The Acute Effects of Psychoactive Drugs on Emotional Episodic Memory Encoding, Consolidation, and Retrieval

Abstract

Psychoactive drugs modulate learning and emotional processes in ways that could impact their recreational and medical use. Recent work has revealed how drugs impact different stages of processing emotional episodic memories, specifically encoding (forming memories), consolidation (stabilizing memories), and retrieval (accessing memories). Drugs administered before encoding may preferentially impair (e.g., GABA_A sedatives including alcohol and benzodiazepines, Δ⁹-tetrahydrocannabinol or THC, ketamine), enhance (e.g., dextroamphetamine and dextromethamphetamine), or both impair and enhance (i.e., ±3,4-methylenedioxymethylamphetamine or MDMA) emotionally negative and positive compared to neutral memories. GABA_A sedatives administered immediately post-encoding (during consolidation) can preferentially enhance emotional memories, though this selectivity may decline or even reverse (i.e., preferential enhancement of neutral memories) as the delay between encoding and retrieval increases. Finally, retrieving memories under the effects of THC, dextroamphetamine, MDMA, and perhaps GABA_A sedatives distorts memory, with potentially greater selectively for emotional (especially positive) memories. We review these effects, propose neural mechanisms, discuss methodological considerations for future work, and speculate how drug effects on emotional episodic memory may contribute to drug use and abuse.

Introduction

Psychoactive drugs are known to impact learning and memory in ways that might affect subsequent drug use and abuse. Prior research has focused on drug effects on neurocognitive systems supporting procedural learning and conditioning (e.g., Belin et al., 2009; Koob & Volkow, 2010; Robbins, 2002). However, such “model-free” forms of learning are typically slower and agnostic to the range of marked subjective effects and idiosyncratic emotional experiences produced by drugs. These individual emotional experiences are also likely contribute to developing addiction (Müller, 2013) and are captured by episodic memory, the conscious reexperiencing of information from the past (Tulving, 2002). In contrast to more incremental forms of learning, episodic memory is “one-shot learning” that may bridge the gap between initial drug experiences and eventual long-term drug-seeking behaviors (Bornstein & Pickard, 2020; Gershman & Daw, 2017).

Recently, there has been growing interest in how psychoactive drugs impact human episodic memory (Bornstein & Pickard, 2020; Doss et al., 2022; Kloft et al., 2021; Müller, 2013). The effects of drugs on emotional episodic memories, specifically, may contribute to emotion regulation and subsequent drug use. Affective and trauma disorders such as depression and posttraumatic stress disorder (PTSD) are characterized by abnormalities in emotional episodic memory (e.g., better memory for emotionally negative events or worse memory for emotionally positive events; Dere et al., 2010; Dillon & Pizzagalli, 2018; Dolcos et al., 2017; Durand et al., 2019), and the prevalence of drug use is high in individuals with these disorders (Brook et al., 2002; Davis et al., 2008; Jacobsen et al., 2001; Leeies et al., 2010). Although such drug use can become maladaptive, recently, several psychoactive drugs, such as cannabis (Mizrachi Zer-Aviv et al., 2016), ketamine (Zarate & Niciu, 2015), ±3,4-methylenedioxymethamphetamine (MDMA; Bahji et al., 2020), psilocybin (Mertens & Preller, 2021), and scopolamine (Drevets & Furey, 2010) are being investigated for their potential to treat affective and trauma disorders, and memory of the subjective effects of some drugs predicts their therapeutic efficacy (Yaden & Griffiths, 2021). Thus, investigating the effects of psychoactive drugs on emotional episodic memory will shed light on the role of drug experiences in both their therapeutic potential and potential for misuse.

Drugs can impact memory at the three main stages of episodic memory: encoding, or formation of memory traces, consolidation, or stabilization of memory traces, and retrieval, or accessing memory traces. Drug effects at each of these phases could contribute to different aspects of emotion regulation, which in turn, could affect future behavior. For example, a drug that reduces the encoding of emotionally negative or enhances the encoding of emotionally positive memories could improve emotional states even when acute drug effects are absent (cf. Warren et al., 2015). Moreover, pleasurable “episodic drug memories” from periods of intoxication may fuel subsequent drug-seeking behavior such as “chasing the first high,” eventually leading to habit formation (Bornstein & Pickard, 2020; Müller, 2013). Alternatively, a drug may retroactively modulate memories of a preceding emotional event (i.e., during their period of consolidation), potentially even enhancing these memories unbeknownst to the individual (e.g., Bruce & Pihl, 1997). Finally, drugs may distort memories of a prior event upon their retrieval, making them more or less emotionally charged, and these distorted memories could persist as such into sobriety.

Here, we review the scientific literature on the acute effects of psychoactive drugs on emotional episodic memory encoding, consolidation, and retrieval, focusing on objective memory tests in humans that included both neutral stimuli and stimuli with negative and/or positive emotional valence (all studies reviewed obtained informed consent). “Psychoactive drugs” include those that induce substantial subjective effects (“altered state of consciousness”), are used for recreational purposes, and have at least some abuse liability (although less common, even hallucinogens such as psychedelics and deliriants can be abused, Chiappini et al., 2022; Heal et al., 2018). Studies that fit these criteria included pharmacological manipulations with sedatives (defined as GABA_A positive allosteric modulators or PAMs including alcohol, benzodiazepines, and zolpidem), Δ9-tetrahydrocannabinol (THC; a CB₁ receptor agonist), stimulants (defined as dopamine and norepinephrine transport inhibitors including dextroamphetamine, dextromethamphetamine, and methylphenidate), MDMA (a serotonin, dopamine, and norepinephrine transporter inhibitor and 5-HT_2A receptor agonist), ketamine (an NMDA receptor antagonist), scopolamine (a muscarinic receptor antagonist), nicotine (a nicotinic receptor agonist), and γ-hydroxybutyric acid (GHB; a GHB and GABA_B receptor agonist). No studies to our knowledge have tested opioids (μ-opioid receptor agonists such as morphine or oxycodone) or psychedelics (5-HT_2A receptor agonist such as psilocybin or lysergic acid diethylamide or LSD) on an emotional episodic memory task. Nevertheless, the effects of MDMA on episodic memory resembles those of psychedelics (Doss et al., 2022) and are mediated by 5-HT_2A activation (van Wel et al., 2011).

