Feature perturbations are no guessing strategy artifact

Supporting evidence has been reported for a model in which identification errors on briefly presented character arrays are attributed to feature perturbations that tend to proceed in a foveal direction. Nevertheless, assuming that subjects guess the most foveal feasible location when they have only a rough idea about where a feature is seemingly could account for the results of these localization tasks. A detection task was thus currently used, with the results subjected to signal detection analysis or correction for response bias. The position of a small parafoveally presented target feature was flanked to the foveal or peripheral side by a strip composed of features that were similar or dissimilar to the target. Detection was poorer in each of three experiments when the strip was nearer the fovea, but only if it contained features similar to the target, indicating that the target feature underwent perturbation to become mixed with features in the strip. A possible explanation involving differential lateral masking rather than feature perturbations was not supported. It was concluded that feature perturbations are not an artifact of guessing biases and tend to proceed in a foveal direction.