Persistent Failure to Recover from Proactive Semantic Interference on the Cognitive Stress Test Differentiates Between Amnestic Mild Cognitive Impairment,Pre-Mild Cognitive Impairment,and Cognitively Unimpaired Older Adults

Abstract

Background:

Susceptibility to proactive semantic interference (PSI) and the inability to ameliorate these difficulties with one additional learning trial have repeatedly been implicated as early features of incipient Alzheimer’s disease (AD). Unfortunately, persistent failure to recover from PSI (frPSI) after repeated learning trials, are not captured by existing memory measures, or been examined in pre-mild cognitive impairment (PreMCI).

Objective:

A novel Cognitive Stress Test (CST) was employed to measure the impact of PSI, initial failure to recover from PSI and persistent effects of PSI, despite multiple learning trials of the new to-be-remembered material (pfrPSI). We hypothesized that PSI deficits on the CST would persist in both PreMCI and amnestic MCI (aMCI) groups over repeated learning trials when compared to cognitively unimpaired (CU) older adults.

Methods:

One hundred fifty older adults (69 CU, 31 PreMCI, and 50 aMCI) underwent a standardized clinical and neuropsychological evaluation. The CST was independent of diagnostic classification.

Results:

Even after adjusting for strength of initial learning, aMCI and PreMCI groups demonstrated greater persistent PSI (pfrPSI) relative to the CU group despite repeated learning trials of List B. Further, the aMCI group made a higher number of semantic intrusion errors relative to the PreMCI and CU groups on all List B Cued Recall trials.

Conclusion:

Persistent PSI appears to be a common feature of aMCI and PreMCI. The possible theoretical mechanisms and empirical implications of these new findings are discussed.

Keywords

Alzheimer’s disease mild cognitive impairment persistent semantic interference proactive semantic interference

INTRODUCTION

Older adults’ diminished ability to inhibit and ignore competing information was first described by Hasher and Zacks [1] and confirmed in subsequent studies [2 –4]. Increasingly, difficulties with inhibitory processes, self-monitoring, and specifically, proactive semantic interference (PSI) have also been characterized as early features of incipient Alzheimer’s disease (AD) [5 –8]. To directly test these hypotheses, both Curiel et al. [9] and Crocco et al. [10] first used a novel paradigm; the Loewenstein and Acevedo Scales for Semantic Interference and Learning (LASSI-L™) that required learning a list of 15 target items representing three semantic categories (fruits, musical instruments, and articles of clothing). On the LASSI-L, maximal learning over two trials was initially facilitated by category cues at both acquisition and during recall. PSI (old learning interfering with new learning) and the failure to recover from PSI (frPSI) were assessed by having the examinee attempt to learn 15 new targets on List B (representing the identical semantic categories used for List A) over two additional learning trials using the identical category cues employed in List A during both acquisition and retrieval.

A growing body of evidence from studies conducted in independent cohorts in the United States and other countries have found that performance deficits on the LASSI-L, particularly those related to initial PSI and frPSI, are superior to several traditionally used memory tests (e.g., list learning measures) in distinguishing between cognitively unimpaired (CU) older adults and those with preclinical AD or early and late-stage MCI [11] in both English and Spanish [12]. Further, other variables such as maximum learning of the initial List A targets, PSI, frPSI, and semantic intrusion errors on the LASSI-L have also been found to be sensitive to discriminate between the aforementioned groupings with AD biomarker confirmation (amyloid PET) [13 –16]; neurodegeneration measured by magnetic resonance imaging (MRI) [17, 18], functional MRI [19], and fluorodeoxyglucose (FDG) positron emission tomography (PET/CT) [20].

The consistent findings that frPSI (even when controlling for initial learning ability) is particularly sensitive to incipient AD raises some important questions. While Loewenstein and colleagues [21] have theorized that persistent failure to recover from proactive semantic interference (pfrPSI) would persist despite additional opportunities to learn two separate semantically competing semantic word list, there has been no means by which to test this hypothesis using the LASSI-L or other existing cognitive paradigms. To empirically test this prediction would necessitate additional opportunities to learn both List A and List B with additional learning trials. The hypothesis that PSI would be persistent despite multiple trials to attenuate its effects is based on the premise that the semantically competing nature of the to-be-learned stimuli challenges the cognitive system to “update” the information it learns efficiently, and to exclude recently learned information that is no longer immediately relevant, a cognitive process requiring the suppression of distracting traces mediated by executive-control processes that are analogous to those involved in overriding prepotent motor responses. This particular deficit has been consistently found among older adults at greatest risk for progression to AD [21, 22].

