Magnocellular facilitation of flanked-letter identification disappears with strong flanker interference

Abstract

Previously, we obtained evidence to suggest that the magnocellular system may reduce interference from flankers during flanked-letter identification. To understand this phenomenon better, we combined the data of our previous experiments, which all used the same flanking letter, and focused on the different target letters that were used. The new analysis showed that after an initial increase, magnocellular facilitation decreased and ultimately disappeared, as the target–flanker combination's level of interference increased. The initial increase was partly while the later decrease was fully replicated in two new experiments that focused on two different target letters while manipulating flanker identity. The outcome of a third new experiment studying the type of interference reduced suggested that although crowding contributed to total interference, it was insensitive to magnocellular mediation. Our results may be understood to reflect the involvement of the magnocellular system in an attentional-selection mechanism that is silenced by surround suppression.

Keywords:

• Crowding
• Developmental dyslexia
• Letter identification
• Magnocellular system
• Surround suppression
• Visual attention

Acknowledgements

We thank three anonymous reviewers for their thoughtful comments on an earlier version of this paper and the Electronic Research Group at the Faculty of Social Sciences at Radboud University Nijmegen for their technical support.

Notes

¹Statistical analysis may perhaps not seem necessary in the present situation, as virtually all letters of the alphabet were included. However, it could be argued that the population consists of all visual stimuli, with the letters of the alphabet representing a small, albeit nonrandom, subset.

²The z-scores were calculated as (x – m)/s, where x is the value of the measure under consideration for a given letter and stimulus condition, and m and s are the sample mean and the sample standard deviation, respectively, of the measure across all stimulus conditions and letters.

³Other than for the v target, for the c target, z was also one of the flanking letters. Because there was no significant difference associated with use of the two different nontarget letters, it was decided to keep both.

⁴We will report uncorrected p-values for our planned comparisons. Although some might argue for the use of Bonferroni or similar corrections, this would not be entirely appropriate for the present situation, where for many comparisons in fact null results were predicted. Moreover, a case could be made as well to adjust p-values downwards (divide by 2; one-tailed test) if results are in the predicted direction.