Publication
Using batteries of visual tests, most studies have found that there are only weak correlations between performance levels of tests in healthy young adults. Factor analysis has confirmed these results. This means that a participant excelling in one test may rank low in another test. Hence, there is very little evidence for a common factor underlying visual abilities. In visual aging research, cross-sectional studies have repeatedly found that older adults' performance deteriorates in most visual tests compared to young adults. However, within the older population, there is no evidence for a visual common factor.
The core work of this thesis seeks to understand where these weak correlations come from.
First, I investigated whether stronger correlations emerge in clinical populations, expected due to the general decline in visual abilities that affects some patients more than others. To this aim, older adults with mild cognitive impairement, young adults with visual snow syndrome, and older adults with age-related macular degeneration were tested with a battery of visual tests. Correlations were stronger only in older adults with age-related macular, suggesting the emergence of a stronger common factor underlying visual abilities although this factor could not account for all the tests included in the battery.
Second, I showed that the visual tests are reliable and individual differences in each test are stable across years.
Third, I examined whether performance in the visual tests correlates with the thicknesses of the retina's sublayers, the nerve fiber layer, or neural correlates as measured with EEG during test performance. Strong correlations were observed only between the structural measures of the retina and visual test performance in older adults with age-related macular degeneration.
Overall, the results highlight the relevance of an individual differences approach to tackle the heterogeneity of the visual system. Similarly, in other fields of research, multidimensionality has been shown to underlie tests commonly used to measure the same construct, suggesting that one test may be less informative than initially thought. Researchers should rethink alternative strategies to embrace the multidimensionality underlying a construct. In this thesis, I discuss the case of schizophrenia and propose two endophenotypes to be included in a larger test battery, aiming to reveal the level of complexity of the disorder.