How to Measure Drug Effects on Emotional Episodic Memory

Testing Emotional Episodic Memory

Most episodic memory tasks contain an encoding phase in which stimuli such as words and pictures are initially presented followed by a retrieval phase in which memory is tested via recall or recognition procedures. Recall tests can vary in the number of cues provided during retrieval. A free recall test requires stating the words or describing the pictures from the encoding phase in any order they are remembered without the help of cues. The primary outcomes are the number of remembered stimuli and intrusions (stimuli not presented in the encoding phase). A cued recall test is similar, except an explicit retrieval cue for each item is presented during retrieval. For example, pairs of stimuli may be presented during encoding, and one stimulus from each pair is presented during retrieval. For each retrieval cue, the non-presented stimulus paired with it from encoding must be recalled. Note that measures of recall are susceptible to floor effects (especially with an amnestic drug manipulation), and picture recall can be more difficult to score, as picture descriptions might be vague or contain some inaccuracies. In contrast, a recognition test presents stimuli from the encoding phase (targets) intermixed with new stimuli (lures), and participants are asked to discriminate between targets and lures (“old” or “new”). The proportion of correctly identified targets (p[“old”|target]) is the hit rate, and the proportion of incorrectly identified lures (p[“old”|lure]) is the false alarm rate. Typically, an impairment in memory will entail smaller hit rates accompanied by larger false alarm rates (due to more guessing), and a composite measure of memory accuracy can be computed by correcting hit rates with false alarm rates (e.g., hit rates - false alarm rates or other signal detection measures, such as d’). However, as will be discussed, some drugs can selectively increase false alarm rates, which can be indicative of a “false memory,” especially when accompanied by high levels of confidence.

Whereas recall primarily requires hippocampally-dependent recollection, recognition can be performed using both recollection and cortically-dependent familiarity (Yonelinas, 2002; Yonelinas et al., 2010). Recollection is the ability to retrieve details such as where or when an event took place, and familiarity is a feeling of knowing a stimulus has been processed in the absence of corroborating evidence, such as recognizing a face but not knowing how one knows this individual. One advantage to recognition tests is simultaneously estimating recollection and familiarity by collecting additional “remember/know” responses or confidence ratings (on a Likert scale) submitted to the dual process signal detection model (DPSD, Yonelinas, 1994, 2002). Confidence ratings can also be used to model metamemory (how well one understands their own memory) by examining the correspondence between confidence and accuracy (using meta-d’ for example; Fleming & Lau, 2014; Maniscalco & Lau, 2012). Finally, false alarms accompanied by high confidence ratings can provide better evidence for the induction of false memories.

One commonly used emotional memory task in psychopharmacological investigations involves testing memory for a story. The encoding phase is composed of a slide show with audio depicting a story containing an emotionally negative event occurring between two relatively neutral events (Cahill et al., 1994). During the retrieval phase, memory can be tested by recalling details from the story (i.e., without cues), recognizing correct statements about the story, and/or recognizing photos from the slide show (these recognition tests have typically used four-alternative forced choice). Although this task models how a real-life emotional event can suddenly occur, it suffers from temporal order effects (the emotional component always takes place in the middle of the story), a lack of emotionally positive stimuli, and a low number of stimuli (several trials can be required to reliably detect drug effects, Adam et al., 2020). Thus, another common task consists of an encoding phase in which several negative, neutral, and positive words or pictures are presented in random order and a retrieval phase in which memory is tested via recall, recognition, or both. In all of these tasks, memory for emotional stimuli is typically superior to that of neutral stimuli, especially when there is a delay between encoding and retrieval (e.g., ≥24 hours), as emotional memories tend to be forgotten more slowly (but see Mather et al., 2016 for nuances; Yonelinas & Ritchey, 2015).

The choice of stimuli is also not a trivial issue in emotional episodic memory paradigms, especially those involving drugs. Some drugs may facilitate the processing of specific emotional stimuli (MDMA and social stimuli, Kamilar-Britt & Bedi, 2015; THC and food-related stimuli, Roberts et al., 2019), and there is a limited range of emotional stimuli that can be administered in multiple arms of a drug study before running into issues related to unbalanced content between valence conditions (e.g., more animals or people in emotional stimuli). Balancing content across valences whenever possible and ensuring that stimuli are distinct from one another is important to avoid unintentional false memory effects that may disproportionately impact a given valence condition. For example, one might false alarm to a snake during retrieval if a different snake was also presented during encoding. Nevertheless, carefully controlling semantic and perceptual overlap between stimuli can be useful in the study of memory precision and distortion (Doss et al., 2019; Leal et al., 2014) and can even be more sensitive to drug effects (cf. Doss et al., 2020).

Pharmacologically Targeting Memory Phases

Two critical factors in pharmacological investigations of memory are when a drug is administered during a memory task and when memory is tested (Figure 1). Many drug studies administer drugs before encoding and test memory shortly afterward (Figure 1a). Although this is a convenient study design (all procedures happen within a single day), it can be difficult to determine whether drug effects (i.e., drug vs. placebo differences) on memory are due to the modulation of encoding, consolidation, retrieval, or some combination of these phases. A delay (e.g., ≥24 hours) between the encoding and retrieval phases to allow for drug effects to dissipate can help resolve this issue (Figure 1b). Nevertheless, there may be cases in which no delay between encoding and retrieval is preferred such as modelling real-world scenarios (e.g., interrogating an individual who was intoxicated at the scene of a crime).

Figure 1.

When a drug is administered during a memory task and when memory is tested can have different implications for which phase of memory was impacted. a) Drug administration takes place before encoding with memory tested immediately after. Drug effects (i.e., drug vs. placebo differences) on encoding, consolidation, and retrieval are confounded. b) Drug administration takes place before encoding with memory tested after a delay (after drug effects have dissipated). Drug effects are confounded between encoding and consolidation. c) Drug administration takes place immediately after encoding with memory tested immediately after. Drug effects are confounded between consolidation and retrieval. d) Drug administration takes place immediately after encoding with memory tested after a delay. Drug effects are isolated to consolidation. e) Drug administration takes place before a delayed memory test. Drug effects are isolated to retrieval. Note, these are heuristics likely with exceptions.

Drug effects do, however, persist beyond the encoding phase, complicating whether a drug effect is on encoding, consolidation, or both. Some have claimed that if a drug effect is not observed with immediate testing but becomes apparent after a delay between encoding and retrieval, then the drug must have modulated consolidation processes (e.g., Curran, 1986; Linssen et al., 2012; Soetens et al., 1993; Zeeuws et al., 2010). However, it is possible for a drug to impact encoding processes that only become apparent after they interact with consolidation processes, without the drug necessarily impacting the consolidation processes themselves. Synaptic tag-and-capture models propose that memory traces for salient stimuli (e.g., emotional stimuli) are “tagged” during encoding (via dopamine and/or norepinephrine) and subsequently “captured” for offline stabilization (i.e., consolidation; Redondo & Morris, 2011). A drug could, for example, prevent this tagging process without altering the capture process, resulting in intact immediate memory but delayed forgetting, as these memory traces cannot be stabilized later. As discussed below, several pre-encoding drugs preferentially modulate emotional compared to neutral memories with these effects only becoming apparent after a delay, thereby providing evidence for a tag-and-capture account.

The only way to determine an isolated consolidation effect is by administering a drug post-encoding followed by a delayed memory test to preclude drug effects on retrieval (Figure 1d). As will be discussed, some drugs can have strikingly different effects when administered pre- vs. post-encoding such as GABA_A sedatives, which impair memory in the former case but enhance memory in the latter (Mednick et al., 2011; Wixted, 2004). Nevertheless, it might be possible for pre- and post-encoding drug administration to produce similar effects (e.g., psychedelics may enhance familiarity-based memory when administered before or immediately after encoding, Doss et al., 2022; Wießner et al., 2022; Zhang et al., 2013). In such cases, it can be difficult to determine if the drug truly impacts encoding, though the magnitude of the effect (pre-encoding administration should contain both encoding and consolidation effects) or neuroimaging could speak to this issue.

