Rodríguez-Rojo IC; Lugo-Marín J; Moreno-Martínez FJ

Review Article

Austin J Clin Neurol 2015;2(7): 1058.

Category Specificity, Alzheimer Disease and Normative Studies: A Review and Several Recent Instruments for Spanish Speakers

Rodríguez-Rojo IC¹, Lugo-Marín J^2,3 and Moreno-Martínez FJ³*

¹Department of Basic Psychology II, Complutense University of Madrid, Spain

²Hospital Universitario Nuestra Señora de la Candelaria, Spain

³Department of Basic Psychology I, U.N.E.D. Madrid, Spain

*Corresponding author: F Javier Moreno-Martínez, Department of Basic Psychology I, U.N.E.D, C/Juan del Rosal, nO 10, 28040-Madrid, Spain

Received: April 06, 2015; Accepted: June 25, 2015; Published: June 30, 2015

Abstract

The study of category specific dissociations has enabled to postulate semantic knowledge has an internal structure that could depend on different neural substrates. Several studies have found that category/domain effects (i.e., a relative impairment of one semantic category/domain respect to the other) are present in patients with Alzheimer Disease. However, there is still some controversy about which semantic domain (living or nonliving things) is mainly affected or not by this selective damage. Some of these inconsistencies could be due to different methodological issues. Throughout this work some of them -such as the lack of control on nuisance variables or the consequences derived from ceiling effects- will be described. Our goal is to highlight the importance of conducting an adequate methodological control, in order to develop suitable assessment tools. Furthermore, we present different normative studies in Spanish that suitability face several methodological problems. It is our intention these works can be useful instruments for those interested in the study of the semantic processing and category specific deficits in Spanish language.

Keywords: Alzheimer’s disease; Category effects; Living-Nonliving things; Normative studies

Introduction

Alzheimer´s disease (AD) is a progressive and degenerative disorder which affects memory, language, motor control, as well as executive functions [1]. At the neuroanatomical level, AD is characterized by the presence of tau neurofibrillary tangles and amyloid plaques in the central nervous system [2,3]. The strongest risk factor for AD is aging; it is estimated that by 2050, about 106 million people around the world could be affected by this disease [4].

One extensively studied issue in AD pathology is the disruption of long term memory systems. The seminal work by Tulving [5] established two types of (explicit) long term memory: episodic and semantic. Semantic memory (SM) is considered as “an organized body of knowledge involving words, concepts, their meanings, their associations, and the rules for manipulating these symbols and concepts” [6]. Although it has been long recognized that AD affects both memory subsystems [7], the present work will be focused on the semantic processing.

From a neuropsychological view, the study of the structure, organization and ulterior- impairment of SM has been guided by the category specific phenomenon. This has been discussed in terms of a functional or anatomical distinction between knowledge of living things (LT; e.g., fruits, vegetables or animals) and non-living things (NLT; e.g., vehicles, tools or buildings); with a relative impairment of the LT domain as the most commonly described profile in the literature [8,9]. The study of category specificity has a great theoretical significance, since it suggests that semantic knowledge has an internal structure, and, additionally, indicates that its components may depend on different neural substrates [10].

Category specific deficits have been mostly reported in focal pathologies, such as traumatic head injury [11] or herpes simplex encephalitis [12]. In addition, they have been also described in people suffering diffuse brain damage [e.g., in patients with AD; [13]]. However, the presence of categorical impairment in AD remains unclear and several studies are inconsistent regarding whether these deficits are present or not in these patients [14].Thus, while some researches stated that AD patients show a selective deficit in the processing of LT items [15,16], others have described impairments of both domains [17]; additionally, several works have found no category effects in AD patients [7,18].

Several of these inconsistencies could be attributed to different causes: the type of task administered the inclusion or exclusion of problematic categories or the image format of the items utilized. Ulterior methodological problems, such as a lack of control on nuisance variables (NV), or the presence of ceiling effects in healthy participants, can be also mentioned as potential causes of disagreement. A description of these points is hereby provided.

Nuisance variables

It has been demonstrated that NV play an important role in predicting naming accuracy and on the living-nonliving thing dissociations [19]. Age of acquisition [AoA; [20]], familiarity [11], manipulability [21], name agreement [22], typicality [23], visual complexity [24], and word frequency of the items [25], are some of the NV most commonly studied. As a rule, LT items used to be harder to process than NLT items because they present lower AoA, manipulability, name agreement, typicality and word frequency; on the contrary, LT items used to have higher visual complexity. Since the ratio of LT to NLT disorders has been estimated at 5:1 [9,26], the higher prevalence of deficits in LT could be at least partially explained by a lack (or no appropriated) control of NV [27].

