Associative-memory Deficit As A Function Of Age And Stimuli Serial Position Part 2

Statistical analysis

All data were analyzed using STATISTICA software (version 12), an advanced analytics software package originally developed by StatSoft and currently maintained by TIBCO Software Inc. 

There is a close relationship between data and memory. Data can help us enhance our memory and enable us to better maintain clear ideas and minds. For many people, mastering large amounts of data requires a strong memory. This data may include rules, formulas, definitions, dates, facts, and names of famous people.

Learning data requires a lot of energy and a certain level of memory. Data is easily forgotten, so we need to make certain connections and memorize them when learning. For example, data can be grouped into topics, and an appropriate schedule developed for learning. We can also deepen impressions by combining data with personal experiences.

In addition to helping us learn and remember data, data can also provide us with broader knowledge and help us expand our horizons. By observing and studying data, we can understand how the world works and predict future developments.

Given the positive impact of data on our memory, we should actively use data to help improve our memory and cognitive abilities. The process of learning data may be difficult, but long-term study and practice can promote our memory ability and expand our ideas in daily life. It can be seen that we need to improve memory, and Cistanche deserticola can significantly improve memory because Cistanche deserticola is a traditional Chinese medicinal material that has many unique effects, one of which is to improve memory. The efficacy of minced meat comes from the various active ingredients it contains, including acid, polysaccharides, flavonoids, etc. These ingredients can promote brain health in various ways.

Click Know to improve short-term memory

Between-group comparisons (young versus older adults) were conducted with Students' T-test and the hypothesized interactions between the independent variables that were used in this study were calculated with F-tests.

Results

Table 1 shows the demographic characteristics of the 22 young and 22 older adult participants that were included in the analysis (including statistical analysis). The groups significantly differed in age but did not differ in years of education and no differences in gender distribution were found between groups. To assess memory accuracy, we computed the "hit minus false-alarm" rate for each participant in each experimental condition. 

A hit occurs when the participant correctly identifies a target test item as a target and a false alarm (FA) occurs when a distracter is erroneously identified by the participant as a target. With this measure, chance level performance (guessing) yields a score of 0.00, and perfect performance yields a score of 1.00. This equated the item and the associative recognition tests concerning the scale used. 

The ADI was calculated as the difference between item and associative recognition (item recognition performance minus associative recognition performance).

To specifically address the role of SSP (beginning/middle/end of the learning list) on memory recognition for items versus associations in younger and older adult participants, we computed a three-way mix design ANOVA with the above factors. 

Reaction-time (RT) means and standard deviations for the item and associative recognition as a function of age and SSP are available in Table 2. The results of the three-way mix design ANOVA (see Table 3 and Fig 2) indicated three main effects. A significant main effect for age [F(1, 42) = 39.93, p<.001, η2 p = .49], with lower overall memory performance in the older group [M = 0.29±0.05SD] compared to the younger group [M = 0.42±0.07SD]. 

A significant main effect for test [F(1, 42) = 136.50, p<.001, η2 p = .76], with higher memory recognition for items [M = 0.45±0.09SD] than for associations [M = 0.26±0.14SD]. A significant main effect for SSP [F (2, 84) = 93.33, p<.001, η2 p = .68], with planned-comparison analysis showing that memory accuracy for stimuli located at the end of the learning list [M = 0.54±0.14SD] was higher than for stimuli located at the beginning of the learning list [M = 0.32±0.16SD] and middle of the learning list [M = 0.21 ±0.08SD] [F(1, 42) = 147.31, p<.001, η2 p = .77]. 

In addition, memory for stimuli located at the beginning of the learning list was significantly higher than for stimuli located at the middle of the learning list [F(1, 42) = 25.17, p<.001, η2 p = .37].

We observed a significant two-way interaction between SSP and age [F(2, 84) = 6.43, p<.001, η2 p = .13], showing a robust memory performance difference between age groups (younger adults>older adults) for stimuli located at the beginning [F(1, 42) = 17.21, p<.001, η2 p = .29] and end of the learning list [F(1, 42) = 26.50, p<.001, η2 p = .38] but not for stimuli located at intermediate positions [F(1, 42) = 1.09, p = .30, η2 p = .02]. The second two-way significant interaction was observed between the test and age [F(1, 42) = 24.58, p<.001, η2 p = .37]. 

