Accuracy of extrapolating a pointer into perceived and imagined space

The authors investigated the ability of people to extrapolate lines in tridimensional space: an experimenter pointed at variously located targets, and the subjects tried to identify them. In an experimental condition, the subjects sat with their backs turned to the target objects, which they had previously inspected during a brief learning period; in two control conditions the targets were in front of the subjects and clearly visible. Overall error (difference in degrees of pointer angle between correct and judged targets) was no worse for objects imagined behind the head than for visible targets, about 2.7 degrees (root-mean-square) in both cases. Variable error was 29% greater for the imagined objects, but constant error was greater on the matched control targets. Some qualitative differences, in constant error patterns, were also found. With due allowance for these differences, the results strongly indicate that imagined space is functionally continuous with perceived space in the representational system.