Effects of texture on visual noise masking.

Visual noise masks, composed of letter fragments scattered randomly over the visual field, disrupt the recognition of briefly viewed words when presented immediately before and/or after these words. According to the widely held feature-similarity assumption, maximal disruption is predicted when the mask's letter fragments are equal in size and shape to the target letters. This assumption is not supported by the present experiment, which measured the masking effectiveness of visual noise patterns varying in texture (achieved by photographic reduction or enlargement). For target letters with stroke widths subtending 7.4' or less, the optimally disruptive mask did not vary with the stroke width of the target letters, but rather, was invariably composed of letter fragments with stroke widths subtending about 7'. For larger target words, however, this mask was not optimally disruptive. These results suggest that recognition of letters less that about 7' is mediated by a single size-tuned mechanism.