Autobiographical emotional memories tend to be recalled more readily, with a stronger sense of familiarity, and more vividly than episodic memories of an impersonal nature (Christianson et al, 1991; Winograd and Neisser, 1992). Autobiographical memory is a complex process that includes many subprocesses: arousal, imagery, attention, executive functions, and semantic processing. The interaction of these cognitive processes may influence the salience of specific content upon encoding and retrieval, thereby contributing to the development of a working autobiographical narrative.
Prior work on temporal patterns of emotional autobiographical memory has used structured cue approaches, such as the use of cue words, to question study participants (Crovitz and Schiffman, 1974). However, it may also be important to use open-ended questioning approaches in research studies to understand the influence of salience, which may be modified by cuing. There have been a few emotional autobiographical studies of the chronological recall of memories that are temporally open ended. Escobedo and Adolphs (2010) examined moral memories by asking about something that made participants feel guilty and something that made them feel proud. They found that people remembered negative moral memories stretching back for years, whereas their positive moral memories were more recent. These investigators thought their results could be explained by the temporal self-appraisal theory (Wilson and Ross, 2000). As noted by Escobedo and Adolphs (2010), this theory posits that people tend to distance themselves from past failures.
In another study of the effects of emotional content on the chronology of autobiographical memories, de Vries et al (2001) asked healthy adults to recall the best and worst times of their life. All age groups identified the immediate past as the best times, but the memories about the worst times were widely distributed across the participants’ life span (de Vries et al, 2001).
Converging evidence from psychological studies, such as depth of processing, disease models, and functional imaging, have consistently demonstrated that emotional experiences have a profound influence on a person’s attention, memory, and decision-making (Dolan, 2002). The emotional context of information has a qualitative impact on how information is encoded, with different functional systems recruited that depend on the interaction of the emotional (valence) response and arousal during memory encoding (Steinmetz et al, 2010). Steinmetz and colleagues demonstrated that information with a negative valence in the context of high arousal elicits an increase in the strength of amygdala-induced activation of the inferior frontal gyrus. In contrast, information with a positive valence in the context of high arousal elicits a decrease in the strength of activation of the amygdala-frontal network.
Prefrontal networks play an important role in memory. Another factor that may influence the characteristics of emotionally related memory encoding and retrieval is the age/development of the person who is encoding or recalling the memories (Rubin et al, 1998). Myelination-connectivity of prefrontal networks continues well into adulthood (Courchesne et al, 2000) and thus may affect the encoding and retrieval of emotional memories of young adults. The reminiscence or memory bump is the ability of people over the age of 40 to recall information encoded during adolescence and early adulthood (Rubin et al, 1998). Further, studies of normal aging using diffusion-weighted imaging revealed an anterior-to-posterior gradient of white matter deterioration (eg, Yoon et al, 2008). Volumetric analyses have also demonstrated a longitudinal decline of the prefrontal cortex with aging (Raz et al, 2005). These age-related changes in brain structure could alter the dynamics of salience and retrieval of emotional memories.
In this study, we assessed the influence of valence and intensity on the age of spontaneously recalled happy and sad memories. We anticipated that the valence of these memories might influence the remoteness of the recalled emotional memory.
Thirteen women and seven men, with an age range of 18–63 years (M=34.6, SD=12.89, median=30.5) and a mean education of 17 years (SD=3.5), volunteered to participate in the study. All of the participants were students or staff at university medical centers. All of the participants were right handed as determined by the Edinburgh Handedness Inventory (Oldfield, 1971), with no history of neurologic disease, psychiatric illness, or organ failure as assessed during an interview. In addition, at the time of testing, no participants were depressed, as determined by the ICD-10 (International Statistical Classification of Diseases and Related Health Problems, 10th edition) major depressive episode criteria (Bech et al, 2001).
All of the participants were provided with a brief description of the experimental protocol before they signed a University of Florida informed consent form that had been approved by the university’s Institutional Review Board. Our data were acquired in the context of a study on emotional state influence on spatial attention (not reported here).
Participants were instructed as follows:
Now what I’d like for you to do is to remember a happy memory. Remember a memory of an event that made you happy. Try to remember the memory in as great a detail as possible. Remember the surroundings and what was happening to you. Try to remember the happy memory as vividly as possible.