Since the LASSI-L paradigm was not sufficient to test (pfrPSI), this required us to employ a different paradigm to assess this potential hypothesis. We used a newer and more challenging paradigm; the Cognitive Stress Test (CST) developed for an NIH funded study. The CST requires the individual to learn 18 target words, all of which belong to one of three semantic categories: occupations, household items, and forms of transportation. Participants are instructed about the three semantic categories to be learned and reads each card at a time to ensure controlled learning and encoding of the to-be-remembered information. Identical category cues were provided during each of the three learning trials as well as during each of the three cued recall trials for each list. The introduction of a second list with 18 different targets that share the same semantic categories as the first list is presented in an identical manner and with the recall elicited by stimulates semantic interference. The CST directly assesses the participants’ ability to recover from persistent retroactive semantic interference over repeated trials, a process that has not carefully examined in aMCI and AD research. In a recent study with the CST [21], it was found that persistent PSI over repeated trials (Cued B3) [pfrPSI] distinguished between a modest number of older adults with aMCI from CU participants. While this was encouraging pilot data, only a limited number of aMCI were examined. Importantly, pfrPSI may not only be a feature of aMCI but might also apply to older adults with PreMCI (a state in-between aMCI and cognitively normal associated with increased risk for progression to a formal diagnosis of aMCI [22] and dementia [23]). The investigation of pfrPSI among the adequate samples of PreMCI and aMCI groups is now possible employing the unique CST paradigm.

In the present investigation, we hypothesized that pfrPSI would occur more frequently among individuals with aMCI and PreMCI regardless of the increased learning trials of the second list of targets that could potentially facilitate the recovery from these deficits. This study using the CST outlined in Fig. 1 represents a unique opportunity to determine whether pfrPSI is a salient marker of early cognitive impairment among older adults at increased risk for aMCI and AD.

Fig. 1

Cognitive Stress Test (CST) administration.

METHODS

Participants

A total of 150 individuals were recruited from an NIA-funded R01 participant pool of community-dwelling older adults receiving a comprehensive standardized clinical and neuropsychological diagnostic battery. This University of Miami Miller School of Medicine investigation is IRB approved. The average age of participants was 71.76 years (SD = 8.0) [range 60–94] (Table 1).

Table 1

Demographic characteristics of sample

	CU (n = 69)	PreMCI (n = 31)	aMCI (n = 50)	F or χ²	p
Age Range (60–94)	71.76	71.56	74.19	1.58	0.211
	SD = 8.0	SD = 7.4	SD = 8.2
Education Range (5–21)	15.93	14.97	14.32	3.39	0.036
	SD = 2.8	SD = 3.1	SD = 3.9
MMSE Range (22–30)	28.68^a	28.07^a	26.26^b	20.51	<0.001
	SD = 1.4	SD = 1.9	SD = 2.6
Sex (Female)	77.9%	45.2%	37.0%	21.6	<0.001
Primary Language (English)	80.9%	66.7%	63.8%	4.67	0.097

Means with different alphabetic superscripts are significantly different at p < 0.05 by the Tukey’s Honestly Significant Difference Test (HSD).

Experienced bilingual clinicians, blind to neuropsychological test results, administered a standard clinical assessment protocol, which included the Clinical Dementia Rating Scale (CDR) [24] and the Mini-Mental State Examination (MMSE) [25] to assess memory and other clinical and cognitive complaints. All participants were community-dwellers, independent in their activities of daily living, had knowledgeable collateral informants, and did not meet DSM-V criteria for Major Neurocognitive Disorder, an active Mood or Psychotic Disorder [26], or any other neuropsychiatric disorder. In cases where there was evidence of memory decline by history and/or clinical examination, the clinician scored the Global CDR as 0.5 and assigned a diagnosis of probable amnestic MCI (aMCI), pending the results of formal neuropsychological testing.

Subsequently, a fluent Spanish/English trained psychometrician blind to the CDR assessment administered a standard neuropsychological battery uniformly across groups independent of the clinical examination in the participants’ dominant and/or preferred language. Participants also received the computerized CST [21] which played no role in diagnostic determination to avoid circularity. The neuropsychological battery used to classify older adults into groups included the Hopkins Verbal Learning Test, Revised (HVLT-R) [28], delayed paragraph recall of the National Alzheimer’s Coordinating Center Uniform Data Set (NACC UDS) [29], Controlled Oral Word Association Test: Category Fluency [30], Block Design subtest of the Wechsler Adult Intelligence Scale, Fourth Edition (WAIS-IV) [31], and the Trail Making Test (Parts A and B) [32].

Amnestic MCI group (aMCI; n = 50)

On the basis of the independent clinical interview and performance on the neuropsychological tests, an individual was classified as aMCI if all the following criteria were met: a) subjective memory complaints by the participant and/or collateral informant; b) evidence by clinical evaluation or history of memory and/or other cognitive decline; c) Global Clinical Dementia Rating scale of 0.5; d) one or more memory measures fell below normal limits (i.e., a score below 1.5 SD or more relative to age, education, and language-adjusted normative data).

