Hearing Loss (HL) and dementia are two of the most prevalent health concerns for the aging population (Kravitz et al., 2012; Olusanya et al., 2014; Rigters et al., 2018; Ogawa et al., 2019). Approximately two-thirds of adults in the United States over the age of 70 have HL and this number is expected to nearly double in the next four decades (Goman et al., 2017). Additionally, an estimated 5.8 million people in the United States and 10% of individuals 65 or older are impacted by Alzheimer’s disease, the most common type of dementia (Zhao, 2020). Mild cognitive impairment (MCI) is a distinct clinical term describing cognitive decline that can precede the formal diagnosis of dementia and is characterized by cognitive deficits not explained by typical aging (Eshkoor et al., 2015). These deficits include difficulties with memory, language, and problem-solving, without the disruption of daily living activities (Petersen, 2011). MCI, Alzheimer’s disease, and other types of dementia are commonly diagnosed by measuring performance on delayed recall tasks, among other cognitive markers (Klages et al., 2005; Dubois et al., 2014; García-Herranz et al., 2016).

Delayed recall is a complex skill involving multiple memory systems. Memory is believed to consist of three storage systems: sensory, short-term memory (STM), and long-term memory (LTM; Murdock, 1967; Atkinson and Shiffrin, 1968). Stimuli move through these systems via three sequential cognitive processes: encoding, consolidation, and retrieval (see Figure 1; Melton, 1963; Brown and Craik, 2000; Baddeley, 2002). Encoding refers to a mental representation or an external perceptual or sensory stimulus in the brain (Tromp et al., 2015). The stimulus is then consolidated when it is actively stored in STM, where, if it remains long enough, will be transferred into LTM, which is understood to have a capacity limited only by its ability to be accessed (i.e., retrieval; Tulving and Pearlstone, 1966). Due to the sequential nature of these processes, stimuli retrieved from LTM (e.g., delayed recall) must be encoded and consolidated first. Given the multiple cognitive processes involved, it is complicated to identify where breakdowns associated with poor delayed recall occur. For example, if stimuli are correctly recalled shortly after presentation, it can be posited that some level of encoding has occurred. Alternately, if the same stimuli are not recalled after a delay period, an impaired consolidation or retrieval mechanism is more likely to be at fault. As such, delayed recall measures are often thought to reflect retrieval abilities, whereas immediate recall tasks are meant to reflect encoding abilities (Delis et al., 2017).

Delayed recall tasks appear to present a greater challenge for individuals with HL compared to individuals without HL (Boxtel et al., 2000; Dupuis et al., 2015; Chandramouli et al., 2019). Specifically, performance on the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005), a test used to assess cognitive functioning and screen for MCI and dementia, demonstrated that individuals with HL struggled to recall delayed recall stimuli more frequently than individuals without HL (Dupuis et al., 2015). This effect is further supported by the association between poorer baseline hearing in both ears and greater declines in delayed verbal memory found over a 2-year period (Armstrong et al., 2020). In fact, Deal et al. (2015) demonstrated an association between HL and a greater longitudinal decline in delayed recall performance over a 20-year period. Recent work suggests that using MoCA alternate scores (i.e., a scoring method whereby specific auditory subtests are systematically removed; Dupuis et al., 2015) may have clinical significance for the hearing-impaired population (Al-Yawer et al., 2019). Our own research has demonstrated that individuals with HL, specifically those with cochlear implants (CIs; surgical implants effective for those with profound hearing loss in which other assistive hearing devices are not appropriate) performed better on the MoCA presented in both a visual and auditory format when delayed recall was removed (Hillyer et al., 2020; Parada et al., 2020). However, removing test items from the total score may also decrease sensitivity (Dupuis et al., 2015). Using the California Verbal Learning Test—Third Edition (CVLT-3; Delis et al., 2017), a neuropsychological assessment of verbal learning and delayed recall, Pisoni et al. (2018) demonstrated more retrieval-induced forgetting of stimuli in delayed recall tasks in experienced CI users compared to those without HL. Additionally, CI users benefited more from semantically cued words than individuals without HL suggesting that semantic cueing allowed individuals with HL to access words that were encoded but not accessible to non-cued retrieval (Chandramouli et al., 2019; Kronenberger and Pisoni, 2019). Taken together, the current literature suggest that an impaired retrieval mechanism may underlie delayed recall deficits in individuals with HL.

