As we approach older age, we tend to view the world more positively. Younger adults often prioritize negative information over positive information (i.e., a negativity bias; Carstensen and DeLiema, 2018; Barber and Kim, 2021), while older adults are less likely to show this bias and can even show the opposite tendency to preferentially process, attend to, and remember positive over negative information (i.e., a positivity bias). Moreover, even in the face of physical and cognitive declines, healthy aging is associated with more positive life experiences and better psychological well-being (Höglund et al., 2020; Korkmaz and Güloğlu, 2021).

Previous studies find that older adults tend to intrinsically orient their attention toward positive stimuli and away from negative stimuli (Mather and Carstensen, 2003; Rösler et al., 2005; Isaacowitz et al., 2006, 2009; Knight et al., 2007). For example, older adults display gaze preferences for happy faces but tend to look away from fearful faces (Noh and Isaacowitz, 2015; Erbey et al., 2020). Research evaluating an event-related potential (ERP) sensitive to attentional engagement to emotional information and emotion regulation, known as the late positive potential (LPP; Hajcak and Nieuwenhuis, 2006; Ferrari et al., 2008; De Cesarei and Codispoti, 2011), found that the amplitude of the LPP decreases in old age when evoked by negative stimuli, yet the LPP amplitude for positive stimuli remains invariable across the lifespan (Wood and Kisley, 2006; Kisley et al., 2007; Langeslag and van Strien, 2009). While young adults show greater LPP amplitude for negative information than positive information, older adults show comparable LPP amplitude in response to both negative and positive stimuli (Wood and Kisley, 2006; Kisley et al., 2007; Mathieu et al., 2014). These behavioral and ERP studies emphasize that the negativity bias commonly seen in younger adults diminishes as individuals age, through a modulation of attention and affective responding to negative stimuli.

The age-related positivity effect also shows up in memory. A meta-analysis demonstrated that older adults are more likely to remember positive than negative stimuli compared to younger adults (Reed et al., 2014). This positivity effect was consistently observed using incidental encoding procedures, in which participants are not explicitly instructed to remember stimuli and are unaware their memory will be tested later (Charles et al., 2003; Leigland et al., 2004; Kensinger, 2008; Piguet et al., 2008; Spaniol et al., 2008). This is likely because the positivity effect is the most robust when older adults process emotional information naturally and spontaneously during encoding.

There are several theories about why the positivity effect occurs as individuals age. The Socioemotional Selectivity Theory, for instance, states that motivations and priorities shift across the life span, such that older adults become more aware that their future time horizons are limited compared to younger adults (Carstensen et al., 1999; Carstensen and Mikels, 2005; Carstensen, 2006). Thus, while younger adults often grapple with uncertainty about their educational and occupational paths, older adults tend to act in accordance with present-focused goals that promote emotional satisfaction (Waszczuk et al., 2016). These shifts in internal goals result in the selective processing of positive over negative information (Carstensen et al., 1999; Carstensen and Mikels, 2005; Carstensen, 2006). In contrast, other theories propose that the positivity effect results from age-related cognitive or neural deficits. The Dynamic Integration Theory states that due to older adults’ natural cognitive decline, they prioritize the processing of positive more than negative information because negative information tends to be more complex (Labouvie-Vief et al., 2010). Relatedly, the aging brain model proposes that the positivity effect arises as a result of amygdala degradation in old age. Given the importance of the amygdala in evaluating and responding to negative emotional information, amygdala degradation may lead to decreased sensitivity to negative information, therefore resulting in a comparative increase in the processing of positive information (Anderson, 2007; Sergerie et al., 2008).

