The present study aimed to uncover the nature of interactions within a network of structures that are associated with emotion processing and that may help to sustain performance on a delayed match-to-sample face recognition task in AD patients but not in healthy older adults. The set of interactions relating to left amygdala activity identified in the AD patients was found to be distinct from that utilized by the controls. In particular, the results indicated strong, positive increases in left amygdala input and output at the 16-sec. delay in the patients relative to controls. Strong, positive output from the left amygdala and from the left hippocampus to left and then right inferior PFC was identified. However, the left hippocampus was not modulated directly by the left amygdala in either group at the 16-sec. delay, and received only weak input from the amygdala in the control group at the 1-sec. delay. These findings suggest that an implicit emotional mechanism, mediated by amygdala connectivity with emotion-related regions, may underlie performance in the patients.
It is unlikely that brain regions that remain structurally intact are functionally insulated from the effects of damage elsewhere in the brain, whether resulting in the disruption or maintenance of performance (Price & Friston, 2002
). Neurodegenerative diseases, such as AD, may differentially affect functional integrity within individual regions that are structurally compromised, such as the hippocampus and amygdala, as well as within networks of regions, and these changes are not always readily captured with univariate approaches that assess activity within each region separately. The few studies that have applied network analysis to understanding AD have reported degraded interactions between the PFC and posterior regions, including the hippocampus during short-term episodic memory (Grady et al., 2001
), visual regions during face perception (Bokde et al., 2006
; Horwitz et al., 1995
), and parietal regions in the resting state (Azari et al., 1992
; Horwitz et al., 1987
; Wang et al., 2007
). In contrast, successful performance on tests of explicit memory in AD patients appears to rely on interactions among a set of PFC regions in isolation of more posterior regions (Grady et al., 2003
; Stern et al., 2000
). Here we show that increased PFC connectivity may not be the only route by which AD patients compensate for cognitive loss due to neural degeneration early in the course of the disease.
Multiple, reciprocal connections among limbic and paralimbic systems with sensory, perceptual, memory, executive, and action systems form an intricate circuitry by which thought is coloured by valence and arousal. Emotion, in turn, is modulated by the ability to comprehend, attend to, maintain, and strategically encode and retrieve information. The network of influences identified in the AD patients but not in the controls may reflect this interplay, allowing for the emergence of an anterior-posterior circuit of primarily left-hemisphere ‘emotional’ regions to offset the earlier finding of a disconnection between ‘cognitive’ dorsolateral PFC and hippocampus (Grady et al., 2001
). Correlated activity with the amygdala, but not hippocampus, was found to underlie better performance in the patients, even though the hippocampus and amygdala are both affected in early stages of AD. The current study extends this finding by showing that the two structures maintain different patterns of connectivity with other structures.
The MTL memory system, though not considered emotional on its own, is known to be strongly influenced by the emotional content of stimuli during explicit encoding and retrieval via its strong reciprocal connections with the amygdala (LaBar & Cabeza, 2006
). Using SEM, Kilpatrick and Cahill (2003)
provided direct evidence for such an effect: increased activity in the right amygdala led to increased activity in right parahippocampus and right inferior PFC during encoding of emotional film clips in healthy young adults. We found a similar influence of left amygdala activity on left inferior PFC activity, which in turn led to an increase in right inferior PFC activity, but no such influence on the left hippocampus was found, suggesting that an explicit emotional memory strategy in the present study was unlikely.
The network of interactions revealed may instead reflect incidental processing of the emotional content of the faces, which included neutral and happy expressions. The amygdala, anterior cingulate, insula, and inferior PFC have all been associated with emotional face perception and memory (Hornak et al., 1996
; Morris et al., 1998
; Phillips et al., 1997
), even when emotional processing is covert (Critchley et al., 2000
; Dolan et al., 1996
; Whalen et al., 1998
). Left amygdala and inferior PFC function have been associated with a wide range of emotions, including processing of neutral and happy facial expressions (Fitzgerald et al., 2006
). Also consistent with the current results are earlier reports of bilateral amygdala, anterior cingulate, and inferior PFC recruitment during nonconscious or incidental perception of happy faces (Gorno-Tempini et al., 2001
; Killgore & Yurgelun-Todd, 2004
; Williams et al., 2004
). A related possibility is that the AD patients recruited an inhibitory system to suppress any emotional response elicited by the faces that is irrelevant to task performance. This is suggested by recent evidence that the role of left inferior PFC in inhibitory processing extends to control of emotional distraction during the delay period of a short-term memory task for neutral faces (Dolcos et al., 2006
; Dolcos & McCarthy, 2006
). Importantly, activity in this region was correlated with that of the amygdala (Dolcos & McCarthy, 2006
). These findings suggest that the amygdala sends input to inferior PFC to signal the presence of emotional distraction, a result that is consistent with the pattern of connectivity seen in our model. It is also possible that the network revealed in the current study reflects differences in the emotional response of the patients and controls (Ressler & Mayberg, 2007
). We view this explanation as unlikely, however, as none of the participants had a history of anxiety or mood disorder, and there was no suggestion of increased arousal in either group during task performance. Future work is needed to differentiate among these alternatives and to determine whether this altered connectivity directly supports short-term memory in AD.
To conclude, a core network of influences within an emotional circuit was more pronounced in AD patients than in healthy older participants during a delayed match-to-sample face recognition task. The direction of influences from left amygdala and left hippocampus on left and then right inferior PFC, in the absence of direct amygdala-hippocampal interactions, may reflect an implicit signaling of emotional content in the faces to increase their memorability or the need to diminish emotional distraction. To our knowledge, this is one of only two studies to apply effective connectivity analysis to neuroimaging of AD (see Horwitz et al., 1987
), and the first to use this approach to characterize the functional interactions that facilitate memory performance. The shift to “hot” emotional processing to achieve what would normally be under the guidance of “cold” cognitive processing may inform intervention strategies for AD patients in early stages of the disease.