A fundamental question about time perception concerns the form in which time is represented. Psychophysical approaches to answering this and related questions have focused on the various scales of representation that might be available to an animal. For example, Stevens (1951) described a hierarchy of measurement scales ranging from nominal to ordinal, interval, and ratio. In this hierarchy, higher order scales contain the properties of lower order scales plus the addition of an extra property. For example, with an ordinal scale, it is possible to represent the order of values on the scale. An interval scale has added to it information about the spacing of successive values, permitting the addition and subtraction of values on the scale. Finally, a ratio scale has added to it information about the relative position of values on a scale with an absolute zero point, permitting the multiplication and division of values on the scale. For example, finalists in a race are ranked first, second, and third. This ordinal ranking conveys no information about how close the finish times were among the finalists. Therefore, there is no guarantee of a meaningful outcome if higher order operations are performed on these ranks. However, in some cases, a meaningful outcome might occur (e.g., when the finish times for the finalists are about equally spaced).Identifying the representation of time is an important issue for theories of timing. For example, in the scalar timing theory (Gibbon, 1991), it is proposed that an accumulator integrates the number of pulses from a pacemaker. According to the behavioral theory of timing (Killeen & Fetterman, 1988), an animal proceeds through a sequence of behavioral states with transitions produced by a pacemaker. Staddon and Higa (1999) proposed that timing abilities are based on the magnitude of a memory trace since the last reinforcement. Theories of timing have generally not explicitly stated what type of measurement scale should be used. In some cases, however, a theory can be more compatible with one measurement scale than with others. For example, an important feature of scalar timing theory is the use of ratio decision rules, which are naturally compatible with a ratio scale. However, in other cases, the compatibility is less clear. For example, although the behavioral theory of timing is based on an ordinal sequence of behaviors, the theory has been used successfully in a variety of tasks that imply higher order scales (e.g., it can predict bisection at the geometric mean, which implies a ratio scale).One approach to identifying the representation of time involves a psychophysical method. For example, in Gibbon and Church's (1981) time-left procedure, animals were required to choose between a standard interval (e.g., a fixed 30 sec) and a comparison interval (e.g., 60 2 t, where t is the elapsed time into the trial when the comparison lever is inserted). For example, in one of their experiments, a trial began with the comparison lever entering the chamber followed by the standard leve...