An encoding-retrieval delay is also required for isolating drug effects to retrieval. Although post-encoding drug administration followed by an immediate memory test (Figure 1c) will minimize consolidation effects, some consolidation can still take place. One example of this is how post-encoding alcohol with a memory test shortly afterward can still enhance memory (Knowles & Duka, 2004), yet alcohol during retrieval on a delayed test typically has no impact (Weissenborn & Duka, 2000). Some have concluded that a pre-encoding drug, specifically ketamine, had an impact on retrieval because free recall was impaired but recognition was not (Ghoneim et al., 1985; Rusted & Warburton, 1989). A more likely explanation is that ketamine impaired recollection but spared familiarity, considering that other studies have found recognition impairments from smaller doses of pre-encoding ketamine (Carter et al., 2013), especially with a >24-hour delay (Becker et al., 2017), suggesting that null effects on recognition are likely due to sensitivity issues (e.g., too few trials or effects on familiarity only being revealed after a delay). Moreover, studies testing ketamine during retrieval have mostly found no impact (Ghoneim et al., 1985; Hetem et al., 2000; Lofwall et al., 2006; Morgan et al., 2003; Oye et al., 1992; Rowland et al., 2005), though all of these studies administered ketamine shortly after encoding. Some have also concluded that if a memory can be retrieved after pre-encoding drug administration but is impaired on subsequent retrieval attempts, this is a retrieval impairment. This could actually be a signature of a strong amnestic effect on encoding, as retrieving a memory is one of the best ways to encode a memory for subsequent retrieval (known as the “testing effect” or “retrieval practice,” Roediger & Butler, 2011). Thus, the only true test of whether a drug impacts retrieval is by implementing a delay between encoding and retrieval and administering the drug just prior to retrieval (Figure 1e). Figure 1 summarizes all the possible ways in which a single drug administration can modulate one or more phases of memory and what other information is needed to infer drug effects on a particular phase.

Drug Effects on Emotional Episodic Memory Encoding

Encoding is the phase of memory most commonly targeted in drug studies. Note that most of the studies below implemented a delay between encoding and retrieval (i.e., Figure 1b), thereby not confounding pre-encoding drug effects with retrieval. Although these effects could still be attributed to consolidation, as will be discussed, most drugs from the studies below do not have the same effect on consolidation (if they have any effect at all).

Many studies of emotional memory also measure emotional reactivity to stimuli during encoding (e.g., physiological response or subjective valence and arousal ratings). Although drugs can impact emotional experiences or reactions (Van Hedger et al., 2021), how they impact various emotional processing measures during encoding is not necessarily predictive of their subsequent effects on memory, thereby highlighting the importance of measuring emotional memory. Many studies below fail to find an impact on emotional reactions during encoding, yet most drugs during encoding preferentially impair or enhance subsequent memory for emotionally negative or positive stimuli when memory is tested after a delay.

Sedatives (GABA_A PAMs)

Sedatives such as alcohol, benzodiazepines, and nonbenzodiazepine hypnotics (or “Z-drugs”) administered during encoding impair the ability to form new memories (Doss et al., 2022; Doss, Weafer, Ruiz, et al., 2018). Functional magnetic resonance imaging (fMRI) studies have also found sedatives to attenuate the amygdala’s response during the processing of emotional stimuli (Del-Ben et al., 2012; Gilman et al., 2008; Paulus et al., 2005), an important structure for the retention of emotional episodic memories (Yonelinas & Ritchey, 2015). Two studies have tested the effects of the benzodiazepine lorazepam (1.5 and 2 mg oral) during the encoding phase of the story task with memory tested seven days later using a multiple-choice test asking questions about the story (Brignell et al., 2007; Kamboj & Curran, 2006). Lorazepam did not impact ratings of emotionality for the story or heart rate and blood pressure changes from the neutral to emotional part of the story. Nevertheless, lorazepam more selectively impaired memory for the emotional component of the story such that there was no longer a mnemonic advantage for emotional information.

Similar effects were reported in a study with alcohol (.6 g/kg oral) in which emotion was not confounded with serial order by creating separate neutral and emotional stories (Brown et al., 2010). Two versions of the story were created by presenting an identical sequence of neutral pictures that were each preceded and proceeded by either neutral pictures or emotional pictures. That is, for the same neutral pictures, the “sandwiching” of neutral or emotional images created either neutral or emotional contexts, respectively. Alcohol did not impact emotionality ratings for the story immediately after the encoding phase or valence and arousal ratings made for all slides immediately after the retrieval phase. However, on a multiple-choice recognition test asking questions about the story 72 hours later, alcohol had its strongest amnestic effect on the encoding of the neutral pictures presented in a negative emotional context that occurred at the beginning of the story. This finding suggests that preferential drug effects on the encoding of emotional information can interact with a primacy effect (Deese & Kaufman, 1957; Murdock, 1962) perhaps because participants habituated to later parts of the emotional story.

The studies just described focused on negative emotional stimuli, leaving the question of how drugs affect positive emotional memories, which tend to be less well remembered even when arousal is comparable to negative memories (Bowen et al., 2018). In one study examining the effects of triazolam (.25 mg oral) on encoding, participants viewed emotionally negative, neutral, and positive pictures paired with short narratives (Buchanan et al., 2003). Memory was tested 48 hours later by recalling details from these picture narratives, recognition of correct statements to questions about the story, and recognition of pictures (using multiple-choice tests for the latter two). Triazolam decreased arousal (but not valence) ratings for neutral and positive stimuli, and there was a trend for attenuated skin conductance responses to negative stimuli. Consistent with the work above, triazolam’s effects on memory were selective for emotional stimuli. Triazolam at encoding impaired the recall of both negative and positive memories but not neutral memories, though performance for neutral stimuli was at floor, and it tended to impair negative memory more than neutral and positive memory on the multiple-choice test asking questions about the story.

In another study testing the effects of alcohol (.8 and .7 g/kg oral for males and females, respectively) on encoding (Weafer et al., 2016), participants received a moderate dose of alcohol before viewing short labels describing negative, neutral, and positive pictures. Half of the labels had an accompanying picture, whereas half did not. Memory for the pictures was tested 48 hours later using a cued recollection test in which participants were presented with labels from the encoding phase and asked whether they had seen a corresponding picture. Despite there being no effect of alcohol on likability, valence, and arousal ratings made for each stimulus, alcohol preferentially impaired the encoding of both negative and positive pictures compared to neutral pictures. A DPSD analysis of the confidence ratings from the cued recollection test similarly found preferential impairments of both recollection and familiarity for emotional stimuli (Doss, Weafer, Ruiz, et al., 2018).