Contrary to the Dynamic Integration Theory and the aging brain model theory, however, some imaging studies observe structural and functional preservation of the amygdala in normal aging, which explains why the amygdala’s responsiveness to positive stimuli does not vary across age (Allen et al., 2005; Leclerc and Kensinger, 2008; Brabec et al., 2010). Though some fMRI studies do find less amygdala activity in older adults while viewing negative information in comparison to their younger counterparts, these studies concurrently observe higher medial prefrontal cortical activity, which may indicate increased recruitment of cognitive control to regulate negative emotions (Mather et al., 2004; Williams et al., 2006; St Jacques et al., 2009; Leclerc and Kensinger, 2011; Samanez-Larkin and Carstensen, 2011). Therefore, increased cognitive control during emotional processing in old age, suggested by increased PFC activation, may underlie the positivity effect, rather than cognitive deficits in processing complex information or impaired amygdala function. Taken together with the aforementioned ERP findings suggesting that attentional engagement to negative but not positive information decreases with age (Wood and Kisley, 2006; Kisley et al., 2007; Langeslag and van Strien, 2009), older adults may show a more deliberate and controlled form of affective regulatory activity that maintains the emotional processing of positive memories while diminishing the processing of negative memories (Nashiro et al., 2011).

Although increasing age is associated with remembering positive experiences better than negative ones, both positive and negative experiences are disproportionately represented in memory compared to neutral ones (McGaugh, 2000; Christianson, 2014; Williams et al., 2022). This is in part due to the influence of emotional arousal at the time of encoding, which triggers the release of stress-related neuromodulators, such as norepinephrine and cortisol, which “tag” emotional aspects of memories that are worth retaining for future recollection (Holland and Kensinger, 2010; Bennion et al., 2015; Payne and Kensinger, 2018). One of the paradigms that has consistently found enhancements in emotional memory is the emotional memory trade-off task, where recognition memory is assessed separately for objects and backgrounds within scenes containing either an emotional or neutral object placed on a neutral background (Payne et al., 2008, 2012, 2015; Payne and Kensinger, 2011; Chambers et al., 2014; Bennion et al., 2017; Alger et al., 2018; Cummings et al., 2018; Vargas et al., 2019; Denis et al., 2022). Studies using this task consistently demonstrate that participants exhibit better memory for emotional objects than any other type of memory measured by the task, which includes neutral objects and the (also neutral) backgrounds paired with either type of object (Payne et al., 2008, 2012, 2015; Payne and Kensinger, 2011; Chambers et al., 2014; Bennion et al., 2017; Alger et al., 2018; Cummings et al., 2018; Vargas et al., 2019; Denis et al., 2022). Previous studies also find that sleep further enhances this emotional memory trade-off effect, potentially because sleep is a critical stage where emotionally salient information is preferentially consolidated (Payne et al., 2008, 2012; Payne and Kensinger, 2011; Cunningham et al., 2014; Bennion et al., 2015, 2017; Denis et al., 2022; Sanders et al., n.d.).

Importantly, however, most prior studies using the emotional memory trade-off task focus on emotional memory for negatively valenced content, with only a few studies examining positive stimuli (Chambers et al., 2014; Denis et al., 2022). Negative stimuli are often used because negative information tends to elicit higher levels of emotional arousal than positive information, resulting in a stronger effect on memory (Kensinger, 2009; Kauschke et al., 2019; Cook et al., 2023). For example, a vicious looking snake might be perceived as more unpleasant and agitating than a cute puppy is perceived as pleasant and exciting, and thus is more likely to be remembered. However, since positive episodic memories are crucial protective factors against adverse mental health outcomes, it is important to clarify whether there is also a memory trade-off effect for positive information (i.e., whether memory for positive emotional content can also be enhanced at the expense of neutral features of the same event; Holland and Kensinger, 2010; Williams et al., 2022). Moreover, knowing when this positivity shift emerges in adulthood can provide insight into the cognitive-affective mechanisms involved in other key psychological functions, including emotional regulation, decision-making, and psychological well-being as we age (Shohamy and Daw, 2015; Isaacowitz, 2022).

While the age-related positivity effect and emotional memory trade-off effect have been well researched separately, whether and how the positivity effect might interact with the emotional memory trade-off across the age-span remains relatively unexplored. It is unclear whether the age-related positivity effect is so strong that fewer attentional resources would be allocated to neutral features of the same scene, eventually leading to impaired memory performance of the peripheral neutral information in the background (Easterbrook, 1959; Reisberg and Heuer, 2004). Moreover, while it is well established that older adults tend to remember positive over negative information as mentioned above (Reed et al., 2014), it is not yet known if the neutral peripheral background of the attention-capturing object is more poorly remembered when the central event is positive compared to negative. Finally, most prior studies on the age-related positive memory bias have tested memory following brief delays, often because they were focused on how the preferential allocation of attentional resources during encoding leads to superior memory for positive information (Barber and Kim, 2021). As a result, delayed memory tasks that capture consolidation-related processes have not been widely employed.

