Decision-making impairments according to history of suicide attempt in depression: A computational model analysis

It is well-established that individuals who have attempted suicide present decision-making impairments. However, the underlying constructs that constitute decision-making are still being explored.

To compare decision-making impairments in patients with a current or past depressive episode based on lifetime suicidal history and current depression.

295 patients with current and past depression were recruited and divided into 4 groups according to history of suicide attempt (suicide attempters (SA) or psychiatric controls (PC)) and severity of depression (high or low). The outcome representation learning (ORL) and the Prospect Valence Learning (PVL) models were applied to the Iowa Gambling Task (IGT) to extract estimations of feedback sensitivity, learning and consistency in the decision-making process. We also explored how childhood trauma moderated the relationship between the parameters and suicide.

The optimal model for our sample was the ORL. Among the five ORL parameters, low depressed PC showed lower reward learning and greater forgetfulness than SA, but higher deck perseverance was observed only when compared to high depressed SA. Moreover, punishment learning was able to differentiate SA groups in terms of depression severity. Furthermore, patients with high childhood trauma, high forgetfulness and deck perseverance were more likely to be in the SA cohort.

Suicide attempt may be linked to heightened feedback sensitivity to environmental contingencies, greater decision-making randomness and higher rumination. Moreover, the role of both forgetfulness and deck perseverance in trauma survivors at risk for suicide needs to be further investigated to refine their cognitive profile.