Stress in the zoo: Tracking the impact of stress on memory formation over time

Highlights

• Stress is known to affect human memory formation, but precisely how remains unclear.

• We tracked stress effects on memory over more than 2 h to unravel the exact time course.

• Two different physiological stress response waves enhanced memory formation.

• Immediate stress effects on memory were linked to the autonomic nervous system.

• Slower enhancing effects of stress were associated with the cortisol stress response.

Abstract

Although stress is well known to modulate human memory, precisely how memory formation is altered by a stressful encounter remains unclear. Stress effects on cognition are mainly mediated by the rapidly acting sympathetic nervous system, resulting in the release of catecholamines, and the slower acting hypothalamus-pituitary-adrenal axis secreting cortisol, which induces its effects on cognition through fast, non-genomic actions and delayed, genomic actions. Importantly, these different waves of the physiological stress response are thought to dynamically alter neural processing in brain regions important for memory such as the amygdala and the hippocampus. However, the precise time course of stress effects on memory formation is still unclear. To track the development of stress effects on memory over time, we tested individuals who underwent a stressful experience or a control procedure before a 2-h walk through a zoo, while an automatic camera continuously photographed the events they encoded. In a recognition memory test one week later, participants were presented with target photographs of their own zoo tour and lure photographs from an alternate tour. Stressed participants showed better memory for the experimental treatment than control participants, and this memory enhancement for the stressful encounter itself was directly linked to the sympathetic stress response. Moreover, stress enhanced memory for events encoded 41–65 min after stressor onset, which was associated with the cortisol stress response, most likely arising from non-genomic cortisol actions. However, memory for events encoded long after the stressor, when genomic cortisol actions had most likely developed, remained unchanged. Our findings provide novel insights into how stress effects on memory formation develop over time, depending on the activity of major physiological stress response systems.

Introduction

Our everyday life is full of challenges, deadlines and demands. For many of us, stress is so common that it is accepted as part of life. Stress, however, can change the way we feel, think, and behave (Diamond et al., 2007, Lupien et al., 2009, Sandi and Haller, 2015). In particular, stress is known to impact learning and memory processes (Diamond et al., 2007, Joëls et al., 2006, Schwabe and Wolf, 2013). Although stress-induced changes in memory formation have major implications for stress-related mental disorders, such as depression or post-traumatic stress disorder (PTSD; Pitman et al., 2012) as well as educational or occupational settings, the literature on the influence of stress on memory formation is heterogeneous, with some studies reporting enhanced and others impaired memory after stress (e.g. Diamond et al., 2007, Payne et al., 2006, Sandi et al., 1997, Schwabe et al., 2008, Zoladz et al., 2011). Thus, exactly how stress affects memory formation is not well understood.

It is commonly assumed that the impact of stress on memory formation depends on the temporal proximity of the stressful encounter and the event that is encoded (Joëls et al., 2011, Schwabe et al., 2012). Stress experienced during or shortly before learning is thought to enhance memory formation whereas stress long before learning is assumed to suppress new encoding to protect the memories formed in the stressful situation from interference (Joëls et al., 2011, Schwabe et al., 2012). These time-dependent effects of stress on memory have been linked to time-dependent physiological and endocrine changes that occur in response to stressful encounters (Joëls and Baram, 2009, Joëls et al., 2011). Within seconds after stressor onset, the autonomic nervous system (ANS) is activated, resulting in the release of catecholamines, such as adrenaline or noradrenaline, which prepare the organism for ‘fight-or-flight’. A second, slower system activated under stress is the hypothalamus-pituitary-adrenal (HPA) axis which leads to the release of corticosteroids (in humans mainly cortisol), reaching peak levels approximately thirty minutes after stressor onset. Upon reaching the brain, corticosteroid hormones operate via two modes of action: a rapid, non-genomic mode mediated by membrane-bound receptors and a slower, genomic mode of action mediated by intracellular receptors that is assumed to set in 60–90 min after a stressful event (Joëls et al., 2012). Thus, there are (at least) three waves of the physiological stress response: rapidly acting ANS activity, early non-genomic cortisol actions, and slow genomic cortisol actions. Critically, it is assumed that these different stress response waves have a distinct, perhaps even opposite impact on memory processes (Joëls et al., 2011). More specifically, rapidly released catecholamines are thought to facilitate memory formation by enhancing neural excitability and glutamatergic transmission in relevant brain structures such as the basolateral amygdala and the hippocampus (Hu et al., 2007, Joëls et al., 2011, Onur et al., 2009). These effects of noradrenaline are thought to be further promoted by non-genomic cortisol effects (Joëls et al., 2011, Karst et al., 2010, Karst et al., 2005). Non-genomic cortisol effects appear to enhance neural activity and noradrenergic input in the amygdala (Roozendaal et al., 2006b, van Stegeren et al., 2007), again leading to enhanced memory performance. In contrast, later genomic effects of cortisol are assumed to suppress new encoding by decreasing and normalizing neural excitability in these structures to allow for enhanced prefrontal cognitive control long after the stressful encounter (Henckens et al., 2010, Karst et al., 2010, Lovallo et al., 2010). Although this time-dependency of stress effects on memory formation appears to be widely accepted in the field, studies that directly assess the dynamics of memory formation after a stressful event are lacking. Rather, this widely held hypothesis is based on a literature primarily comprising studies that investigated memory formation at defined time points after a stressor, e.g. 0, 10, or 30 min after a stressful event (Domes et al., 2002, Payne et al., 2006, Schwabe et al., 2008, Smeets et al., 2009, Zoladz et al., 2011). However, in order to understand when stress effects on memory formation arise, how long they last, and to what extent they are linked to the activity of major stress response systems, memory processes need to be assessed continuously during and after a stressful event.

Moreover, to date virtually all studies investigating stress effects on memory were performed in strictly controlled laboratory settings, using by and large arbitrary learning material (Domes et al., 2002, Schwabe et al., 2008, Smeets et al., 2009). But can these findings actually be translated to real-life settings? Given that stress effects on memory have major implications, for example, for educational or clinical settings, the translation of findings to natural environments is essential. Thus, the aims of this experiment were twofold. First and foremost, we aimed to unravel the dynamics of stress effects on memory formation by testing how stress affects the formation of memories during the first hours after a stressful event. Second, we aimed to assess the impact of stress on memory formation in a natural environment. To this end, we used a unique paradigm in which participants first underwent a stress or control manipulation before they encoded experiences on a 2-h tour through a zoo. Both during the stress/control manipulation and during the zoo tour participants were carrying a camera that automatically took pictures from their first-person perspective. This paradigm allowed us to exert control over the encoding of real-world events and to determine when exactly after stressor onset a specific event was encoded. Memory performance was tested one week later in a recognition test that included photographs of the participant’s own stress/control experience and zoo tour, as well as lures from an alternative tour through the zoo. In order to track activity of the ANS and the HPA axis, we assessed participants pulse and blood pressure in the context of the stress (or control) manipulation and measured salivary cortisol every 6 min across the entire encoding session. We predicted that stress would enhance memory for the stressful event itself, when autonomic arousal is high and catecholamines facilitate neural activity in the amygdala and hippocampus, and for events encoded during the rapid mode of cortisol action, about 30 min post-stress, when non-genomic cortisol effects support the effects of catecholamines. However, memory was expected to be impaired for events that were encoded about 90 min after stress, when the genomic effects of cortisol decrease activity of amygdala and hippocampus.