A Comparison of Contour Interaction and Crowding

Purpose: The ability to correctly identify a target is significantly reduced by the presence of flanking distractors. The term “crowding” is often used to encompass a large variety of target and flanker types. A major aim of this study was to enumerate the differences between the impairment of target identification produced by flanking bars (contour interaction, “CI”) and flanking letters (crowding, “CW”). Both the magnitude and extent of CI and CW were studied in the context of foveal and peripheral viewing, for various background luminances, and in color congruent versus incongruent conditions.

Methods: Subjects (n=4) viewed Sloan letters (C, D, H, K, N, O, R, S, V, Z) and were asked to identify such target. Flanking objects were placed at various distances from the target to induce CI or CW. Subjects repeated this task for a variety of target eccentricities (foveal versus 1.25, 2.5, 5.0, and 10 degrees in the inferior visual field), background luminances (photopic, mesopic, high scotopic, and low scotopic), and color-congruent (flanker and target either matched or did not match in color) conditions. Average percent correct letter identification was plotted as a function of flanker-to-target separation (in arcmin).

Results: A significant interaction effect was found in all conditions. In the fovea, a change from photopic to mesopic luminance produced a significant decrease in both CI and CW. In scotopic conditions, peripheral CI also was reduced. No significant effect of color congruency was found. The extent of foveal CW and CI was approximately 5 arcmin. During peripheral viewing the extent ranged from 6-40 arcmin for CI and from 10 to greater than 40 arcmin for CW, with the extent increasing as targets were viewed further from the fovea.

Conclusion: This study suggests that the use of flanking bars versus letters should not both be subsumed within the term “crowding” and that future comparisons must consider flanker type. The effects of luminance are consistent with a center-surround receptive-field neural model for CI. The results regarding flanker type and color congruency raise further questions on the role of flanker complexity and “top-down” influences on CW and CI.