Planned-comparison analysis showed that older and younger adult participants did not significantly differ in memory recognition for items [F(1, 42) = 3.33, p = .07, η2 p = .13] while a sharp group difference in performance was evident for associative recognition [F(1, 42) = 58.84, p<.001, η2 p = .58]. The third significant two-way interaction was found between the test and SSP [F(2, 84) = 8.40, p<.001, η2 p = .16]. 

Planned-comparison analysis showed a significant difference between item and associative recognition at early list positions [F(1, 42) = 23.27, p<.001, η2 p = .35], middle list positions [F(1, 42) = 20.35, p<.001, η2 p = .32], and later/end list positions [F(1, 42) = 92.42, p<.001, η2 p = .68]. Despite significant differences in all three comparisons, the effect size of the later end of the list positions was the largest (.68, compared to the two other conditions: early; .35 and middle; .32 list positions). This pattern replicates our previous findings [55] and enables us to generalize the reported results of greater associativedecline for stimuli located at the end of the learning list to older adult participants.

Although the two-way interaction between test and age was significant under all serial positions (primacy, F(1, 42) = 4.81, p = .033, η2 p = .10; middle, F(1, 42) = 5.31, p = .026, η2 p = .11; and recency, F(1, 42) = 12.75, p = .000, η2 p = .23), the differences were additive and the threeway interaction was not significant [F<1]. To specifically assess our hypothesis regarding greater ADI scores under the recency portion in older adults, we also computed a two-way mixed design ANOVA with SSP (beginning/middle/end of the learning list, a within-subjects factor) X age (young/old, a between-subjects factor) as independent variables and ADI as the dependent variable. This interaction was not significant [F<1]. Together, these statistical analyses emphasize that the associative deficit is significantly augmented under the recency portion, but with a similar outcome for both young and old adults.

It is worth noting that higher ADI scores can be evident due to a decrease in hit rates or increase in FA responses or both. To assess the locus of the associative deficit we computed a four-way mix design ANOVA with the following factors: test (item versus associative recognition; a within-subject variable), X age (younger versus older adults; a between-subject variable), X SSP (beginning/middle/end of the list, a within-subject variable) and X response (Hits versus FAs; a within-subject variable). The four-way interaction did not reach significance (F<1). 

The results reported here are for the highest interactions that include the test factor and that reached significance; the three-way interaction between test X age X response [F(1, 42) = 24.57, p<.001, η2 p = .37]. Planned-comparison analysis on this interaction showed that while the two-way simple interaction between age and response in the item condition was not significant [F(1, 42) = 3.33, p = .07, η2 p = .07], the two-way simple interaction between the same factors was significant under the associative condition [F(1, 42) = 58.84, p<.001, η2 p = .58]. 

Further analysis of the later interaction revealed that while no difference between younger and older adults was evident for Hits responses [F < 1], significantly higher FA rates were evident for older adult participants under the associative recognition condition [F(1, 42) = 10.30,p<.001, η2 p = .19], see Fig 3. Another three-way interaction that was found to be significant was between test X SSP and X response [F(2, 84) = 8.40, p<.001, η2 p = .17]. 

Planned-comparison analysis showed that the well-established interaction between test (item versus association) and response (Hits versus FA) was most robust for the recency position [F(1, 42) = 92.42, p<.001, η2 p = .68] compared to primacy and middle positions [F(1, 42) = 23.25, p<.001, η2 p = .35; F(1, 42) = 20.35, p<.001, η2 p = .32; respectively].

Discussion

The objective of the current study was to test the separate and joint effect of both SSP and aging on memory recognition for items versus associations. The results of the current experiment replicate our previous results that showed greater associative memory decline (measured as the ADI score; i.e., reflecting the difference between item and associative recognition performance, regardless of age) for stimuli presented at the end of the learning list [55] and extend them to older adult participants. 

