Many factors influence which moments from our past are remembered best, and the affect experienced during an event is an important contributor. As Brown and Kulik (1977) noted in their seminal paper on “flashbulb memories,” events of emotional significance are more likely to be recalled vividly than mundane experiences, and neurobiological research has confirmed that the occurrence of affective responses can increase the likelihood that an event is stored in memory (LaBar & Cabeza, 2006; McGaugh, 2004). Yet my own research, and that of other laboratories, has revealed that there is a complex interplay between affect and memory. This research has suggested that universal memory enhancement is not the best description for how affective responses influence memory; rather, affective responses often lead to memory trade-offs, enhancing memory for select features of an event while impairing memory for other aspects. Moreover, even when events elicit similar intensities of affect, the valence of the experience (whether it is positive or negative) can impact the details remembered. The impact of valence on memory is further complicated when we add aging into the mix: We have found that an adult's age impacts some, but not all, effects of valence on memory (Mather & Carstensen, 2005). Below, I will discuss some of the research leading to each of these revelations.
Affective Responses Lead to Memory Trade-offs
Most memories are incomplete. Even a seemingly vivid memory of a dinner at a restaurant is unlikely to include details such as the color of the napkin, the design of the silverware, or the earrings of the waitress. Memories of arousing experiences are no exception to this rule. Despite labeling “flashbulb memories” as such, Brown and Kulik (1977) noted that these emotional memories were not wholly like a photograph: “An actual photograph, taken by flashbulb, preserves everything within its scope; it is altogether indiscriminate. Our flashbulb memories are not…In short, a flashbulb memory is only somewhat indiscriminate and is very far from complete.” (Brown & Kulik, 1977, pg. 75). Not only are emotional memories susceptible to omissions, but the elicitation of an affective response may serve to increase the likelihood that details are forgotten. As one example, a “weapon focus” effect has been described, with witnesses to a crime remembering the weapon used but not other aspects of the crime scene (reviewed by Kihlstrom, 2006; Heuer & Reisberg, 2007). Empirical studies have confirmed that memory can be narrowed onto select aspects of an emotional event, impairing memory for temporally or spatially proximate information (reviewed by Holland & Kensinger, 2011; Levine & Edelstein, 2009; Mather & Sutherland, 2011; Reisberg & Heuer, 2004). For instance, after studying an image of a snake in the forest, participants have good memory for the snake but poor memory for the forest. In fact, their memory for the forest is worse if they saw an emotional item, such as a snake, than if they saw a neutral item, such as a squirrel, in the forest (Kensinger et al., 2005; Kensinger, Gutchess, & Schacter, 2007). My colleagues and I have referred to this memory pattern as a trade-off, because the memory for the background context (the forest) seems to be traded in favor of memory for the emotional item (the snake). This memory trade-off can occur for positive as well as for negative stimuli, and it manifests in adults of all ages (Kensinger, Gutchess, & Schacter, 2007; Waring & Kensinger, 2009).
Using functional magnetic resonance imaging (fMRI) to examine the neural activity engaged during the initial processing of these scenes, we have found that increased engagement of regions associated with affective processing – including the amygdala and orbitofrontal cortex – corresponds with a person’s ability to remember the emotional item but does not boost memory for the associated background (Waring & Kensinger, 2011; see also Kensinger, Garoff-Eaton, & Schacter, 2007b). Thus, affective processing during encoding seems to lead to selective memory benefits and not to a universal benefit in memory for all event details. In fact, we have proposed that affectively orienting toward material during encoding may enhance the probability of a trade-off being revealed, whereas processing the scenes in a more elaborative or controlled fashion during the initial study phase may enable a more balanced encoding of all scene elements (Kensinger, Gutchess, & Schacter, 2007; Steinberger, Payne, & Kensinger, in press; Waring et al., 2010).
These findings emphasize that the way the event is initially processed will affect whether a trade-off is revealed, yet memory trade-offs are about more than what happens as the event is initially experienced. The discrepancy in memory for the emotional item and for its background tends to increase over a delay, particularly if that delay includes time spent asleep (Payne et al., 2008), revealing that the trade-off can evolve over time. Moreover, a single night of sleep is sufficient to change the circuitry engaged during retrieval; after a night of sleep there is increased activity and connectivity within the amygdala and the ventromedial prefrontal cortex – two regions consistently implicated in affective processing – than after a day spent awake (Payne & Kensinger, 2011). These results suggest that the trade-off emerges both because of how information is initially processed and also because of how that information is stabilized in memory. The occurrence of affective responses does not seem to create a “picture perfect” memory, but rather serves to ensure only that the affectively-relevant aspects of an experience become part of a lasting memory representation.
Valence-Based Effects on Memory Specificity
Although the trade-off can occur for both positive and negative stimuli, other aspects of memory differ depending on the valence of the stimuli. Across a number of studies, my colleagues and I have noted that memory for negative information often includes more item-specific visual details than memory for positive or neutral information. People have a hard time remembering which specific balloon or butterfly (both positive) they have seen, whereas they find it relatively easy to remember which snake, or gun, or dirty toilet they have seen (Kensinger et al., 2006; Kensinger, Garoff-Eaton, & Schacter, 2007a). Negative items tend to be encoded with more perceptual processing than positive information (Mickley & Kensinger, 2008; Mickley Steinmetz & Kensinger, 2009), which may explain this mnemonic benefit for their visual details.