Similar instructions were given for the sad memory. The order of happy versus sad memory recall was counterbalanced across participants. The description of the happy and sad memories was provided orally by each participant to the researcher. After each memory condition, participants were also asked to rate the intensity of the memory on a scale of 1–6 and to indicate the age of the memory.
We analyzed descriptive statistics first. One outlier was identified (participant reported happy and sad memories >10,000 days old) and was removed from the parametric analyses. The distributions of both the happy and sad emotional memory ages were mildly positively skewed (happy memory skewness=1.004, standard error=0.524; sad memory skewness=1.351, standard error=0.524). Neither distribution showed abnormal kurtosis (happy memory age kurtosis=−0.603, standard error=1.014; sad memory age kurtosis=1.211, standard error=1.014).
Unsurprisingly, the SDs were quite large (see below) and, therefore, we evaluated the data using both parametric and nonparametric approaches. Effect sizes are reported as appropriate to the statistical analysis approach used (see literature for qualitative evaluation of size of effects, eg, Fritz et al, 2012).
To test the relationship between the valence of self-selected memories and the age (remoteness) of the memories, we used a paired-samples t test to compare the ages of the provided happy and sad memories. The participants’ sad memories were significantly older than their happy memories: t(18)=−2.705, P=0.014 (mean age of sad memory=7.6 years, SD=7.6 years; mean age of happy memory=3.4 years, SD=4.3 years); Cohen’s d=0.62 (Figure 1).
Further, of the 20 participants, 16 recalled a more remote sad than happy memory. We also performed a χ2 analysis (0=sad memory older than happy memory, 1=happy memory older than sad memory). This analysis confirmed that a sad memory was significantly more likely to be older (more remote) than a happy memory: χ2(1, N=20)=7.2, P<0.007, φ=0.6. Of note, there were no significant differences in the reported intensity of retrieved happy or sad memories in the present: t(18)=−0.782, P=0.443, nor was there a correlation between the intensity of memories and the reported age of the memories (r=0.284, P=0.248, happy; r=0.145, P=0.555, sad).
Although this sample does not have a large representation of ages in all decades of adulthood, it is distributed across a wide age range (18–63 years). Thus, in a secondary analysis to test the impact of the participants’ ages on differences in the recency-remoteness of reported memories, we used two regression models (dependent variables=time between recalled event and current date of happy and sad memories, predictor=age of participant). The age of the participant was a significant predictor of the age (remoteness) of the reported sad memories: F1,18=7.858, P=0.012, r2=0.316. That is, older participants reported older sad memories. However, age did not predict the recency-remoteness of reported happy memories: F1,18=0.741, P=0.401, r2=0.042. Further, the ratio of happy to sad memory age was not significantly different as a function of age: F1,18=0.211, P=0.652, r2=0.012. This finding was likely a product of the small sample size. Because of the small sample size, we did not analyze differences by sex.
We qualitatively evaluated the themes that participants reported for their self-generated happy and sad memories. Happy memories included relatively recent events such as nice vacations, whereas sad memories were predominantly related to either major life course changes (eg, a participant reported a significant move from country of origin to the United States) or, the most frequent response, the loss of a loved one (death). Table 1 lists the classification/quality of the participants’ emotional memories.
Our study revealed that self-selected sad memories are significantly older (more remote) than self-selected happy memories. In addition, the ratio of happy memories to sad memories is unchanged by a participant’s age, although, overall, the older participants in our study recalled older sad memories but not older happy memories.
Previous research has reported an enhanced memory primacy effect for verbal learning of affective material with a negative valence and an enhanced recency effect for verbal learning of affective material with a positive valence (Demaree et al, 2004; Snyder and Harrison, 1997). These results suggest that material with a positive valence is more likely to remain in working memory, and material with a negative valence is more likely to be encoded in declarative memory networks. In our study of retrograde autobiographical memories, there was no opportunity to use working memory, and yet positive memories were more recent than those with a negative valence. The current study suggests a retrieval salience effect whereby older sad memories are more likely to be retrieved than newer sad memories, and more recent happy memories are more likely to be retrieved than older happy memories.