The current study seeks to investigate the effects of aging on the selective consolidation of emotionally positive and negative central objects at the potential cost of their associated neutral backgrounds. We examined this question using a large sample of middle-aged and older adults (n = 274). Because we were interested in whether consolidation processes would influence these variables, we tested recognition memory after a significant delay (M = 11.86 h). This experiment is part of a larger study examining the impact of sleep (vs. wakefulness) on memory. As such, the memory delay interval spanned either a night of sleep or a day of wakefulness. However, because we were less interested in the specific impact of sleep here and more interested in the impact of a consolidation delay (i.e., the general passage of time), we collapsed these groups for maximum power. We hypothesized that increasing age would be associated with positive and neutral (but not negative) information being rated more positively, and with better memory performance following the delay for neutral and positive (but not negative) objects at the cost of associated neutral backgrounds. Although we did not have a firm prediction about when in the age span these positivity biases would emerge, we suspected that they would arise sometime in late middle age or early older age.

Our findings show that individuals interpret both positive and neutral information more positively as they approach older age, demonstrating the expected age-related positivity effect on valence ratings (i.e., positive interpretation bias). The age-related positivity effect also extends to memory. While none of the age groups exhibited a preference for remembering negative over positive information, older adults were more likely to remember positive than negative information. This was revealed as better memory for positive objects compared to negative objects on separate scenes, and descriptively speaking, greater positive than negative memory trade-off magnitude (i.e., the degree to which emotional objects were remembered better at the cost of their neutral peripheral backgrounds on the same scenes).

Consistent with prior research (Mather and Knight, 2005; Grühn and Scheibe, 2008; Gallo et al., 2009; Gilet et al., 2012; Schryer and Ross, 2012), we found that older age is correlated with perceiving positive events more positively. It is intriguing that older adults in our study also rated neutral scenes more positively, even though these stimuli have been previously normed as neutral in younger adults (Payne et al., 2008, 2012; Payne and Kensinger, 2011; Cunningham et al., 2014; Bennion et al., 2015, 2017; Denis et al., 2022; Sanders et al., n.d.). These results align with a previous study that this age-related positive interpretation bias also extends to neutral information (Mather and Knight, 2005). One possible mechanism underlying this interpretation bias is related to emotion regulation strategies that evolve with age. Specifically, compared to younger adults, older individuals tend to employ cognitive reappraisal to reframe their life experiences with positive interpretations (Liang et al., 2017; Nakagawa et al., 2017). This use of cognitive reappraisal and increased positive interpretation bias may contribute to a more positive overall outlook in older age, potentially promoting psychological resilience and satisfaction throughout the aging process (Höglund et al., 2020; Korkmaz and Güloğlu, 2021).

However, we failed to find a parallel association of increasing age and perceiving negative information less negatively. This is possibly because our negative stimuli consisted of highly negative content (e.g., snake, weapon, blood, vomit) that may have reached a functional ceiling with older participants, limiting the extent to which middle-aged adults could perceive them even more negatively. This potential ceiling effect might explain why universally negative information is not consistently perceived less negatively as individuals approach older age (Grühn and Scheibe, 2008; Streubel and Kunzmann, 2011; Gilet et al., 2012; Lee et al., 2023). In addition, our findings show that while middle-aged adults rated positive scenes as more arousing compared to neutral scenes, this pattern did not persist in older adults. Although age-related differences in arousal ratings did not reach significance, this may still point to a subtle shift in perceived arousal with age, consistent with previous findings on age-related decreases in perceived arousal to positive information (Keil and Freund, 2009; Gilet et al., 2012; Lee et al., 2023).

