Qualitatively different forms of pure alexia

Abstract

In this study we investigated two patients with pure alexia, F.C. and L.D.S., in order to make inferences about how processes and levels involved in the early stage of visual word recognition are organized and how they can be selectively damaged. Moreover, we investigated whether pure alexia can be caused by different functional deficits. F.C. and L.D.S. were presented with tasks of letter processing and tasks of orthographic integration. There was a clear double dissociation between the pattern of performance of F.C. and L.D.S. F.C. was able to process single letters rapidly and accurately, but was unable to group together the letters that he had correctly identified. By contrast, L.D.S. was slower and more impaired at letter identification, but she could use letter groups to assist reading. Thus, two different forms of pure alexia emerged: F.C. has a higher level deficit in integrating letters, whereas L.D.S. has a lower level deficit in letter processing. The results support the assumption of a functional organization of the reading process that involves a series of orthographic units (i.e., single letters, sublexical letter groups, and the lexical unit), which can be selectively damaged. Finally, our data present difficulties for models of pure alexia that assume all patients to have a low-level processing deficit.

We are grateful to F.C. and L.D.S. and their matched controls for their active cooperation and patience over many hours of testing. We wish to thank Dr. Lucia Tedesco for enabling us to make contact with F.C. and the neuroradiologists Dr. Marco Grimaldi and Dr. Andrea Falini for their assistance with MRI scans. We should also thank Prof. Rick Hanley and two anonymous referees for their helpful comments on early drafts of the paper.

Notes

¹ For example the word “take” is represented by activation of units (bigrams) representing TA, TK, TE, AK, AE, and KE.

² Type 1 patients who had a disconnection deficit (Patterson & Kay, 1982) showed a high proportion of “clear misidentification errors” (M.W. 40%, C.H. 69%).

³ In particular, Behrmann et al. (1998) wrote (p. 23): “The findings from the review [of 57 published cases of LBL reading] suggests that there is no single participant for whom letter recognition is definitely normal.”