Constant errors occur in matched reproduction of angles even when likely biases are eliminated

The literature on the reproduction of visually perceived angles often seems to contain an explicit or implicit assumption that systematic errors are caused by misperception of the sizes of the angles. We argue that in the absence of procedural artifacts no consistent error should arise from purely perceptual causes. Accordingly, an experimental procedure was designed to eliminate or evaluate a number of artifacts that probably affected earlier experiments. A computer-controlled display generated both a standard angle and a variable one that could be adjusted by a subject to match the standard. The data analysis followed Cox’s (1970) procedure for the analysis of binary data to focus on the numbers of positive and negative errors rather than on their magnitudes and so eliminate the numerically distorting effects of asymmetrical error distributions. Nevertheless, the results showed what has usually been reported before: acute angles were matched with too large a comparison angle, and obtuse ones with too small an angle. Thus, there is a real effect to be explained, although the explanation must invoke response processes.