Suicide is known to be frequent in depression, and in most of the psychiatric diseases. But as it can occur in patients with no psychiatric illness and doesn't occur in every patients with psychiatric illness, it has to be considered henceforth as a specific vulnerability.

This trial will study two fMRI paradigms, emotional prosody recognition and decision making, in order to characterize emotional and cognitive trait factors in a population of patients vulnerable to suicide. Four different groups will be constituted : depressed suicide attempters, depressed patients with past history of suicidal acts, depressed patients with no history of suicidal acts and healthy controls.

The main goal will be to correlate fMRI activation during the two paradigms in subjects vulnerable to suicide. The secondary goals will be to characterize emotional and cognitive trait factors in these subjects, to demonstrate that those characteristics are independent from depression and to correlate these trait factors with socio-demographic and clinical features with fMRI activations.

Inclusion Criteria:

• Right-handed,
• Depressed suicide attempters for the first group,
• Depressed patients with past history of suicidal acts for the second group,
• Depressed patients with no history of suicidal acts for the third group,
• Healthy controls for the fourth group.

Exclusion Criteria:

• Another psychiatric trouble than depression,
• Another psychotrope treatment than SSRI/SNRI and/or benzodiazepine at inclusion and/or one month before,
• For the first and secand groups : more than fous suicidals acts within the past 3 years,
• Deficits that do not allow test realisation (visual audition troubles, cognitif deficits,
• Prosopagnosia,
• Pregnancy,
• Severe central nervous system disease or somatic disease,
• MRI contra-indication,
• Major benefiting from a legal protective measure,
• No coverage care.