Gravar-mail: How Does the Attention System Learn From Aversive Outcomes?