Thus, several studies have found that the disproportionate deficit in the processing of LT disappeared once the items were carefully controlled in NV across domains [18]. Nevertheless, it must be considered that others authors have continued reporting LT impairments after a careful control of NV [28]. In addition, recent works with AD patients have highlighted NV are better predictors of longitudinal deterioration than the semantic domain of the items [i.e., LT/NLT; [29]]; this is true both in picture naming [30] and in verbal fluency tasks as well [19].

Ceiling effects

The occurrence of ceiling effects in healthy participants is closely related to the fact their level of performance is close to 100% of accuracy in some tasks. Accordingly, is not unusual that data gathered from AD patients have been compared with those from controls performing at or near to ceiling [31]. This could have produced spurious results i.e., masking the presence of category effects in controls especially in studies on category specificity [32].

In an attempt to control for ceiling effects and additionally, to optimize the reliability of the results, several authors have increased the difficulty of the tasks; for example, by using low frequency items [33,34]. In others studies, more semantic categories than the six (more) commonly used to evaluate patients (i.e., animals, fruits and vegetables for LT and tools vehicles and furniture, for NLT) have been also considered [35]. In recent studies, not only the number of categories, but also the number of items included in each one has been progressively increased to optimize the reliability of the results [36,37].

Atypical categories, differences in manipulability and “true” semantic domains

As aforementioned, the vast majority of the studies have used the categories of animals, fruits and vegetables for evaluating the LT domain, and tools, vehicles and furniture for assessing the NLT domain [8,38]. However, there are two categories involved in theoretical controversy: body parts and musical instruments [8]. It is reasonable to consider that body parts belong to the LT domain; likewise, apparently musical instruments belong to the NLT domain. Nevertheless, in terms of damage, different studies have observed that both categories fall within the opposite domain [8]. That is, body parts used to fall within NLT domain and musical instruments within LT domain. This controversy has done different studies have excluded these two categories from their observations [39].

Other important topic into the category specific arena has been remarked by Caramazza and co-workers [27,40]. This group of authors has proposed the existence of three “true” semantic domains: Animals, Plant life and Tools. According to this, evolutionary pressures would have led to domain-specific organization of conceptual system. Thus, a semantic structure with plants (source of food and medicine), animals (potential predators and additional sources of food) and tools (manufactured according to different functional purposes) are represented in separate (potentially independent) systems. Accordingly, the only pure category/domain specific impairment will involve these three domains; although, it is possible body parts can also own a separate domain [40].

Additionally, it has been observed dissociations between small manipulable objects (e.g., tools), and large outdoor objects [e.g., buildings; [41]]. Thus, clear differences in terms of grasping can be observed between, e.g., a saw and a house. Indeed, it has been reported that the manipulability of one object is a variable which impacts on its identification [21]. In order to investigate this topic, is advisable studies include categories and items varying in grasping/ manipulability [38]. Additionally, it seems reasonable that “atypical” categories (i.e., body parts and musical instruments) are not excluded from but included in the studies, in order to elucidate their specific role on category specific effects.

Format of the items utilized

As stated by Laws and collaborators [14], many studies on category specificity in AD have used the corpus of items by Snodgrass and Vanderwart [42]. These simple line drawings of 260 familiar everyday LT/NLT objects have been a useful tool extensively used by studies examining language memory and object processing. Nonetheless, the ecological validity of the mentioned items has been recently questioned [43]. In addition, due to their simplicity for example, to be named for healthy participants the mentioned corpus has been also associated with problems derived from ceiling effects [9]. Thus, the number of works using colored stimuli (e.g. color photographs) has been progressively increasing [35-37,44]. In addition, color stimuli allow, for example, working with categories of theoretical significance, which are difficult to be represented by using line-drawings [e.g., different types of trees or insects; [27,45]].

Type of tasks utilized

The picture naming task has been the most extensively used, and, sometimes, the only one to report category specific effects [31,37,46]; this is clearly related to the fact anomia is one of the earliest hallmarks for many neurological pathologies. However, the presence of “paradoxical dissociations”, where a patient can show impairment of one domain on one task (e.g. naming to description), and the reverse pattern on another domain (e.g. semantic fluency), suggests that the direction of category effects could depend on the task performed by the subjects [12]. Thus, it is interesting to use different tasks, in order to increase inter-task consistency; as well as to corroborate the ulterior presence of category effects in tasks other than picture naming [47,48].

Semantic Batteries: Fitness for Use

One of the best strategies for studying category specificity is utilizing semantic batteries [38,49,50,51]. These instruments own important characteristics, such as using the same items presented under different modalities [i.e., visual vs. verbal; [38,51]]. Certainly, the development of a useful neuropsychological tool focused on evaluating semantic knowledge should critically consider the aforementioned theoretical points.