Cognitively unimpaired group (CU; n = 69)

On the basis of the independent clinical interview and performance on the neuropsychological tests, an individual was classified as CU if all the following criteria were met: a) no memory complaints by the participant and/or collateral informant; b) no evidence by clinical evaluation or history of memory and/or other cognitive decline; c) Global Clinical Dementia Rating scale of 0; d) all memory measures less than 1.0 SD below age, education, and language-adjusted normative data.

Pre-mild cognitive impairment group (PreMCI; n = 31)

This group of older adults could not be categorized as aMCI or CU based on the aforementioned criteria. Instead, this particular group met National Alzheimer’s Coordinating Criteria for what is deemed “impaired-not MCI” [29] which, for the purposes of this study was be categorized as PreMCI. These participants fell within two categories which have relatively equivalent progression rates over time to MCI [22]. The first group of individuals meet all clinical criteria for aMCI (CDR Global score = 0.5), but their neuropsychological evaluation was considered within normal limits (i.e., no score below 1.0 SD) (n = 16). The second group was considered clinically unimpaired (CDR global score = 0) but scored in the impaired range on one or more neuropsychological memory tasks (at or below 1.5 SD below expected levels) (n = 15). While PreMCI participants do not meet formal criteria for MCI, they are not CU either. Ample evidence demonstrates that individuals with both types of these PreMCI profiles are at much greater risk for progression to MCI and dementia [21 , 27]. In the current sample, there we no statistically significant differences between the two PreMCI groups with regards to demographic characteristics including education and CST scores.

Administration of the Cognitive Stress Test

We employed a novel computerized measure called The Cognitive Stress Test (CST) that expands upon our previous work with the widely studied Loewenstein-Acevedo Scale for Semantic Interference and Learning (LASSI-L) [21], including its computerized version the LASSI-BC, which has evidenced high test-retest reliability and discriminative validity [27]. The CST has also demonstrated reliability and validity in previous studies and employs the following: 1) semantic cueing at both acquisition and retrieval of 18 List A targets representing three semantic categories (occupations, household items, or types of transportation) over three initial learning trials, 2) three consecutive presentations of a second list of 18 new targets (List B) representing the same categories as the first list to examine PSI, frPSI, and persistent frPSI, and 3) use of category cues to elicit recall of List A targets to assess RSI, with an additional learning trial to examine failure to recover from retroactive semantic interference (frRSI). The CST as depicted in Fig. 1, represents a new approach examination of the role of persistent failure to recover from proactive interference which has not been possible with existing paradigms in preclinical and prodromal AD assessment. Further, the CST builds upon our previous work and is a fully computer-administered web-based task, which facilitates remote deliverability, reduces the need for a skilled psychometrist, and allows for automatic scoring of correct responses, semantic intrusion errors, and other errors. The CST is available in English and Spanish. While the CST is totally administered on the computer which uses powerful voice recognition software, to be conservative, we relied on our experienced psychometrist’s scoring of all patient responses to make this investigation consistent with previous work.

Statistical analyses

Statistical analyses were conducted using SPSS Version 26. First, age, sex, education, and language of testing and then global cognitive function were evaluated between diagnostic groups using one-way ANOVAs with Tukey’s Honestly significant difference test (HSD) for comparison of group means. Chi-square analyses with Yate’s Correction for Discontinuity were also employed. CST cued recall and intrusion errors were compared using ANOVA, then adjustments were made for demographic factors that were distributed differently between diagnostic groups using ANCOVA analyses following statistically significant F-values of p < 0.05, adjusted using the Bonferroni correction for multiple comparisons of means. To examine initial learning with regards to PSI, we calculated a PSI ratio (List B1 recall divided by List A1 recall), for the failure to recover from proactive interference, a frPSI ratio was also calculated (List B2 recall divided by List A2 recall). Finally, a persistent frPSI ratio was calculated (List B3 recall divided by List A3 recall). While all Cued A2 recall measures (maximum strength of acquisition) could have served as denominators for different Cue B2 recall trials, a tremendous strength of this paradigm was the ability to directly compare the relative responses of List A1, List A2 and List A3 of the CST against these identical learning trials on corresponding List B targets and to determine how repeated trials of Cued B recall was related to t semantic interference effects.

Since all subjects received all CST subscales and high intercorrelations were observed between both List A and List B repeated recall trials of the identical targets within subjects, performance on the PSI, frPSI and persistent frPSI ratios, were treated as a within subjects variable (cued recall of identical targets over multiple time points for each subject) while diagnostic group (CU versus aMCI) or (CU versus PreMCI) was treated as a between subjects variable. Both main effects as well as interaction terms were examined.

