 |
     |
|
|
|
|
Mapping semantic interference during picture naming: an fMRI study
Tatiana T. Schnur, Elizabeth A. Hirshorn, and Sharon L. Thompson-Schill
The selection of a word for production has been argued to be a competitive process: selection is determined by the word with the highest activation level in comparison to other activated words (Dell, 1986; Levelt, Roelofs, & Meyer, 1999; Roelofs, 1992). The blocked picture-naming paradigm provides a window on production by magnifying the competitive nature of lexical selection. In this paradigm participants take longer to name a repeated series of pictures when the pictures are from the same semantic category vs. mixed categories presumably due to increased competition for lexical selection from semantically related competitors (Damian, Vigliocco, & Levelt, 2001). Thompson-Schill et al. (1998) argued that the posterior portion of the left inferior frontal gyrus (LIFG) subserves selection of representations in the presence of competing alternatives. Consistent with this, anterior aphasics produced more errors in the homogeneous compared to mixed condition of blocked naming (McCarthy & Kartsounis, 2000; Schnur et al., 2004; Wilshire & McCarthy, 2002). In order to test the prediction that the LIFG supports processes involved in the blocked naming paradigm, we measured fMRI activity while healthy participants overtly named semantically homogeneous and mixed pictures. Data showed significantly greater activity of the LIFG and left temporal cortex in semantically homogeneous naming consistent with the hypothesis that the LIFG is involved in the semantic blocking effect. Further investigation is needed to understand whether the activated areas differentially subserve component processes involved in the blocking effect: activation spread to competitors and lexical selection when demand is high.
Effects of frontal lobe damage on verbal fluency and picture naming
Irene P. Kan, Rachel A. Caravella & Sharon L. Thompson-Schill
Selection of an appropriate response among competing alternatives is one of many facets of cognitive control, and the link between this mechanism and the lateral prefrontal cortex has been demonstrated in both neuroimaging and neuropsychological studies. Specifically, it has been suggested that the left frontal operculum (BA 44/45) plays an important role in selection among competitors. In this experiment, we explored selection abilities in patients with unilateral frontal damage in two semantic tasks: verbal fluency and picture naming. During verbal fluency, subjects generated members of given categories, and selection demands were manipulated in terms of response set size. That is, categories with larger set sizes (e.g., "animals") have higher selection demands than categories with smaller set sizes (e.g., "farm animals"). In picture naming, selection demands were manipulated as a function of name agreement, which refers to the extent to which a given picture evokes a single, reliable name. The lower the name agreement, the higher the selection demands. We compared fluency and naming performance of patient NJ, whose focal damage included the left frontal operculum, with three control groups: left frontal patients whose damage did not include that region, right frontal patients, and neurologically intact controls. We found that frontal patients performed within normal range across different selection conditions. Critically, NJ was significantly impaired on both high selection tasks, but he performed within normal range on both low selection tasks. Our data support previous findings that the left frontal operculum is necessary for selection among competing alternatives.
Category-specific differences in conceptual knowledge between congenitally blind and sighted subjects
Connolly, Andrew C., Thompson-Schill, Sharon L., Gleitman, Lila R.
A leading theory of semantic memory, the sensory-functional hypothesis (Warrington & Shallice, 1984), claims that category-specific deficits in patients with focal brain damage arise from damage to modality-specific semantic representations. Accordingly, categories believed to be more dependent on the damaged modality become selectively impaired. Neuroimaging studies have supported this modality-specific view of semantic organization (Thompson-Schill et al., 1999, inter alia). Our study tests this hypothesis using congenitally blind subjects. Specifically, we test the prediction that the conceptual similarity space of blind subjects, compared to sighted controls, will be altered for ≥visual≤ categories (e.g., fruits and vegetables) more so than for ≥non-visual≤ categories (e.g., furniture). Subjects participated in two types of similarity judgment tasks: (1) a triad task where subjects chose the odd-one-out from a triad of auditorially presented words; (2) a card-sorting task where subjects sorted cards each containing a single printed word (in Braille for blind subjects). Similarity scores for items within a given category were derived from the outcome of these tasks. In line with the stated hypothesis, analyses reveal a difference in the similarity space of blind subjects for the visual category fruits and vegetables compared to sighted controls. Cluster analysis using an individual differences additive clusters model (Carroll & Arabie, 1983) reveals that this difference is driven primarily by differences along the color dimension. Thus, although blind subjects can report visual characteristics of common objects, these data indicate that the lack of direct sensory experience with these features alters the conceptual similarity space for certain categories.
Proactive and reactive cognitive control during semantic retrieval
Elizabeth A. Hirshorn, Geoffrey K. Aguirre, Sharon L. Thompson-Schill
Recent studies have revealed a left inferior frontal gyrus (LIFG) response to increasing demands for cognitive control during semantic processing. In some cases, demands for control were manipulated by blocked instructions (e.g., global or specific decision criteria; Thompson-Schill et al., 1997), whereas in other cases, demands for control were manipulated by trial-specific competition among response choices (e.g, strength of association; Wagner et al., 2001). Different cognitive control mechanisms (i.e., proactive or reactive) might be deployed to resolve competition in these two cases. The current study investigated transient and sustained effects of cognitive control on LIFG activity during semantic similarity judgments. In each block, participants chose which of two words is more similar to a target, based on either global similarity or one specific criterion (e.g, color). Within each block, item-specific competition was manipulated by varying global similarity of the foil and target. Crossing these two factors yielded four conditions of interest (e.g., specific-high conflict: deciding that a sunflower is more similar in color than a strawberry to a banana). As a nonsemantic control, subjects made visual identity judgments to false font strings (that also varied in item-specific competition from the foil). Item-specific competition had an effect on transient LIFG responses during global blocks but not during specific blocks (or for nonsemantic blocks). However, sustained LIFG activation in the specific trials was greater than in the identity trials. These results have implications for the role of LIFG in proactive and reactive cognitive control.
The role of selection demands in left inferior frontal gyrus activation during semantic and phonological tasks
Hannah Snyder, Keith Feigenson, Sharon L. Thompson-Schill
Recent studies of the function of posterior, left inferior frontal gyrus (pLIFG) have yielded two quite different interpretations: One hypothesis holds that pLIFG plays a domain-specific role in phonological processing (e.g., Devlin et al., 2003), while another hypothesis describes a more general function of pLIFG in selecting among competing alternatives (e.g., Thompson-Schill et al., 1997). There are numerous reports of activation in pLIFG during phonological processing of words (e.g., Price et al., 1997). However, because word meaning may be accessed automatically, activation during phonological processing of words could reflect demands for selection from competing semantic information. Consistent with this conjecture, comparisons of phonological processing of non-words to semantic processing have not supported the hypothesis that pLIFG is activated by phonological processing (Gold et al., 2002; Barde & Thompson-Schill, 2002). In the current study, we tried to distinguish effects of increasing selection demands from effects of phonological processing by comparing phonological tasks involving words and pronounceable non-words, and semantic tasks with varying selection demands. Participants decided which of two words is more similar to a probe. In phonological tasks, the decision was based on vowel sound; in semantic tasks, the decision was based on either overall similarity or on a specific feature. The fMRI data showed significantly greater pLIFG activation in the phonological word than non-word condition, and in the semantic specific than global condition. These results support the hypothesized role for pLIFG in selection among competing alternatives, and suggest an alternative explanation of previously reported phonology-specific effects in this region.
