., Buschkuehl et al., 2008; Li et al. 2008; Olesen et al., 2004), sometimes with a verbal WM component (Jaeggi et al., 2008; Pe rsson & Reuter-Lorenz, 2008), and have included a plethora of transfer tasks both ne ar transfer to tasks similar to the trained domains (e.g., between similar visual WM ta sks; Buschkuehl, 2008) and far
42 transfer to tasks in other areas of cognition (e.g ., WM training to fluid cognition; Jaeggi et al., 2008), but none to real-life tasks. Training programs teaching mnemonic strate gies as well as those targeting participants with cognit ive impairment, such as ADHD in children and dementia in older adults, have been mo re likely to include practical, real-life measures such as overall daily functioning (Ball et al., 2002, and Willis et al., 2006, with their self-reports of participants degree of indep endence when performing daily activities; Hofmann et al., 1996, and Kawashima et al., 2005 with their assessments of general cognitive ability) and more specific observ ational measures of WM control (Klingberg et al.s, 2002 and 2005, tracking of hea d movements to measure hyperactivity). However, there is a dearth of trai ning research with healthy older adults that have included transfer tasks measuring everyda y functioning. In the present study, participants underwent 5 wee ks of training on 3 standard WM tasks (operation span, reading span, and alphabe t span). Training was conducted over 5 sessions in a group setting, once a week, fo r about 1.5 hours each session. However, unlike the above WM training studies that measured transfer only to other laboratory-based cognitive tasks, the present study investigated transfer to a real-life task: grocery shopping. The comparison of a skill within the context of a real-life environment and a decontexted lab environment is similar to the single-domain studies conducted by Hambrick and colleagues (1999, 2002) in that simila r skills were tested both within the context of the skill (baseball knowledge or crosswo rd solving skills) and decontexted from the game in standard lab tests (memory, reason ing, general knowledge, etc.). In this way, like the present study, the skills were measur ed within the studied or trained task as well as in a transfer task. However, the present s tudys focus on skill of a basic cognitive
43 process like WM (related to fluid cognition) differ ed from skill in, for example, crossword solving (related to crystallized cognitio n). This distinction is important in older adults as crystallized cognition remains stab le as one ages, while fluid cognition declines into old age (Craik & Bialystok, 2006). The goal of the training program was to improve pa rticipants WM skills as well as to transfer these skills to a real-life grocery shopping task, which required WM and LTM usage. WM is closely related to LTM (Park et a l., 2002). Specifically, WM mediates much of the variance in LTM, such that cha nges in LTM are related and likely due largely to changes in WM. In this way, trainin g to improve WM ability (through WM training) has the potential to transfer to a LTM task (grocery shopping task). To measure this transfer, participants performed in a LTM-WM (a combination of LTM and WM) task in a grocery store for a list of grocery i tems before and after training. Figure 1 depicts the model of WM and LTM assumed to be impli cated in these two sets of tasks. For the WM training task, participants were asked t o store a list of words while carrying out another processing task such as solving math eq uations, reading sentences, or alphabetizing a list of words. Rehearsal of the wo rds was inhibited by the high demands for attention of the concurrent processing task. R ecall of the words occurred immediately after their presentation so there was neither oppor tunity nor incentive to encode the words in LTM. However, in the grocery shopping task part icipants were asked to store over twice as many words, but did not have the competing processing demands as in the WM training task. Participants were able to rehearse the list of words they had to remember (remembering some items in this manner without enco ding) and, due to the long duration
44 and nature of the task, likely encoded several of t he words in LTM which they retrieved (either through recognition or recall) during the t ask. Unlike previous studies that utilized randomly-gen erated words to train their WM tasks (e.g., Jaeggi et al., 2008; Persson & ReuterLorenz, 2008), the present study manipulated the word lists by theme. The lack of s trong effects in performance-based skills or daily functioning for Ball et al. (2002) and Willis et al.s (2006) training studies may have resulted from the mere 10% of explicit, re al-life application training. The present study sought to investigate the potential b enefits of explicit training more directly with the inclusion of two training groups: explicit and non-explicit. The non-explicit training group received training with the standard WM tasks, such as remembering a random list of nouns (e.g., shadow diamond etc.) while reading a series of sentences. The explicit training group received similar traini ng except that the stimuli word lists contained all food-related nouns (e.g., apricot, bread, etc.). It was hypothesized that while both groups would improve in their trained WM tasks from preto post-training, the explicit training groupand maybe to a lesser e xtent the non-explicit training groupwould improve on the grocery shopping transfe r task from preto post-training in that the content of the training and transfer tasks would be similar and facilitate transfer effects. These two training groups were compared t o a no-training control group that was just tested on the preand post-grocery shopping t asks. No improvement on the postgrocery shopping task was expected for the control group. A final transfer measure was a daily functioning m easure taken from Ball et al. (2002) that asked participants to self-report on a variety of common daily activities (e.g., managing finances, taking medications, etc.). They responded about the degree to which
45 they performed each activity independently (complet ed it by oneself) or dependently (relied on someone else to complete it for them). It was hypothesized that on this transfer measure of daily functioning both training groups w ould experience improvement, though to a lesser extent than the explicit groups improvement in the grocery shopping transfer task, and that there would be no improveme nt experienced by the control group. Therefore, along with measuring improvement within the lab context on the trained WM tasks, improvement at the grocery store and in over all daily functioning were also measured as transfer tasks. As such, this study te sts the ecological validity of a cognitive training program, and thus its usefulness for facil itating everyday activities. In summary, the following hypotheses were posited for improvements from preto post-training measurements: 1) significant impro vement on the trained WM tasks for both training groups, 2) significant improvement on the grocery shopping task for the explicit training group (though perhaps moderate im provement for the non-explicit group) above that of the control group, and 3) some what lesser improvement (than trained and grocery shopping transfer tasks) on the daily functioning measure for both training groups and no change in the control group. Method Participants The participants of the present study were 32 olde r adults (24 women, 8 men, age range: 59 to 87 years, Mage = 74.25 years) recruited via flyers and recommenda tions at a local senior center in Florida. Inclusion in the s tudy was based on the following criteria: over 55 years old, no medical diagnosis of a memory or attention disorder, and no regular participation in any other cognitive training compu ter program or class. Participants gave
46 informed consent and were not paid for their partic ipation in the study. They were assigned to one of the two training groups (explici t: n = 8, non-explicit: n = 9) based on how their personal schedules corresponded with the session times. Participants in the control group ( n = 4) were recruited after the initial recruitment for the experimental groups and so were assigned automatically to the co ntrol group. While participants in the training groups knew the full trajectory of the stu dy, those in the control group were not informed of the training portion but rather thought it was merely a study on everyday memory in a real-life context (just the grocery sho pping portions). Eleven participants (6 women, 5 men, age range: 59 to 87 years, Mage = 75.89 years) were excluded from data analysis due to eith er withdrawal at some point during the study or exclusion due to lack of engagement in the training tasks ( n = 1). For participants who withdrew from the study, they had scheduling or health issues preventing weekly attendance at the training sessio ns. Therefore, 21 participants (18 women, 3 men, age range: 65 to 87 years, Mage = 76.95 years) were included in the final data analyses. There were no significant differenc es between the two training groups (explicit and non-explicit) and the control group o n several measures of participant characteristics (see Table 1 for group means), Wilk s Lambda = .01, F (2,18) = 2.33, p > .05. Participants in all groups tended to be the s ame age, have the same level of education, walk at the same rate, and participate i n cognitive, social, physical, and foodrelated activities just as frequently. Additionall y, in a pre-test measure of daily functioning (see Table 2 for group means), particip ants in all groups were equally very independent when performing several everyday activi ties, F (2,18) = .11, p > .05. There were two non-native English speakers (Spanish was t heir native language), which
