**5. Conclusions**

We can conclude that age is a significant predictor of visuospatial memory decline. Accordingly Rabbit et al. [36] reported that age predicted the results from Spatial Working Memory test.

The two age groups did not differ significantly in the performance of house and cube drawing and of Block design task. Data exist about worse perception and presentation of three-dimensionality in cube drawing task by elders [29], but we could not find studies of house and cube drawing in different late-life groups. It could be supposed that the interindividual variability, characteristic of old age in this comparatively small sample, influenced our results. The size of the sample has also prevented the use of a more detailed statistical analysis of the perfor-

A possible explanation of the results concerning the cube drawing task score and BVTR total outcome measures could be the complexity of the tasks and in particular the three-dimensionality, as a mandatory feature of the cube drawing. These results could be partially explained as well by the structure of the BVRT task, which involves reproduction of geometric shapes by memory. The task of drawing a cube and a house also requires reproduction, but long-term memory is involved here, while Benton test assesses short-term memory. Another difference between Benton test and the drawing of cube and house is related to BVRT patterns themselves—part of them are new, unknown spatial models, and the other part are well-known figures (triangle, square, circle, trapezoid) engaging long-term representations. A comprehensive cognitive model of adults' drawing ability has not yet been developed. What is well known is its "multicomponential nature" [[51], p. 117] confirmed in this study by a principal component analysis of the results from testing healthy adults and individuals with CIND

The global functioning or the intelligence together with attention, sensory, motor, and executive functions is fundamental for the visuospatial and visuoconstructive abilities, following R. Mapou's [43] hierarchical model. The correlations found reflect the relationship between these basic functions and the capabilities required for specific (constructional and spatial) cognitive functions. This explanation is supported by the principal component analysis, according to the results of which global and executive functioning are required for the performance, assessed by a large number of study variables. Interpretation of principal component analysis reveals at the same time the specificity of constructive and spatial functions, based on

As elements of the multiple regression model, education of participants predicted the total number of errors in the group of healthy subjects and the number of omissions in CIND group. In another study without consideration of type of errors, the level of performance of normal older adults aged 61–97 showed dependence on education. In the same paper, in the group of normals with memory concerns from 64 to 74, less educated had worse performance, the difference found not reaching significance over 75. As for the gender effects on BVRT performance, there are no evidences about significant differences between men and women, from most research results available [70]. Resnick et al. [74] reported sex differences for omissions and rotations in subjects from 20 to 102, but they account for very low percentage of the variance (1%). Our multiple regression results gave a gender effect only on distortions, made by healthy subjects.

mance of these three tests.

184 Gerontology

over 60 years of age (four factors extracted).

visuospatial analysis and perception.

The basic objective of this paper was to analyze the performance of constructive and visuospatial tasks in healthy and in CIND subjects.

The results confirm our hypothesis about significant differences in the level of performance in drawing and construction between persons with CIND and normally aging individuals over 60 years.

We found a prevalence of omissions and distortions in the error profile of CIND and significant difference between CIND and normal aging regarding these two types of errors.

In both diagnostic groups, age of participants showed a significant effect on BVRT omissions, when fluid intelligence, education, and gender were also considered.

Results proved discriminative sensitivity of BVRT general scoring criteria and the separate error types (omissions and distortions) in the preclinical stages of dementia.

We tested a modification of Moore and Wike [52] scoring system for house and cube drawing task in elders, and this study confirmed its diagnostic sensitivity. Drawing of cube and house could be used for quick screening of CIND in subjects over 60.

Results from the principal component analysis (oblimin rotation) reaffirmed the multicomponent structure of the visuospatial and constructive abilities in old age.

The main limitation of this study is the small number of participants with complete neuropsychological testing and the lack of detailed clinical and neuroimaging examination. For the future, it might be interesting to carry out a similar analysis using more detailed description of subjects, including neuroimaging with functional MRT that could give the possibility to conclude about brain structures involved in different task performance.
