According to the salient-features coding hypothesis, S-R translation is more efficient when an S-R mapping maintains a structural correspondence of the salient features between the stimulus and response sets (Weeks & Proctor, 1990). As "above" and "right" are the salient features in vertical and horizontal axes respectively, an above-to-right/below-to-left mapping has an advantage over the opposite mapping when vertically arrayed stimuli are mapped to horizontally arrayed responses (orthogonal stimulus-response compatibility effect). However, the above-right/below-left advantage may not be restricted to the intentional S-R translation. With parallel S-R arrangements, response selection is faster and more accurate when the stimulus and response positions correspond even if the stimulus location is task-irrelevant (Simon effect). We investigated whether a stimulus automatically activates a response corresponding in the saliency dimension. In Experiment 1, participants responded with right or left key-press to the color of the stimulus presented above or below the fixation. Although the stimulus location was task-irrelevant, the above-right/below-left advantage was observed (orthogonal Simon effect). In Experiment 2, we manipulated the salient feature in the horizontal response dimension by varying the position of the response set horizontally because the side on which a response set is positioned becomes the salient feature (Proctor & Cho, 2003). An orthogonal Simon effect emerged with response set on the right side. In contrast, a reversed orthogonal Simon effect (i.e., above-left/below-right advantage) emerged with the response set on the left side. It was confirmed that the orthogonal Simon effect is based on the S-R correspondence of the salient features. We conclude that the salient and non-salient stimulus features automatically activate the corresponding salient and non-salient response features.