Neurobiology of Emotional Dysfunction in Schizophrenia: New Directions Revealed Through Meta-Analyses

A ffective dysfunction is a prominent feature of schizophrenia psychopathology. Behavioral studies converge on several distinct aspects of emotional dysfunction, namely, 1) emotion expression (1–3); 2) recognition of facial expressions and emotional classification (4,5); and 3) anticipation of hedonic experience (6,7). However, precise neural deficits across these domains remain obscure. For a decade, neuroimaging studies have probed affective dysfunction in schizophrenia under a variety of conditions, but no clear consensus emerged. These equivocal findings revealed a clear need for meta-analytic summaries of reported effects: over the past year, three groups have independently conducted meta-analyses of the functional neuroimag-ing literature of emotional processing in schizophrenia (8 –10), with somewhat different, but complementary goals: 1) Li et al. examined evidence for differences in activation peaks across studies of facial emotion; 2) Anticevic and colleagues quantified the magnitude of activation differences in the amygdala; and 3) Taylor and colleagues expanded the search across all cortical and subcortical regions. The three studies highlight how slight variation in meta-analytic strategies can produce both complementary and divergent results. For instance, Li and colleagues reported reduced amygdala and fusiform gyrus activity in schizophrenia in response to emotional faces. However, they employed a version of activation likelihood estimation that samples activation peaks across studies, but treats each peak as independent. Such results can be biased by studies that report a greater number of peaks (11), although a recent modification of activation likelihood estimation corrects this problem (12). In contrast, the region of interest– based effect-size analysis employed by Anticevic et al. allowed in-depth probing of a given region—the amygdala— and permitted a search for variables that moderate the magnitude of observed effects in this region. This study confirmed reduced amygdala activation in schizophrenia but did not examine differences elsewhere in the brain. The most recent study, that by Taylor and colleagues, expands on the aforementioned strategies via multilevel kernel density analysis, a tool designed for voxelwise examination of task contrasts (11). Although this approach does not measure the magnitude of between-group effects in a given study, it can reveal regions that show consistent differences across the entire brain, while overcoming peak bias. Such an approach has the potential to provide a window on novel aspects of compromised circuitry involved in emotional processing in schizophrenia. The overall pattern of results reported by Taylor and colleagues was critical in that it ruled out the possibility that all differences in the …