2 Review TRENDS in Cognitive Sciences Vol.9 No.5 May 2005 l ing in emotional appraisal systems such as the amygdala. (i)deliberate attempts to regulate pain in order to e manipulated the amount of attention ional control have sh that limiti on modulation of amygdala responses,have produced but the contexts mechanism activation de when participants attend to and valuate ing than ptual features there has compared with subliminal presentations of negative) Ame hcan faces dging the of conceived 6 happy faces [231, or rating their emotional response to judgments or tasks.For studies of selective attention ther than ing them passively 4 wever, a more important problem might be an over activity to be invariant with res ct to attention to rroborating behavioral or physiological meas & ants judged the gender ort the at en th tion ha ed hous dged the stimuli)has made it difficult to det and/o ppy, and fea [27],judge the ag ure to famous ing highly arousing(painful)stimuli,question people (e.g. sely what processes are being carried out by find clear,but two Summary be w).First,some judgments might impose a Cognitive change ing of e long-ter and me also limits a and reasoning has been a pants might some the primar cognitive i making good/bad evaluations of valenced concepts interactions between prefrontal systems that sup or ion),right atrdlateralpreironta control posterio pants indicated in po ratings that they hade ed (e g visual spatial auditory)information 18 361 In the 22 P activity is also found in ntext lti motio have in a ng how as dis Thes es can be why similar reciprocal PFC-amygdala relationships have can com a nge both to otional response wl en none was ongoing or to regulate an already triggered response emotion uses a distracting ondary task to limit Controlled generation ttention to emotional stimu The use of fcognitive change to generate an emotional e roach has e ined th netral correlate oop task 341,or simply being asked to'think of of anticipatory re that precede expected emotion ve 0 ents ical a implicated in mental state attribution [41],which migh gray orbi ofrontal e orteto emotional stimuli or their features modulates processing in emotional appraisal systems such as the amygdala. Selective attention Several studies have manipulated the amount of attention paid to emotional stimuli by asking participants to selectively judge either their emotional or their perceptual features. These studies, which have focused particularly on modulation of amygdala responses, have produced strikingly discrepant results. On one hand, some studies have shown that amygdala activation decreases when participants attend to and evaluate emotional features, including matching emotional faces or scenes based on emotional labels rather than perceptual features [20,21], viewing supra- as compared with subliminal presentations of (presumably negative) African American faces [22], judging the expression rather than the gender of fearful, angry, or happy faces [23], or rating their emotional response to aversive scenes rather than viewing them passively [24]. On the other hand, studies have shown amygdala activity to be invariant with respect to attention to emotional features when participants judged the gender of fearful faces rather than judging aspects of simultaneously presented houses [19,25], judged the gender as compared with expression of happy and disgusted [26], or happy, sad, disgusted and fearful faces [27], judged the age or trustworthiness of normatively untrustworthy faces [28], or the age or goodness of normatively ‘bad’ famous people (e.g. Hitler) [29]. The reasons for these discrepant findings are not yet clear, but two possibilities stand out (see [18], and Critical Summary below). First, some judgments might impose a greater attentional load, which more strongly limits processing of perceptual inputs and as a consequence also limits amygdala responses (cf. [30]). Second, participants might in some cases actively regulate their responses. In keeping with the latter suggestion, when making good/bad evaluations of valenced concepts (e.g. abortion), right ventral lateral prefrontal cortex (LPFC) activation was found on trials for which participants indicated in postscan ratings that they had exerted control [22]. Right ventral LPFC activity is also found in combination with amygdala deactivation during cognitive change, as discussed below. These results could explain why similar reciprocal PFC–amygdala relationships have been observed when participants judged emotional compared with perceptual properties of stimuli [20,21]. Attentional distraction A second approach to interactions between attention and emotion uses a distracting secondary task to limit attention to emotional stimuli. These studies have focused primarily on responses to pain (however, see [31]), and have found that performance of a verbal fluency task [32], the Stroop task [33,34], or simply being asked to ‘think of something else’ [35] diminishes the aversiveness of pain, reduces activity in cortical and subcortical pain-related regions, including midcingulate cortex, insula, thalamus and periacqueductal gray, and activates orbitofrontal cortex (OFC), anterior cingulate cortex (ACC) and medial and lateral PFC regions related to cognitive control. It is not yet clear, however, whether these activations reflect (i) deliberate attempts to regulate pain in order to facilitate performance of the distractor task and/or (ii) processes supporting performance of that task directly. Critical summary Studies of attentional control have shown that limiting attention to emotional stimuli can limit responses in appraisal systems, but the contexts and mechanisms governing this regulatory effect are not clear. For example, studies of selective attention have used primarily emotional face stimuli whereas studies of distraction have used painful stimuli, confounding type of attentional control and type of stimulus. Furthermore, there has been lack of clarity concerning the underlying processing demands – whether conceived as attentional load or some other type of cognitive operation – imposed by specific judgments or tasks. For studies of selective attention, however, a more important problem might be an overreliance on brain activation changes – in the absence of corroborating behavioral or physiological measures – to support the inference that emotion regulation has taken place. That fact (coupled with the use of low arousal, face stimuli) has made it difficult to determine whether amygdala modulation reflects regulatory success and/or the failure to elicit a strong response. Although studies of attentional distraction have avoided these pitfalls by using highly arousing (painful) stimuli, questions remain about precisely what processes are being carried out by control systems. Cognitive change The use of higher cognitive abilities such as working memory, long-term memory and mental imagery to support learning, judgment and reasoning has been a primary focus of research in cognitive neuroscience. In general, these abilities have been shown to depend upon interactions between prefrontal systems that support control processes and posterior cortical and subcortical systems that represent different types of modality specific (e.g. visual, spatial, auditory) information [8,36]. In the context of emotion, researchers have begun asking how these abilities can be used to construct expectations for, select alternative interpretations of, and/or make different judgments about emotional stimuli [18,36] that can change both behavioral and neural responses to them. Cognitive change might be used either to generate an emotional response when none was ongoing or to regulate an already triggered response. Controlled generation The use of cognitive change to generate an emotional response has been studied in three ways. The first approach has examined the neural correlates of anticipatory responses that precede expected emotional events. Such anticipation has been associated with activation of dorsal medial PFC (MPFC) regions [37–40] implicated in mental state attribution [41], which might reflect cognitive expectations for pleasant or unpleasant experiences, in combination with activation of regions 244 Review TRENDS in Cognitive Sciences Vol.9 No.5 May 2005 www.sciencedirect.com