THE EMOTIONAL BRAIN AND AGING

The neural processing of emotionally salient events involves many brain regions (Phan, Wager, Taylor, & Liberzon, Reference Phan, Wager, Taylor and Liberzon2002) including automatic processes in the amygdala and controlled processes in the prefrontal cortex (PFC). The amygdala is critical for the detection of emotions and the generation of physiological responses. Furthermore, the amygdala optimizes perception and memory of emotional events through its dense connections with cortical and subcortical areas (Amaral & Price, Reference Amaral and Price1984), including the modulation of memory consolidation (LaBar & Cabeza, Reference LaBar and Cabeza2006). In contrast, the PFC exerts control over emotions (Ochsner & Gross, Reference Ochsner and Gross2005), with medial and lateral subregions supporting different neural processes. First, medial PFC (Brodmann’s Area 10), including nearby anterior cingulate cortex (ACC; Brodmann’s Area 32, 24), coordinates the generation and regulation of affect (Phan et al., Reference Phan, Wager, Taylor and Liberzon2002). Anterior medial PFC regions (particularly BA 10) are also associated with self-referential processes (Amodio & Frith, Reference Amodio and Frith2006), which refers to relating information to one’s self. Medial PFC is part of the default network, which refers to a set of regions that tend to be deactivated during attention demanding tasks and activated during internally focused tasks, such as during rest (Gusnard, Akbudak, Shulman, & Raichle, Reference Gusnard, Akbudak, Shulman and Raichle2001). The default network is implicated in emotionally meaningful events (Cabeza & St Jacques, Reference Cabeza and St Jacques2007) and malfunctions in depressed patients (Mayberg, Reference Mayberg1997; Sheline et al., Reference Sheline, Barch, Price, Rundle, Vaishnavi and Snyder2009). Second, lateral PFC regions (Brodmann’s Area 44/46, 46/9) are more generally related to controlled processes (Miller & Cohen, Reference Miller and Cohen2001) that contribute to emotional regulation (Ochsner & Gross, Reference Ochsner and Gross2005).

Compared with structural and functional decline in other brain regions, the amygdala and some subregions of the PFC might be less vulnerable to aging. First, the structural integrity of the amygdala remains relatively preserved in aging (Good et al., Reference Good, Johnsrude, Ashburner, Henson, Friston and Frackowiak2001; Grieve, Clark, Williams, Peduto, & Gordon, Reference Grieve, Clark, Williams, Peduto and Gordon2005), as does the functional recruitment of this region to novelty (Wright et al., Reference Wright, Negreira, Gold, Britton, Williams and Barrett2008; Wright, Wedig, Williams, Rauch, & Albert, Reference Wright, Wedig, Williams, Rauch and Albert2006). Second, there is evidence for structural integrity of the medial PFC and ACC (Salat, Kaye, & Janowsky, Reference Salat, Kaye and Janowsky2001; also see Raz & Rodrigue, Reference Raz and Rodrigue2006; although see Petit-Taboue, Landeau, Desson, Desgranges, & Baron, Reference Petit-Taboue, Landeau, Desson, Desgranges and Baron1998). In fact, older adults sometimes show less deactivation in medial PFC regions during cognitive tasks (Grady, Springer, Hongwanishkul, McIntosh, & Winocur, Reference Grady, Springer, Hongwanishkul, McIntosh and Winocur2006), which has been linked to their difficulty in disengaging default mode activity during these attention demanding tasks. One interpretation of this finding is that older adults chronically engage self-referential processes (Kensinger & Leclerc, Reference Kensinger and Leclerc2009), which might interfere or benefit performance depending upon the particular cognitive task (Gutchess, Kensinger, Yoon, & Schacter, Reference Gutchess, Kensinger, Yoon and Schacter2007b). In contrast to the medial PFC, aging has a large impact on the structural integrity of the lateral PFC (Raz et al., Reference Raz, Lindenberger, Rodrigue, Kennedy, Head and Williamson2005). Functional neuroimaging studies have shown that lateral PFC regions are over-recruited in aging (Grady, Reference Grady2008). In particular, greater age-related recruitment of the PFC is typically coupled with a reduction in the recruitment of posterior regions (Posterior-Anterior Shift in Aging, PASA; Dennis & Cabeza, Reference Dennis, Cabeza, Craik and Salthouse2008), a pattern of activation linked to compensation (Davis, Dennis, Daselaar, Fleck, & Cabeza, Reference Davis, Dennis, Daselaar, Fleck and Cabeza2008).

In summary, the relative preservation of the structure and function of the amygdala and specific frontal regions in older adults might underlie the maintenance and enhancement of emotional processes in aging.