The finding that older adults remembered positive objects better than negative ones aligns with prior research that advancing age increases the positivity effect on memory (Reed et al., 2014). Notably, although this shift to a positivity memory bias is consistently found in studies where participants are instructed to passively view image or word stimuli (Charles et al., 2003; Leigland et al., 2004; Kensinger, 2008; Piguet et al., 2008; Spaniol et al., 2008), it is not observed in paradigms that impose informational processing constraints while viewing the stimuli (Kensinger et al., 2002; Denburg et al., 2003; Comblain et al., 2004; Grühn et al., 2005; Emery and Hess, 2008; Majerus and D’Argembeau, 2011; Leshikar et al., 2015; Eich and Castel, 2016). For example, when participants are explicitly instructed to remember stimuli (i.e., intentional encoding; Kensinger et al., 2002; Grühn et al., 2005; Emery and Hess, 2008; Majerus and D’Argembeau, 2011; Eich and Castel, 2016), or to make judgments about the stimuli during encoding, such as evaluating the complexity of visual images and determining whether verbal stimuli are self-referential (Comblain et al., 2004; Majerus and D’Argembeau, 2011), their ability to freely and naturally process emotional information may be hindered (Grühn et al., 2005; Gallo et al., 2009; Majerus and D’Argembeau, 2011). Therefore, the positivity effect may be attenuated when participants’ goals or focus are influenced by the experimental instruction (Carstensen and DeLiema, 2018; Sakaki et al., 2019).

Although we failed to find a linear association between older age and increasing positive memory bias, this is in accordance with prior research that age-related differences are only found when directly comparing older and younger, but not middle-aged adults (Fung et al., 2010; Reed et al., 2014). Critically, because most previous studies showing an age-related positivity effect focus on the comparisons between young and older adults (Leigland et al., 2004; Mather and Carstensen, 2005; Mikels et al., 2005; Kensinger, 2008; Piguet et al., 2008; Spaniol et al., 2008), it is not yet known whether this positivity effect already arises in middle age. The current study revealed that both early and late middle-aged adults do not show better memory for negative than positive information. This may suggest that the negativity bias and its influence on memory diminish in middle age, although a full transition to a positivity effect has not yet taken place. Another possibility for the null association between older age and increasing positive memory bias is that age-associated memory impairments counteract this positivity effect. Specifically, older adults tend to experience memory-related deficits due to declines in the structure and functionality of the hippocampus (Daselaar et al., 2006; de Flores et al., 2015). Although we focused on gist memory due to its resilience against forgetfulness, age-related memory deficits might still increase exponentially during the 12-h delay period (Davis et al., 2003; Titz and Verhaeghen, 2010; Matorina and Poppenk, 2021). Future research should incorporate retention intervals of varying lengths to explore the optimal delay condition during which positive aspects of memory are selectively consolidated more than accelerated forgetfulness impairs overall memory performance.

We replicated previous research findings on emotional memory trade-off effects, such that negative and positive objects were remembered better than both neutral objects on separate scenes and paired neutral backgrounds on the same scenes (Payne et al., 2008, 2012; Payne and Kensinger, 2011; Cunningham et al., 2014; Bennion et al., 2015, 2017; Denis et al., 2022; Sanders et al., n.d.). Interestingly, the typical pattern in younger adults, where neutral information in the background is less well-remembered when paired with a negative central object compared to a positive one (Denis et al., 2022; Sanders et al., n.d.), is not observed with middle-aged and older adults in the current study. Instead, in older age, the trade-off in emotional memory is more pronounced when the emotional object is positive as opposed to negative. This is potentially because emotionally positive events captures more attention than negative ones in older age (Noh and Isaacowitz, 2015; Erbey et al., 2020), thereby overshadowing the neutral and less salient aspects of the same event to a greater extent (Easterbrook, 1959; Reisberg and Heuer, 2004).

It is important to acknowledge that we did not explore how highly arousing, emotional information might have differential effects on memory bias from non-arousing, emotional information, especially because arousal and valence may influence memory processes through distinct mechanisms (Kensinger, 2004; Kensinger and Corkin, 2004). Prior research has shown that emotional memory induced by highly arousing information involves automatic processes and is well preserved with age (Kensinger, 2008), especially for negative stimuli (Mather and Knight, 2006; Leclerc and Kensinger, 2008). In contrast, emotional memory induced by non-arousing, negatively-valenced information tends to decline with age (Nashiro and Mather, 2011; Lee et al., 2023), as older adults may be utilizing cognitive control mechanisms via prefrontal cortex pathways to down-regulate their negative affect processing (Williams et al., 2006; St Jacques et al., 2009; Peña-Gómez et al., 2011). In line with prior research, our findings show that negative information is better remembered among middle-aged and older adults when it is perceived as arousing, but not when it is perceived as unpleasant. These results suggest that the age-related shift from negative to positive memory bias is most robust when comparing non-arousing or low-arousing, negative vs. positive information.

