On the costs and benefits of repeating a nonspatial feature in an exogenous spatial cuing paradigm