Comparing a visual and verbal semantic memory test on the effects of gender, age and education as assessed in a cognitively healthy sample

For years the number of people diagnosed with dementia has been steadily increasing in Austria. According to the Austrian Alzheimer’s Association the number of people suffering from dementia is around 100,000 with the projected numbers for 2050 being more than twice as many [1]. This is said to be due to the aging population in this country [2] as the incidence and prevalence of dementia increase with age [1]. This real increase in dementia patients is accompanied by an increase of the number of people fearful of developing dementia in part due to the heightened public awareness of the disease. A great number of people attending memory clinics present with perceived cognitive problems, despite unimpaired performance on cognitive tests which is termed subjective cognitive decline (SCD). Previous research has demonstrated, however, that SCD may indicate the first symptomatic manifestation of Alzheimer’s dementia (AD) [3].

Due to the incline in dementia patients and the increase of people interested in monitoring their cognitive state, the need for efficient and easy tools to help distinguish normal cognitive status from cognitive impairment in patients is more pressing than ever. Due to time and financial restrictions on medical professionals as well as restrictions on patients such as feasibility of travel to clinics and medical practices and a possibility for bias in testing due to an unfamiliar testing environment, a self-administered, home-based testing solution would be the most effective way to screen a large number of potential patients.

Over the past decades, computerized neurocognitive tests have gained popularity for assessing dementia and cognitive decline [4]. Overall, computerized tests are said to be superior to traditional paper–pencil testing in that they offer precise standardization of administration [4, 5]. They are also more time and cost efficient than paper–pencil tests and can be used in large-scale testing more easily [5]. With tablet-based tests, technological innovation has been further integrated into neuropsychological testing. Tablets may be advantageous to computers as touch screens have been argued to be more intuitive for older adults than computers using a mouse or a keyboard [6]. They also allow more mobility in testing and may improve engagement and therefore patient compliance [6]. One study evaluating the feasibility of the Computerized Cognitive Composite for Preclinical Alzheimer’s Disease (C3-PAD)—a home-based tablet test done by healthy elderly over the course of a week (multiple measurements)—demonstrated feasibility of this type of assessment, with authors reporting high reliability and validity of cognitive data recorded in a home environment as compared with in-clinic assessment [7].

The background of this current study is the development of a proprietary tablet-based neurocognitive battery—the International Neurocognitive Profile (INTP)—that should be useful to assess cognitive impairment remotely and to differentiate SCD and mild cognitive impairment (MCI) cases. While current cognitive assessments for the detection of preclinical dementia use composites of different cognitive domains specifically sensitive for AD pathology, this new battery should test all six neurocognitive domains as proposed in the Diagnostic and Statistical Manual of Mental Disorders V (i.e. language, learning and memory, executive function, complex attention, social cognition and perceptual–motor function) [8]. This should lead to a more detailed depiction of impairment/resources and could therefore provide information on preclinical levels of other dementia types aside from the common AD, which would set this battery apart from other batteries currently available [9, 10]. Another goal for battery development is to produce a repeatable measure useful for monitoring cognitive development over time, which is important for detecting and determining the progression of impairment. This could improve our understanding of the progression from the preclinical to the clinical stage of cognitive impairment. As a first step in the development of the INTP, subtests need to be newly developed and psychometrically tested. Thus, in this pilot study, two novel tests assessing verbal and visual semantic memory and implicit learning that were completed by a cognitively healthy sample were examined regarding the effects of participant characteristics such as gender, age and education.

Semantic memory is responsible for the individual’s ability to acquire and maintain general knowledge about the world [11]. It encapsulated all memory on facts, concepts, words, associations and meanings that were not encoded with episodic associations [12]. Semantic memory, unlike episodic memory does not show impairment in healthy aging [11‐13] but is impaired in dementia [14]. In fact, one study reported it to be one of the first functions to be impaired in AD patients with retrospective analyses determining a lower semantic memory task score in future AD developed than the reference group 12 years prior to diagnosis [15]. Semantic memory measures have been reported to differentiate between levels of cognitive impairment. Aside from countless studies, describing the significant differences found on measures of faces [16, 17], supportive data has been reported for measures using buildings and world capitals differentiating between healthy controls, MCI and AD [17]. In a healthy sample, semantic memory should be unaffected by age due to this function remaining relatively stable over the life span (as compared to episodic memory) [11, 13]. It should relate to education with more highly educated people reportedly performing better on semantic memory tasks [13, 18]. Concerning gender, previous research was divided with some papers stating an advantage for female participants [19], while others reported equality of performance between genders [11, 13] and one reported a male advantage in semantic memory measures [20].