While the CVLT-3 offers a more comprehensive assessment of delayed recall than the MoCA (e.g., allows for the distinction of short and long delay free and cued recall), it does not provide distinct measures of individual memory processes (i.e., encoding, consolidation and retrieval). Where traditional metrics of the CVLT-3 (e.g., learning slope, recognition-hits) reflect an overlap between memory processes (Delis et al., 1991), the Item-Specific Deficit Approach (ISDA, Wright et al., 2009) was developed with the goal of providing more distinct indices of encoding, consolidation and retrieval. The ISDA is a scoring method that evaluates list-learning performance at the item level rather than by overall trial performance across immediate recall and subsequent delayed recall trials. For example, the CVLT-3 calculates scores as a summation of total words recalled within each trial whereas the ISDA takes into account the amount of times each word has been recalled across multiple trials. This item-level approach also aims to compensate for the effects of inattention, which may prevent a participant from initially encoding a target word for later recall. This scoring method may be similarly helpful for participants with HL who may not encode a target word due to mishearing or not hearing, thus affecting their overall performance.

The aim of this study was to further explore our previous findings from the MoCA where the largest change in passing rate was observed by removing the delayed recall subtest, suggesting that delayed recall is more challenging for this patient population (Parada et al., 2020). Given the potential clinical utility of the alternate MoCA scoring methods for people with HL, we considered original and alternate MoCA scores in relation to a more comprehensive delayed recall test: the CVLT-3. As such, we administered both the MoCA as well as the CVLT-3 to gain a clearer understanding of the underlying memory processes associated with delayed recall. Given that delayed recall is a complex cognitive process consisting of encoding, consolidation, and retrieval, we also used CVLT-3 scores to produce individual ISDA indices reflective of these processes. We predicted that higher delayed recall scores on the MoCA would relate to better performance on the CVLT-3. Given that the CVLT-3 and ISDA utilize the same raw scores, we also expected that respective measures of encoding and retrieval would relate to equivalent ISDA deficit indices.

Performance on the standard MoCA, including the MoCA delayed recall subtest, did not relate with performance on the immediate or delayed recall subtests of the CVLT-3, regardless of MoCA scoring method used. CVLT-3 subtest performance scores thought to reflect encoding abilities correlated with the ISDA encoding index, with the exception of learning slope. The ISDA retrieval index was not associated with any CVLT-3 measures; however, the alternate ISDA retrieval scoring method was related to the CVLT-3 long-delay free recall subtest. Consolidation abilities, which are only distinctly defined by the ISDA, were associated with CVLT-3 immediate recall trials 2 and 3 as well as both short and long delay cued recall subtests and recognition-hits, but not the delay free recall subtests.

The primary aim of this study was to further examine delayed recall performance in experienced, high-performing CI users to better understand the underlying memory processes characteristic of this group. The CVLT-3 was employed as a comprehensive test of delayed recall, in contrast with the MoCA, which includes only one delayed recall subtest. Incongruent with our predictions, performance on the MoCA delayed recall subtest did not relate to performance on any CVLT-3 subtests. While the CVLT-3 and MoCA are both widely used by clinicians, the lack of relationship between delayed recall performance as measured by the two tests highlights their differences when applied to CI users.