RESULTS

An inspection of Table 1 indicates that there were no statistically significant differences between aMCI, PreMCI, or CU diagnostic groups with regards to age [F(2,147) = 1.58; p = 0.211] or primary language of the evaluation (English versus Spanish) [χ^2 .(d = 2) = 4.67; p = 0.097]. There were group differences with regards to sex [χ^2 .(d = 2) = 21.6; p < 0.001] and MMSE scores [F(2,147) = 20.51; p < 0.001]. There was a predominance of females in the CU group, and as expected both CU and PreMCI groups had higher mean MMSE scores than the aMCI group.

As shown in Table 2, diagnostic groups were differentiated by both correct responses and intrusion errors on all the three initial Cued A and Cued B2 trials. (p < 0.001) and continued to be statistically significant (p < 0.001) after entering demographic factors such as education, sex, and MMSE score.

Table 2

CST scores among different diagnostic groups

	CU (n = 69)	PreMCI (n = 31)	aMCI (n = 50)	F	p	eta-squared
CST Cued A1 Recall	10.48^a (SD = 2.8)	8.88^b (SD = 2.6)	6.82^c (SD = 2.7)	25.50	<0.001	0.25
	Range (3–17)	Range (3–14)	Range (4–15)
CST Cued A2 Recall	14.07^a (SD = 2.2)	12.87^b (SD = 2.2)	9.70^c (SD = 2.7)	49.10	<0.001	0.41
	Range (8–18)	Range (7–17)	Range (5–15)
CST Cued A3 Recall	15.32^a (SD = 1.9)	14.23^a (SD = 1.8)	10.92^b (SD = 3.1)	52.43	<0.001	0.44
	Range (10–18)	Range (4–15)	Range (4–16)
CST Cued B1 Recall	8.16^a (SD = 3.5)	5.42^b (SD = 2.8)	3.84^c (SD = 2.2)	32.02	<0.001	0.31
	Range (0–18)	Range (1–13)	Range (0–9)
CST Cued B1 Intrusions	1.01^b (SD = 1.3)	1.74^b (SD = 2.0)	3.24^a (SD = 2.6)	18.68	<0.001	0.21
	Range (0–5)	Range (0–7)	Range (0–7)
CST Cued B2 Recall	12.24^a (SD = 3.2)	9.13^b (SD = 3.3)	6.37^c (SD = 2.8)	51.50	<0.001	0.43
	Range (3–18)	Range (3–18)	Range (1–13)
CST Cued B2 Intrusions	1.29^b (SD = 1.5)	2.13^b (SD = 1.5)	3.33^a (SD = 2.2)	16.14	<0.001	0.18
	Range (0–7)	Range (0–7)	Range (0–9)
CST Cued B3 Recall	13.91^a (SD = 2.8)	11.26^b (SD = 3.2)	7.98^c (SD = 3.4)	53.47	<0.001	0.45
	Range (5–18)	Range (4–18)	Range (1–15)
CST Cued B3 Intrusions	1.26^b (SD = 1.5)	1.84^b (SD = 1.5)	2.80^a (SD = 2.1)	13.08	<0.001	0.14
	Range (0–8)	Range (0–6)	Range (0–10)
CST Cued A4	9.22^a (SD = 3.5)	7.63^ab (SD = 3.8)	6.40^b (SD = 3.2)	9.60	<0.001	0.12
	Range (1–18)	Range (0–15)	Range (0–14)
CST Cued A4 Intrusions	2.56 (SD = 3.3)	2.77 (SD = 4.0)	3.44 (SD = 3.1)	.971	0.38	0.02
	Range (0–15)	Range (0–14)	Range (0–10)
CST CUED A5	12.19^a (SD = 3.0)	10.93^a (SD = 3.4)	8.79^b (SD = 2.6)	18.57	0.001	0.21
	Range (2–18)	Range (4–18)	Range (3–14)
CST CUED A5 Intrusions	2.43 (SD = 2.4)	2.47 (SD = 2.4)	3.33 (SD = 2.2)	2.36	0.098	0.04
	Range (0–10)	Range (0–8)	Range (0–8)

Means with different alphabetic superscripts are significantly different at p < 0.05 by the Tukey’s Honestly Significant Difference Test (HSD).

Post-hoc tests of unadjusted means statistically by the Tukey HSD indicated statistically significant difference between CU, PreMCI, and aMCI on tests susceptible to proactive semantic interference Cued B1 (PSI), Cued B2 (frPSI), and Cued B3 (pfPSI), showing a decrease in performance between these study groups on tasks sensitive to PSI as well as persistent proactive interference across trials. In Table 2, this pattern of results was also observed for Cued A1 and Cued A2 recall, although CU and amnestic amCI participants did not differ with regards to Cued A3 recall. Similarly, while F values were statistically significant for Cued A4 and Cued A5, (p < 0.001) post-hoc tests of means showed that only aMCI could be statistically distinguished from CU.

Examination of Table 2 indicated that Intrusion errors for tests susceptible to PSI, frPSI, and pfrPSI were higher for aMCI than the other diagnostic groups. These intrusion errors are thought to represent a failure to inhibit semantic inhibitory control on all Cued B trials.