In our study, where most negative stimuli were rated as highly arousing relative to positive stimuli, it is not surprising that older age was not associated with worse memory for negative scenes, as they are likely still protected by the arousal pathway. Nevertheless, older adults still demonstrated superior memory for positive compared to negative information, which may suggest that age-related positivity effect observed in our study is likely driven primarily by valence-related emotional processing. Still, more studies should incorporate a broader spectrum of valence and arousal levels in their negative and positive stimuli, which is essential for exploring the nuanced impact of valence and arousal, both independently and interactively, on memory-related processes.

We failed to find sleep-dependent benefit for gist memory, which aligns with previous research indicating that meaning-based general properties in gist memory, relative to episodic-like details in specific memory, are less susceptible to forgetting over time and are therefore not as protected by sleep-related consolidation processes (Stickgold and Walker, 2013; Lutz et al., 2017; Matorina and Poppenk, 2021). Another plausible explanation could involve the fact that most participants in the sleep condition did not complete the encoding session immediately before their bedtime, whereas prior research demonstrates that sleep may only enhance memory consolidation when it occurs closely after learning (Solano et al., 2023). In addition, contrary to what we initially hypothesized based upon prior research, sleep did not magnify the emotional memory trade-off effect to result in better memory for emotional objects at the expense of associated neutral backgrounds, as has been seen before (Payne et al., 2008, 2012; Payne and Kensinger, 2011; Cunningham et al., 2014; Bennion et al., 2015, 2017). However, it is worth noting that most of these previous studies examined college-aged adults (Payne et al., 2008, 2012; Payne and Kensinger, 2011; Cunningham et al., 2014; Bennion et al., 2015, 2017), and the functional link between sleep and memory likely weakens with advancing age (Tucker et al., 2011; Scullin, 2013). Overall, this is in line with the existing literature that the prioritization of emotional memory during sleep is more of an exception than the norm (Prehn-Kristensen et al., 2009, 2013; Baran et al., 2012; Morgenthaler et al., 2014; Göder et al., 2015; Bolinger et al., 2018; Kurz et al., 2019; Davidson et al., 2021). Additionally, previous studies indicate that the proportion of rapid eye movement sleep and slow-wave sleep (Cairney et al., 2015) and the local coupling between sleep spindles and slow oscillations (Solano et al., 2021) both play crucial roles in memory consolidation. Future research with polysomnography-monitored sleep data is needed to investigate how different aspects of sleep architecture may facilitate age-related shifts in positive memory bias.

With a large online sample drawn from various U.S. communities, our findings provide better generalizability compared to laboratory studies with participants limited to college students or a single community. However, our sample may not fully represent the broader U.S. population as it consisted of healthy adults who were primarily White and had the technological literacy to participate using Prolific. It is also important to note that the attrition rate (50%) in our study was relatively high compared to in-person laboratory studies, as half of the eligible participants did not complete the follow-up experimental sessions. Although our attrition rate is still considered typical for longitudinal online studies, it may still impact the generalizability of our findings (Khadjesari et al., 2011; Rübsamen et al., 2017).

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://doi.org/10.17605/OSF.IO/8Q2VW.

The studies involving humans were approved by The Institutional Review Board at the University of Notre Dame. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

XN: Conceptualization, Data curation, Formal Analysis, Methodology, Project administration, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. MU: Conceptualization, Methodology, Project administration, Writing – original draft, Writing – review & editing. KS: Data curation, Investigation, Methodology, Resources, Software, Supervision, Visualization, Writing – review & editing. EK: Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing. JP: Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing. DD: Data curation, Investigation, Methodology, Resources, Software, Visualization, Writing – review & editing.