47 presented some issue with recall of the words durin g the training sessions. Only one of these participants was excluded from analysis as th e difficulty caused the scores to be significantly lower than the groups. Working Memory Tasks All participants in the training groups received t raining in three WM span tasks: alphabet span (Alpha), operation span (OSPAN), and reading span (RSPAN). The word lists used for these WM tasks varied between the tw o training groups. The non-explicit group trained with a random list of words (e.g., shadow diamond etc.) generated from online word databases (MRC Psycholinguistic Databas e: http://www.psy.uwa.edu.au/MRCDataBase/uwa_mrc.htm; Paivio Word Pool: http://www.math.yorku.ca/SCS/Online/paivio/). The words obtained from the databases were limited by the following criteria: noun as par t of speech, 1 to 3 syllables, 3 to 10 letters, high frequency and concrete. The explicit group trained with a list of food-related words grouped specifically for this study. Food-re lated words were compiled using the following criteria: noun as part of speech, 1 to 3 syllables, and 3 to 10 letters. They varied from typical food names (e.g., apricot, bread ), to cooking appliances/utensils (e.g., blender spatula ), to more obscure foods (e.g., tofu scone ), and other food-related items (e.g., farm vitamin breakfast ). The span tasks were all presented using a timed Mi crosoft PowerPoint presentation with a projector onto a blank wall. F or all tasks, the time allowed for recall of the words was group-paced and so advancement to the next trial was based on all participants indicating they were ready. Participa nts wrote their answers on paper answer
48 sheets. For all of the span tasks, there were 2 po ints possible for each word: 1 for the correct word and 1 for each word recalled in the co rrect order. Alphabet span. In the Alpha task, borrowed from Waters and Caplan (2003), participants saw a series of words, each for 5 seco nds, which they were instructed to remember (see Figure 2 for a depiction of stimulus presentation and participant response sequence). After the final word was presented, par ticipants were cued to RECALL the words in alphabetical order from that trial. If pa rticipants were unable to recall a word, they wrote an X in its place. This task varied in difficulty by the number of words presented per trial: easy (2 words), medium (3-4), and hard (5-7). Operation span. The OSPAN task was adapted from Turner and Engle ( 1989). It involved presentation of a series of basic mathemat ical equations paired with words (see Figure 3 for depiction of stimulus presentation and participant response sequence). Equations (e.g., (5x4)+2=22 ) included a multiplication or division problem fol lowed by an addition or subtraction problem, with a 1or 2digit number solution. Each equation was displayed for 15 seconds and was followed immed iately by a 1-second presentation of the paired word. While each equation was being presented, participants decided whether the solution provided was correct or incorr ect, circling Yes or No on their answer sheets. Then, after the final word displaye d, participants were cued to RECALL the words in the order in which they were presented If participants were unable to recall a word, they wrote an X in its place. This task varied in difficulty by t he number of words presented per trial: easy (1-2 words), medium (3-4), and hard (5-7). Reading span. The RSPAN task, adapted from Daneman and Carpente r (1980), was similar to the OSPAN except that there were sentences paired with words, and each
49 sentence was shown for 10 seconds followed by a 1-s econd presentation of the paired word (see Figure 4 for depiction of stimulus presen tation and participant response sequence). While each sentence was being presented participants decided whether it was grammatically correct (all words were in correct gr ammatical order) or incorrect (some words were mixed up and not in correct grammatical order) by circling Yes or No on their answer sheets. After the final word was disp layed, participants were cued to RECALL the words in the order in which they were presente d. If participants were unable to recall a word, they wrote an X in its place. Sentences for the RSPAN were gener ated such that they were 10 to 15 words in length (e.g., Corrie was excited to receive a Christmas card from her mother. ). The last several words in each sentence were ju mbled to create the grammatically incorrect stimuli (e.g. Corrie was excited to receive a card Christmas mother her from. ). This task varied in difficulty by the number of words presented per trial: easy (1-2 words), medium (3-4) and hard (5-7). Grocery Shopping Task All participants underwent a preand post-shoppin g task in the grocery store for a list of grocery items (see Table 3 for two shopping lists). The lists were generated such that the items were distributed throughout the stor e for each shopping session (see Figure 5 for grocery layout and item distribution) but ran domized within each list. To ensure equivalence between the two lists, the lists were t ested with college students ( N = 22) in the context of a psychology class. Half of the par ticipants saw the pre-test list, while the other half saw the post-test list. The words were presented in a similar manner to the Alpha task, with each word presented for 5 seconds. After the last word was presented, participants were cued to RECALL the words on their answer sheets. However,
50 participants were not required to recall the words in a particular order. If a given item was completely correct participants received 2 poin ts, and if the item was partially correct they received 1 point. The total score possible wa s 30 points (15 items worth 2 points each). Partial correctness entailed either of the following: 1) correct recall of the item but not its indicated brand name (e.g., juice instead of Welchs juice), or 2) recall of an item from the presented items category but not the exact item presented (e.g., any cereal instead of Cheerios). The latter criterion was n ot important for responses from the pilot test of the shopping lists but was important for th e grocery shopping sessions. There was no difference between the pre-test ( M = 17.73, SD = 3.55) and post-test ( M = 18.45, SD = 4.37) lists, t (20) = -.43, p > .05. Therefore, participants who were exposed to the pre-test list did not recall more items than those who were exposed to the post-test list. The grocery store for this study was chosen for it s rather small size and low number of customers during the day. Each aisle had a sign with the items it contained, in addition to individual labels down each aisle for t he exact locations. In addition to ensuring a fair distribution throughout the store, list items were chosen only if they did not appear exactly on either the aisle signs or wit hin-aisle labels. This was to ensure that participants who used the aisle signs and/or labels more frequently did not obtain an advantage over those who used them less frequently. In addition, the shopping carts provided aisle aides on the handlebars, though very few participants used them as assessed by experimenter notes, and they were not e specially helpful for the shopping lists tested.
51 Questionnaires Daily activities. The Daily Activities questionnaire was modified fro m Richards, Hardy, and Wadsworth (2003) for the present study. It inquired about the frequency with which people participated in a number of physical a nd leisure (social and cognitive) activities (see Table 4 for questionnaire items). In addition to the activities listed, participants were given the opportunity to list up to 4 other physical and leisure activities in which they were involved. Participants indicate d the frequency with which they participated in each of the activities: not at all , once a year, once a month, once a week, more than once a week, and every day. T he purpose of the questionnaire was to gather information about the level of activity o f participants in many domains of life. General demographic information (age, sex, highest level of education, and occurrence of a medically diagnosed memory or attention disorder) frequency of participation in cognitive training computer programs and classes, a nd familiarity with the grocery store used for testing were also gathered in the context of this questionnaire. Daily functioning. The Daily Functioning questionnaire (modified Instr umental Activities of Daily Living; IADL; see Table 5 for q uestionnaire items) was also modified for the present study from Willis et al. (2006). I t inquired about the independence with which participants completed several activities at home or in the community during the previous 7 days. Participants indicated on a scale of 1 to 6 the degree to which they completed each activity completely dependently (r elied on someone else to complete the task for them) or completely independently (c ompleted the task by oneself), or if the activity did not occur. The purpose of this self-report of daily functioning was to see if training had positive effects on overall dai ly life, as found in Willis et al.
