The stability of error‐related brain activity with increasing trials