On the Interpretation of Across-Patient Variability

Abstract

Consider the sets of errors displayed in Table 1. They were produced by three patients who premorbidly were adequate spellers of their language but developed a spelling disorder after sustaining brain damage. At face value, there is a close stimulus—error relationship in all the reported examples. All the incorrect responses are nonwords, and most of them are very close to the stimulus item (they can be construed as letter substitutions, insertions, deletions and transpositions). However, other features of these errors suggest that they might differ under crucial respects. For example, if the incorrect spelling responses are read aloud, the errors produced by I.G.R. “sound very similar” to the stimulus, the errors produced by patient J.G. “sound identical” to the stimulus, and the errors produced by patient L.B. “sound somewhat similar” to the stimulus. Should more relevance be given to the similarities or to the differences observed across patients? A motivated answer to this question can be provided only if the investigation proceeds under explicit assumptions, that unequivocally legitimate the conclusion in favor of one of the two alternatives. The aim of the present chapter is to demonstrate that an explicit cognitive model is a necessary prerequisite for theoretically driven analyses of pathological behavior, and in particular for the principled interpretation of across-subject performance variation.

This research was supported in part by NIH grant NS 22201 nad by a grant from the Ministro della Sanità. I also wish to thank Roberta Ann Goodman and Alfonso Caramazza for making available the corpus of errors produced by patient J.G.