52 Strategy use. The Strategy Use questionnaire was developed for th e present study to inquire about participants subjective experienc es of the training tasks. They reflected on the various tasks and reported what, if anything they found difficult about each trained WM task, as well as the preand post-groce ry shopping tasks. Additionally, they reported any strategies they used to help them to r emember the words for each task (training and grocery shopping) or to help them com plete the task overall. Participants also reported the frequency with which they partici pated in various food-related activities in the context of this questionnaire: grocery shopp ing, cooking, taking a cooking class, eating at a restaurant, and ordering take-out from a restaurant. In addition to the activities listed, participants could report up to three other food-related activities in which they were involved. They reported their frequency of pa rticipation as follows: not at all, once a year, once a month, once a week, more than once a week, and every day. Procedure Participants in all groups followed a similar time line (see Figure 6 for study timeline). Participants in the two training groups experienced the timeline as depicted. Training sessions 1-5 occurred over the course of 5 consecutive weeks. The grocery shopping tasks occurred within 2 weeks before and a fter training for the training groups, and simply 5-6 weeks apart for the control group. If participants were absent for a training session, they were required to make it up before the next weeks session. This occurred either one-on-one or in a small group sett ing, depending on scheduling and the number of participants that were absent for that we eks session. Over the course of the 5 weeks of training, the number of participants that made up sessions was as follows (the
53 number of absent participants is in parentheses): s ession 1 (2), session 2 (3), session 3 (2), session 4 (6), and session 5 (4). Participants who were absent reported health issues or memory slips, the latter of which made it especially important for those participants to make up the training sessions. In addition to the training and grocery shopping t asks, participants completed a number of questionnaires as listed above. The exce ption to the depicted timeline was that the control group completed the Daily Activities Qu estionnaire during the post-grocery shopping task. All questionnaires were answered in a paper format. Training sessions: Working memory. As stated before, training occurred over the course of 5 weeks and consisted of 5 approximat ely 1.5-hour group training sessions. Sessions occurred on the same day and at the same t ime every week in a classroom at a local senior center (except in the cases where abse nt participants made up sessions as noted above). Support for the duration, number, an d group style of the sessions came from Verhaeghen et al.s (1992) meta-analysis, in w hich the most successful training interventions used a similar format. The training aspect of the program entailed incr easing the WM tasks in difficulty over the course of the 5 weeks (see Figu re 6 for study timeline): an increase in the number of words to be recalled indicated an inc rease in difficulty. Each time a new WM task was introduced, it was trained that day fir st at an easy-medium level and then at a medium level. The following session, that partic ular WM task was trained at an increased medium-hard level. This same pattern of introduction and increase in difficulty occurred with each WM task. Both training groups e xperienced the same training format
54 and tasks except that the word lists used were grou p-specific (i.e., food-related words or random words). The training sessions commenced with the experimen ter explaining briefly the WM tasks that would be trained on that day. In add ition to this brief introduction, before each new WM task began, the experimenter explained in detail how the given WM task would progress and oriented the participants to the ir answer sheets. The experimenter answered any questions the participants had about t he task, and then began the practice trials. Each practice trial set included 5 trials, each 2 words in length. After the practice set, participants corrected their own answers accor ding to the answers provided on the presentation in order to check their understanding of the task procedure. The experimenter again clarified any issues participant s had with the task procedure before progressing. Then, the experimental trials began an d the number of trials and word list lengths varied according to the training session (s ee Figure 6 for the number of trials per task set). At the completion of the program, participants wer e awarded Certificates of Achievement for their perseverance through the rig orous memory training program. Transfer task: Grocery shopping. The grocery shopping sessions occurred oneon-one with the participant and experimenter. The shopping task had several components. First, participants were brought into the grocery store to the bench (see Figure 5 for grocery store layout) and the experime nter explained to them briefly about the grocery shopping task (an explanation of the en tire study was given when participants initially signed up for the study). Participants s igned an informed consent form and received their own copy. Then, in order to account for familiarity with the grocery store,
55 a brief familiarization session occurred in which t he participants were walked around the store and shown the aisles in which they would be t ested. A baseline walking rate was then taken down Aisle 5 of the store because, unbek nownst to the participants, that aisle did not contain any items on the shopping lists (as seen in Figure 5). The experimenter timed how long it took each participant to stroll d own the aisle. Then the participants were brought back to the front of the store to the bench and the experimenter explained in detail the testing procedure: 1) they would be show n a list of 15 items and have 2 minutes to study the list; 2) then the list would be taken away and they would have as much time as they needed (though they were encouraged to shop as swiftly as possible) to shop for the items by placing one of each item into their sh opping cart; and 3) there was one rule: on their first trip through the store they could on ly go in one direction to increasing number aisles (from Produce to Aisles 1-15). They were not required to travel down each aisle, and they were allowed to travel up and down within each aisle as much as they wanted, but they could not return to any aisles the y had already passed. However, once they reached Aisle 15, they were allowed to go back and visit any aisles they wanted as much as they wanted. This directional rule was inc luded to try to account for different shopping styles (i.e., participants who would shop aisle-by-aisle vs. those who would only go to necessary aisles). Since all participan ts were required to travel in a similar path, these shopping differences would be somewhat reduced. Additionally, it aided the experimenter in tracking participants shopping pat hs and progress throughout the task. Also when the participants reached Aisle 15, they w ere asked by the experimenter if they could remember any items they were unable to find a nd those items were counted as if they were found by the participants, as the purpose of the task was not store familiarity
56 but rather recall of the items. These later-recall ed items never totaled more than 4 items per participant and usually were only 1 or fewer. The experimenter followed the participants through the store by waiting out of sight at the end of each aisle. When a participant passed the experimenter to go to the next aisle, the experimenter recorded the items tha t had been found as well as the path of the participant through the aisles. The latter mea sure was also used to determine whether participants retraced and made a second trip to obt ain forgotten or recently-remembered items, as well as their method of shopping: just to the necessary aisles, aisle-by-aisle, or a combination of the two methods. The criterion for each shopping method was as follows: 1) just necessary aisles: participants skipped at least one aisle 2 or more times, 2) aisle-by-aisle: participants either skipped no ai sles or only one aisle, or 3) combination: participants followed the just nece ssary aisles method but also went to over half of the aisles (9 or more aisles). In add ition to these measures, the experimenter also recorded the overall time each participant sho pped. In the same way that the piloted shopping lists were scored above (see Grocery Shopp ing Task), the items in the grocery shopping task were scored. Just as a reminder, if a given item was completely correct participants received 2 points, and if the item was partially correct they received 1 point (total score possible was 30 points, 15 items worth 2 points each). Partial correctness could have entailed either of the following: 1) cor rect recall of the item but not its indicated brand name (e.g., any juice instead of W elchs juice), or 2) recall of an item from the presented items category but not the exac t item presented (e.g., any cereal instead of Cheerios).
57 Results The hypotheses of the present study posited for im provements from preto posttraining measurements were as follows: 1) significa nt improvement on the trained WM tasks for both training groups, 2) significant impr ovement on the grocery shopping transfer task for the explicit training group (thou gh perhaps moderate improvement for the non-explicit group) above that of the control g roup, and 3) somewhat lesser improvement (than on the trained and transfer tasks ) on the daily functioning measure for both training groups and no change for the control group. The analyses were divided in this section in the following manner: 1) WM trainin g tasks, and 2) grocery shopping and daily functioning transfer tasks. One participant was excluded from the analyses, despite completing all WM and grocery shopping tasks, due t o disengagement with the WM training tasks (discussed further below). Strategi es that participants used in each task as well as gender differences are also discussed in re ference to their effects on performance. All analyses were assessed at an alpha level of .05 Training Tasks: Working Memory Overall, participants within each training group di d not improve from preto posttest on their trained WM tasks (see Table 2 for gro up means): Alpha ( t (16) = -.55, p > .05), OSPAN ( t (16) = -.74, p > .05), or RSPAN ( t (16) = -1.963, p > .05). However, for all participants, improvement in the RSPAN task app roached significance (p=.07). When separated by group, participants in each group stil l did not improve significantly on any WM span task. For the explicit training group, par ticipants performed just as well preand post-training: Alpha ( t (7) = -.16, p > .05), OSPAN ( t (7) = -1.92, p > .05), or RSPAN ( t (7) = -.91, p > .05). However, improvement in the OSPAN task was moderate (p=.10).
58 This lack of significance was the same for the nonexplicit training group: Alpha ( t (8) = .64, p > .05), OSPAN ( t (8) = 1.04, p > .05), or RSPAN ( t (8) = -2.05, p > .05). However, improvement in the RSPAN task also approached signi ficance (p=.07). While there were no group improvements in the WM training task, ther e were some individual differences in progress throughout training. Figure 7 depicts participant-by-participant performance on the WM span tasks. Participant 3Bs performance on the WM tasks (see Figure 8) represent the typical pattern of results for partic ipants, such that they were engaged in the secondary equation/sentence/alphabetizing task (nea r 100% correct) but did not improve on all measures of the preto post-test scores on the primary task (word recall). A lack of engagement in the WM tasks was seen only in one par ticipant, Participant 31B (see Figure 8 for participants performance), whose scor es were at or below 50% on the secondary task, which was a yes/no choice task and so 50% indicated chance level. Using a mnemonic strategy to aid memory for the wor ds did not help participants to perform significantly better on any of the preor post-test WM span tasks (see Table 6 for group means). Participants who did use a strat egy ( n = 5) performed just as well as those who did not use a strategy ( n = 12). There were no significant differences between the two training groups on any of the WM span tasks from preto post-test, Wilks La mbda = .01, F (1,15) = 5.22, p > .05. Therefore, participants in the explicit training gr oup performed just as well as the nonexplicit group on the WM tasks for the pre-test (Al pha ( F (1,15) = .02, p > .05), OSPAN ( F (1,15) = .46, p > .05), and RSPAN ( F (1,15) = .97, p > .05)) and post-test (Alpha ( F (1,15) = .00, p > .05), OSPAN ( F (1,15) = 1.51, p > .05), and RSPAN ( F (1,15) = .64, p > .05)).