To better understand the lack of relationship between MoCA and CVLT-3 performance observed here, the differences between two tests need to be explored. Delayed recall consists of 45.5% of the overall score of the CVLT-3, whereas it only accounts for 16.6% of the overall score of the MoCA. Additionally, the delayed recall periods are different, with the CVLT-3 instructing for an approximately 20 min delay period between short-delay cued recall and long-delay free recall subtests, and the MoCA instructing for a 5 min delay period between initial immediate recall memory and delayed recall subtests (see Table 2). These differing delay periods have the potential to allow for different levels of consolidation and forgetting of stimuli based on the capacity of each memory system. This is because the assumed storage capability of STM is about 30 s with a capacity of about seven numeric digits (Miller, 1956; Craik and Lockhart, 1972). The longer delay period in the CVLT-3 may result in greater forgetting of target stimuli (i.e., the 16 words the participant is attempting to remember). Conversely, the word list in the CVLT-3 is read to the participant five times, allowing for further consolidation of the stimuli through repetition, while the MoCA word list is only read twice. Additionally, the CVLT-3 includes scoring for delay cued recall, whereas the 7.1-3 versions of the MoCA do not. The inclusion of delay cued recall, which utilizes semantic categories for recalling stimuli, allows for the exploration of other types of delayed recall that may use different cognitive processes from delay free recall. Taken together, differences in scoring methodology, delay periods, repetition and cueing complicate comparing delayed recall performance measured of the MoCA v. the CVLT-3.

Previous literature has examined cognitive performance in relation to hearing abilities for the MoCA (Dupuis et al., 2015; Ambert-Dahan et al., 2017; Lim and Loo, 2018; Hillyer et al., 2020; Parada et al., 2020; Shen et al., 2020; Utoomprurkporn et al., 2020) and CVLT-3 (Kramer et al., 2018; Moseley, 2018; Pisoni et al., 2018; Chandramouli et al., 2019), demonstrating that differences in cognitive abilities due to sensory impairments like HL should be taken into account during test administration. Our current findings present the first application of the ISDA to a population with HL. One potential strength of the ISDA is the ability to counteract the effects of inattention by measuring performance at the item level rather than by overall trial performance (Wiegner and Donders, 1999; Wright et al., 2009). This item-level approach may also offer the ability to account for the fact that not all 16 words are always properly encoded by tabulating scores based only on the words successfully recalled. The original CVLT-3 scoring method tabulates each score as a proportion of all 16 words, whether all 16 words were encoded at some point or not. In this study, our participants repeated 91% or approximately 15 of the 16 target words at least once across the five immediate recall trials, whereas they repeated 78% or approximately 13 out of the 16 target words at least once across all delayed recall subtests. These results support similar findings by Parada et al. (2020) suggesting that poorer delayed recall performance may be a result of an impaired consolidation and/or retrieval mechanism than an inability to properly hear and encode the words during the initial immediate recall trials. While the majority of words were properly encoded in our participant population, individuals with poorer speech discrimination could benefit from this item-level approach, as it may mitigate failures to encode test stimuli due not hearing or mishearing. More research is needed to address this.

In line with our secondary aim to relate ISDA indices with equivalent CVLT-3 subtests, we found that performance on the CVLT-3 immediate recall measures thought to reflect encoding abilities were related to the ISDA encoding index, whereas consolidation measures demonstrated a different pattern. CVLT-3 cued delay recall performance related to the ISDA consolidation index, but CVLT-3 free delay recall performance did not. These results may be driven by the ISDA scoring criteria for consolidation, which requires participants to recall a target word on at least one of the four delayed recall subtests in order to not receive a point toward their consolidation index score. In other words, the inability to recall the target word on any delayed recall subtests (i.e., score of 0 out of 4) would result in an increased consolidation deficit. Consequently, it stands to reason that the ISDA consolidation index may relate to performance on delayed recall subtests where stimuli were remembered most frequently, which in our study were the delay cued subtests. Previous research in individuals with normal hearing has demonstrated that the cognitive processes associated with cued and free recall are different (Nobel and Shiffrin, 2001; Brainerd et al., 2002; Nyberg et al., 2002; Padilla-Walker and Poole, 2002; Ivanoiu et al., 2005; Cerciello et al., 2017). Similarly, in CI users, a potential difference in the cognitive processes associated with cued recall could be that cued recall, much like a recognition task (e.g., recognizing target words among other distractor words), is both a measure of familiarity and recall (Bastin and Van der Linden, 2003). Studies involving recognition tasks have indeed suggested that familiarity and recall are processes occurring independently of one another in the brain (Aggleton and Brown, 2006). In the context of the CVLT-3, the categorical cues given during the delay cued recall subtests may tap into a pre-existing system of familiar words already existing in the participant’s memory. With regards to our study, an increase in performance was observed when semantic cues were provided (1 extra word on average recalled; range of 0–7 words). Participants who may have used the provided CVLT-3 semantic cues to recall more words demonstrated proper encoding; however, they may have had more difficulty with retrieval since the benefit of semantic cues has been shown to demonstrate a retrieval deficit (Farrer and Drozdick, 2020). Participants who did not demonstrate an increase in performance (and therefore did not benefit from provided CVLT-3 cues) may have already been using their own semantic strategies to recall words, such as constructing unique semantic categories or using rehearsal devices. In future administrations of the CVLT-3, asking participants whether they constructed their own semantic cues or not prior to providing the CVLT-3 semantic cues would offer additional insight.