No group differences were observed for intrusion errors on trials susceptible to retroactive interference or failure to recover from retroactive interference.

As depicted in Table 3, PSI, frPSI, and pfrPSI ratios were calculated by the formulas outlined in the Methods, and all demonstrated statistically significant effects at p < 0.01 after adjustment for covariates such as education and sex. Post-hoc Bonferroni measures indicated that both PreMCI and aMCI groups showed non-statistically significant PSI, frPSI, and persistent PfrPSI effects adjusted for initial learning but each of these diagnostic groups exhibited a higher percentage of intrusion errors relative to those exhibited by CU participants (p < 0.001).

Table 3

CST Cued B scores among different diagnostic groups adjusting for initial learning performance

	CU (n = 69)	PreMCI (n = 31)	aMCI (n = 50)	Adjusted F-Value	p	eta-squared
CST Cued B1 Recall adjusted for Cued A1 performance	7.48^a (SE = 0.35)	5.38^b (SE = 0.49)	4.79^b (SE = 0.42)	12.21	p < 0.001^*	0.43
CST Cued B2 Recall adjusted for Cued A2 performance	10.85^a (SE = 0.33)	8.72^b (SE = 0.44)	8.55^b (SE = 0.42)	11.24	p < 0.001^*	0.65
CST Cued B3 Recall adjusted for Cued A3 performance	12.74^a (SE = 0.34)	10.80^b (SE = 0.49)	10.03^b (SD = 3.1)	12.00	p < 0.001^*	0.63

^*Adjusted Means with different alphabetic superscripts are significantly different at p < 0.05 by the Bonferroni Procedure.

As presented in Table 4 and Fig. 2, we subsequently conducted direct comparisons and plotted these PSI, frPSI, and persistent frPSI effects that were obtained for each individual in aMCI versus CU groups using the same B targets adjusted for List A targets on a similar trial as described in the Methods. Education and sex were treated as covariates in the model. The F-Value for Diagnosis = 25.02 (p < 0.001) and the F-Value for the multiple recall trials = 8.70 (p < 0.001) demonstrated that aMCI participants in general had greater PSI effects across all List B learning trials. When MCI and CU participants were pooled together, interference effects generally improved across each List B learning trial. Particularly instructive is the lack of statistical significance of the Diagnosis and Trial interaction (F value = 0.31; p = 0.680) suggesting that although there was a difference in magnitude of PSI, frPSI, and persistent frPSI effects, these differences remained relatively constant with repeated administrations of List B for both aMCI and CU groups.

Table 4

Percentage of cued B targets recalled across trials relative to initial learning

	CU (n = 69)	aMCI (n = 50)	F -Value Diagnosis = 25.02 (p < 0.001)
			F-Value Trial = 8.70 (p < 0.001)
			F-Value Diagnosis×Trial = 0.31 (p = 0.680)
Cued B1 Targets Recalled as Percentage of Cued A1 Targets Recalled	0.814 (SD = 0.37)	0.609 (SD = 0.36)	NA
Cued B2 Targets Recalled as Percentage of Cued A2 Targets Recalled	0.865 (SD = 0.18)	0.655 (SD = 0.23)	NA
Cued B3 Targets Recalled as Percentage of Cued A3 Targets Recalled	0.917 (SD = 0.15)	0.748 (SD = 0.29)	NA

Fig. 2

Comparison of percentage of Cued B targets relative to Cued A cognitively unimpaired versus amnestic MCI groups over repeated CST trials.

Comparisons were also made between PreMCI and CU groups using a similar set of analyses using the same B targets adjusted for A targets on a similar trial as described in the Methods. The F-value for diagnosis was 11.27; p = 0.01 and the F-value for multiple trials were 8.57; p = 0.004. The F-value for the interaction term was 0.33; p = 0.51. This pattern of results was similar to the aMCI and CU group comparisons described above. Although Trials were initially treated as a repeated measure since it represented within subject performance on repeated learning trials with identical targets, we also conducted post-hoc analyses using a Diagnosis X Trials treating Trials as a non-repeated measure. Similar results were obtained when employing this approach and when education, sex and MMSE scores as covariates in the model.