59 Transfer Tasks: Grocery Shopping and Daily Function ing Preand post-grocery shopping tasks. All participants reported they went grocery shopping once a week or more. In reference to familiarity with the grocery store used in the present study, no participants reported they shopped there regularly. Overall, participants did not improve from preto post-test on the grocery shopping task (see Table 2 for group means): accuracy of items ( t (20) = -.89, p > .05) or overall time ( t (20) = -.85, p > .05). When separated by group, participants in e ach group still did not improve significantly on the grocery shopping task. For the explicit training group, participants performed just as well on the preand post-grocery shopping task: accuracy in items ( t (7) = .31, p > .05) or overall time ( t (7) = -1.41, p > .05). These results were the same for the non-explicit training group who perfor med equally well preand posttraining: accuracy in items ( t (8) = -.50, p > .05) or overall time ( t (8) = 1.26, p > .05). The control group also did not change significantly fro m the preto post-grocery shopping task in items ( t (4) = -2.85, p > .05) or time ( t (4) = -1.65, p > .05), though they tended to retrieve more items on the postas compared to the pre-test ( p =.07). There were no significant differences between the two training groups on either of the shopping measures from preto post-test (see T able 2 for group means), Wilks Lambda = .01, F (2,18) = 2.33, p > .05. Therefore, participants in the explicit tra ining group performed just as well as the non-explicit gr oup on the grocery shopping task for the pre-test (accuracy of items ( F (2,18) = .02, p > .05) and overall time ( F (2,18) = .36, p > .05)) and post-test (accuracy of items ( F (2,18) = .36, p > .05) and overall time ( F (2,18) = 3.63, p > .05)). However, for all participants, improvemen t in the overall time approached significance ( p =.08), such that participants in the control group tended to
60 spend longer shopping than the non-explicit trainin g group. While there were no group improvements in the WM training task, there were so me individual differences in progress throughout training. Figure 9 depicts par ticipant-by-participant performance on the grocery shopping task measures. However, there were some cases of improvement on both sets of measures. For example, Participant s 21A and 30B (see Figure 10 for their performance) increased from preto post-test in th e number of items recalled, and 30B even decreased in the overall time it took to find the items. These were participants who also improved on one or more WM tasks from preto post-test. Using a mnemonic strategy to aid memory for the li st items did not help participants to perform significantly better on eit her of the preor post-test shopping measures (see Table 6 for strategy type means). Th erefore, participants who used a strategy ( n = 10) performed just as well as those who did not use a strategy ( n = 11). However, in the post-grocery shopping task particip ants who used a strategy tended to perform slightly better on the grocery shopping ite ms than those who did not use a strategy, though this result only approached signif icance. The type of shopping method participants used did not affect grocery shopping t ask performance significantly (see Table 7 for shopping method means). In the pre-gro cery shopping task, participants who just went to necessary aisles ( n = 6), those who shopped aisle-by-aisle ( n = 5), and those who used a combination of the two methods ( n = 10) all performed just as well on the grocery shopping task. However, in the pre-grocery shopping task, participants who shopped aisle-by-aisle tended to shop for more time overall than those who only went to the necessary aisles, though not to a significant d egree. Participants in the post-grocery shopping task who just went to necessary aisles ( n = 5), those who shopped aisle-by-aisle
61 ( n = 8), and those who used a combination of the two methods ( n = 8) all performed just as well on the shopping task. Also, in the post-gr ocery shopping task, participants who shopped aisle-by-aisle tended to recall more items than those who only went to the necessary aisles, though this result only approache d significance. Gender differences were also assessed in the grocery shopping task (se e Table 8 for means by gender) and overall showed moderate effects, Wilks Lambda = .6 2, F (1,19) = .2.47, p =.09. However, there were significant differences for som e measures between genders. For both the preand post-test, males took significant ly longer to shop than females and tended to find fewer items, though this latter effe ct was only moderate. Daily functioning. Participants did not differ within each group or ac ross groups on the questionnaire on Daily Functioning (see Tabl e 2 for group means). Within each group, participants did not improve on their level of independence for completing daily activities: explicit training ( t (7) = 1.00, p > .05), non-explicit training ( t (8) = .48, p > .05), or control (all participants scored the highest pos sible both preand post-test). Looking between groups, the explicit training group reporte d the same degree of independence with which they completed daily activities as the n on-explicit and control groups on the pre( F (2,18) = .11, p > .05) and post-test ( F (2,18) = .45, p > .05). Discussion The WM training program developed and utilized in t he present study was ineffective in causing improvements both within the trained domain and to a far transfer task. Participants within each training group, exp licit and non-explicit, did not improve to a significant degree from preto post-training on the three WM tasks, and there were
62 no differences between groups on these tasks. The main strength of the study was the equivalence of its participants across groups. Des pite the small sample size and convenience sampling procedures and group assignmen t, participants seemed to be well matched on age, education, gender, and cognitive, s ocial, and physical activity participation across groups. Effects of Training within the Trained Domain While there were no group improvements in the WM t raining task, there were some individual differences in progress throughout training. Figure 7 depicts participantby-participant performance on the WM span tasks. A cross participants there was not much change from preto post-test on the Alpha, an d a bit more variation (increases and decreases) on the OSPAN and RSPAN. Additionally, p articipants scored along almost the full range of scores possible, which indicates large individual differences on these span tasks both preand post-test. Like most part icipants, Participant 3B (see Figure 8 for participants performance) was engaged in the t ask, as measured by high performance on the secondary equation/sentence/alphabetizing ta sk (near 100%), but this participant was not able to improve on all measures of the preto post-test scores on the primary task, as measured by remembering the words in the W M span tasks. A lack of engagement was seen only in one participant, Partic ipant 31B (see Figure 8 for participants performance), whose scores were at or below 50% on the secondary task, which was a yes/no choice task and so 50% indicated chance level. This is likely because this participant was not a native English s peaker and reported difficulty completing the task due to difficulty with the lang uage. Some degree of improvement was experienced by several participants, and this e ffect may have been more pervasive
63 with a longer training period or more rigorous trai ning tools. For example, Participant 3B improved the RSPAN preto post-test score by over 15 points, and Participant 21A improved the OSPAN preto post-test score by over 35 points. One suggestion may be that this improvement was due to increased familiar ity with the task. However, some participants remained the same or decreased from pr eto post-test, despite repetition of these same tasks for 5 sessions, so familiarity is probably not the cause of 3Bs and 21As improvement. There are several explanations for the ineffective ness of the WM training tasks to improve post-test scores. One main reason is most likely the lack of an adaptive program. An adaptive program like the ones used in Jaeggi et al. (2008), Buschkuehl et al. (2008), and Olesen et al., (2004)) may have bee n more effective. Their WM training program caused participants to train at the edge of their abilities, and facilitated individual advancement to greater levels of difficu lty at an appropriate rate. The predetermined increases in difficulty in the WM tasks across the 5 training sessions of the present study likely did not create the steady rate of increasing difficulty like Jaeggi et al. Instead, the difficulty may have been too easy at t he start (when word lists were as little as 1-4 items long) and quickly have become too hard (when word lists reached 7 items long). This assertion is supported by Participant 3Bs performance (see Figure 8 for participants performance), who performed near perf ectly on the Alpha and OSPAN tasks until the medium-hard difficulty level when the sco re reduced to 60%. While the predetermined difficulty increases were necessary beca use the training occurred in a group setting, they did not seem to be effective training tools. An adaptive program is not a requirement for a successful WM training program as Li et al. (2008) used pre-
64 determined increases in difficulty. However, Li et al.s training sessions occurred only 15 minutes per session, for 45 sessions, and so the set increases in difficulty likely seemed more adaptive in nature to participants cur rent skill levels than the large jumps in difficulty of the present study across sessions. Additionally, another successful aspect of Jaeggi et al., which was not a quality of the present study, and the other WM training progra ms cited above (Buschkuehl et al., 2008; Li et al., 2008; Olesen et al., 2004; and Per sson & Reuter-Lorenz, 2008) was that the training was independently conducted by partici pants on a computer. Therefore, there was not a pre-determined time duration to, for exam ple, solve a math equation or read a sentence. Instead, as soon as participants respond ed to the correctness of the equation/sentence, the next set of stimuli appeared in a participant-paced manner. In this way, the training program was fast-paced, adding to the rigor of the training, and prevented participants from rehearsing and, related ly, implementing strategies to help them to remember the items. Instead, they were for ced to strictly rely on their WM storage capacity. In the present study, participan ts who solved the math equations and read the sentences quickly had extra time to rehear se the words, which may have prevented enduring expansion of the WM capacity har dware in favor of an artificial, temporary expansion of WM capacity through implemen tation of strategy software, such as grouping words or associating them with ones li fe. Participants did not work to cause a more enduring change in the capacity of their WM hardware, but rather superimposed strategy software to circumvent the WM capacity lim itations. However, results showed that participants who used some strategy did not pe rform better on the WM tasks, though the small sample size may have prevented detection of this effect. Buschkuehl et al.
