Is it that simple? The use of linear models in cognitive neuroscience