Visual Confusion Matrices: Fact or Artifact?

On tactile and visual recognition

“…While not totally clear-cut, it appears that the differences between either of the two Fisher et al (1969) matrices and anyone of Townsend's (1971aTownsend's ( , 1971b three matrices are only slightly larger than the differences within either subset, and substantially smaller than the differences between anyone of these five and either the Craig (1979) or data. Mewhort and Dow (1979) point out that Gilmore et al's main diagonal vector correlates -.873 with the number of dots used in the representation of each character.…”

“…Since they used tachistoscopic presentations of capital letters, it seemed plausible that, if they could be compared on some common basis to Townsend's (1971aTownsend's ( , 1971b, the results should be similar. The difficulty is that Townsend controlled his average correct recognition to .5, whereas it was .59 and .78 for Fisher et al's (1969) two matrices. As an approximation, these two matrices were normalized to .5 by multiplying all main diagonal elements by the appropriate factors (.85 and .64) and proportionally correcting all off-diagonal cells.…”

“…In a further effort to understand this question, Fisher, Monty, and Glucksberg's (1969) data were also analyzed. Since they used tachistoscopic presentations of capital letters, it seemed plausible that, if they could be compared on some common basis to Townsend's (1971aTownsend's ( , 1971b, the results should be similar.…”

“…As an approximation, these two matrices were normalized to .5 by multiplying all main diagonal elements by the appropriate factors (.85 and .64) and proportionally correcting all off-diagonal cells. These reconstructed matrices were then compared to Townsend's (1971aTownsend's ( , 1971b), Craig's (1979), and Fisher et al's (1969) matrices. The results are summarized in Table 1.…”

Comment on Recognition and confusion of the lowercase alphabet

“…Maddox, Burnette, and Gutmann (1977) present data for uppercase dot matrix letters, which indicate that letter style may be an important variable altering the pattern of confusions. Further, research presented by Fisher, Monty, and Glucksberg (1969) suggests that comparison across confusion data derived from different procedures may not be valid. The comparison made by Geyer (1977) rests on the assumption that a basic "pattern of confusion" exists between the lowercase letters of the alphabet.…”