The results of the previous and the current study together provide behavioral support for the existence of an associative-binding deficit which is predominantly evident for stimuli with later list positions (i.e., recency positions). However, we did not find evidence for an augmented associative deficit at recency positions in older age which would have supported our hypothesis. 

Furthermore, in a planned analysis conducted on the Hit versus FA rates, we were able to focus on the locus of the associative deficit in older adults to significantly increase FA rates under the associative condition. These results support the conclusion that generally, older adults were more likely to respond that a recombined pair (namely, a distracter) had appeared in the learning list, especially if the tested material was located at the end of the learning list.

The dual-process theory asserts that single and paired (i.e., associated) units of information are differently treated in memory [57]. The theory claims that two independent contributing processes underlying memory: are familiarity and recollection. Familiarity is the recognition of information in the absence of any specific details. It is considered relatively automatic and is often defined as a sense of "being familiar" with the stimuli to be recognized. 

Recollection refers to the memory of events that is accompanied by specific details and associations. It is considered to involve executive functioning and is associated with an obvious sense of remembering. The difference between these processes is often tested behaviorally by using item-association memory paradigms, suggesting that while item recognition is linked to both familiarity and recollection processes, associative recognition relies solely on recollection [10].

Our results are in line with the dual-process theory [57] and the view that while familiarity is sufficient for the correct recognition of items, correct recognition of associations requires recollection. To correctly recollect associative information, it is not enough to be familiar with its components (i.e., items), rather enough details must be retained to determine whether those items were presented together during learning. 

Aging appears to disrupt recollection, but not familiarity processes, hence, we attribute older adults' tendency to mistakenly judge recombined pairs as having appeared in the learning list, to the deterioration in cognitive abilities occurring in older age. 

Consistent with previous literature [10] we postulate that with age, learning becomes more of a familiarity-based process rather than a recollection-based process because relying on familiarity is not enough for correct recognition of associations. This claim is exhibited by the pronounced associative (i.e., recollection-based) recency deficit found in the current study.

While most studies have investigated the serial position effect using a free recall paradigm, the current study benefits from three exceptional advantages of using a recognition-based memory paradigm. Firstly, it deals with the lack of experimental control in the free recall paradigm (in the free recall paradigm the investigator cannot control the retrieving order of the subject). Secondly, using a recognition-based memory paradigm allows the ability to compare the associative performance to the single items that form it. 

That is, the ADI score enables the comparison between item and associative recognition, while in the free recall task, this comparison is not possible. Thirdly, stimuli at the test phase were not presented randomly on the screen but rather according to their specific location in the learning list (beginning/middle/ end of the list). This procedure enabled us to control primacy and recency effects. 

Each learning list was followed by only one short test according to the 6 experimental conditions (3X2); although this controlled design is not parsimonious (as it requires a large number of tests) it does, however, decrease the sequential effect (i.e., the effect of the previous stimulus on the upcoming response for the current stimulus) and the preparation of the participant to a specific test.

Studies of the serial position curve assert that in the case of the recency effect, stimuli at late serial positions benefit from higher retrieval probability since they are still "stored" in WM and thus can be effortlessly retrieved [30, 46, 50–52]. In the current study, we were interested in the involvement of WM in the (specific) associative decline in older age. 

We tested the separate and joint impact of SSP and aging on memory for single versus paired units of information. Our results show that the recency effect does not fully explain associative memory recognition performance for material presented at later list positions. Despite the overall benefit of information located at the end of a learning list (compared to the rest of the list), for both younger and older adults the associative deficit was largest for end-of-list information. The results of the current study clearly show that single and paired stimuli (i.e., associative material) cannot benefit the same from higher retrieval probability due to their presence in WM. 

Recognition for associative material decreased dramatically for stimuli from later list positions (compared to memory recognition for single items at similar positions). These findings raise the possibility of the existence of binding deficits already at WM phases as bound stimuli are still maintained in WM during immediate recognition testing.

