Testing computational models of letter perception with item-level event-related potentials

Abstract

In the present study, online measures of letter identification were used to test computational models of letter perception. Event-related potentials (ERPs) were recorded to letters and pseudoletters revealing a transition from feature analysis to letter identification in the 100–200-ms time window. Measures indexing this transition were then computed at the level of individual letters. Simulations with several versions of an interactive-activation model of letter perception were fitted with these item-level ERP measures. The results are in favour of a model of letter perception with feedforward excitatory connections from the feature to the letter levels, lateral inhibition at the letter level, and excitatory feedback from the letter to the feature levels.

Keywords:

• Letter perception
• Event-related potential
• Computational modelling
• Feedback
• Lateral inhibition

We are grateful to Fermı´n Moscoso Del Prado Martı´n for his useful advice.

Notes

¹ Abbreviations used for parameters are as follows: α and γ stand for excitatory and inhibitory connections, respectively; F and L represent the feature and letter levels, respectively.

² Following Thorpe et al. (1996), we used the mastoid as the reference electrode. Although this choice does affect the spatial distribution of ERPs, it does not affect the timing of the difference between letters and pseudoletters. Since we were interested in getting precise information about timing (and not localization), we decided to use the same reference as that in the study by Thorpe et al. (1996).

³ Note that the grand mean for pseudoletters is only used here as a common reference in order to visualize variability between ERPs for individual letters. A direct comparison between this ERP grand mean and individual-letter ERPs is meaningless because the whole set of pseudoletters was not matched on visual features with each of the individual letters.

⁴ As already noted in the introduction, ERP measures obtained in the masked priming paradigm with letter stimuli led to slightly later estimates (Petit et al., 2006). This is probably due to a general slowing of the target letter identification process following a pattern mask and prime stimulus. This possibility is consistent with the findings of Ziegler et al. (2000) who demonstrated, using the incremental priming technique proposed by Jacobs, Grainger, and Ferrand (1995), that most of the repetition priming effect found for letter targets in an alphabetic decision task was in fact driven by a reduction in the interference effect of unrelated primes.