Some studies have failed to find preferential effects of sedatives on emotional memory encoding. One study found weak evidence for alcohol (.65 g/kg oral) at encoding to preferentially impair the recall of emotional pictures (Knowles & Duka, 2004). Surprisingly, this study also found valence (but not arousal) ratings for all stimuli to be greater in the alcohol group despite sedatives in the other studies reducing or having no impact on emotional processing. Another study found no evidence for alcohol (.67 g/kg and .56 g/kg oral in males and females, respectively) at encoding to preferentially impair recall of emotional pictures (Ray et al., 2012). This study contained only 12 participants per drug group and presented stimuli of different valences within separate blocks, which could make preferential emotional effects difficult to overcome serial order effects in a small sample. Two other studies failed to find any impairment of the benzodiazepine diazepam (5 mg oral) at encoding on recall or recognition of a task containing only negative and positive words (Murphy et al., 2008; Pringle et al., 2016). A test with only emotional stimuli might be expected to be particularly sensitive to amnestic effects if sedatives preferentially impact the encoding of emotional memories, though habituation to emotional content may also become more likely, rendering the processing of these stimuli to be more emotionally neutral. One commonality across these studies that did not find evidence for preferential emotional effects is that participants were tested shortly after encoding. Not only does such a design confound drug effects between encoding and retrieval, a delay between encoding and retrieval may heighten emotional selectivity via interactions between encoding and consolidation. As discussed, emotional memory enhancements are typically larger after a delay (Yonelinas & Ritchey, 2015) perhaps due to these memory traces being tagged and subsequently processed offline (Redondo & Morris, 2011). Sedatives may impair this tagging process at encoding, thereby resulting in a larger number of emotional compared to neutral memories being forgotten over time. Nevertheless, consolidation processes themselves may be altered by sedatives. Rapid eye movement (REM) sleep, which is important for emotional memory consolidation (Cunningham et al., 2022; Groch et al., 2013, 2015; Wagner, 2001), is suppressed by sedatives (Belyavin & Nicholson, 1987; Ebrahim et al., 2013). Thus, a period of sleep in which REM sleep is attenuated might also contribute to preferential emotional effects of sedatives at encoding.

THC

Like sedatives, THC impairs episodic memory encoding (Ranganathan & D’Souza, 2006), attenuates the amygdala’s response to emotional stimuli (Phan et al., 2008), and reduces the duration of REM sleep (Babson, Sottile, & Morabito, 2017). Only one study has examined the effects of THC (7.5 and 15 mg oral) during the encoding of emotional episodic memories (Ballard et al., 2013). Participants encoded negative, neutral, and positive pictures and were tested on a recognition memory test 48 hours later. THC tended to broadly increase valence and arousal ratings during encoding, though only the lower dose significantly increased negative valence ratings for neutral stimuli (negative and positive valence were rated on separate scales) and arousal ratings for both neutral and positive stimuli. Despite these increased ratings, THC at encoding preferentially impaired memory for negative and positive pictures. Valence and arousal ratings were also collected during retrieval, though there was no drug effect.

Stimulants (Catecholaminergic Transport Inhibitors)

Stimulants administered during encoding generally enhance memory (Linssen et al., 2012; Soetens et al., 1993; Zeeuws et al., 2010) and the amygdala’s response to emotional stimuli (Hariri et al., 2002). Thus, it might be expected that they preferentially enhance the encoding of emotional stimuli. Indeed, in the same study that found THC to more selectively impair the encoding of emotional stimuli, the opposite pattern was found with dextroamphetamine (10 and 20 mg; Ballard et al., 2013). Dextroamphetamine generally increased positivity and arousal ratings for all stimuli during encoding, yet both doses enhanced memory for negative and positive pictures more than neutral pictures on the recognition test 48 hours later. Like THC, dextroamphetamine did not impact valence and arousal ratings during retrieval. Preferential emotional memory enhancements were also obtained in a similar study with dextromethamphetamine (10 and 20 mg; Ballard et al., 2015) at encoding. Although there was no impact on valence and arousal ratings at encoding or retrieval, dextromethamphetamine, especially at the higher dose, preferentially enhanced the encoding of negative and positive pictures but only in participants reporting adequate sleep on the night following dextromethamphetamine administration. Dextromethamphetamine has a longer half-life than dextroamphetamine, and those participants reporting impaired sleep from the higher dose had worse memory, especially for negative stimuli. Like sedatives and THC, stimulants may particularly reduce REM sleep (but see Feinberg, 1974; Rechtschaffen & Maron, 1964), suggesting that emotionally selective encoding enhancements may depend on intact consolidation processes during REM sleep.

In another study, dextroamphetamine (20 mg oral) at encoding had no effect on memory for emotional or neutral images tested 48 hours after encoding (Weafer et al., 2014). Unlike the previous two stimulant studies, memory was tested with the cued recollection test described above and used a between-subjects design that may have been underpowered. Nevertheless, in a reanalysis of the confidence data from this study with meta-d’ modelling to estimate metamemory (Maniscalco & Lau, 2012), dextroamphetamine at encoding preferentially enhanced metamemory for negative and positive pictures compared to neutral pictures (Doss et al., 2022).

In contrast to the above studies, some studies have found stimulants to selectively enhance the encoding of neutral memories. In one of these studies, participants encoded negative and neutral pictures (no positive pictures) in the morning followed by two recognition tests (Whitehurst & Mednick, 2020). Half the stimuli were tested approximately 12 hours after encoding, and the other half were tested the following morning after a night’s sleep in the laboratory. On the first test, dextroamphetamine (20 mg oral) at encoding tended to enhance memory only for neutral pictures. On the second test, using a forgetting score by looking at the difference between the first and second test, dextroamphetamine enhanced memory for negative pictures, consistent with the idea that a period of sleep is needed for revealing preferential encoding enhancements of emotional memories. There was also evidence for dextroamphetamine at encoding to impair memory for neutral pictures on this second test that was associated with a latency to sleep. Some caution is warranted in these findings, however, as the placebo condition’s memory for neutral pictures on the second test was abnormally high such that it was greater than neutral memory on the first test and negative memory on the second test.

In another study finding preferential neutral memory enhancements, methylphenidate (40 mg oral) at encoding enhanced recognition on a multiple-choice test seven days later only for the neutral components of the story task (Brignell et al., 2007). One peculiar finding potentially related to this inconsistency with the above studies was that emotionality ratings of the story were lower in the methylphenidate group. It is also possible that drug effects interacted with serial order and/or consolidation, considering that post-encoding epinephrine has been found to enhance a primacy effect (Cahill & Alkire, 2003). Finally, there are differences in the pharmacology of methylphenidate compared to dextroamphetamine and dextromethamphetamine. Whereas dextroamphetamine and dextromethamphetamine inhibit and reverse the action of all monoamine transporters, methylphenidate only inhibits catecholamine transporters with no reversal of their action.