Evidence regarding an age-related associative-memory deficit in short-term/WM could reveal the locus of the deficit, as the existence of the deficit already in short-term/WM phases, would support an age-related variance in associative encoding abilities. However, the absence of a deficit in short-term/WM would support a long-term age-related associative deficit that could be a result of age-related differences in consolidation and forgetting rates of associations over time [25]. Our results do not fall under one of these options solely. Our findings do point to a larger associative deficit for associative material presented at later list positions (i.e., stimuli that are still maintained in WM and lack a trace in LTM), but additively for both young and old participants. These results support general variance in associative encoding abilities, independent of age.

Several limitations of the current study must be acknowledged. Firstly, in the current study, no formal-objective cognitive assessment was used, thus we cannot rule out the confounding of an undiagnosed MCI as a contributor to the memory deficits shown in the older adults group. Further studies are encouraged to use a formal-objective cognitive assessment to ensure the general cognition of the participants and to avoid potential confounds arising from general cognitive decline. 

Secondly, although the current study provides behavioral support to well-known aging and memory processes that are largely considered as relying on the brain structures mentioned it did not include imaging. Further neuroimaging studies are warranted to test the presence of associative-binding deficits in both young and old adults and to provide empirical neuro-anatomical and functional connectivity evidence to support or contradict the behavioral findings at the brain level. 

Lastly, in the current study, we relied on the widely accepted historical view that supports the division and interpretation of the serial position curve as beginning/middle of list stimuli as corresponding to LTM and end of list stimuli as corresponding to short-term/WM [30–36, 58]. Notwithstanding, other views also exist [59–62], thus our interpretation of the results must be taken with caution.

Whereas the study of aging has traditionally been carried out by searching for specific brain regions that are susceptible to neural degeneration or a decrease in brain volume, it is crucial to study their interactions with diverse cognitive paradigms, allowing us to identify not only the presence of a deficit but also its locus and core characteristics. 

Deep behavioral investigation of the core features of associative binding can shed light on the precise nature of, for example, encoding and retrieval deficits, and thus set the ground for neuroimaging studies to test the reported deficits at the brain level. As seen in the current study, we were able to emphasize the locus of associative-memory demands to increased FA rates already at WM stages. Since cognitive dysfunctions are a hallmark of older age, such understanding can serve as the basis for the detection of potential biomarkers for diagnosis and prognosis of aging individuals, as well as the development of standard behavioral and functional-imaging protocols and tools aimed to adequately assess these (lacking) abilities in older age.

References

1. Erickson CA, Barnes CA. The neurobiology of memory changes in normal aging. Experimental gerontology. 2003; 38(1–2):61–9.

2. Craik FIM, Byrd M. Aging and cognitive deficits: The role of attentional resources. In: Craik FIM, Trehub SE, editors. Aging and Cognitive Processes: Plenum Press, New York; 1982. p. 191–211.

3. Light LL. Memory and aging: Four hypotheses in search of data. Annual Review of Psychology. 1991; 42:333–76.

4. Salthouse TA. The processing-speed theory of adult age differences in cognition. Psychological review. 1996; 103(3):403–28.

5. Hasher L, Zacks RT. Working memory, comprehension, and aging: A review and a new view. In: Bower GH, editor. The psychology of learning and motivation. 22: San Diego, CA: Academic Press; 1988. p. 193–225.

6. Nyberg L, Maitland SB, Ronnlund M, Backman L, Dixon RA, Wahlin A, et al. Selective adult age differences in an age-invariant multifactor model of declarative memory. Psychology and aging. 2003; 18 (1):149–60.

7. Ronnlund M, Nyberg L, Backman L, Nilsson LG. Stability, growth, and decline in adult life span development of declarative memory: cross-sectional and longitudinal data from a population-based study. Psychology and aging. 2005; 20(1):3–18.

8. James T, Rajah MN, Duarte A. Multielement Episodic Encoding in Young and Older Adults. Journal of cognitive neuroscience. 2019; 31(6):837–54.

For more information:1950477648nn@gmail.com