Abstract
Although previous studies on detection or reading have not consistently shown significant polarity effects, our studies of background effects on text readability have shown that, on a gray background, dark text is more readable than light text. Our readability tasks included searching for target words within a paragraph of text, word identification, word discrimination, and letter identification. For all tasks, positive contrast task responses were slower and less accurate than those for negative contrast tasks. Polarity effects are often explained as the result of separate channels for the two polarities. Effects in reading-related tasks might result from the increased experience people have reading dark letters on a light background. To differentiate between these explanations of the polarity effect, we surrounded gray letters with a small positive or negative pedestal. The pedestal polarity determines the channel polarity, opposite the apparent polarity of the letter. Stimulus-response confusion matrices and response latencies were collected from 14 observers for 4 repetitions of 12 letters presented randomly at 6 letter or pedestal contrasts (−0.4, −0.2, −0.1, 0.1, 0.2, 0.4). Performance was computed as information transmitted per second from the confusion matrices pooled over repetitions and the corresponding average latency. The letters without the pedestals replicated earlier effects, the information rate in bits per second was 3.86 bits/sec for the negative contrast and 3.76 bits/sec for the positive (F = 6.29, p[[lt]].05). The information rate was also larger for the negative contrast versus the positive contrast pedestal (3.82 bits/sec vs. 3.74 bits/sec; F = 6.99, p[[lt]].05). The pedestal results are in the direction of the pedestal contrast, rather than the perceived contrast of the letter, supporting the channel explanation over the experience explanation of the polarity effect.