[Visual pattern analysis and reasoning: Ravens' Coloured Progressive Matrices in old-age and very-old-age adults].

Sixty-five non-demented elderly adults, born between 1895 and 1918 (mean age +/- sd: 80.0 +/- 5.4) were tested with Raven's Coloured Progressive Matrices (CPM). Subjects were recruited from homes for the aged and were rated by the staff as free from any symptoms of dementia or other psychiatric disease. Mean number of years of education was 8.2 (range 5 to 18). CPM scores ranged from 9 to 35 with mean +/- sd of 25.9 +/- 5.9. Subjects who had received more formal education performed better on the CPM (Pearson's r: 0.49). When education was controlled for in the analysis, the correlation between age and performance was attenuated and did not reach statistical significance (Pearson's r: -0.24). Test results appeared to be specific for generations, regardless of age. Mean performance in this sample was significantly higher than less recently published norms suggest. Analysis of item content revealed that the CPM consists of three main types of problems: two of a predominantly visuospatial type (12 items of simple continuous pattern-completion and 15 concrete items showing progressive changes in one or two directions) and 9 items of an abstract reasoning type. The concrete visuospatial items appear very useful in the assessment of visuoperceptive dysfunction, as for example in visual apperceptive agnosia. The abstract matrices were very difficult for most of our elderly subjects, so that these items cannot be used to detect deviations from normal old age.