AGING AND EMOTIONAL PERCEPTION

Functional neuroimaging studies of age-related differences in emotional perception are generally consistent with the FADE pattern. Below we discuss the findings relevant to the amygdala and frontal cortices and the interaction between these regions.

One of the most consistent findings from functional neuroimaging studies of emotional perception is an age-related reduction in amygdala activation for negative stimuli (see Table 1). For example, in the first functional magnetic resonance imaging (fMRI) study of emotion and aging, Iidaka et al. (Reference Iidaka, Okada, Murata, Omori, Kosaka and Sadato2002) found less left amygdala activity for negative faces in older adults when compared with young adults, and other studies have since replicated this finding (Fischer et al., Reference Fischer, Sandblom, Gavazzeni, Fransson, Wright and Backman2005; Gunning-Dixon et al., Reference Gunning-Dixon, Gur, Perkins, Schroeder, Turner and Turetsky2003; Tessitore et al., Reference Tessitore, Hariri, Fera, Smith, Das and Weinberger2005; also see Erk, Walter, Abler, Reference Erk, Walter and Abler2008). However, in the aforementioned studies (Gunning-Dixon et al., Reference Gunning-Dixon, Gur, Perkins, Schroeder, Turner and Turetsky2003; Iidaka et al., Reference Iidaka, Okada, Murata, Omori, Kosaka and Sadato2002; Tessitore et al., Reference Tessitore, Hariri, Fera, Smith, Das and Weinberger2005), older adults reported negative stimuli valence ratings that were less negative than the ones given by young adults and associated with the standardized ratings (Lang, Bradley, & Cuthbert, Reference Lang, Bradley and Cuthbert1997). Thus, one possible interpretation is that the age-related reduction in amygdala activity reflects an inability of the stimuli to elicit strong negative emotions in older adults rather than a deficit in amygdala function. Consistent with this idea, others and we have found that amygdala activity can be as strong in older as in young adults when stimuli were classified according to participants’ own ratings (Leclerc & Kensinger, Reference Kensinger2008a; Mather et al., Reference Mather, Canli, English, Whitfield, Wais and Ochsner2004; St Jacques, Dolcos, & Cabeza, Reference St Jacques, Dolcos and Cabeza2008; Wright et al., Reference Wright, Wedig, Williams, Rauch and Albert2006) or when there were no age-related differences in emotional ratings (Fischer et al., Reference Fischer, Sandblom, Gavazzeni, Fransson, Wright and Backman2005). Furthermore, we (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008) found that amygdala activity was indeed reduced for negative stimuli that older adults subjectively rated as neutral. Interestingly, age-invariant activity in the amygdala is typically observed for positive stimuli (Gutchess, Kensinger, & Schacter, Reference Gutchess, Kensinger and Schacter2007a; Leclerc & Kensinger, Reference Kensinger2008b; Mather et al., Reference Mather, Canli, English, Whitfield, Wais and Ochsner2004; although see Williams et al., Reference Williams, Brown, Palmer, Liddell, Kemp and Olivieri2006) and might reflect a shift in the preference of amygdala activation in aging (Mather et al., Reference Mather, Canli, English, Whitfield, Wais and Ochsner2004; also see Wright et al., Reference Wright, Wedig, Williams, Rauch and Albert2006; Wright, Dickerson, Feczko, Negeira, & Williams, Reference Wright, Dickerson, Feczko, Negeira and Williams2007). In sum, the function of the amygdala remains intact in aging. Below we discuss age-related findings in the frontal cortex and how this might interact with the amygdala findings.

In contrast with the amygdala, functional neuroimaging studies of emotional perception have generally found an age-related increase in the recruitment of the frontal cortex, especially in medial PFC regions (see Table 1), although age-related increases in lateral PFC regions are also prevalent (Gunning-Dixon et al., Reference Gunning-Dixon, Gur, Perkins, Schroeder, Turner and Turetsky2003; Gutchess et al., Reference Gutchess, Kensinger and Schacter2007a; Tessitore et al., Reference Tessitore, Hariri, Fera, Smith, Das and Weinberger2005). We and others (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008; Williams et al., Reference Williams, Brown, Palmer, Liddell, Kemp and Olivieri2006) found an age-related increase in the recruitment of medial PFC during the perception of negative versus neutral pictures, while other studies have observed that the age-related increases in medial PFC recruitment are greater for positive versus negative stimuli (Gutchess et al., Reference Gutchess, Kensinger and Schacter2007a; Leclerc & Kensinger, Reference Kensinger2008a). These apparent differences might be due to the dissociable effects of arousal and valence in dorsal versus ventral aspects of the medial PFC (e.g., Dolcos, LaBar, & Cabeza, Reference Dolcos, LaBar and Cabeza2004). For example, Leclerc and Kensinger (Reference Kensinger2008a) found age-invariant activity in dorsomedial regions reflecting emotional arousal, whereas, the age-related differences observed in ventromedial regions (specifically ventral ACC) reflected differences in positive versus negative valence. Future research is needed to tease apart potential age-related differences in the neural correlates subserving emotional arousal and valence.

