Effects of emotion on word perception

In the last few years, human brain mapping and ERP studies revealed that extrastriate cortex regions located in the ventral and lateral parts of the inferior temporal lobe respond selectively to

words (e.g., Cohen et al., 2000; Gaillard, Naccache, Pinel, Clemenceau, Volle et al., 2006b;

Nobre et al., 1994; Petersen, Fox, Snyder, & Raichle, 1990). Neural activity in these visual brain regions is sensitive to a word`s lexical and semantic features (e.g., Gaillard et al., 2006b; Nobre et al., 1994), to the degree of attention elicited by the presented words (e.g., Nacchache, Blandin,

& Dehaene, 2002), occurs within the first 200 – 300 ms after word presentation (e.g., Gaillard et al., 2006b; Martin-Loeches et al., 2001) and presumably reflects early semantic processing (e.g., Martin-Loeches et al., 2001; Nobre et al., 1994; Vandenberghe et al., 1996).

Lesion studies and imaging data imply that the amygdala facilitates the recognition of emotional words in the visual cortex. Most of these studies employed unpleasant and neutral words. Studies with amygdala-lesioned patients show that amygdala damage impairs enhanced perceptual awareness for aversive words in an attentional blink paradigm: Whereas healthy subjects are better at detecting a second target occurring in close proximity to a first target stimulus if it was an emotionally aversive word (e.g., rape, bastard), damage to the amygdala abolishes this effect (Anderson et al., 2001). The authors hypothesized that the amygdala modulates attentive processing of emotional words by enhancing activity in brain regions associated with word recognition. Imaging studies with healthy subjects corroborate this hypothesis reporting enhanced activity in the amygdala and the visual cortex during attentive processing of emotional words. Isenberg and colleagues (1999) found enhanced bilateral amygdala and left parahippocampal/lingual and fusiform gyrus activation for threat words in contrast to neutral words in a modified emotional stroop paradigm. Additionally, threatening words activated premotor regions (BA 6) in the left frontal cortex, indicating preparation for action in response to perceived threat. Using a shallow encoding (subjects had to decide whether the first letter of a noun had an enclosed space) and a deep encoding condition (semantic decision of whether nouns belong to a living or non-living category), Strange and colleagues (2000) reported increased activity in the left amygdala and the fusiform gyrus, bilaterally, as well as in the left prefrontal cortex for emotionally deviant unpleasant nouns in an oddball paradigm. Prefrontal cortex activation was strongest during semantic encoding. By contrast, activation in the amygdala and the fusiform gyrus occurred during both encoding conditions supporting the view that early perceptual processing of unpleasant nouns is independent from depth of processing. Yet another fMRI study convinces of amygdala engagement in facilitated perception of emotional unpleasant words. Tabert and colleagues (2001) found increased right amygdala activity for unpleasant words (nouns and adjectives) being correlated with increased BOLD responses in bilateral secondary and primary visual cortices (BA 17, 18, 19) during an emotional decision task.

Subjects (females) had to respond to the most unpleasant or neutral words. Time course analysis

revealed sustained amygdala and visual cortex activation over blocks of unpleasant word presentations indicating enhanced perceptual vigilance to words with strong unpleasant connotations (e.g., cancer, kill, ugly, agony, hell, etc.).

These findings demonstrate impressively that the amygdala enhances selective processing of emotional words by its modulatory influence on word perception and attention in the visual cortex. The findings also suggest that input from successive stages of visual word processing in the ventral visual processing stream is sufficient for the amygdala to further emotional word recognition and word encoding. At present, results are constrained to the processing of unpleasant, highly aversive words. Of course, there are imaging studies that have employed unpleasant and pleasant words (e.g., Hamann et al., 2002). However, many of these studies were designed to examine effects of emotion on semantic and memory processing (e.g., Cato et al., 2004; Crosson et al., 1999, 2002; Maddock, Garret, & Buonocore, 2003).

Only a paucity of imaging studies using both unpleasant and pleasant words aimed to investigate emotion effects on word perception and attention and results appear less consistent than those reported in studies using aversive words only. Using a blocked stimulus design, both emotionally unpleasant and pleasant words enhanced activation in bilateral occipital cortices and the left inferior temporal lobe when emotional words were contrasted against baseline conditions of viewing plus signs (Beauregard et al., 1997). In contrast to viewing blocks of concrete and abstract but emotionally neutral words, emotional words additionally activated areas in the left orbito-frontal and medio-frontal cortex suggesting an impact of fronto-limbic structures in the evaluation of emotional words. In this positron emission tomography (PET) study male student subjects were explicitly instructed to simply view the words ‘without thinking too much’ about the presented items (cf. Beauregard et al., 1997; pp. 459). Prior to word viewing participants were warned about the upcoming word category by the presentation of a plus sign indicating whether a word (abstract, concrete or emotional) or a random letter string will be expected.

Emotional categories were not analyzed for differential effects between unpleasant and pleasant words. One frequently used paradigm to explore effects of affective word recognition is the lexical decision tasks (LDT) (e.g., Siegle, 1996 for an overview). Using fMRI and LDT, two functional imaging studies investigated brain responses elicited by words with unpleasant, pleasant and neutral valence and pronounceable nonwords. One study investigated healthy student subjects of both sexes (Kuchinke et al., 2005) and one study has focused on between-group effects comparing depressed females against normal controls (Canli, Sivers, Thomason, Whitfield-Gabrieli, Gabrieli et al., 2004). Across studies reaction time data of healthy subjects indicated significantly faster discrimination of pleasant as opposed to unpleasant and neutral

words. In the fMRI, better discrimination of emotional compared to neutral words was associated with stronger increase in the amygdala for pleasant words in healthy subjects than in depressed females (Canli et al., 2004) and in a number of other limbic (e.g., orbito-frontal cortex, hippocampus) and cortical (e.g., inferior and superior prefrontal cortex, anterior and posterior cingulate cortex and middle temporal cortex) brain regions in the second study yielding enhanced semantic and memory encoding of pleasant in contrast to unpleasant words (Kuchinke et al., 2005). The time it takes to make a lexical decision is thought to be a measure of lexical and semantic access and the cognitive effort/resources expended on the task (e.g., Balota &

Chumbley, 1984; Coltheart, Davelaar, Jonasson, & Besner, 1977). Thus, when LDT paradigms are used in imaging studies, brain activation patterns may reflect the interaction of several processes associated with lexical and semantic processing and decision making. Accordingly, the results of the affective LDT paradigm studies described above suggest that these processes are enhanced for emotional, particularly pleasant words, in contrast to unpleasant and neutral words and associated with enhanced activation in cortico-limbic structures.