65 (2008) notes that it is important for a training pr ogram to be adaptive and minimize strategy development in order to be effective and f acilitate transfer to other tasks. The pre-timed nature of the stimulus presentation also caused some difficulty for participants who were not vigilant and missed the words in the O SPAN and RSPAN tasks, for they were flashed for only 1 second each. However, thes e difficulties occurred for a minority of participants and trials. Based on subjective reports of the participants in a questionnaire, there was a difference in approach for the Alpha task (particip ants were shown words and asked to recall them in alphabetical order). Some participa nts reported rehearsing the words in the order in which they were presented and alphabetizin g them only as they were writing them down. Other participants reported that they a lphabetized the words as they were presented to them. Given that the task was meant t o be a simultaneous storage and processing WM task, the intention was for participa nts to alphabetize as each word was presented. The approach participants used in the A lpha task approach was not reported by many participants, so this incongruity could not be measured across participants as impacting performance. However, it is clear that t he participants who waited to alphabetize until recall did not engage in a dual-c omponent WM task, neglecting the processing component until recall. While there were some crucial disadvantages to per forming these tasks in a group setting, Verhaeghen et al.s (1992) meta-analysis s uggested that group settings aided training programsor at least those teaching mnemon ic strategiesand were beneficial to performance due to social comparison, within-gro up support, and increased motivation for participants. WM training programs may benefit from being independently conducted
66 on a participant-by-participant basis, though anecd otally, participants in the present study seemed to enjoy the group setting and reported bein g motivated to outperform their friends and other fellow participants. However, th is social comparison may also have led to an increase in participants anxiety. This was reported by some participants as an explanation especially for their poorer performance on the post-grocery shopping task. A few participants reported feeling anxious about per forming better on the post-grocery shopping task such that it interfered with their re call of the lists. While performance on the WM and shopping tasks was not reported to parti cipants until the end, it was easy to track ones progress as performance was easily meas ured by oneself, and so progression or lack thereof may have induced such anxiety. Another issue with the classroom setting was the l evel of noise both by participants and outside of the room. The WM span task required a great amount of attention that could be easily disrupted by noise i nterference. While participants were instructed to remain quiet during the trials, some participants did talk or make noises on some trials. This notably disrupted performance fo r some participants on some trials. Also, the classroom in which the training sessions were held was at a local senior center. The center is known for its daily musical performan ces, which sometimes coincided with session times. The classroom used is distanced fro m the musical playroom but sometimes the music was distracting to participants especially if the songs were familiar to them. However, it is unlikely that this noise i nterference had a great affect on WM training as there were no significant group differe nces on performance across sessions or from preto post-test. In fact, the noise interfe rence may have forced participants to focus their attention more intensely on the tasks. Inclusion of preand post-tests of
67 attention and inhibition would have revealed the po tential for these non-targeted training effects. Effects of Training on Two Far-Transfer Tasks Neither training group, nor the control group, imp roved from the preto postgrocery shopping or daily functioning transfer task s, and there were no differences between the groups on scores. The lack of transfer to the grocery shopping task is likely due to the lack of training effects. The WM traini ng program was not strong enough to improve performance consistently for any group, mak ing potential transfer nonexistent (Buschkuehl et al., 2008). Finding transfer effect s in older adults has also proved a difficult feat for other training programs (as cite d in Buschkuehl et al., 2008), so perhaps younger adults would have experienced some transfer effects in the same training program. Figure 9 depicts participant-by-participa nt performance on the grocery shopping task measures. Across participants there was not much change from preto post-test on either the grocery items or grocery ti me measures. Additionally, there were many individual differences on the grocery items sc ore, with participants scoring along almost the full range of scores possible. The tota l time to complete the shopping task was less variable, with most participants spending abou t 10-15 minutes shopping, although there was one outlier who took about 15 minutes lon ger to shop than the others. It should be noted that participants who experienced improvem ent, as evidenced by positive slopes, on the WM tasks did not necessarily experie nce improvement in the grocery shopping task. However, while this was not the nor m, there were some cases of improvement on both sets of measures. For example, Participants 21A and 30B (see Figure 10 for their performance) increased from pre to post-test in the number of items
68 recalled, and 30B even decreased in the overall tim e it took to find the items. These were participants who also improved on one or more WM ta sks from preto post-test. Explicit strategy use, such as grouping items or associating them with ones life, did not account for the increase in success in the grocery shopping task. The control group did not significantly improve fr om their preto post-test in the grocery shopping task, so it seems that the somewha t greater familiarity with the grocery store after having shopped for the pre-test several weeks before did not enhance performance on the post-test. Unexpectedly though, the control group tended to retrieve more items on the postas compared to the pre-test (p=.07). This effect may have been significant with a bigger sample size. It is uncle ar as to the reason for improvement without training, except that perhaps greater famil iarity with the store for this group did have beneficial effects on the task. It is likely that the training groups overall did not improve because of an increased level of anxiety an d pressure to increase performance from preto post-test in the grocery store. Parti cipant performance was very easily tracked throughout the task as participants knew th e number of items they had retrieved during the pre-test and perhaps felt anxious about performing equivalently and/or better on the post-test. Some participants reported this increased anxiety to perform on the post-test, which supports this claim. The control group was not informed about the inclusion of the training groups and did not hold t he expectation that they needed to improve on the post-test, and mostly likely did not experience anxiety. A factor in the grocery shopping task that may hav e affected performance was the method of shopping used: just to necessary aisles, aisle-by-aisle, or a combination of those two methods. The label just to necessary ai sle refers to participants perception
69 of the necessary aisles as these participants were the ones who used the aisle labels to determine which aisles to go down and so objectivel y skipped more aisles, as opposed to the aisle-by-aisle shoppers who simply went to ev ery aisle. Participants who used a combination of the two methods likely adopted an ai sle-by-aisle approach but still looked at aisle labels, skipping only those they were cert ain did not contain the test items. These different shopping methods potentially created two types of memory tasks: recognition and recall. While participants may have continuall y rehearsed some of the 15 items, due to the average duration of the task (about 15 minut es) as well as participants reports (some reported not remembering the items until they saw the aisle labels or the actual item in the shelf, which triggered their memory for it), many of the items were likely encoded in LTM. These differential conditions were the same as those cited by Ross, Spencer, Linardatos, Lam, and Perunovic (2004) in t heir memory test in a grocery store: participants experienced recognition when they scanned the shelves or aisle labels for no particular item and then remembered a test item whe n they saw it, and recall when they remembered an item and then sought it out in the st ore. This dichotomy of memory tasks corresponds with the dichotomy of shopping methods such that just-to-necessary-aisle shoppers likely used recall memory to find items, w hile aisle-by-aisle shoppers likely used recognition memory to find items. However, wh ile these pairings of memory task and shopping method appear plausible, it is difficu lt to determine whether aisle-by-aisle shoppers relied on recognition memory or if they tr uly recalled the items and just appeared to be shopping aisle-by-aisle because ther e were test items in almost every aisle of the store (see Figure 5 for distribution of item s throughout store).