Consequently, one basic factor should be the inclusion of a sufficient number of color images belonging to a wide range of categories; categories with theoretical interest should be also included; thus, body parts and musical instruments have to be considered in order to clarify their role into living-nonliving thing dissociations; on the other hand, categories such as trees, flowers or insects [27,45]; and NLT which varying in their degree of manipulability [52,53] must be also taken into account. It is also particularly important that items from LT and NLT are closely matched across domain in all the NV knowing to affect semantic processing. Furthermore, the selected items should make difficult healthy people get ceiling effects.

Additionally, the norms on the items (e.g., those from NV) should be gathered from the target population who will be assessed (e.g. young vs. old participants; Spanish vs. French). It is worth considering that several studies in the field have no contemplated this aspect. For example, the Snodgrass and Vanderwart corpus [42] were exclusively designed for American population and, thereby, is likely their cultural characteristics are not directly applicable to other populations [54].

In this context, our group has conducted several normative studies both with pictorial and verbal material focused on Spanish population. This theoretically derived material faces the aforementioned issues and is described hereafter.

Pictorial material

A set of 112 visual items

Moreno-Martínez and Peraita [54] presented a set of 112 items with norms of several NV gathered from a group of elderly Spaniards; this was one of the main novelties of the study, because it should be emphasized many normative studies have been conducted with young students.

The authors selected 14 semantic categories for theoretical and methodologically significant reasons. They included problematic/ atypical categories, such as body parts and musical instruments [55,56]; plant life categories, such as flowers, fruits, trees and vegetables [27] as well as categories differing in their degree of manipulability, such as buildings, kitchen utensils or tools [52]. Seven categories from the LT domain (animals, body parts, insects, flowers, fruits, trees and vegetables); and seven from the NLT domain (buildings, clothing, furniture, kitchen utensils, musical instruments, tools and vehicles) were included. Subsequently, a set of fifty-six visual stimuli (color photographs) for each of the two domains was selected; photographs for each item were obtained by one of the authors (FJMM) who photographed several items; the remainders were obtained via online sources.

This study gathered norms from AoA, familiarity, manipulability, name agreement and visual complexity; lexical frequency and typicality indexes, from other Spanish studies, were also provided. Concerning validity, the study showed high NV correlations with previous normative works that used color as well as black and white images [42,57,58,59,60]; this means the study shows high convergent validity with other databases. High rates of reliability were found among the NV: Cronbach’s alpha coefficients ranged between .83 (name agreement) to .98 (visual complexity and familiarity). In addition, differences in categorical processing with respect to gender were observed: women showed higher ratings for familiarity and manipulability (LT); and with visual complexity (NLT).

A set of 140 visual items: the Nombela naming test

Moreno-Martínez, Montoro et al. [37] Presented the Nombela Naming Test, a set of 140 color visual stimuli. The aim of this study was twofold, (i) to present the items providing different difficulty levels with the objective of avoiding problems relate to ceiling effects; (ii) to provide standards from a group of healthy controls in seven psycholinguistic variables: AoA, familiarity, manipulability, name agreement, typicality, visual complexity, plus lexical frequency indexes derived from internet search hits. Differently to the previously described stimuli set [54], the stimuli by Moreno-Martínez, Montoro et al. [37] were located on a white background to avoid possible influence of context (Figure 1).

Figure 1: Examples of items of the Nombela Naming Test. (Moreno-Martínez, Montoro & Laws, 2011).

    
    
    Figure 1:  Examples of items of the Nombela Naming Test. (Moreno-Martínez, Montoro & Laws, 2011).

The authors reported high correlations with similar works carried out in English [35,42]; this suggests the study has high convergent validity with similar works. Reliability coefficients for NV were also high: Cronbach’s alpha coefficients oscillated between .91 (name agreement) to .97 (manipulability and familiarity). The study reported higher familiarity, lexical frequency and typicality for LT items, as well as higher AoA and manipulability for NLT. No domain differences were found for name agreement or visual complexity.

A set of 360 visual items

A more ambitious study was afterward conducted by Moreno- Martínez and Montoro [36]; the author’s standardized 360 visual color stimuli (Figure 2) from 23 semantic categories (Table 1). The wide range of semantic categories studied–belonging to LT, NLT and elements of nature is one of the mains contributions of this work. The authors studied seven NV: AoA, familiarity, lexical frequency, manipulability, name agreement, prototipicity, and visual complexity. The 360 items consist of color photographs directly taken by the first author of the study; FJMM owns their copyright and allows the utilisation of the items for scientific proposals.

LT	NLT	NATURE
1. Animals	11. Buildings	23. Nature
2. Birds	12. Clothing
3. Body parts	13. Desk material
4. Flowers	14. Food
5. Fruits	15. Furniture
6. Insects	16. Pieces of jewellery
7. Marine creatures	17. Kitchen utensils
8. Nuts	18. Musical instruments
9. Trees	19. Sports/Games
10. Vegetables	20. Tools
	21. Vehicles
	22. Weapons