In visual backward masking (VBM), a target is followed by a mask. Over the past decade, a VBM task, called the Shine-Through paradigm (ST), has been shown to be a very sensitive endophenotype of schizophrenia. In the present thesis, the electrophysiological correlates (EEG) of this endophenotype are investigated. Patients with schizophrenia show strongly reduced amplitudes of the Global Field Power (GFP, standard deviation across all electrodes of the EEG) 200 ms after target onset. Here, I show, first, that patients with first episode psychosis also have masking deficits and that the associated EEG amplitudes are between those of controls and patients with schizophrenia. In detail, when the task is challenging the neural correlates are similar to patients with schizophrenia and when the task is easy the neural correlates are similar to controls. This result suggests that patients with first episode psychosis are in an intermediate state where the disease is not fully developed yet. In addition, the neural signature is stable over at least one year meaning that the fully developed deficits must occur over durations longer than one year. Second, I show that healthy participants with high schizotypal traits, in particular scoring high in the cognitive disorganization dimension, also have GFP amplitude reductions 200 ms after target onset compared to participants with low schizotypal traits. Quantitatively, the difference between high and low scoring participants is much smaller than between patients and controls. Nonetheless, the result shows that deficits are qualitatively similar along the schizophrenia continuum even in a non-clinical population. Third, patients with 22q11 deletions syndrome, who also suffer from psychosis in 30% of the cases, do not show masking deficits. Deficits were expected but potentially because of high co-morbidity with other disorders, I did not find the same neural signature. Fourth, patients with major depressive disorder are not impaired in the ST, however, their neural correlates are lower compared to controls, though not as low as schizophrenia patients. This is surprising given the good behavioral performance. Despite the GFP reduction, the neural signature of depressive patients differs from the one of schizophrenia patients. They do not show the amplitude increment as a function of task difficulty seen in controls, siblings of patients with schizophrenia, and patients with schizophrenia. Finally, I describe preliminary results from on-going studies on healthy aging, autism and bipolarity. The objective is to make all populations fit in a unified framework.