The spatial–temporal association of response codes (STEARC) effect reflects a space-related representation of time (mental time line) with a genuine left-to-right orientation. Alternatively, it may simply reflect an over learned motor association between time and manual responses. In this study, we want to explore STEARC effect under different response conditions, i.e. manual and saccadic response. There are two experiments in this study. Experiment 1a and Experiment 1b tested the STEARC effect using horizontally arrayed left/right responses by manual and saccadic responses respectively. Experiment 2a and Experiment 2b used vertically arrayed bottom/top responses by manual and saccadic responses respectively. The stimuli consisted of the two types of time words (past vs. future), which were displayed in the center of a computer monitor. Manual or saccadic response latencies were recorded. The Lenovo computer was used to records manual data and the SMI EyeLink-2000 systems was used to track saccadic response. In experiment 1, there was a interaction between response side and time words. Left-hand responses were faster than right-hand responses of the past words, Right-hand responses were generally faster than left-hand responses of the past words. There was also a interaction between saccadic and time words, past words responses were faster to the left saccadic, future words responses were faster to the right saccadic. In experiment 2, there was a interaction between response side and time words. Past words responses were faster of the top key, future words responses were faster of the bottom key, there was also a interaction between saccadic and time words, past words responses were faster to the up saccadic, future words responses were faster to the down saccadic. In addition, the error rate results support our hypothesis. The result suggest that that the time–space associations consistently found by manual and saccadic response conditions, which cannot be caused by preferred and overlearned manual motor associations. Our findings support the interpretation that in two response conditions, space-related magnitude representations are automatically activated and modulate efficiency a horizontal or a vertical motor response. The present results also suggest that time map is a preferred explanation for STEARC, not a mental timeline.