pubmed.ncbi.nlm.nih.gov

Feature contingencies when reading letter strings - PubMed

Feature contingencies when reading letter strings

Daniel R Coates et al. Vision Res. 2019 Mar.

Abstract

Many models posit the use of distinctive spatial features to recognize letters of the alphabet, a fundamental component of reading. It has also been hypothesized that when letters are in close proximity, visual crowding may cause features to mislocalize between nearby letters, causing identification errors. Here, we took a data-driven approach to investigate these aspects of textual processing. Using data collected from subjects identifying each letter in thousands of lower-case letter trigrams presented in the peripheral visual field, we found characteristic error patterns in the results suggestive of the use of particular spatial features. Distinctive features were seldom entirely missed, and we found evidence for errors due to doubling, masking, and migration of features. Dependencies both amongst neighboring letters and in the responses revealed the contingent nature of processing letter strings, challenging the most basic models of reading that ignore either crowding or featural decomposition.

Keywords: Crowding; Feature migration; Letter recognition; Peripheral vision; Redundancy masking.

Copyright © 2019 Elsevier Ltd. All rights reserved.

PubMed Disclaimer

Figures

Figure 1:
Figure 1:

Summary of errors for each subject. In the figure legend, an “a” indicates a correct answer in the corresponding position (order: left, middle, right), while an ”X” indicates an error in that position. Incorrectly identifying the middle letter and correctly identifying the two outside letters was by far the most common error for all subjects, constituting nearly half of the error trials for each subject (or approximately a quarter of the total trials for each subject). Total number of trials are shown on the left axis, while the proportion of total trials are shown on the right axis. Each subject made an error in 42–55% of trials.

Figure 2:
Figure 2:

Aggregate confusion matrices for left, middle, and right letters, respectively. Cells are colored by occurrence from light/pastel to dark (see the color bar on the far right for values of occurrence). Letters presented are shown in rows while responses are shown in columns.

Figure 3:
Figure 3:

Proportion correct for each observer identifying each of the 26 letters presented in each position. Proportion is colored as indicated by scale bar, with darker colors indicating worse performance.

Figure 4:
Figure 4:

Illustration of Monte Carlo approach to statistical robustness determination of error contingencies (Section 3.1.4). The conditional likelihood of a subject making a left letter error is shown by the two vertical lines, conditioned on whether the middle letter was correctly identified (green vertical line) or erroneous (red vertical line). A left letter error was about 40% more likely when a middle letter error occurred, and simulations from the independence model (colored histograms) show that this difference was extremely unlikely to happen by chance.

Figure 5:
Figure 5:

For trials with an error in the middle position and at least one outer letter, the summed change in complexity of the outer letters is plotted against the change in complexity of the middle letter. A positive number indicates that complexity is gained in the response. Each point represents a single trial. The correlation coefficient, given in the upper right corner, indicates a negative correlation—complexity gain/loss is balanced between letter positions.

Figure 6:
Figure 6:

The average confusability of letters for each pair of middle and outer letters, plotted against the number of correct letters reported in the pair. Points and lines show the empirical data, while colored “violin plots” show the distributions resulting from 1000 simulations. Correct reports are negatively correlated with overall confusability, in both the data and the simulations.

Similar articles

Cited by

References

    1. Attneave F, Arnoult MD (1956). The quantitative study of shape and pattern perception. Psychological bulletin, 53, 452. - PubMed
    1. Ağaoğlu MN, Chung STL (2016). Can (should) theories of crowding be unified? Journal of Vision, 16, 10–10. - PMC - PubMed
    1. Bernard J-B, Chung STL (2011). The dependence of crowding on flanker complexity and target–flanker similarity. Journal of Vision, 11, 1–16. - PMC - PubMed
    1. Bouma H (1970). Interaction effects in parafoveal letter recognition. Nature, 226, 177–178. - PubMed
    1. Bouma H (1971). Visual recognition of isolated lower-case letters. Vision Research, 11, 459–474. - PubMed

Publication types

MeSH terms