Two tasks were presented in a row. In the first task, the manual tilting task, the participants actually
tilted the glasses to an angle where the imaginary water inside the glass would reach the rim. The
second task was a judgment task in which no movement took place. In this task, the angle at which the
water would reach the rim needed to be indicated by attaching a pointer to the panel at the appropriate
angle. The order of the two tasks was not counterbalanced so as to avoid provoking erroneous
thoughts before the manual tilting task (see Schwartz, 1999). In both tasks, the dependent variable
was the angle of tilt in degrees. Thus, in both tasks, the dependent measure was metric and interval
scaled, allowing for quantitative inferences about the underlying mental representations. In addition,
the response was nonverbal and, therefore, suitable for young children at preschool age because it did
not require special verbal skills. Before each task, two practice trials were presented in counterbalanced
order with intermediate water levels that did not occur later in the experimental trials.