Bernstein and Rubin (2002) reported that happy memories are most likely to be recalled from events that occurred when people were young adults. Although these results appear to conflict with our results, their questions were centered on the “most important” or “happiest/saddest” memories, whereas in our study, we simply requested a happy and a sad memory. We left the selection to the participant—a method that allows for salience assessment. This difference in cuing method may account for the difference in results. In our study, the request was open ended, requiring the participants to provide their own strategy for retrieval.
Although our study is purely behavioral and does not allow for evaluation of the mechanism of these findings, neurophysiological differences in memory encoding may be relevant. Recalling a particular experience might involve partial reactivation of the functional anatomic network that was active during the original encoding (Buckner and Wheeler, 2001). Negative events including death and illness, such as those recalled by our participants, are often associated with “fight” or “flight” responses that are dependent on portions of the limbic system such as the amygdala. The amygdala is part of the salience network. Retaining negative memories about illness, injury, and death may have adaptive-survival value: “never forget.” In addition, older sad memories are likely to be more novel by virtue of less life experience and, therefore, possibly more traumatic than newer sad memories. For example, when we asked our participants to provide a sad memory, death was the most common theme. Although older people experience more deaths than younger people, both groups recalled old (remote) deaths.
Most of the sad autobiographical memories reported in this study were of deaths of a loved one; grief responses were probably quite strong at encoding. Participants reported intensities of their positive and negative memories similar at the time of assessment; unfortunately, we did not ask them about the intensity (arousal) of the experience at the actual time of the event. Indeed, Bradley et al (1992) presented normal participants with pictures from the International Affective Picture System (Lang et al, 2005) and measured valence and arousal. They found that it was the level of arousal at the time of viewing that most influenced recall.
In 1948, Paul Yakovlev described the basolateral limbic circuit (also called Yakovlev’s network) as including the orbitofrontal cortex, which is connected to the anterior temporal lobe by means of the uncinate fasciculus, and the anterior temporal lobe, which is connected to the amygdala. The amygdala projects to the dorsomedial thalamic nucleus by means of the ventral amygdala-fugal pathway, and the dorsomedial nucleus projects back to the frontal lobe. Yakovlev’s circuit is important in the control of emotions, and emotional arousal/visceral reactions are attenuated by more robust prefrontal networks. LaBar and Cabeza (2006) noted that emotional events attain a privileged status in memory, and the amygdala is a brain structure that directly mediates aspects of emotional learning and facilitates memory operations in other regions, such as the hippocampus.
Because prefrontal networks continue developing well into adulthood, younger people may have greater amygdala arousal to news of a death, illness, and injury than do older people. This might partially explain why the sad memories were of much older events (again, mostly deaths); in other words, arousal/visceral reaction is attenuated by more robust prefrontal networks as we mature. At present, we could find no articles comparing arousal-autonomic/visceral responses to death, illness, and injury in children versus adults, though it is reported that children display a more extended grief response compared to adults in the context of a pet’s death (Jarolmen, 1998). Therefore, this postulate is speculative and will need to be tested in future experiments.
The tendency for aging adults to voluntarily recall older sad autobiographical memories relative to happy memories may be related to the right hemisphere hemi-aging hypothesis (Lapidot, 1987), as negative emotions tend to be more lateralized to the right hemisphere, and it is possible that with right hemi-aging, the impact of, or the learning of, new negative memories may be affected. However, it is notable that, in our small sample, there was no difference in the ratio of positive to sad memories as a function of age.
Several right-left hemispheric dichotomies of emotional processing have been reported. These asymmetries are influenced by various features of emotions. Hemispheric laterality of emotional processing is complex, and a full discussion is beyond the scope of this paper. But, laterality features may be relevant to our results. Emotional communication using prosody and facial expressions appears to be mediated by the right hemisphere (Heilman and Valenstein, 2012; Ross and Monnot, 2011), and emotional communication using propositional language is performed primarily by the left hemisphere. Further, Ross et al (1994) divided emotions into primary (eg, anger, fear, surprise) and social (eg, pity, envy, silly) and provided evidence that social emotions are mediated by the left hemisphere and primary emotions are mediated by the right hemisphere.