70 This explanation of differential memory tasks acro ss shoppers is not clearly supported by the present data as all participants, despite shopping method, performed equally well on item retrieval. On the post-test, participants who used an aisle-by-aisle method tended to find more items, though this was o nly to a moderate, insignificant degree. However, if the methods created two memory task versions, one recall and one recognition, participants who relied on recall (jus t-to-necessary-aisle shoppers) would have been disadvantaged as recall memory is worse i n older adults than recognition memory (Schonfield & Robertson, 1966). In a test o f remembering a list of nouns and adjectives, older adults performed equally well as younger adults on a recognition task but performed worse on a recall task. The study in cluded about 20 participants from each decade 20-60 years and older, and found a steady de crease in recall memory as age increased, while recognition memory was stable desp ite age. This age-related difference in memory tasks is due to the increased attentional demand required for recall tasks, which require un-cued retrieval from LTM, while rec ognition tasks do not require the same attentional demands (Craik & McDowd, 1987) Another age-related difference in memory for lists refers to the frequency of the words (the rating of occurrence in a language along a low to high continuum). The word lists used for each WM and grocery shopping task in cluded words of both high and low frequency. First, in reference to the WM test list s, the explicit training group saw word lists that were mostly low frequency words. Just b y nature of the food-theme requirement for these lists, words such as scone and skillet that are low frequency in the English language comprised most of the lists. However, for the non-explicit training group, since the word lists did not have a required theme, they were comprised of all high
71 frequency words, such as bread and table, extra cted from an online word database. This distinction in word lists between the training groups may be important in light of research in word frequencies and age-related change s in recognition and recall. Older adults tend to be slower to recognize low frequency words than younger adults (Allen, Wallace, & Waag, 1991), which likely has detrimenta l implications for older adults recall of those words. However, if word lists are mixed with high and low frequency words, as the lists of the explicit group, low freq uency words are recalled just as well and sometimes better than high frequency words, at leas t in younger adults (MacLeod & Kampe, 1996). The equivalence of the words lists i s supported by the equivalent performance of the two groups preand post-test. However, it would be interesting to measure differential training effects of memory tas ks using purely low or high frequency lists or different proportions of highand low-fre quency words. The grocery shopping lists also included mostly lo w but also some high frequency words. Shopping items such as apple and orange can be considered high frequency, while linguine and Smuckers jam are obviously lower frequency. The grocery lists were pilot tested with younger adults, who do not s how the same degree of differential recognition and recall of low and high frequency wo rds (Allen et al., 1991) and who when shown lists of mixed high and low frequency wo rds recall them equally well (MacLeod & Kampe, 1996). Therefore, the two test l ists may have been less equivalent than the pilot test suggested since the present stu dy included older adults. This inequality in lists may have affected the preand post-grocer y shopping task performances of older participants.
72 While female participants ( n = 18) composed a great majority of the sample, mal e participants ( n = 3) performed significantly differently than the females on some aspects of the grocery shopping task. For both shopping se ssions, males took significantly longer to shop than females and tended to find fewer items though this latter effect was only moderate but may have been more pronounced with a l arger sample size. The reason for the greater, or at least more efficient, performanc e by females is unclear as all participants reported grocery shopping at least onc e a week. However, as was noted in Ross et al.s (2004) study, while two participants (husband and wife) were present for the shopping trips, one participant was usually more re sponsible for retrieving the items, and this participant was usually the female participant Therefore, the female participants may be more familiar with the grocery shopping expe rience, allowing them to shop more swiftly than males. There were some additional unforeseen issues with the grocery store itself and the ways in which people navigated its layout. Though not apparent from Figure 5s layout, it was easy for participants to pass Aisles 2 and 3 and start on Aisle 4. Aisles 2 and 3 each contained a test item for both sessions, but w ere somewhat hidden because half of each of the aisles, and the portion of the aisles t hat participants saw when approaching them from the produce section, contained exclusivel y wine, which was never a test item. When participants saw the wine in the beginning of the aisles, they tended to bypass the two aisles entirely. Later some participants repor ted being confused and/or frustrated that they could not find the items, which sometimes caus ed them to forget that item or other items. Additionally, some participants looked for items in places where they expected them to be but were not, which also caused confusio n and/or frustration and sometimes
73 led to forgetfulness by the end. Despite these set -backs in the grocery stores design, participants were mostly very successful at finding items, and the inclusion of the experimenter asking participants if there were item s they could not find once they reached the final aisle also served to ease the unf amiliarity with the store. Regardless of the lack of training effects, the la ck of transfer to the daily functioning measure is likely due to the ceiling ef fects of the pre-test measure in which all participants in the control group as well as a number of participants in the training groups self-reported complete independence when per forming daily activities (i.e., they received the highest score possible on the measure) If participants were already functioning at the highest level of independence, t here would have been no room for improvement. In this way, participants in the pres ent study seem to be especially high functioning older adults, though participants in Wi llis et al.s (2006) study, who did show improvement in this area, were reportedly living in dependently and of good functioning and cognitive ability too. Future Research The keys to an effective WM training study, based on previous researchs success and the present studys lack of success, seem to be 1) adaptivity and 2) rigor. The adaptivity component forces participants to train a t the edge of their ability, being neither overly easy nor overly difficult for the individual Participants are not permitted to become bored at a relatively simplistic level, and are also not pushed too far beyond their ability such that they lose motivation. Also, trai ning programs need to be rigorous. This quality stems partly from the challenging nature of the adaptive program, but also to the distribution and duration of training sessions. Se ssions should not be too long or sparsely
74 dispersed throughout the training program, so that gains can be acquired without fatigue or decay (loss over time). With these two componen ts, WM training programs have great potential, with younger and especially older adults This potential exists to improve performance not o nly in the trained domain but to near and far transfer tasks. There exists no pract ical benefit to becoming an expert in being able to remember a sequence of emphasized blo cks paired with words, so transfer to other domain-general tasks that have real-life, practical effects is the goal. For example, Jaeggi et al. (2008) included a lab-measur ed far transfer task of reasoning skill (Ravens Progressive Matrices measuring fluid cogni tion), which has been positively correlated in other studies with learning and adapt ing to new situations in the real world (Carpenter, Just, & Shell, 1990 as cited in Jaeggi et al., 2008). Therefore training within the lab has practical, real-life applications outsi de the lab. However, while reasoning skill is positively correlated with benefits in rea l life, it remains uncertain if training to affect reasoning would produce these same real-life effects, or if the WM link to real-life fluid cognition is immutable. Future training rese arch should include real-life transfer tasks to address this question. Conclusion While the present study did not successfully impro ve most peoples performance in either the trained WM domain or transfer LTM dom ain, there were a few examples of improvement in both areas, and inclusion of the gro cery shopping task added to a gap in the research with practical, real-life transfer tas ks. Performance of older adults in the shopping task united several areas of research on a ge-related differences in memory performance, such as word frequency research and re call vs. recognition research; and so
75 the grocery store seems to be both a practical and useful context to test many aspects of memory. The potential of WM training programs, as evidenced by recent research, is great though their methods have thus far been quite diverse. Further study of WM trainings mechanisms and effects is necessary, and the present study highlighted some methodological concerns involved in this type of co gnitive hardware training program.
76 References Allen, P. A., Wallace, B., & Waag, E. (1991). Effec t of imagery ability on letter-level and word-level processing. Perception & Psychophysics, 49 295-300. Ball, K., Berch, D. B., Helmers, K. F., Jobe, J. B. Leveck, M. D., Marsiske, M., et al. (2002). Effects of cognitive training interventions with older adults: A randomized controlled trial. Journal of the American Medical Association, 288 2271-2281. doi: 10.1001/jama.288.18.2271 Borella, E., Carretti, B., & De Beni, R. (2008). Wo rking memory and inhibition across the adult life-span. Acta Psychologica, 128 33-44. doi: 10.1016/j.actpsy.2007.09.008 Buschkuehl, M., Jaeggi, S. M., Hutchison, S., Perri g-Chiello, P., Dpp, C., Mller, M., Perrig, W. J. (2008). Impact of working memory training on memory performance in old-old adults. Psychology and Aging, 23 743-753. doi: 10.1037/a0014342 Craik, F. I. M., & Bialystok, E. (2006). Cognition through the lifespan: Mechanisms of change. Trends in Cognitive Sciences, 10 131-138. doi: 10.1016/j.tics.2006.01.007 Craik, F. I. M., & McDowd, J. M. (1987). Age differ ences in recall and recognition. Journal of Experimental Psychology: Learning, Memor y, and Cognition, 13 474479. doi: 10.1037/0278-73188.8.131.524 Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of Verbal Learning and Verbal Behavior, 19, 450-460. doi: 10.1016/S0022-5371(80)90312-6
77 Gibson, Gondoli, & Grundy (in press). Fluid IQ as a potential mediator of working memory training for attention-deficit/hyperactivity disorder. Journal of Child Psychology and Psychiatry Hambrick, D. Z., & Engle, R. W. (2002). Effects of domain knowledge, working memory capacity, and age on cognitive performance: An inve stigation of the knowledgeis-power hypothesis. Cognitive Psychology, 44 339-387. doi: 10.1006/cogp.2001.0769 Hambrick, D. Z., Salthouse, T. A., & Meinz, E. J. ( 1999). Predictors of crossword puzzle proficiency and moderators of age-cognition relatio ns. Journal of Experimental Psychology: General, 128 131-164. doi: 10.1037/0096-34184.108.40.206 Hofmann, M., Hock, C., Kuhler, A., & Muller-Spahn, F. (1996). Interactive computerbased cognitive training in patients with Alzheimer 's disease. Journal of Psychiatric Research, 30 493-501. doi: 10.1016/S0022-3956(96)00036-2 Hultsch, D. F., Hertzog, C., Small, B. J., & Dixon, R. A. (1999). Use it or lose it: Engaged lifestyle as a buffer of cognitive decline in aging. Psychology & Aging, 14 245-263. doi: 10.1037/0882-79220.127.116.11 Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perri g, W. J. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, USA, 105 6829-6833. doi: 10.1073/pnas.0801268105 Kawashima, R., Okita, K., Yamazaki, R., Tajima, N., Yoshida, H., Taira, M.,...Sugimoto, K. (2005). Reading aloud and arithmetic calculation improve frontal function of people with dementia. Journals of Gerontology: Series A: Biological Scie nces and Medical Sciences, 60 380-384.