Interestingly, our research suggests that there is a minimum duration of affective engagement or of visual processing required before mnemonic modulation occurs. The benefit in memory for negative items typically occurs only if participants are given sufficient time to process the items; the benefit exists when stimuli are presented for 500ms each but not for 250ms (Kensinger et al., 2006). However, the necessary duration may be reduced if stimuli are presented in a way that optimizes affective processing and the encoding of visual details. In an attempt to optimize these processes, we took advantage of the fact that the right hemisphere has been proposed to be critical for the processing of negative affect (e.g., Davidson, 1992) and also for the processing of visually-specific exemplar details (e.g., Kosslyn, 1987; Marsolek, 1995). We postulated that if information were presented directly to the right hemisphere (via presentation in the left visual hemi-field), this would optimize both the affective processing of the negative stimuli and also the processing of the item’s visual details, leading to a mnemonic benefit for negative information even at a rapid 250ms presentation speed. Indeed, this was the result obtained (Kensinger & Choi, 2009). Using fMRI, our research has revealed that there is a stronger link between the engagement of the right fusiform gyrus and the successful encoding of item-specific visual details for negative items than for positive ones (Kensinger & Schacter, 2008), and more generally, that negative information enhances the connectivity between the amygdala and the fusiform gyrus, a region important for visual processing (Addis et al., 2010). These findings connect clearly to the behavioral findings, insofar as fusiform activity is often associated with memory for exemplar details (Garoff-Eaton, Slotnick, & Schacter, 2006). More generally, the findings confirm that the valence of an experience can impact the way an event is initially encoded into memory, thereby affecting the details that can later be remembered about the event. Our ongoing research is examining how these valence-based differences at encoding affect the way that information is stabilized in memory or later retrieved.
Valence and Age Sometimes Interact to Influence Memory
There appears to be much continuity across the adult lifespan in how affective responses guide memory. Adults of all ages show memory trade-offs, remembering emotional items while forgetting their background contexts (Kensinger et al., 2005; Kensinger, Gutchess, & Schacter 2007; Waring & Kensinger, 2009; see also Denburg et al., 2003). The enhanced retention of item-specific details of negative information may also be preserved across the adult lifespan. Older adults show a better ability to distinguish seen from imagined items (Kensinger, O’Brien, Garoff-Eaton, & Schacter, 2007) and to distinguish among different item exemplars (Kensinger, Garoff-Eaton, & Schacter 2007a) when the items are negative rather than positive. They also show enhanced recruitment of the fusiform gyrus during the encoding of negative as compared to positive items (Kensinger & Schacter, 2008) and strong connectivity between the amygdala, hippocampus (necessary for long-term memory), and fusiform gyrus during the encoding of negative items (Addis et al., 2010).
Despite these similarities, older adults sometimes show a “positivity effect” in memory, remembering proportionally more positive events than younger adults (Mather & Carstensen, 2005). This effect is not always seen, and there has been significant discussion about the sets of circumstances that may lead to its occurrence (Murphy & Isaacowitz, 2008; Kensinger, 2009). My colleagues and I have noted that this interaction between valence and age is more likely to occur if participants only need to remember the “gist” or the general theme of previously presented information. If they need to remember the item-specific details, older adults often perform more similarly to younger adults by showing a memory enhancement for negative as compared to positive and neutral items (Kensinger, Garoff-Eaton, & Schacter, 2007a, 2007b).
In our studies, when the interaction between age and valence on memory performance occurs, it has been paralleled by an interaction between age and valence on recruitment of the medial prefrontal cortex (Kensinger & Schacter, 2008; Leclerc & Kensinger, 2011), a region implicated in self-evaluative processing. Whereas young adults tend to recruit the medial prefrontal cortex more during the processing of negative than positive information, older adults show the opposite effect. Middle-aged adults seem to display an intermediate pattern, recruiting the medial prefrontal cortex equally for positive and negative information (Allard & Kensinger, 2011). Ongoing research in our laboratory is aimed at identifying the reason for this shift. Our working hypothesis is that older adults are more likely than young adults to think about positive information in a self-referential and affective-evaluative way, though it remains possible that age-related differences in emotion regulation (discussed by Mather, 2006; Williams et al., 2006) explain the effects.
Summary
Whether it is a harrowing drive along a treacherous road or a joyous celebration of a friend’s accomplishment, life is filled with moments that elicit affective responses. These responses, in turn, can influence the details that are remembered. Although we often think of affect as enhancing memory, it may be better to think about memory trade-offs as the rule-of-thumb, with affect leading to a combination of memory enhancements and deficiencies. Still, there is variability in how affective responses guide memory, some of which is related to the type of affective response experienced. I have focused on the importance of valence here, but other types of motivations and goals may play an essential role as well (discussed by Levine & Edelstein, 2009). Other variability is linked to the types of details that must be recalled, and to the age of the person recalling those details. These findings underscore the complexity of the associations between affect and memory. While we seem to have a long road ahead of us to understand this interplay, I believe it is an important road to follow. I am motivated by the basic scientific importance of these issues because understanding how affective responses influence memory processes is a prerequisite to fully understanding how human memory operates in daily life, but I also hope that this research will have relevance to understanding the memory biases and distortions that accompany many affective disorders.