Proactive and reactive recruitment of cognitive control: Comment on Hikosaka and Isoda