DISCUSSION

The primary finding is in this investigation is the persistence of failure to recover from proactive interference (pfrPSI) in older adults at higher risk of AD. Using a novel cognitive stress test (CST paradigm), this study represents the first attempt to assess the effects of persistent PSI (pfrPSI) despite multiple learning trials to dissipate those effects in both aMCI and PreMCI groups. As clearly indicated in Table 2, all Cued B cued recall trials tapping PSI, frPSI, and pfrPSI evidence a monotonic increase in these semantic interference effects from cognitively normal, to PreMCI and finally aMCI groups. Further, after adjusting for performance on similar trials on List A in statistical models, cued recall deficits on Cued B1, Cued B2, and Cued B3 indicate that PSI, frPSI, and persistent frPSI (measured by the number of correct responses) continued to be observed more so among persons diagnosed with PreMCI and aMCI relative to CU controls. Post-hoc analyses revealed that when Cued B1, Cued B2, and Cued B3 were expressed as a percentage of Cued A3 recall similar effects were obtained. This supports the notion that difficulties with proactive interference continues with repeated opportunities to overcome pfrPSI effects. Not only does this finding represent the first data that older adults with aMCI and PreMCI have deficits related to their ability to recover from PSI on the CST, but it also shows that these PSI deficits persist despite repeated opportunities to learn (e.g., frPSI). Thus, this investigation is the first study to demonstrate pfrPSI effects in at-risk groups versus cognitively unimpaired aging adults. The findings suggest that not only PSI but persistent frPSI may be an early feature of those PreMCI and aMCI individuals who are at higher risk of eventual progression to clinical AD.

It is important to note that repeated measures mixed models Diagnostic Group (CU versus aMCI or CU versus PreMCI) across multiple recall trials of the same List B targets, failed to show a statistically significant interaction term, further supporting the hypothesis that persons with PreMCI and aMCI continued to exhibit difficulties in recovering from the effects of PSI over extended learning trials. We were also interested in intrusion errors on Cued B1, Cued B2, and Cued B3 trials since these types of errors have been consistently associated with amyloid pathology in those with incipient AD [13 –16] and are viewed as deficits in semantic inhibitory control. In the present study the aMCI group had a higher number of semantic intrusion errors than the PreMCI and CU groups on all List B cued recall trials.

Prior work with the LASSI-L [20] and LASSI-BC [27] have consistently shown that frPSI on one additional learning trial is a unique and common feature of MCI due to AD and has been associated with amyloid load [14 –16] and neurodegeneration [17, 18] in AD prone regions in a number of laboratories in the United States and abroad [20]. However, prior to the current investigation, it was unknown whether frPSI was a persistent cognitive deficit that occurred in either PreMCI or aMCI. The use of the CST paradigm [21] allowed for greater number of learning targets (n = 18 per list) with three initial learning trials for List A targets and three subsequent learning trials for List B targets. Since targets on List A and B belong to three shared semantic categories provided to the participant at both acquisition and retrieval, this provided a unique opportunity to examine the effects of PSI and the subsequent effects of frPSI and pfrPSI on subsequent learning trials.

As indicated in Fig. 2, persistent deficits were observed with aMCI in learning List B trials compared to the CU group. The data suggest that both groups benefit from additional Cued B2 learning trials, but that even with time and repeated learning attempts, performance on Cued B3 remained more than 25% lower than Cued A3 (maximum learning) for aMCI participants, while CU individuals greatly closed this gap during repeated opportunities to learn the new material. As previously noted, deficits in PSI, frPSI, and persistent frPSI are observed even after adjustments for initial learning strength, suggesting that these impairments are not solely a function of initial learning ability. Moreover, intrusion errors observed on PSI, frPSI, and persistent frPSI trials for those with aMCI confirm that this group had significant difficulties with inhibition, source memory and monitoring that are associated with different but semantically related word lists and these difficulties persist despite multiple learning trials for List B targets. The provision of semantic cues at both acquisition and retrieval with 18 targets ensures that ceiling or floor effects with the CST are not an issue. Controlled learning also minimizes the possibility of individualized learning strategies that may serve as compensatory mechanisms masking underlying deficits.

Taken together, the present findings indicate that PSI and pfrPSI deficits occur in PreMCI and aMCI older adults who are at higher risk for AD. The CST paradigm offers the unique ability to access the continuing effects of PSI and RSI over repeated trials, which is unique to this cognitive paradigm. Difficulties regarding susceptibility to PSI and frPSI and early difficulties with memory binding and accessing the semantic memory system are two key features of early AD that have been found to be predictive of disease progression [11 , 33].

The CST computerized paradigm may be particularly useful in deconstructing cognitive processes that in future studies can be compared to structural and functional MRI, amyloid and tau PET imaging, and emerging plasma biomarkers of AD using advanced technologies such as SiMoA or mass spectroscopy. Future research is required with larger samples with longitudinal follow-up over time. Nonetheless, the CST paradigm extends findings from the LASSI-L and the LASSI-BC in more thoroughly evaluating the effects of frPSI and persistent frRSI. The investigation of the persistence of PSI deficits has promise as a paradigm in future research to further elucidate specific types of memory breakdown in older adults at risk for AD and related neurodegenerative disorders. A comparison involving an equal number of learning trials for both List A and List B semantic categories that is unique to the CST, has promise in future research. Further directions include replicating these findings using a larger sample, including biomarker data to correlate with observed findings, and the continued longitudinal observation of those with PreMCI to estimate the rate of progression. The obtained findings also raises the question as to those brain mechanisms that not only are responsible for PSI, but importantly, the persistence of these deficits despite additional learning trials to mitigate these effects. Future studies can assist in more closely examining neural mechanisms underlying the obtained results and may provide invaluable insights into both AD and different aspects of brain function. This is certainly an area worthy of further research.

