According to the spatial–temporal association of response codes (STEARC) effect, time can be spatially represented from left to right. However, exploration of a possible STEARC effect along the vertical axis has yielded mixed results. Here, in six experiments based on a novel paradigm, we systematically explored whether a STEARC effect could emerge when participants were asked to classify the actual temporal duration of a visual stimulus. Speeded manual responses were provided using a vertically oriented response box. Interestingly, although a top-to-bottom time representation emerged when only two temporal durations were employed, an inverted bottom-to-top time representation emerged when a denser set of temporal durations, arranged along a continuum, was used. Moreover, no STEARC effects emerged when participants classified the shapes of visual stimuli rather than their temporal duration. Finally, three additional experiments explored the STEARC effect along the horizontal axis, confirming that the paradigm we devised successfully replicated the standard left-to-right representation of time. These results provide supporting evidence for the notion that temporal durations can be mapped along the vertical axis, and that such mapping appears to be relatively flexible.