MDMA

MDMA is typically administered as the racemate, and despite its similarity to typical stimulants, pre-encoding MDMA impairs rather than enhances verbal recall (de Sousa Fernandes Perna et al., 2014; Kuypers et al., 2008, 2011, 2013; Kuypers & Ramaekers, 2005). This encoding impairment is perhaps due to the R-enantiomer’s action at the 5-HT_2A receptor, as the amnestic effect is blocked by the 5-HT_2A antagonist ketanserin (van Wel et al., 2011). Another difference from typical stimulants is that MDMA attenuates amygdalar responses to some negative stimuli (angry faces) but enhances striatal responses to some positive stimuli (happy faces; Bedi et al., 2009).

In perhaps the most striking example of preferential effects on the encoding of emotional memories, MDMA (1 mg/kg) both selectively impaired and enhanced the encoding of emotional memories dependent on the specific memory process (Doss, Weafer, Gallo, et al., 2018a). Participants encoded negative, neutral, and positive labels sometimes paired with pictures and were tested 48 hours later via cued recollection. MDMA did not impact valence and arousal ratings at encoding or overall memory, but a DPSD analysis of the confidence data revealed both impairments and enhancements of recollection and familiarity, respectively, similar to the psychedelic 5-HT_2A agonist psilocybin (Doss et al., 2022). Importantly, MDMA at encoding selectively impaired recollection of negative and positive pictures, whereas there were trends for selective enhancements of familiarity for negative and positive pictures (Figure 2).

Figure 2.

MDMA administered prior to memory encoding with sober retrieval 48 hours later resulted in impairments of recollection and trending enhancements of familiarity only for emotional memories (reproduced using data from Doss, Weafer, Gallo, et al., 2018a). * = p < .05, † = p ≤ .10, MDMA = ±3,4-methylenedioxymethylamphetamine, Enc = encoding.

Ketamine

Ketamine is another drug that is sometimes administered as the racemate, with the S-enantiomer having greater affinity for the NMDA receptor (and thus, greater dissociative effects). Like sedatives and THC, ketamine reliably impairs encoding (Morgan & Curran, 2006). One study tested the effects of racemic ketamine (2 mg/ml continuous intravenous infusion to reach a plasma level of 100 ng/ml) on emotional episodic memory (Becker et al., 2017). Participants were scanned with fMRI during the encoding of negative, neutral, and positive pictures, and their recognition memory was tested five days later outside of the scanner. Ketamine impaired encoding across all valences but to a somewhat greater extent for positive stimuli despite selectively increasing arousal ratings for positive stimuli at encoding. Importantly, ketamine enhanced the amygdala’s response during the encoding of negative but not neutral or positive pictures, suggesting that this enhanced processing spared negative stimuli from a more selective amnestic effect. However, another study that administered specifically S-ketamine found it to reduce the amygdala’s response to negative and neutral pictures to a greater extent than positive pictures (Scheidegger et al., 2016), suggesting potential differences between the racemate and S-enantiomer.

Cholinergic Modulation via Nicotine and Scopolamine

Different psychoactive drugs can bidirectionally modulate the cholinergic system with subsequent bidirectional effects on episodic memory encoding. Scopolamine is a deliriant that blocks muscarinic acetylcholine receptors and impairs encoding (Ebert & Kirch, 1998). Using the story task with recognition memory tested on a multiple-choice test seven days later, scopolamine (.6 mg subcutaneous) at encoding preferentially impaired memory for the emotional component of the story task without impacting emotionality ratings (Kamboj & Curran, 2006). Nicotine, on the other hand, activates nicotinic acetylcholine receptors and can enhance encoding (Valentine & Sofuoglu, 2018). One study tested the effects of nicotine on emotional episodic memory in smokers and non-smokers (14 and 7 mg transdermal patch in smokers and non-smokers, respectively; Froeliger et al., 2009). Participants completed three verbal oddball tasks that served as encoding phases, each followed by a recognition test for the oddball words.

Oddball tasks included non-italicized neutral words with italicized neutral oddballs, nonliving objects with living object oddballs, and neutral words with negative and positive emotional oddballs. Thus, emotional stimuli were on separate encoding and retrieval blocks from the two types of neutral stimuli. Nicotine enhanced memory for all stimuli in non-smokers, but this enhancement was not larger for emotional stimuli. In addition to a delay potentially revealing preferential emotional memory enhancements, the lack of delay confounds drug effects between encoding and retrieval. As will be discussed, some drugs at retrieval increase false alarms, especially for emotional stimuli. Although this study did not report false alarm rates, one possibility is that nicotine increased false alarms for emotional stimuli, thereby counteracting any enhancements that would otherwise be observed in a measure of memory accuracy.

Drug Effects on Emotional Episodic Memory Consolidation

In the previous section, we discussed how pre-encoding drug administration impacts encoding but also potentially memory consolidation. Although it is difficult to completely decouple the effects of drugs administered at encoding from those on subsequent consolidation, drug effects on consolidation can be studied by administering drugs with rapid absorption immediately after the encoding phase. Note that a drug with rapid absorption is preferable because we expect cellular consolidation, rather than systems consolidation, to be relevant to these post-encoding drug manipulations. Synaptic learning stabilizes within hours, whereas systems consolidation is thought to occur over years (Dudai, 2004; Kandel et al., 2014; Mednick et al., 2011) if it even exists at all (Yonelinas et al., 2019). Most drug manipulations discussed below occurred immediately after encoding, and memory was tested after a delay (i.e., Figure 1d).

Although many psychoactive drugs seemingly fail to impact consolidation (dextromethamphetamine, Ballard et al., 2015; ketamine, Feld et al., 2013; GHB, Kaestner et al., 2013; cannabis, Parker et al., 1980; scopolamine, Rasch et al., 2006), GABA_A sedatives are a notable exception. Despite robustly impairing memory when administered before encoding, they paradoxically enhance memory when administered immediately after encoding (Mednick et al., 2011; Wixted, 2004). This phenomenon has been found with multiple GABA_A sedatives (alcohol, benzodiazepines, zolpidem), encoding-retrieval delays (immediate testing to 48 hours), and stimuli including emotional stimuli. In this section, we review these studies of emotional memory and then discuss the few studies that have administered other classes of psychoactive drugs after the encoding phase of an emotional memory paradigm.

Retrograde Facilitation via GABA_A Sedatives

“Retrograde facilitation” can refer to the any post-encoding manipulation that retroactively enhances memory (e.g., stress, Shields et al., 2017), but it has been used most often in reference to episodic memory enhancements from post-encoding GABA_A PAMs (i.e., alcohol, benzodiazepines, and zolpidem). One explanation for this effect is that sedatives simply impair the encoding of new information, thereby reducing retroactive interference while newly strengthened synapses stabilize (Mednick et al., 2011; Wixted, 2004). Although this may partially explain such enhancements, other drugs that impair encoding should also result in retrograde facilitation when they are administered immediately post-encoding. In contrast, post-encoding cannabis (Parker et al., 1980), GHB (Kaestner et al., 2013), and scopolamine (but see Pomara et al., 2010 for retrograde facilitation via the muscarinic antagonist trihexyphenidyl; Rasch et al., 2006) have not been found to impact episodic memory.

