Sandali Liyanagoonawardena¹, Simona Adele Garobbio¹, Michael Herzog¹; ¹École Polytechnique Fédérale de Lausanne

Visual tests are crucial to assess visual abilities in vision research. Usually, we implicitly assume that the performance in one visual test correlates with that of similar tests. However, despite good test re-test reliability, attempts to show strong correlations between visual tests have proven unsuccessful. Even more surprisingly, visual tests that showed clear-cut differences between young and old observers did not correlate strongly in the older population. These results are surprising because one would have expected that, for example, a mild lens clouding should affect all visual functions, leading to correlations. Yet, observations convey that inter-individual differences are increased during aging. Here, we asked whether children, whose brains are not fully matured, exhibit correlations. We tested 52 children in 8 visual tests and 10 illusions, over 3 visits which were 6 months apart. Each test was measured twice at each visit to examine test-retest reliability. We analyzed the correlations between tests within a single visit and the longitudinal variations in children's vision. Furthermore, we investigated the correlations between performance changes across tests and between age and individual performance. The correlations between the performance of most visual tests remained below 0.4 and failed to reach the moderate range of Spearman normative values. Similarly, insignificant correlations were shown between the performance in each test and the age at each visit. When analyzing the individual performance changes between the three visits, the latent difference scores of visual tests were largely uncorrelated. However, there were significant variances in the individual performance change over time. Our results suggest higher-than-expected inter-individual variability of visual assessments, questioning the validity of visual tests. We encourage the application of a battery of tests, where the tests are carefully chosen, to replace the conventional use of a single visual test.