78 Klingberg, T., Fernell, E., Olesen, P. J., Johnson, M., Gustafsson, P., Dahlstrm, K., Westerberg, H. (2005). Computerized training of working memory in children with ADHD: A randomized, controlled trial. Journal of the American Academy of Child & Adolescent Psychiatry, 44 177-186. doi: 10.1097/00004583200502000-00010 Klingberg, T., Forssberg, H., & Westerberg, H. (200 2). Training of working memory in children with ADHD. Journal of Clinical and Experimental Neuropsycholo gy, 24 (6), 781-791. Li, S., Schmiedek, F., Huxhold, O., Rcke, C., Smit h, J., & Lindenberger, U. (2008). Working memory plasticity in old age: Practice gain transfer, and maintenance. Psychology and Aging, 23 731-742. doi: 10.1037/a0014343 MacLeod, C. M., & Kampe, K. E. (1996). Word frequen cy effects on recall, recognition, and word fragment completion tests. Journal of Experimental Psychology, 22 132-142. doi: 10.1037/0278-7318.104.22.168 Olesen, P. J., Westerberg, H., & Klingberg, T. (200 4). Increased prefrontal and parietal activity after training of working memory. Nature Neuroscience, 7 75-79. doi: 10.1038/nn1165 Park, D. C., Lautenschlager, G., Hedden, T., Davids on, N. S., Smith, A. D., & Smith, P. K. (2002). Models of visuospatial and verbal memory across the adult life span. Psychology and Aging, 17 299-320. doi: 10.1037/0882-7922.214.171.1249 Persson, J., & Reuter-Lorenz, P. (2008). Gaining co ntrol: Training executive function and far transfer of the ability to resolve interference Psychological Science, 19 881888. doi: 10.1111/j.1467-9280.2008.02172.x
79 Richards, M., Hardy, R., & Wadsworth, M. E. J. (200 3). Does active leisure protect cognition? Evidence from a national birth cohort. Social Science & Medicine, 56 785-792. doi: 10.1016/S0277-9536(02)00075-8 Ross, M., Spencer, S. J., Linardatos, L., Lam, K. C H., & Perunovic, M. (2004). Going shopping and identifying landmarks: Does collaborat ion improve older peoples memory? Applied Cognitive Psychology, 18 683-696. doi: 10.1002/acp.1023 Salthouse, T. A., Berish, D. E., & Miles, J. D. (20 02). The role of cognitive stimulation on the relations between age and cognitive function ing. Psychology and Aging, 17 548-557. doi: 10.1037/0882-79126.96.36.1998 Schonfield, D., & Robertson, B. (1966). Memory stor age and aging. Canadian Journal of Psychology, 20 228-236. doi: 10.1037/h0082941 Schooler, C., & Mulatu, M. S. (2001). The reciproca l effects of leisure time activities and intellectual functioning in older people: A longitu dinal analysis. Psychology and Aging, 16 466-482. doi: 10.1037/0882-79188.8.131.526 Schooler, C., Mulatu, M. S., & Oates, G. (1999). Th e continuing effects of substantively complex work on the intellectual functioning of old er workers. Psychology and Aging, 14 483-506. doi: 10.1037/0882-79184.108.40.2063 Turner, M. L., & Engle, R. W., (1989). Is working m emory capacity task dependent? Journal of Memory and Language, 28 127-154. doi: 10.1016/0749596X(89)90040-5 Verhaeghen, P., Marcoen, A., & Goossens, L. (1992). Improving memory performance in the aged through mnemonic training: A meta-analytic study. Psychology and Aging, 7 242-251. doi: 10.1037/0882-79220.127.116.11
80 Waters, G. S., & Caplan, G. (2003). The reliability and stability of verbal working memory measures. Behavior Research Methods, Instruments, & Computers 35 550-564. Willis, S. L., Tennstedt, S. L., Marsiske, M., Ball K., Elias, J., Koepke, K. M., Wright, E. (2006). Long-term effects of cognitive training on everyday functional outcomes in older adults. JAMA, 296 2805-2814. doi : 10.1001/jama.296.23.2805 Wilson, R. S., Mendes de Leon, C. F., Barnes, L. L. Schneider, J. A., Bienias, J. L., Evans, D. A., & Bennett, D. A. (2002). Participatio n in cognitively stimulating activities and risk of incident Alzheimer disease. JAMA, 287 742-748. doi: 10.1001/jama.296.23.2805
81 Figure 1 Model of Working Memory and Long-Term Memorys Rela tions to Training and Transfer Tasks TRANSFER TASK: Grocery Shopping Storage Processing WORKING MEMORY Rehearsal LONG-TERM MEMORY words Encoding/ Retrieval TRAINING TASK: WM Spans Storage Processing WORKING MEMORY Rehearsal for the OSPAN and RSPAN tasks Processing solve math equation read sentence alphabetize read aisle label plan shopping route Storage stimulus words to be recalled WORKING MEMORY
82 Figure 2 Alphabet Span Stimulus and Participant Response Seq uence bread apricot etc. RECALL 5 s 5 s Stimulus Sequence watch watch etc. recall words 5 s 5 s Participant Response Sequence
83 Figure 3 Operation Span Stimulus and Participant Response Se quence (5x4)+2=22 bread (12/3)-1=3 15 s 1 s 15 s apricot etc. RECALL 1 s Stimulus Sequence circle Yes or No watch circle Yes or No 15 s 1 s 15 s watch etc. recall words 1 s Participant Response Sequence
84 Figure 4 Reading Span Stimulus and Participant Response Sequ ence circle Yes or No watch circle Yes or No 10 s 1 s 10 s watch etc. recall words 1 s Participant Response Sequence Corrie was excited to bread Spending his vacation near 10 s 1 s 10 s apricot etc. RECALL 1 s Stimulus Sequence
85 Figure 5 Layout of Grocery Store with Food Items from Prea nd Post-Tests Key Entrance Produce Check-out Lines 1 2 3 4 5 6 7 8 9 10 11 12 13 15 14 Natural Food A B A B A B A B A B B A B A B A B A B A B A B A B B A B A A A B Bench = Pre-test = Post-test Note The shape indicators do not indicate the exact lo cation of the items within each aisle, but rather t hat the item was located somewhere in the aisle.
86 Figure 6 Timeline of Study Training set Span task Difficulty level Pre-test all span tasks easy, medium, hard (40) Pre-test (continued) all span tasks easy, medium, hard (40) Set 1 Alpha Alpha easy, medium (12) medium (12) Set 2 OSPAN easy, medium (12) OSPAN medium (12) Alpha medium, hard (15) Set 3 RSPAN easy, medium (12) RSPAN medium (12) OSPAN medium, hard (15) Set 4 RSPAN medium, hard (15) Post-test all span tasks easy, medium, hard (40) Session 1 Session 2 Session 3 Session 4 Session 5 Grocery Pre-Test Grocery Shopping Task Pre-test Daily Functioning Questionnaire Pre-test Grocery Shopping Task Post-Test Daily Functioning Questionnaire Post-Test Strategy Use Questionnaire Grocery Post-Test *Daily Activities Questionnaire Note The number in parentheses beside each task diffic ulty level indicates the total number of trials.
87 Reading Span0 10 20 30 40 50 60 70 80 90 100 110Pre-testPost-test Total words recalled Operation Span0 10 20 30 40 50 60 70 80 90 100 110Pre-testPost-test Total words recalled Alphabet Span0 10 20 30 40 50 60 70 80 90 100 110Pre-testPost-test Total words recalled Figure 7 Participant-by-Participant Performance on WM Span T asks Preand Post-Test Note A shaded circle represents the explicit training group and an unshaded circle represents the non-exp licit training group. The three span task scores represent total n umber of words recalled.