To the extent that post-encoding emotional and neutral information is held constant, preferential enhancements of either emotional or neutral episodic memories could also speak against an account of retrograde facilitation that strictly relies on reducing interference. In one study, participants encoded negative, neutral, and positive pictures followed by the administration of placebo or alcohol (.65 g/kg oral) and a recall test approximately 160 minutes thereafter (Knowles & Duka, 2004). Moreover, another set of neutral and emotional stimuli was presented just after drug administration to examine the impact of alcohol on encoding (as discussed above). Despite these intervening stimuli being matched to those that were encoded pre-drug, enhancements from alcohol were larger for the recall of emotional pictures (collapsed across negative and positive valence) compared to neutral pictures. Nevertheless, the preferential enhancement of emotional over neutral memories may have been driven by alcohol’s greater impairment of the encoding of subsequently interfering emotional over neutral items. Moreover, one could argue that outside of experimentally controlled stimuli, there is typically more interfering neutral information in one’s environment, making it difficult to truly match the level of interference for neutral and emotional memories. Although this idea might suggest that neutral memories should benefit more from reducing retroactive interference, such levels of interference would be more likely to mask boosts in memory from post-encoding alcohol. Another complication in this study was that drug effects on consolidation and retrieval were potentially confounded, as some acute effects likely persisted during the memory test. Importantly, sedatives do not typically impact true memory when drug effects are isolated to retrieval (Lister, 1985; Weissenborn & Duka, 2000), and as discussed below, drug manipulations of retrieval tend to increase false alarms. Thus, it can be inferred that this post-encoding manipulation was likely impacting consolidation.

In addition to reductions in interference from new encoding, it may be that some sedatives, specifically zolpidem, can enhance sleep-based consolidation processes, especially for emotional information (Kaestner et al., 2013). In one study, participants encoded negative, neutral, and positive pictures followed by the administration of a placebo or the hypnotic zolpidem (10 mg oral). After drug administration, participants took a nap with electroencephalography recordings that included one REM cycle. Four hours after waking up, allowing for drug effects to mostly dissipate, recognition memory was tested. Zolpidem at consolidation enhanced memory for negative pictures and arousing pictures, suggesting contributions of both valence and arousal in these preferential enhancements of emotional memories. Furthermore, zolpidem increased sleep spindle density, a measure associated with consolidation (Antony et al., 2019), though these increases were not associated with the memory enhancements from zolpidem at consolidation. A subsequent study found evidence for similar mnemonic and neural effects even when zolpidem (10 mg oral) was administered 12 hours after encoding followed by a full night’s sleep (Simon et al., 2021). It is unclear, however, if these effects of zolpidem on sleep-based consolidation processes apply to other sedatives, as hypnotics tend to suppress REM sleep to a lesser extent (Besset et al., 1995).

In contrast to the above work, the selectivity of retrograde facilitation for emotional memory appears to diminish or even reverse with longer delays between encoding and retrieval, such that neutral memories may receive greater retrograde facilitation effects relative to emotional memories. In one study, participants encoded negative, neutral, and positive verbal statements followed by placebo or alcohol (.75 g/kg oral) administration, and memory was tested 24 hours later on recall and recognition tests (Bruce & Pihl, 1997). On the recall test, alcohol at consolidation enhanced memory across all valences with only a slightly larger effect for negative statements, though performance was near floor. On the recognition test, alcohol at consolidation enhanced memory significantly for only negative and positive statements. In a similar study without neutral statements, alcohol (.75 g/kg oral) at consolidation was found to enhance the recall of positive but not negative statements (Bruce et al., 1999). This increase in positive relative to negative memory was related to increases in heart rate from alcohol during the ascending limb (a measure of initial stimulant-like effects of alcohol). Finally, in a study that presented negative, neutral, and positive labels sometimes paired with pictures during encoding and tested memory 48 hours later via cued recollection, alcohol (.8 and .7 g/kg oral for males and females, respectively) at consolidation enhanced memory only for neutral pictures (Weafer et al., 2016). A DPSD analysis of the confidence data from this study found alcohol at consolidation to enhance both recollection and familiarity of neutral memory with a trending enhancement of recollection of positive memory (Doss, Weafer, Ruiz, et al., 2018). Figure 3 summarizes the findings from these studies on retrograde facilitation of emotional episodic memory.

Figure 3.

Graphical summary of drug studies testing the effects of sedatives (GABA_A positive allosteric modulators) on the consolidation of emotional and neutral episodic memories (Bruce & Pihl, 1997; Kaestner et al., 2013; Knowles & Duka, 2004; Weafer et al., 2016). Studies in the red zone found preferential emotional memory enhancements (negative and positive memory except Kaestner et al., 2013, which found a selective negative memory enhancements), and studies in the blue zone found preferential neutral memory enhancements. Longer arrows represent a greater bias in such enhancements (qualitatively drawn to scale such that longer arrows represent studies in which either only emotional or only neutral memory was enhanced). Longer delays between encoding and retrieval tend to result in less enhancements of emotional memories and larger enhancements of neutral memories (timeline is drawn to scale). Note that studies testing memory shortly after drug administration could be confounded by drug effects on memory retrieval. M = males, F = females.

Other Post-Encoding Drug Manipulations

Although some post-encoding drugs have also been found to enhance memory (the deliriant/muscarinic antagonist trihexyphenidyl, Pomara et al., 2010; the psychedelic/5-HT_2A agonist LSD, Wießner et al., 2022), few studies have tested them on an emotional memory task. The same study that found zolpidem at consolidation to enhance arousing and negative memories after a nap also tested the effects of GHB (2.5 g oral, Kaestner et al., 2013). Despite also modulating the GABAergic system, specifically GABA_B receptors, GHB at consolidation did not significantly impact memory. Like zolpidem, GHB numerically enhanced negative and arousing memories, but in contrast to zolpidem, it reduced sleep spindle density.

Some have speculated that post-encoding NMDA antagonism should enhance memory like GABA_A sedatives because recently strengthened synapses are no longer NMDA-dependent and should benefit from reductions in interference (Mednick et al., 2011; Wixted, 2004). In contrast to this hypothesis, one study found that inhaling nitrous oxide (an NMDA antagonist) administered immediately after watching a traumatic video reduced memory intrusions for the video the following day (Das et al., 2016). However, objective memory for events in the video was not affected by nitrous oxide at consolidation, and there was no neutral stimulus to see if the effect on memory intrusions was emotionally selective.

Because dopamine is involved in the consolidation of memory (Gruber et al., 2016; Redondo & Morris, 2011) and post-learning amphetamine enhances aversive and reward conditioning in mice (Janak & Martinez, 1992; Simon & Setlow, 2006), one study tested the effects of dextromethamphetamine (10 and 20 mg oral in a liquid suspension for rapid absorption) on the consolidation of emotional episodic memory (Ballard et al., 2015).

