Gravar-mail: Functional biases in attentional templates from associative memory