Nice to hear from you again, MBM!
Regarding the change in scores: remember that testing at just six is not yet considered entirely stable. Also, you note a Dx of dyslexia. Both of those can affect how his VCI unfolds over time. Additionally, the larger context is that his GAI appears likely to be quite similar to past testing (assuming BD remained somewhat similar); the more general composites (e.g., FSIQ, GAI) tend to be more stable over time than the lower level composites or subtests.
As to how dyslexia may affect the VCI over time: as we move up through the age norms, more and more of the VCI is derived from language acquired by reading, and not through oral language, which means that, for many individuals with discrepancies between their verbal cognition and their reading achievement (aka, dyslexia), their ability to reason abstractly with language begins to diverge from their access to acquired language skills (rich vocabulary, complex language structures found mainly in print, knowledge acquired from text), due to (relatively) restricted access to text. Of the two subtests comprising the VCI, one is mainly a measure of reasoning (Similarities), and one is mainly a measure of acquired knowledge (Vocabulary)--but both are affected by vocabulary. Consequently, Vocabulary usually gets dragged down more than Similarities, but both can be depressed by prolonged limitations in access to text. Typically, I see this divergence become more pronounced later (generally at the secondary level), but in a high cognitive young learner, the comparison group is other young high cognitive learners without unexpectedly lower reading ability, who are probably accessing secondary level text. So the effect shows up much earlier.
IOW, the VCI could be lower because it really is lower (i.e., the earlier result was on the high side, and he is now regressing to his own mean), or because it is being artifactually lowered by his relatively limited access to text. Or some other reason.
Do you know the subtest scores? Were other measures of language cognition or achievement administered? I could probably offer higher-quality hypotheses on the VCI-FRI difference with a little more detail.