Participants encoded negative, neutral, and positive pictures, and recognition memory was tested 48 hours later. Dextromethamphetamine at consolidation was not found to impact emotional or neutral memories. One reason for this null effect might have been that memory performance was near ceiling in participants whose sleep was not disrupted by dextromethamphetamine (this was the same study that found preferential encoding enhancements from dextromethamphetamine in participants who did not report sleep disturbances). Another possibility is that memory consolidation could have been enhanced, but dextromethamphetamine simultaneously enhanced encoding that occurred after the encoding phase of the experiment, thereby increasing retroactive interference. In a meta-d’ analysis of the confidence data from this study, dextromethamphetamine at consolidation impaired metamemory, though this effect was found for emotional and neutral stimuli (Doss et al., 2022).

Drug Effects on Emotional Episodic Memory Retrieval

Like consolidation, many drugs appear to have little impact on retrieval (diazepam and scopolamine, Ghoneim & Mewaldt, 1975; ketamine, Lehmann et al., 2021; alcohol, Weissenborn & Duka, 2000), though some of these studies confounded drug effects between consolidation and retrieval (i.e., Figure 1c). Only a handful of studies have properly isolated drug effects to retrieval on an objective emotional memory task (i.e., Figure 1e). One commonality across these studies is that drugs at retrieval can increase false memories. Evidence for preferential distortions of emotional compared to neutral memories is currently limited, though these effects tend to be larger for positive memories.

THC

Although over 50 years ago the effects of cannabis during memory retrieval was found to increase false recognition (Abel, 1971), only recently was this effect replicated with THC (15 mg oral) and extended to an emotional memory paradigm (Doss, Weafer, Gallo, et al., 2018b). Participants encoded negative, neutral, and positive labels sometimes paired with pictures, and 48 hours later, they completed cued recollection and recognition memory tests while under the effects of THC or placebo. Whereas true memory was unaffected, THC during retrieval increased false alarms on both tests. This effect included high confidence false alarms of emotionally distinct material on the cued recollection test, consistent with an increase in false memories. Although THC at retrieval similarly drove both emotional and neutral false memories, the effects were numerically larger for positive stimuli. When two participants with extreme changes under THC were excluded (i.e., 0% false alarm rate under placebo, ≥60% false alarm rate under THC; note that this diminishes the overall false memory effect), THC at retrieval selectively drove medium to high confidence false alarms for emotional stimuli, especially positive stimuli (Figure 4a).

Figure 4.

THC (a) and MDMA (b) during retrieval increased false memories (reproduced using data from Doss, Weafer, Gallo, et al., 2018a, 2018b). Compared to the original publications, two participants were excluded with extreme changes under THC (0 to ≥60% false alarm rate from placebo to THC), and one participant in the MDMA at retrieval group was excluded for being an outlier (>60% false alarm rate; note that these exclusions reduce the overall false memory effect under drug). These false memory effects were selective for emotional stimuli, especially positive stimuli. Note that the THC study was a repeated measures design, hence the “strings” on violin plots (darker strings where data overlapped), whereas the MDMA study was an independent groups design and thus, potentially underpowered. * = p < .05, ** = p < .005, THC = Δ⁹-tetrahydrocannabinol, MDMA = ±3,4-methylenedioxymethylamphetamine.

Dextroamphetamine

Over 30 years ago, methamphetamine at retrieval was found to increase false memories (Mewaldt & Ghoneim, 1979), but only recently was that effect replicated and extended (albeit with dextroamphetamine) using an emotional memory task. In this study, participants encoded negative, neutral, and positive words and pictures in separate encoding phases, and 48 hours later, recall and recognition memory was tested (Ballard et al., 2014). Like THC, dextroamphetamine (10 and 20 mg oral) did not impact true memory but rather increased recall intrusions of pictures and words and false recognition of words. Increases in false recognition of words was larger for emotional words, especially negative words, though the interaction was not significant. Moreover, for each recalled item, participants rated its valence as negative, neutral, or positive. Dextroamphetamine at retrieval increased the recall (collapsed across true recall and intrusions) of items that were perceived as positive with a similar numerical increase in the recall of items perceived as negative.

In another study, participants encoded negative, neutral, and positive labels sometimes paired with pictures, and their memory was test 48 hours later on a cued recollection test (Weafer et al., 2014). Dextroamphetamine at retrieval was not found to have any impact on true or false memory, but in a meta-d’ analysis of the confidence data, dextroamphetamine at retrieval enhanced metamemory for negative pictures, with a trending enhancement for positive pictures. Thus, both studies with dextroamphetamine at retrieval found evidence for preferential effects on emotional memory, though these effects were at odds with each other. Whereas one study found a deleterious effect of dextroamphetamine at retrieval, an increase in emotional false memories, the other study found a beneficial effect, a better understanding of one’s emotional memories.

MDMA

One study tested the effects of MDMA (1 mg/kg oral) on retrieval of an objective emotional memory task (Doss, Weafer, Gallo, et al., 2018a). Participants encoded negative, neutral, and positive labels sometimes paired with pictures followed by a cued recollection test 48 hours later. MDMA at retrieval did not impact true memory, but there were trends for MDMA during retrieval to increase false alarms, especially medium to high confidence false alarms for positive stimuli (Figure 4b). A DPSD analysis of these data found numerical decreases in recollection estimates for negative and positive stimuli that were interpreted as MDMA impairing memory retrieval. However, high-confidence false alarms can bias DPSD recollection estimates toward zero, a finding also observed when true memory is in fact poor (e.g., hippocampal amnesics; Yonelinas et al., 1998). Considering that MDMA did not impact true memory across multiple measures, these reductions in recollection estimates are likely due to a false memory effect.

Although MDMA shares some pharmacology with dextroamphetamine (i.e., inhibition and reversal of monoamine transporters), it is possible that MDMA’s psychedelic pharmacology (i.e., 5-HT_2A agonism) could also drive emotionally selective false memory effects. Psychedelics enhance mental imagery (de Araujo et al., 2012; Kraehenmann et al., 2017), and mental imagery can drive false memories (Weinstein & Shanks, 2008, 2010). The psychedelic psilocybin has been found to increase the subjective vividness of retrieved positive autobiographical memories (no neutral or negative condition in this study; Carhart-Harris et al., 2012), and MDMA was found to selectively increase the subjective vividness of positive but not negative autobiographical memories (no neutral condition in this study; Carhart-Harris et al., 2014). Although more research is needed, a tendency to create favorable memories, regardless of their veracity, could be viewed as an enhanced optimism bias (Sharot, 2011) that could contribute to the therapeutic potential of these drugs (Bahji et al., 2020; Mertens & Preller, 2021).

Lorazepam

Sedatives were once used during the “Satanic panic” to supposedly recover repressed traumatic memories when they were instead likely driving false memories (Pendergrast, 2017; Perry & Jacobs, 1982; Stocks, 1998). In the lab, however, sedatives during retrieval have had minimal if any effect on false memory (benzodiazepines, Lister, 1985; alcohol, Weissenborn & Duka, 2000). This could be because studies that properly isolated sedative effects to retrieval (i.e., by including a delay between encoding and retrieval to avoid retrograde facilitation) used recall tasks in which intrusions can be near floor and are sometimes not even reported. Nevertheless, one study found evidence for lorazepam (.038 mg/kg) to increase false memories, especially those with high emotionality (Pernot-Marino et al., 2004). In this study, participants recorded in a diary two true events, one altered event, and one false event that happened to them every day over a period of two months. Two months after the last diary entry, they completed a recognition memory test containing true events, altered and false events made by the participant, and altered and false events made by the experimenter. Participants were to only accept true events as old, and they also made several ratings for each memory such as “remember/know” responses (Yonelinas, 2002), emotionality, and personal significance. Lorazepam during retrieval increased subjective recollections (based on “remember” responses), emotionality, and personal significance for all events including altered and false events. Although this study did not directly manipulate emotion (e.g., by asking participants to report specific, negative, neutral, and positive events), increased ratings of emotionality and personal significance were found to mediate the increase in subjective recollections.