Footnotes

ACKNOWLEDGMENTS

This research was supported by the National Institute of Aging Grants R01 AG061106-04 (David Loewenstein, PI); 1R015R01AG055638-05 (Rosie Curiel Cid, PI). Funding was also provided by the Florida Department of Health Ed and Ethel Moore Grants awarded to Drs. Loewenstein and Curiel. This study was IRB approved and met all national and international standards for the protection of human subjects.

The CST measure was developed by and is intellectual property held by Drs. Loewenstein and Curiel at the University of Miami.

Authors’ disclosures available online ().

References

Hasher

, Zacks

(1988) Working memory, comprehension, and aging: A review and a new view. In Psychology of Learning and Motivation, Bower

, ed. Academic Press, pp. 193–225.

Amieva

, Phillips

, Della Sala

, Henry

(2004) Inhibitory functioning in Alzheimer’s disease, Brain 127, 949–964.

Collette

, Amieva

, Adam

, Hogge

, Van der Linden

, Fabrigoule

, Salmon

(2007) Comparison of inhibitory functioning in mild Alzheimer’s disease and frontotemporal dementia, Cortex 43, 866–874.

Clapp

, Gazzaley

(2012) Distinct mechanisms for the impact of distraction and interruption on working memory in aging, Neurobiol Aging 33, 134–148.

Belleville

, Bherer

, Lepage

, Chertkow

, Gauthier

(2008) Task switching capacities in persons with Alzheimer’s disease and mild cognitive impairment, Neuropsychologia 46, 2225–2233.

Loewenstein

, Acevedo

, Luis

, Crum

, Barker

, Duara

(2004) Semantic interference deficits and the detection of mild Alzheimer’s disease and mild cognitive impairment without dementia, J Int Neuropsychol Soc 10, 91–100.

Dewar

, Pesallaccia

, Cowan

, Provinciali

, Della Sala

(2012) Insights into spared memory capacity in amnestic MCI and Alzheimer’s disease via minimal interference, Brain Cogn 78, 189–199.

Aurtenetxe

, García-Pacios

, Del Río

, López

, Pineda-Pardo

, Marcos

, Delgado Losada

, López-Frutos

, Maestú

(2016) Interference impacts working memory in mild cognitive impairment, Front Neurosci 10, 443.

Curiel

(2013) A new scale for the evaluation of proactive and retroactive interference in mild cognitive impairment and early Alzheimer’s disease, J Aging Sci 2013, 1.

10.

Crocco

, Curiel

, Acevedo

, Czaja

, Loewenstein

(2014) An evaluation of deficits in semantic cueing and proactive and retroactive interference as early features of Alzheimer’s disease, Am J Geriatr Psychiatry 22, 889–897.

11.

Loewenstein

, Curiel

, Duara

, Buschke

(2018) Novel cognitive paradigms for the detection of memory impairment in preclinical Alzheimer’s disease, Assessment 25, 348–359.

12.

Curiel Cid

, Loewenstein

, Rosselli

, Matias-Guiu

, Piña

, Adjouadi

, Cabrerizo

, Bauer

, Chan

, DeKosky

, Golde

, Greig-Custo

, Lizarraga

, Peñate

, Duara

(2019) A cognitive stress test for prodromal Alzheimer’s disease: Multiethnic generalizability, Alzheimers Dement 11, 550–559.

13.

Kitaigorodsky

, Curiel Cid

, Crocco

, Gorman

, González-Jiménez

, Greig-Custo

, Barker

, Duara

, Loewenstein

(2021) Changes in LASSI-L performance over time among older adults with amnestic MCI and amyloid positivity: A preliminary study, J Psychiatr Res 143, 98–105.

14.

Cid

REC

, Curiel Cid

, Crocco

, Duara

, Garcia

, Rosselli

, DeKosky

, Smith

, Bauer

, Chirinos

, Adjouadi

, Barker

, Loewenstein

(2020) A novel method of evaluating semantic intrusion errors to distinguish between amyloid positive and negative groups on the Alzheimer’s disease continuum, J Psychiatr Res 124, 131–136.

15.

Loewenstein

, Curiel

, Greig

, Bauer

, Rosado

, Bowers

, Wicklund

, Crocco

, Pontecorvo

, Joshi

, Rodriguez

, Barker

, Hidalgo

, Duara

(2016) A novel cognitive stress test for the detection of preclinical Alzheimer disease: Discriminative properties and relation to amyloid load, Am J Geriatr Psychiatry 24, 804–813.