In summary, consistent with the FADE pattern, emotional perception studies of aging converge on the observation of an age-related alteration in amygdala activation coupled with an increase in frontal cortex activation, which suggest that older adults engage more controlled processing during the perception of emotional stimuli. It is also critical to understand how the effects of aging on emotional processing ramify through cognitive operations, such as memory processes.

AGING AND EMOTIONAL EPISODIC MEMORY

In young adults, memory is usually better for emotional than neutral stimuli. Although this enhancement for emotional memory is relatively preserved in healthy aging, there is an age-related alteration in the extent of the enhancement, especially for negative information. Only recently have functional neuroimaging studies begun to explore the neural bases of the age-related changes in emotional memory, and the available evidence is consistent with FADE. Below we discuss the age-related findings in the amygdala and PFC.

Despite differences in methodology, all three available functional neuroimaging studies of memory encoding show age-invariant amygdala activity coupled with age-related increases in PFC activity. For example, we (St Jacques, Dolcos, & Cabeza, Reference St Jacques, Dolcos and Cabeza2009) found that amygdala activity predicted subsequent memory of negative versus neutral pictures in both young and older adults, but that older adults recruited additional frontal activity to support memory formation as visual cortex activity declined. Similarly, Murty et al. (Reference Murty, Sambataro, Das, Tan, Callicott and Goldberg2008) found age-related increases in the recruitment of left dorsolateral PFC (DLPFC) during the encoding of negative versus neutral picture blocks, but they did not observe age-related differences in the amygdala. Consistent with behavioral findings of preserved enhancement for emotional memory in aging, Murty et al. (Reference Murty, Sambataro, Das, Tan, Callicott and Goldberg2008) suggested that the increased recruitment of the PFC might reflect compensation. Complementing these findings, Kensinger and Schacter (Reference Kensinger and Schacter2008) observed age-invariant recruitment of the amygdala during successful memory encoding of objects, irrespective of the particular negative or positive valence, but older adults elicited greater medial PFC activity during successful encoding of positive objects. Furthermore, others and we have observed an age-related decrease in functional connectivity between the amygdala and the areas that typically support memory formation (i.e., hippocampus) but an age-related increase in functional connectivity with PFC areas involved in controlled processing (Murty et al., Reference Murty, Sambataro, Das, Tan, Callicott and Goldberg2008; St Jacques et al., Reference St Jacques, Dolcos and Cabeza2009). Currently, only one study (Murty et al., Reference Murty, Sambataro, Das, Tan, Callicott and Goldberg2008) has examined age-related effects during memory retrieval, where they found that older adults showed a reduction in the recruitment of the amygdala during memory retrieval of negative versus neutral blocks coupled with an age-related increase in right DLPFC.

In summary, the results of these initial fMRI studies suggest that aging leads to an increased reliance on controlled processes in the PFC that support and maintain enhanced memory for emotional materials.

THE FADE PATTERN

Evidence from age-related functional neuroimaging studies across both emotional perception and memory domains generally show a pattern of activity consistent with FADE, although some studies provide only partial support potentially due to methodological differences (Fischer et al., Reference Fischer, Sandblom, Gavazzeni, Fransson, Wright and Backman2005; Gutchess et al., Reference Gutchess, Kensinger and Schacter2007a; Iidaka et al., Reference Iidaka, Okada, Murata, Omori, Kosaka and Sadato2002) and to focused region of interest analyses on the amygdala (Mather et al., Reference Mather, Canli, English, Whitfield, Wais and Ochsner2004; Wright et al., Reference Wright, Wedig, Williams, Rauch and Albert2006, Reference Wright, Dickerson, Feczko, Negeira and Williams2007). Below we focus on the evidence that supports the FADE pattern and provide possible interpretations of this empirical regularity.

Although age-invariant activation in the amygdala was observed during both emotional perception and the encoding and formation of emotional memories, suggesting that the function of the amygdala remains intact in aging, the amount of amygdala activity elicited in older adults might differ depending upon the recruitment of controlled processes involving the frontal cortices. There are at least three potential interpretations of the increase in frontal recruitment associated with the FADE pattern. First, the frontal increase could be an instance of the posterior anterior shift in aging (PASA), which is frequently observed in nonemotional domains (Dennis & Cabeza, Reference Dennis, Cabeza, Craik and Salthouse2008). Consistent with this interpretation, others and we have found that the increase in frontal activity is coupled with a decrease in the recruitment of posterior regions (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008, Reference St Jacques, Dolcos and Cabeza2009; Tessitore et al., Reference Tessitore, Hariri, Fera, Smith, Das and Weinberger2005; also see Gunning-Dixon et al., Reference Gunning-Dixon, Gur, Perkins, Schroeder, Turner and Turetsky2003; Iidaka et al., Reference Iidaka, Okada, Murata, Omori, Kosaka and Sadato2002). Furthermore, in keeping with the compensatory account of PASA, we also found a significant correlation between the age-related increase in frontal activity and the reduction in visual cortex activity during emotional perception (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008) and that the age-related frontal increase was predictive of subsequent memory for negative stimuli (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2009).