Implicit learning is characterized by a lack of awareness during the acquisition of knowledge [21]. It is the counterpart of explicit or intentional learning, wherein the person is actively and consciously gathering knowledge. In terms of neuropsychological testing, implicit learning can be tested by not informing participants on later memory evaluation [22]. Likewise semantic memory reports regarding gender influences on implicit learning show inconsistent results with some researchers reporting a slight advantage of women over men [23]. Conversely, a recent paper suggested that gender does not show any effect on incidental memory [22]. Age has been shown to affect incidental learning, with previous research finding a clear advantage of younger versus older participants [22]. The impact of education on implicit learning has not been addressed in the scientific literature. As intelligence, which is positively correlated with education level [24], has been found to lack an association with implicit learning, it is plausible that education also does not show any association or effect on implicit learning.

Because a visual and verbal memory test were used to assess both semantic memory and incidental learning, this study aims to compare the results from these two tests. Due to the lack of comparative studies, this study is therefore using a more exploratory approach to compare these measures. Focussing on the effects of demographic variables on semantic memory and incidental learning, based upon previous research, this study aimed at testing the following hypotheses: (1) Gender differences may be seen in semantic memory and incidental learning. Due to inconsistent findings, no hypothesis on the existence or direction of effect of gender can be made. (2) Age is expected to impact incidental learning with younger participants performing better than older participants. Age is not expected to affect semantic memory. (3) Education is expected to affect semantic memory with more highly educated individuals outperforming less educated participants, but not incidental learning.

This study aimed at comparing two tablet-based tests of verbal (CK) and visual (FK) semantic memory on the impacts of descriptive variables gender, age and education. Although CK and FK were assumed to measure the same cognitive functions (semantic memory, incidental learning) this study found differences between these tests. Overall, CK seemed to have been more difficult than FK with participants showing lower performance scores in all rounds on CK compared to FK (Table 1). Using repeated measure ANOVA (analysis of variance) it was shown that CK was influenced significantly by gender and age. Both variables were assessed with post hoc tests and were found to affect the test in the hypothesized ways. As the same model using FK data failed to show significant effects of any variable this study cannot speak to the influences of gender or age on FK over and above correlational analysis results. Possible explanations for the lack of significance of the effects on the ANOVA could be an inappropriate sample due to the small size or participants’ general competence level on this test.

Gender was found to influence CK semantic knowledge, with men outperforming women. This finding was peculiar but not unexpected, as the scientific literature had been divided on the influence of gender on semantic memory [11, 13, 19]. It was interesting that women scored lower in both CK and FK semantic memory measures in this sample with additional correlational analysis revealing significant relationships of gender and first round performances on both tests. Understanding this male superiority was challenging—one possible explanation could have been a better knowledge level of men in the field of geography. In this study gender did not influence incidental learning measures in CK or FK. This could be interpreted as incidental learning being independent of gender.

Hypotheses regarding age were confirmed in this study. In CK, age group was found to significantly impact incidental learning measures but not the semantic memory measure. Due to previous findings on semantic memory remaining stable in healthy aging [11, 13], this finding was expected. Incidental learning was impacted by age group with younger participants ‘out-learning’ older participants. This finding was also consistent with the scientific literature [22]. Association with incidental learning measures but not with semantic memory measures were also found in correlative analysis using both CK and FK data. Overall, age did seem to impact semantic memory tests, which may be due to differences in learning abilities rather than semantic memory changes with age.

Education was only assessed using correlational analysis due to the broad range of education levels and the small sample size. In this study, education was found to relate to semantic memory measures of CK and FK which concurred with the posed hypothesis. It was also found to have no relationship with incidental learning on CK but did have a relationship with the second-round performance in FK. However, because FK data were skewed and the analysis of variance did not result in any significant effects, this finding should be treated with caution. Overall education was shown to relate to semantic memory and, in most computations, did not relate to incidental learning. This result is suggestive of education level affecting semantic memory tests.

Findings from this study need to be considered when creating normative data for CK and FK as norms must be adapted for gender, age and education. Overall CK and FK resulted in different performance data with this healthy sample performing better on FK than CK in all three rounds, which may hint at that test being too easy. FK may therefore be the preferable test to CK for a clinical population, as it may better distinguish the cognitively fit from the cognitively impaired. However, a recent study, advocated the use of both verbal and visual tests of semantic memory in the assessment of (amnestic) cognitive impairment [28]. The authors stated that in using only verbal tests, which did find the majority (60%) of amnestic MCI patients, 27% of patients who showed only impairment in visual tests were overlooked. This was especially problematic, since patients showing impairments on visual memory tests were also more likely to have multidomain aMCI and were had a higher rate of progression to AD [28]. Showing differing results between CK and FK in a healthy sample may justify the inclusion of both tests in the future tablet- based test.

In conclusion, this study showed differential outcomes of two novel tests of semantic memory in a healthy sample of participants, therefore hinting at a benefit of keeping both tests in future INTP versions. It also evidenced that the demographic variables of gender, age and education affected participant performance on these tests. It further demonstrated that effects were due to variables affecting either semantic memory storage or incidental learning. Overall, all three variables were shown to affect CK and FK and should therefore be considered in the creation of normative data for these tests.