ABSTRACT
According to most models of language production, to name a picture one must first map semantic features onto lexical items. Even if both sets of representations are intact, problems in mapping semantic to lexical representations can impair production. Individuals with this problem, sometimes referred to as “access deficit”, often demonstrate evidence of preserved semantic knowledge (e.g., good comprehension), increased rate of lexical (usually semantic) errors in production, and inconsistent accuracy on naming the same picture on different occasions. In this paper, I argue that access deficit can have two distinct etiologies. I will present a case of double dissociation between two individuals with chronic post-stroke aphasia, one of whom shows a profile compatible with impaired activation of the target lexical item from semantic features (activation deficit), while the other shows a profile compatible with impaired inhibition of competing lexical items (inhibition deficit). These results have three key implications: (a) they provide support for the theoretical separation between activation and selection processes in computational models of word production, (b) they point to the critical role of inhibitory control in lexical selection, and (c) they invite a closer inspection of the origin of semantic errors in individuals with access deficit in order to choose the best treatment option.
Acknowledgments
I would like to thank Michael Freund for collecting some of the data, Christopher Hepner for his valuable input, and Dr. Brenda Rapp for sharing the scans of the XR and QD with me. This work was supported in part by the Therapeutic Cognitive Neuroscience Fund, and in part by a Science of Learning Initiative grant from Johns Hopkins University awarded to NN. The scans were collected under the grant NIH- DC012283 awarded to Brenda Rapp.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 Note that while semantic similarity increases the activation of the competitors, in systems with feedback from lexical to semantic representations, the activation has the greatest convergence on the target.
2 The reason for examining the effect of semantic similarity in the reversed as opposed to straight naming was that the two participants were fairly high-level, and made few errors in the straight conditions (i.e., there was a ceiling effect in accuracy). The reversed condition increased task difficulty and allowed differences to show up.