In this study, we did not examine social emotions or emotional communication. However, we did study valence, and other researchers have suggested that the right hemisphere, which mediates sympathetic activity, appears to play a dominant role in the mediation of emotions with negative valence, and especially those with high arousal (Heilman and Valenstein, 2012).
In regard to memory, Buchanan et al (2006) studied patients who had undergone right or left temporal lobectomy as well as healthy controls. They found that the healthy controls and the left-temporal lobectomy group generated similar numbers of pleasant and unpleasant memories. In contrast, the right-temporal lobectomy group produced significantly fewer memories of unpleasant events. When memories were further categorized according to pleasantness and intensity, the right-temporal lobectomy group produced significantly fewer unpleasant, high-intensity memories than the other groups. These results are consistent with the notion that the right anteromedial temporal lobe may be preferentially involved in the retrieval of negatively valenced, high-intensity memories.
Charles et al (2003) found that when compared with younger adults, older adults were more likely to forget negatively valenced emotional stimuli, called the “positivity effect” by Mather and Carstensen (2005). Perhaps it is this change in right-hemisphere function with aging that makes it less likely for older adults than younger adults both to encode and to retrieve newer, negatively valenced autobiographical memories.
Although there are several prior studies on the differences in the recall of positive and negative emotional events/information, many focus on psychological explanations of the phenomena (eg, people push sad information away from them). Our discussion focuses on possible neurophysiological mechanisms that may account for the dissociation we report. Although our study did not directly assess brain physiology, we have advanced some hypotheses that may account for the dissociations we found and that are not described in the current literature. Based on the results of our study and the hypotheses raised by these results, we are hopeful that future studies will be able to investigate the neurologic bases of these dissociations. Our study was limited by sample size; thus, results such as the lack of a difference in the ratio of happy to sad memories but a significant difference in the age of sad memories with age may reflect a lack of power. Further, replication is necessary.
Unlike prior studies that evaluated the temporal properties of positive and negative memories, we used an unstructured memory query, allowing us to assess the impact of emotional valence on memory salience. Our results do not necessarily suggest that people have a propensity to remember older sad events or newer happy events better than newer sad events or older happy events, but they do suggest that when asked to recall a happy event and a sad event, people are more likely to reach further back in their memory for a sad event than for a happy one, if they are not given additional context for recall. Future research may focus on evaluating arousal/intensity differences in younger and older people when confronted with sad events, particularly the death of a loved one.
Bech P, Rasmussen NA, Olsen LR, et al. 2001. The sensitivity and specificity of the Major Depression Inventory, using the Present State Examination as the index of diagnostic validity. J Affect Disord. 66:159–164. doi:10.1016/S0165-327(00)00309-8
Bernstein D, Rubin DC. 2002. Emotionally charged autobiographical memories across the life span: the recall
of happy, sad, traumatic, and involuntary memories. Psychol Aging. 17:646–652. doi:10.1037/0882-7918.104.22.1686
Bradley MM, Greenwald MK, Petry MC, et al. 1992. Remembering pictures: pleasure and arousal in memory. J Exp Psychol Learn Mem Cogn. 18:379–390. doi:10.1037/0278-7322.214.171.1249
Buchanan TW, Tranel D, Adolphs R. 2006. Memories for emotional autobiographical events following unilateral damage to medial temporal lobe. Brain. 129:115–127.
Buckner RL, Wheeler ME. 2001. The cognitive neuroscience of remembering. Nat Rev Neurosci. 2:624–634. doi:10.1038/35090048
Charles ST, Mather M, Carstensen LL. 2003. Aging and emotional memory: the forgettable nature of negative images for older adults. J Exp Psychol Gen. 132:310–324. doi:10.1037/0096-34126.96.36.1990
Christianson SA, Loftus EF, Hoffman H, et al. 1991. Eye fixations and memory for emotional events. J Exp Psychol Learn Mem Cogn. 17:693–701. doi:10.1037/0278-73188.8.131.523
Courchesne E, Chisum HJ, Townsend J, et al. 2000. Normal brain development and aging: quantitative analysis at in vivo MR imaging in healthy volunteers. Radiology. 216:672–682.