Second, age-related increases in medial PFC could reflect an augmentation of self-referential processes (Kensinger & Leclerc, Reference Kensinger and Leclerc2009), which are processes previously associated with this region (Northoff & Bermpohl, Reference Northoff and Bermpohl2004). One line of evidence supporting this interpretation is that medial PFC recruitment in older adults has been shown to vary as a function of valence (Kensinger & Schacter, Reference Kensinger and Schacter2008; Leclerc & Kensinger, Reference Kensinger2008a). For example, Leclerc and Kensinger (Reference Kensinger2008a) found an age-related reversal in the medial PFC, such that older adults recruited this region more for positive stimuli and less for negative stimuli (although see Williams et al., Reference Williams, Brown, Palmer, Liddell, Kemp and Olivieri2006). They suggested that older adults might interpret positive stimuli in a more self-relevant way. In keeping with this interpretation, a study directly interrogating age-related differences in self-referential processing for emotional stimuli found that older adults recruited medial PFC regions to a greater extent for self-related positive words (Gutchess et al., Reference Gutchess, Kensinger and Schacter2007a). Given that older adults tend to engage the default network more during cognitive tasks (Grady et al., Reference Grady, Springer, Hongwanishkul, McIntosh and Winocur2006), one possibility is that the positivity shift observed in perception and memory studies might result from an age-related increase in the tendency to interpret information in a self-relevant manner (cf. Kensinger & Leclerc, Reference Kensinger and Leclerc2009).

Third, age-related frontal increases could reflect emotional regulation. Consistent with FADE, in many studies, the age-related increase in frontal activity was coupled with a reduction in amygdala activity during perception (Gunning-Dixon et al., Reference Gunning-Dixon, Gur, Perkins, Schroeder, Turner and Turetsky2003; Tessitore et al., Reference Tessitore, Hariri, Fera, Smith, Das and Weinberger2005; also see Fischer et al., Reference Fischer, Sandblom, Gavazzeni, Fransson, Wright and Backman2005; Iidaka et al., Reference Iidaka, Okada, Murata, Omori, Kosaka and Sadato2002; Samanez-Larkin et al., Reference Samanez-Larkin, Gibbs, Khanna, Nielsen, Carstensen and Knutson2007; St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008; although see Williams et al., Reference Williams, Brown, Palmer, Liddell, Kemp and Olivieri2006) and retrieval (Murty et al., Reference Murty, Sambataro, Das, Tan, Callicott and Goldberg2008) of negative stimuli. Thus, one possibility is that older adults’ enhanced emotional regulation strategies lead to the recruitment of PFC-mediated control processes that dampen amygdala responses for negative stimuli. For example, we examined age-related differences in the functional connectivity with the amygdala and found an age-related increase in the functional connectivity between the ventral ACC and the amygdala (St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008). Importantly, we found a negative correlation between these regions during the perception of negative pictures that older adults subjectively rated as neutral and a subsequent decrease in amygdala activity, which suggests the engagement of emotional regulation. Urry et al. (Reference Urry, van Reekum, Johnstone, Kalin, Thurow and Schaefer2006) directly asked older adults to regulate emotions while viewing negative pictures and found that compared with passive viewing, when older adults were asked to decrease their emotional responses they recruited greater ventromedial PFC activity coupled with a reduction in the recruitment of the amygdala. In fact, emotional well-being in aging has been associated with a shift from automatic processing to more controlled processing of emotions via the recruitment of medial PFC (Williams et al., Reference Williams, Brown, Palmer, Liddell, Kemp and Olivieri2006).

In summary, age-related increases in frontal activity might reflect compensation, self-referential processing, or emotional regulation. It is important to note, however, that these accounts are not incompatible with each other. For example, emotional regulation can be seen as a form of compensation (cf. St Jacques et al., Reference St Jacques, Dolcos and Cabeza2008), and self-referential processing could be seen as an emotional regulation strategy (cf. Kensinger & Leclerc, Reference Kensinger and Leclerc2009). Yet, not all forms of emotional regulation are beneficial for performance (i.e., compensatory) and self-referential processing is not necessarily an effective regulation strategy. Thus, understanding the specific contributions of each of these processes and their interactions is a major challenge for future research.