Error patterns in a chord keyboard response task

Errors in a serial choice-reaction time experiment employing two-finger chord responses to alphabetic stimuli are shown to be distributed in a significantly non-random fashion. Two factors are postulated which are relevant to prediction of the error response. These are the presence or absence of a common finger, and the difference in finger separation, between the actual and the required response. It is suggested that the chord response is coded in terms of a pattern (the distance between the two response components). Responses which mirror the correct response about the centre of the hand are also shown to be significantly more frequent.