Crovitz HF, Schiffman H. 1974. Frequency of episodic memories as a function of their age. Bull Psychon Soc. 4:517–518.
de Vries B, Blando J, Southard P, et alKeyon G, Clark P, de Vries B. 2001. The times of our life. Narrative Gerontology: Theory, Research and Practice. New York, New York: Springer; 137–158.
Demaree HA, Shenal BV, Everhart DE, et al. 2004. Primary and recency effects found using affective word lists. Cogn Behav Neurol. 17:102–108.
Dolan RJ. 2002. Emotion, cognition, and behavior. Science. 298:1191–1194. doi:10.1126/science.1076358
Escobedo JR, Adolphs R. 2010. Becoming a better person: temporal remoteness biases autobiographical memories for moral events. Emotion. 10:511–518. doi:10.1037/a0018723
Fritz CO, Morris PE, Richler JJ. 2012. Effect size estimates: current use, calculations, and interpretation. J Exp Psychol Gen. 141:2–18.
Heilman KM, Valenstein E, eds. 2012. Clinical Neuropsychology, 5th ed. New York, New York: Oxford University Press.
Jarolmen J. 1998. A comparison of the grief reaction of children and adults: focusing on pet loss and bereavement. Omega: J Death and Dying. 37:133–150.
LaBar KS, Cabeza R. 2006. Cognitive neuroscience of emotional memory. Nat Rev Neurosci. 7:54–64.
Lang PJ, Bradley MM, Cuthbert BN. 2005. International Affective Picture System (IAPS): Affective Ratings of Pictures and Instruction Manual. Gainesville, Florida: University of Florida.
Lapidot MB. 1987. Does the brain age uniformly? Evidence from effects of smooth pursuit eye movements on verbal and visual tasks. J Gerontol. 42:329–331.
Mather M, Carstensen LL. 2005. Aging and motivated cognition: the positivity effect in attention and memory. Trends Cogn Sci. 9:496–502. doi:10.1016/j.tics.2005.08.005
Oldfield RC. 1971. The assessment and analysis of handedness: the Edinburgh Inventory. Neuropsychologia. 9:97–113. doi:10.1016/0028-3932(71)90067-4
Raz N, Lindenberger U, Rodrigue KM, et al. 2005. Regional brain changes in aging healthy adults: general trends, individual differences and modifiers. Cereb Cortex. 15:1676–1689. doi:10.1093/cercor/bhi044
Ross ED, Homan RW, Buck R. 1994. Differential hemispheric lateralization of primary and social emotions. Neuropsychiatry Neuropsychol Behav Neurol. 7:1–19.
Ross ED, Monnot M. 2011. Affective prosody: what do comprehension errors tell us about hemispheric lateralization of emotions, sex and aging effects, and the role of cognitive appraisal. Neuropsychologia. 49:866–877.
Rubin DC, Rahhal TA, Poon LW. 1998. Things learned in early adulthood are remembered best. Mem Cognit. 26:3–19. doi:10.3758/BF03211366
Snyder KA, Harrison DW. 1997. The affective Auditory Verbal Learning Test. Arch Clin Neuropsychol. 12:477–482. doi:10.1093/arclin/12.5.477
Steinmetz KR, Addis DR, Kensinger EA. 2010. The effect of arousal on the emotional memory network depends on valence
. Neuroimage. 1:318–324.
Wilson AE, Ross M. 2000. The frequency of temporal-self and social comparisons in people’s personal appraisals. J Pers Soc Psychol. 78:928–942. doi:10.1037/0011-35184.108.40.2068
Winograd E, Neisser U, eds. 1992. Affect and Accuracy in Recall: Studies of “Flashbulb” Memories
(Emory Symposia in Cognition, Book 4). New York, New York: Cambridge University Press. doi:10.1017/CBO9780511664069.005
Yakovlev PI. 1948. Motility, behavior and the brain: stereodynamic organization and neural correlates of behavior. J Nerv Men Dis. 107:313–335.
Yoon B, Shim YS, Lee KS, et al. 2008. Region-specific changes of cerebral white matter during normal aging: a diffusion-tensor analysis. Arch Gerontol Geriatr. 47:129–138.