Silvia Guidi¹, Benjamin Wolfe^1,2, Anna Kosovicheva^1,2; ¹University of Toronto, ²University of Toronto Mississauga

Despite our extensive reading experience, people vary greatly in how well they read. These differences in reading performance can partly arise from differences in how a reader copes with the limitations of their visual system. We investigated readers’ ability to identify visually degraded stimuli in order to estimate individual susceptibilities to these manipulations. We tested whether there are common mechanisms for identifying degraded text and, if not, what forms of degradation relate to each other. Participants (N=11) completed two tasks: a lexical decision task and a non-lexical control task with six-character strings. We visually degraded the stimuli for both tasks with one of four methods by blurring, spatially distorting, reducing the contrast, or reducing the size of the stimuli, and varied the strength of the manipulation on each trial. We determined the stimulus level that corresponded to the 80% threshold for each condition in each task. In the lexical decision task, thresholds were moderately correlated with each other across manipulations (mean pairwise Fisher’s z = 0.41). Blur and acuity were the most strongly correlated, such that participants who were more susceptible to blur also showed the largest performance decrement with small text sizes. Correlations were similarly strong in the non-lexical task (Fisher’s z = 0.49), with the strongest correlation between size and contrast; participants who were better able to identify low contrast stimuli also had better acuity. These correlations show that susceptibility to different visual degradations is partially degradation-specific. Moreover, shared patterns of correlations across the tasks point to non-language specific mechanisms in play. We also correlated the difference in thresholds between tasks and found an average Fisher’s z of 0.66. suggesting that language-specific mechanisms contribute to performance with degraded stimuli. Taken together, our results point to both language-specific and non-language specific mechanisms impacting the identification of degraded stimuli.

Acknowledgements: This work was supported by a SSHRC Insight Grant to AK and BW.