It would be expected that CVLT-3 retrieval performance be related to the ISDA retrieval index; however, in our study, we found that ISDA retrieval deficit calculated with the original scoring method did not relate to performance on any CVLT-3 retrieval measures. Alternately, our less stringent scoring method to calculate the ISDA retrieval deficit did reveal this expected relationship. This alternate ISDA retrieval index was calculated with relaxed criteria; specifically, a participant could recall a word across three or four subtests and avoid increasing their overall retrieval deficit score. In other words, a participant’s retrieval deficit score was not increased if they failed to recall a word on one subtest, but rather, they were given a point toward their retrieval deficit score if they failed to recall a word for two or more subtests. This alternate index still required the participant to recall a target word on at least one short delay and long delay condition. Our data indicated that 8.33 words on average satisfied the original ISDA retrieval index criteria and were recalled across all four delayed recall subtests, whereas with the less restrictive retrieval index, 10.44 words were recalled on at least three or more delayed recall subtests. While the mean difference between the original and alternate retrieval indices is only two words, the less restrictive retrieval index significantly related with a CVLT-3 retrieval measure (long-delay free recall). Our results highlight that the alternate ISDA retrieval index may be beneficial in capturing an element of retrieval that is not part of the ISDA original score. Indeed, long-delay free recall is often considered to be a more pure measure of retrieval abilities based on its lack of interference from distractor words (i.e., list B; Ebert and Anderson, 2009; Farrer and Drozdick, 2020). To understand the relationship between our alternate retrieval measure and the CVLT-3 short and long delay delayed recall conditions, we calculated which of the four conditions a participant was most likely to forget a target word. We found that when a word was recalled three out of the four possible times that the condition where a word was most likely forgotten was long-delay free recall. Specifically, we determined that target words were forgotten 21, 11, 27, and 14 times across all participants on the CVLT-3 short-delay free, short-delay cued, long-delay free, and long-delay cued subtests, respectively. With the original ISDA scoring method, these forgotten words would count toward an overall retrieval deficit without accounting for the fact that the majority of these instances of forgetting occurred during the long-delay free recall subtest. Perhaps our less stringent ISDA retrieval scoring method provided more sensitivity to the presence (or absence) of a retrieval deficit and therefore revealed this expected significant relationship with long-delay free recall ability. However, additional work is needed to determine the clinical utility of this alternate score in a CI patient population as well as expanding this scoring method to a normal hearing population.

While CVLT-3 and ISDA measures did not relate with the MoCA, our work indicates that the ISDA can successfully be applied to CI users to quantify delayed recall ability. Specifically, the advantage of the ISDA is that it provides a discrete measure of consolidation, although our results also highlight that an alternate ISDA retrieval score may be needed. Our work, however, should be considered preliminary as additional work is needed to assess the clinical utility of the original and alternate ISDA scoring methods in both a normal hearing and more expansive CI patient population.

The datasets presented in this article are not readily available because the Auditory Research Laboratory is part of a hospital system that does not allow for data sharing due to patient privacy requirements. Requests to access the datasets should be directed to corresponding author.

The studies involving human participants were reviewed and approved by the Swedish Medical Center Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

NB, JP and AP-C completed the statistical analysis of data. NB created the figures and tables. NB, EE and AP-C wrote the manuscript. JP and JH contributed to editing the manuscript. All authors approved the submitted version, contributed to the study design, and participated in data collection.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.