88 0 20 40 60 80 100 Pretest (easy-hard) Easy-medMediumMed-hardPosttest (easy-hard)Difficulty Level/SessionWM Tasks: % Correct0 5 10 15 20 25 30 Shopping Task 0 20 40 60 80 100 Pretest (easy-hard) Easy-medMediumMed-hardPosttest (easy-hard)Difficulty Level/SessionWM Tasks: % Correct0 5 10 15 20 25 30 Shopping Task Figure 8 Performance of Participants 3B and 31B on Training and Transfer Tasks Note Alpha = alphabet span, OSPAN = operation span, RS PAN = reading span, Grocery items = total item scor e, and Grocery time = total shopping time. 3B 31B
89 Grocery Time0 5 10 15 20 25 30 35Pre-testPost-test Total time (in minutes) Grocery Items0 5 10 15 20 25 30 35Pre-testPost-testItem scoreFigure 9 Participant-by-Participant Performance on the Groce ry Shopping Task Preand PostTest Note A shaded circle represents the explicit training group and an unshaded circle represents the non-exp licit training group. The Grocery Items is scored accordi ng to the procedure described in the text, and the Grocery Time is the total shopping time in minutes.
90 Figure 10 Performance of Participants 21A and 30B on Training and Transfer Tasks 0 20 40 60 80 100 Pretest (easy-hard) Easy-medMediumMed-hardPosttest (easy-hard)Difficulty Level/SessionWM Tasks: % Correct0 5 10 15 20 25 30 Shopping Task Note Alpha = alphabet span, OSPAN = operation span, RS PAN = reading span, Grocery items = total item scor e, and Grocery time = total shopping time. 0 20 40 60 80 100 Pretest (easy-hard) Easy-medMediumMed-hardPosttest (easy-hard)Difficulty Level/SessionWM Tasks: % Correct0 5 10 15 20 25 30 Shopping Task 21A 30B
91 Table 1 Participant Characteristics and MANOVA Analysis Explicit training ( n = 8) Non-explicit training ( n = 9) Control ( n = 4) MANOVA M SD M SD M SD F (2,18) p Age 79.00 5.29 75.33 7.95 77.00 9.20 .54 .59 Education 2.88 .99 2.67 1.12 2.25 .50 .53 .60 Baseline walk 21.88 5.69 21.67 3.16 21.25 4.92 .03 .98 Cognitive 12.50 5.71 10.78 3.19 11.75 3.30 .33 .72 Social 9.69 4.30 10.00 2.83 12.00 2.16 .67 .53 Physical 9.81 6.80 8.11 4.26 9.25 3.86 .22 .80 Food 12.19 1.13 11.78 2.05 11.88 2.02 .12 .89 Note. Education (1= high school, 2=some college/not 4-ye ar college, 3=college, 4=post-graduate), Baseline walk = baseline walking rate (in seconds), Cognitiv e = cognitive daily activities, Social = social dai ly activities, Physical = physical daily activities, a nd Food = food-related daily activities.
92 Table 2 Performance on Training and Transfer Tasks by Group Explicit training Non-explicit training Control M SD M SD M SD Pre-test Alpha 64.25 17.69 62.89 20.28 --OSPAN 63.75 21.27 70.56 19.99 --RSPA 59.62 15.30 50.11 23.20 --Grocery items 19.50 4.81 19.89 5.97 18.75 2.99 Grocery time 14.12 5.94 12.78 2.99 15.25 7.97 Daily func 34.88 3.18 34.22 4.60 36.00 .00 Post-test Alpha 64.88 14.65 64.78 16.88 --OSPAN 77.50 12.10 65.11 26.06 --RSPAN 64.25 17.28 56.44 22.20 --Grocery items 19.00 4.96 20.67 6.36 22.50 1.29 Grocery time 16.00 7.25 11.11 2.52 19.75 13.05 Daily func 34.62 3.16 33.33 4.56 36.00 .00 Note. Alpha = alphabet span, OSPAN = operation span, RSP AN = reading span, Grocery items = total item score, Grocery time = total shopping time, and Dail y func = Daily Functioning Questionnaire score. Fo r Alpha, OSPAN, and RSPAN the scores represent the me an number of words recalled on all trials. The Grocery items is scored according to the procedure described in the text, and the Grocery time is the mean time in minutes. A -- indicates that data for th is measure was not applicable to the group and was not included in the analyses.
93 Table 3 Grocery Shopping Lists for Preand Post-Test Pre-test Post-test pecans linguine folgers coffee apple smucker's jam honey cough drops toilet paper cheerios elmer's glue eggs gatorade jello mix listerine ranch dressing mac n' cheese oatmeal crest toothpaste dove soap hershey's syrup goldfish tide detergent cornbread mix yoplait yogurt vinegar kleenex welch's juice orange chocolate chips canned tuna
94 Table 4 Daily Activities Questionnaire Items Physical activities Leisure activities Sports (for example, golf, tennis, etc.) Jogging/running Walking Swimming Sailing, boating, or similar activity Yoga Fishing Dancing Gardening + Church or religious activities + Go to the movies, theater, or concerts + Visit friends, attend a party, or similar activity + Volunteer with a club, local government, or other organization Read a book, the newspaper, or other reading material Play a musical instrument Chess, bridge, or similar games Crossword puzzles, Sudoku, or similar games Computer games Learn a language Note A + indicates a social leisure activity and a - indicates a cognitive leisure activity.
95 Table 5 Daily Functioning Questionnaire Items Activities Meal preparation (such as planning meals, cooking, assembling ingred ients, setting out food and utensils) Housework (such as doing dishes, dusting, making bed, tidying up, laundry) Managing finances (such as paying bills, balancing a checkbook and ho usehold expenses) Managing medications (such as taking medications at prescribed time and taking correct drug dosages) Shopping (such as selecting items, managing money) Telephone use (such as making and receiving phone calls)
96 Table 6 Effect of Strategies on Performance on Training and Transfer Task Means t test Strategy No strategy t (15)* t (19)** p Pre-test Alpha* 67.00 62.08 -.49 .63 OSPAN* 65.00 68.33 .30 .77 RSPAN* 59.60 52.00 -.66 .52 Grocery items** 20.20 18.91 -.60 .56 Grocery time** 14.00 13.55 -.20 .85 Post-test Alpha* 64.80 64.83 .00 1.00 OSPAN* 80.40 67.00 -1.21 .24 RSPAN* 69.40 56.25 -1.27 .22 Grocery items** 22.00 18.91 -1.40 .18 Grocery time** 13.80 15.36 .46 .65 Note. Alpha = alphabet span, OSPAN = operation span, RSP AN = reading span, Grocery items = total item score, and Grocery time = total shopping time. For Alpha, OSPAN, and RSPAN the scores represent the mean number of words recalled on all trials. The G rocery items is scored according to the procedure described in the text, and the Grocery time is the mean time in minutes. The df for each set of analyses, training (Alpha, OSPAN, and RSPAN) and transfer (Gr ocery items and Grocery time), is indicated by * and ** respectively.
97 Table 7 Effect of Shopping Methods on Performance on Shoppi ng Task Means MANOVA Necessary Aisle-by-aisle Combo F (2,18) p Pre-test Grocery items 16.50 21.40 20.40 1.82 .19 Grocery time 12.17 16.60 13.30 1.10 .35 Post-test Grocery items 16.20 22.75 20.63 2.97 .08 Grocery time 10.40 17.50 14.38 1.40 .27 Note. Necessary = just to necessary aisles method, ais le-by-aisle = aisle-by-aisle method, Combo = combination of previous two methods, Grocery items = total item score, and Grocery time = total shoppi ng time. The Grocery items is scored according to the procedure described in the text, and the Grocery t ime is the mean time in minutes.
98 Table 8 Effect of Gender on Performance on Shopping Task Me ans MANOVA Males Females F (1,19) p Pre-test Grocery items 16.33 20.06 1.53 .23 Grocery time 19.33 12.83 4.93 .04* Post-test Grocery items 16.33 21.06 2.28 .15 Grocery time 24.33 13.00 7.56 .01* Note. Necessary = just to necessary aisles method, ais le-by-aisle = aisle-by-aisle method, Combo = combination of previous two methods, Grocery items = total item score, and Grocery time = total shoppi ng time. The Grocery items is scored according to the procedure described in the text, and the Grocery t ime is the mean time in minutes. *p <.05