16.

Loewenstein

, Curiel

, DeKosky

, Bauer

, Rosselli

, Guinjoan

, Adjouadi

, Peñate

, Barker

, Goenaga

, Golde

, Greig-Custo

, Hanson

, Li

, Lizarraga

, Marsiske

, Duara

(2018) Utilizing semantic intrusions to identify amyloid positivity in mild cognitive impairment, Neurology 91, e976–e984.

17.

Loewenstein

, Curiel

, DeKosky

, Rosselli

, Bauer

, Grieg-Custo

, Penate

, Li

, Lizagarra

, Golde

, Adjouadi

, Duara

(2017) Recovery from proactive semantic interference and MRI volume: A replication and extension study, J Alzheimers Dis 59, 131–139.

18.

Loewenstein

, Curiel

, Wright

, Sun

, Alperin

, Crocco

, Czaja

, Raffo

, Penate

, Melo

, Capp

, Gamez

, Duara

(2017) Recovery from proactive semantic interference in mild cognitive impairment and normal aging: Relationship to atrophy in brain regions vulnerable to Alzheimer’s disease, J Alzheimers Dis 56, 1119–1126.

19.

Sánchez

, Abulafia

, Duarte-Abritta

, de Guevara

MSL

, Castro

, Drucaroff

, Sevlever

, Nemeroff

, Vigo

, Loewenstein

, Villarreal

, Guinjoan

(2017) Failure to recover from proactive semantic interference and abnormal limbic connectivity in asymptomatic, middle-aged offspring of patients with late-onset Alzheimer’s disease, J Alzheimers Dis 60, 1183–1193.

20.

Matias-Guiu

, Cabrera-Martín

, Curiel

, Valles-Salgado

, Rognoni

, Moreno-Ramos

, Carreras

, Loewenstein

, Matías-Guiu

(2018) Comparison between FCSRT and LASSI-L to detect early stage Alzheimer’s disease, J Alzheimers Dis 61, 103–111.

21.

Loewenstein

, Curiel Cid

, Kitaigorodsky

, Crocco

, Zheng

, Gorman

(2021) Amnestic mild cognitive impairment is characterized by the inability to recover from proactive semantic interference across multiple learning trials, J Prev Alzheimers Dis 8, 181–187.

22.

Loewenstein

, Greig

, Schinka

, Barker

, Shen

, Potter

, Raj

, Brooks

, Varon

, Schoenberg

, Banko

, Potter

, Duara

(2012) An investigation of PreMCI: Subtypes and longitudinal outcomes, Alzheimers Dement 8, 172–179.

23.

Crocco

, Cid

, Kitaigorodsky

, Grau

, Garcia

, Duara

, Barker

, Chirinos

, Rodriguez

, Loewenstein

(2021) Intrusion errors and progression of cognitive deficits in older adults with mild cognitive impairment and PreMCI states, Dement Geriatr Cogn Disord 50, 135–142.

24.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules, Neurology 43, 2412–2414.

25.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state.”, J Psychiatr Res 12, 189–198.

26.

American Psychiatric Association (2015) DSM-5 Classification, American Psychiatric Association.

27.

Curiel Cid

, Crocco

, Kitaigorodsky

, Beaufils

, Peña

, Grau

, Visser

, Loewenstein

(2021) A novel computerized cognitive stress test to detect mild cognitive impairment, J Prev Alzheimers Dis 8, 135–141.

28.

Benedict

RHB

, Schretlen

, Groninger

, Brandt

(1998) Hopkins Verbal Learning Test – Revised: Normative data and analysis of inter-form and test-retest reliability, Clin Neuropsychol 12, 43–55.

29.

Beekly

, Ramos

, Lee

, Deitrich

, Jacka

, Wu

, Hubbard

, Koepsell

, Morris

, Kukull

(2007) The National Alzheimer’s Coordinating Center (NACC) database: The uniform data set, Alzheimer Dis Assoc Disord 21, 249–258.

30.

Binetti

, Magni

, Cappa

, Padovani

, Bianchetti

, Trabucchi

(1995) Semantic memory in Alzheimer’s disease: An analysis of category fluency, J Clin Exp Neuropsychol 17, 82–89.

31.

Wechsler

(2012) Wechsler Adult Intelligence Scale–Fourth Edition. PsycTESTS Dataset.

32.

Reitan

(1958) Validity of the Trail Making Test as an indicator of organic brain damage, Percept Mot Skills 8, 271–276.

33.

Markova

, Mazancova

, Jester

, Cechova

, Matuskova

, Nikolai

, Nedelska

, Uller

, Andel

, Laczó

, Hort

, Vyhnalek

(2022) Memory Binding Test and its associations with hippocampal volume across the cognitive continuum preceding dementia, Assessment. doi: 10.1177/10731911211069676.