Implicit Measures of Receptive Vocabulary Knowledge in Individuals With Level 3 Autism

Abstract

Implicit measures of cognition are essential for assessing knowledge in people with Level 3 autism because such individuals are often unable to make reliable overt behavioral responses. In this study, we investigated whether three implicit measures-eye movement (EM) monitoring, pupillary dilation (PD), and event-related potentials (ERPs)-can be used to reliably estimate vocabulary knowledge in individuals with Level 3 autism. Five adults with Level 3 autism were tested in a repeated-measures design with two tasks. High-frequency 'known' words (eg, bus, airplane) and low-frequency 'unknown' words (eg, ackee, cherimoya) were presented in a visual world task (during which EM and PD data were collected) and a picture-word congruity task (during which ERP data were collected). Using a case-study approach with single-subject analyses, we found that these implicit measures have the potential to provide estimates of receptive vocabulary knowledge in individuals with Level 3 autism. Participants differed with respect to which measures were the most sensitive and which variables best predicted vocabulary knowledge. These implicit measures may be useful to assess language abilities in individuals with Level 3 autism, but their use should be tailored to each individual.

FIGURE 1

Examples of ‘known’ and ‘unknown’ stimuli.

FIGURE 2

Illustration of the nine electrode clusters used for the EEG analysis.

FIGURE 3

Results for D.L. A: Bar graphs comparing ‘known’ and ‘unknown’ word trials for each of the eye movement variables. B: Comparisons of ‘known’ and ‘unknown’ word trials for each of the pupillometry variables. Error bars indicate standard error of the mean. C: Event-related potential data for all conditions at the nine electrode cluster sites. Negative is plotted up.

FIGURE 4

Results for H.D. A: Bar graphs comparing ‘known’ and ‘unknown’ word trials for each of the eye movement variables. B: Comparisons of ‘known’ and ‘unknown’ word trials for each of the pupillometry variables. Error bars indicate standard error of the mean. C: Event-related potential data for all conditions at the nine electrode cluster sites. Negative is plotted up.

FIGURE 5

Results for W.F. A: Bar graphs comparing ‘known’ and ‘unknown’ word trials for each of the eye movement variables. B: Comparisons of ‘known’ and ‘unknown’ word trials for each of the pupillometry variables. Error bars indicate standard error of the mean. Significant differences between ‘known’ and ‘unknown’ words, based on permutation tests with Bonferroni corrections, are indicated by asterisks (*Significant at P<0.05; **Significant at P<0.01). C: Event-related potential data for all conditions at the nine electrode cluster sites. Negative is plotted up. The orange bar beneath the waveforms indicates significant differences between congruent and incongruent conditions for ‘known’ words, as determined by permutation tests with a cluster-based family-wise error correction at P<0.05.

FIGURE 6

Results for S.E. A: Bar graphs comparing ‘known’ and ‘unknown’ word trials for each of the eye movement variables. B: Comparisons of ‘known’ and ‘unknown’ word trials for each of the pupillometry variables. Error bars indicate standard error of the mean. Significant differences between ‘known’ and ‘unknown’ words, based on permutation tests with Bonferroni corrections, are indicated by asterisks (*Significant at P<0.05). C: Event-related potential data for all conditions at the nine electrode cluster sites. Negative is plotted up.

FIGURE 7

Results for P.B. A: Bar graphs comparing ‘known’ and ‘unknown’ trials for each of the EM variables. B: Comparisons of ‘known’ and ‘unknown’ trials for each of the pupillometry variables. Error bars indicate standard error of the mean. Significant differences or trends toward significance between ‘known’ and ‘unknown’ words, based on permutation tests with Bonferroni corrections, are indicated by asterisks (*Significant at P<0.05; **Significant at P<0.01; ‡Statistical trend at P<0.10). C: Event-related potential data for all conditions at the nine electrode cluster sites. Negative is plotted up. The orange bar beneath the waveforms indicates significant differences between congruent and incongruent conditions for ‘known’ words, as determined by permutation tests with a cluster-based family-wise error correction at P<0.05.

FIGURE 8

Summary descriptions of individual participant data. A: ‘Unknown’−‘known’ difference scores (scaled z scores) for each eye movement and pupil dilation variable. Variables on the left were predicted to be larger for ‘unknown’ than ‘known’ word trials, and so the ‘unknown’−‘known’ difference score should be negative. Variables on the right were predicted to be larger for ‘known’ than ‘unknown’ word trials, and so the ‘unknown’−‘known’ differences should be positive. B: Topographic plots of the event-related potential incongruent – congruent difference for ‘known’ and ‘unknown’ words in 50-millsecond windows from 200 to 800 milliseconds after sound presentation.