RTs were obtained in a task in which a red or a green stimulus could appear at either of two locations ‐ left or right ‐ and subjects responded by pressing a green button on the left or a red button on the right. Two forms of instruction were used, one mapping stimulus colour onto response location, the other mapping stimulus location onto response colour. Faster performance was hypothesized in the former because it entailed fewer operations in response selection. This was confirmed. Faster performance was also hypothesized when the S‐R relations in the task for both colour and location could be described by a single recoding rule (both “same” or both “alternative”). This was not confirmed, faster performance being obtained only when colour and location were both “same”. Whilst no specific superiority of performance under a single recoding rule was demonstrated, the faster performance under a colour to position mapping was interpreted as showing the general importance of underlying recoding processes as d