Conclusions and Future Directions

This review surveyed the literature on the effects of psychoactive drugs on emotional episodic memory encoding, consolidation, and retrieval with several key patterns worthy of further exploration. First, drugs that impair encoding are typically more amnestic for emotional memories, and drugs that enhance encoding can be more selective for emotional memories, though this latter effect was less consistent. Importantly, both effects may require a delay (≥24 hours) between encoding and retrieval and/or a period of sleep. Second, there is mixed evidence for preferential emotional memory enhancements with sedatives administered at consolidation with shorter delays (≤24 hours) between encoding and retrieval selectively enhancing emotional memory and longer delays (i.e., 48 hours) selectively enhancing neutral memory. Third, in contrast to these patterns on true memory, THC, dextroamphetamine, MDMA, and lorazepam during retrieval increased false memory with somewhat larger effects for positive memories. Together, these findings suggest that under certain conditions, psychoactive drugs can more selectively modulate emotional compared to neutral memories. Table 1 provides a summary of these findings on objective memory tests and what remains to be tested.

Table 1.

Summary of Psychoactive Drug Effects on Emotional Episodic Memory

	Encoding (True Memory)			Consolidation (True Memory)			Retrieval (False Memory)
	Negative	Neutral	Positive	Negative	Neutral	Positive	Negative	Neutral	Positive
Sedatives (GABAA PAMs; i.e., alcohol, benzodiazepines, zolpidem)	↓	↓	↓	↑ *	↑ *	↑ *	?	-	?
THC (a CB1 agonist)	↓	↓	↓	?	-†	?	↑	↑	↑
Stimulants (DAT/NET inhibitors; i.e., dextroamphetamine, dextromethamphet amine, methylphenidate)	↑ *	↑	↑	-	-	-	↑ *	↑	↑ *
MDMA (a SERT/DAT/NET inhibitor and 5- HT2A agonist)	↓/↑	↓†	↓/↑	?	?/↑	?	-	-	↑
Ketamine (a NMDA antagonist)	↓	↓	↓	?	-	?	?	-†	?
Scopolamine (a muscarinic antagonist)	↓	↓	?	?	-/↑	?	?	-†	?
Nicotine (a nicotinergic agonist)	↑†	↑†	↑†	?	?	?	?	?	?
GHB (a GHB and GABAB agonist)	?	↓	?	-	-	-	?	?	?

Note. Drug effects on encoding and consolidation are for true memories, whereas drug effects on retrieval are for false memories. Up and down arrows indicate increases and decreases, respectively. Both an up and a down arrow indicate increases and decreases on different processes (i.e., enhancement of recollection and impairment of familiarity). Larger relative to smaller arrows indicate at least one study finding larger effects in a valence condition compared to another (e.g., larger effects in negative and/or positive vs. neutral, larger effects in neutral vs. negative and/or positive). An asterisk (*) with a larger bold arrow indicates mixed evidence. A hyphen (−) indicates no effect. A question mark (?) indicates a lack of data. A dagger (†) indicates that memory phases were confounded (i.e., a drug administered prior to encoding with retrieval taking place during acute drug effects or a drug administered immediately postencoding with retrieval taking place during acute drug effects). An additional slash (/) and up arrow indicate increases have been found with pharmacologically similar drugs (i.e., in the case of MDMA and scopolamine, LSD and trihexyphenidyl, respectively, enhance consolidation). PAM = positive allosteric modulator, THC = Δ⁹-tetrahydrocannabinol, DAT = dopamine transporter, NET = norepinephrine transporter, MDMA = ±3,4- methylenedioxymethamphetamine, SERT = serotonin transporter, GHB = γ-hydroxybutyric acid.

In contrast to recent work highlighting how different classes of psychoactive drugs uniquely modulate episodic memory processes (Doss et al., 2022), the findings reviewed here allude to some potential similarities between classes. Preferential impairments or enhancements of encoding emotional memories may come from modulation of a tagging process in synaptic tag-and-capture accounts (Redondo & Morris, 2011), considering that these effects depend on a delay between encoding and retrieval. Modulation of amygdalar processing during encoding may also be shared between drugs, as drugs that impair (MDMA, Bedi et al., 2009; sedatives, Del-Ben et al., 2012; Gilman et al., 2008; Paulus et al., 2005; THC, Phan et al., 2008) and drugs that enhance (stimulants, Hariri et al., 2002) amygdalar activation preferentially impair and enhance, respectively, emotional memory encoding. Nevertheless, there are likely differences between the effects of different drugs on emotional memory encoding. Whereas most drugs preferentially impacted the encoding of both negative and positive stimuli compared to neutral stimuli, ketamine at encoding preferentially impaired only positive stimuli (Becker et al., 2017). In fact, ketamine increased amygdalar activation to negative stimuli, potentially preserving these stimuli from a more selective memory impairment. MDMA may also have valence-specific, as well as process-specific effects on emotional memory encoding. MDMA has been found to decrease amygdalar activation to negative emotional faces but increase striatal responses to positive emotional faces (Bedi et al., 2009). Moreover, although MDMA similarly impacted the encoding of negative and positive stimuli, it uniquely impaired the encoding of recollection-based memory but enhanced the encoding of familiarity-based memory, and these effects were numerically larger for negative and positive memories, respectively. The potential for drug effects that interact with both valence and mnemonic process is underscored by the tendency for negative emotion to particularly modulate recollection (Yonelinas & Ritchey, 2015), perhaps due to a maintenance of hippocampal-dependency (Duszkiewicz et al., 2019), and positive emotion’s association with familiarity driven by processing fluency (Duke et al., 2014; Reber et al., 1998; Verde et al., 2010).

Although task differences may underlie whether sedatives at consolidation preferentially enhance emotional memories, there are potential explanations as to why emotional episodic memories are less likely to benefit from retrograde facilitation over longer delays. One explanation may be that post-encoding sedatives preferentially stabilize emotional memories initially, perhaps by increasing γ oscillations (Campbell et al., 2014; Hall et al., 2009; Saxena et al., 2013), which are thought to support emotional memory reactivations and synaptic strengthening (Paré & Headley, 2023). However, emotional memory stabilization during sleep may be disrupted due to less emotional memory-supporting REM sleep (Belyavin & Nicholson, 1987; Ebrahim et al., 2013; Groch et al., 2013, 2015; Wagner, 2001). Additionally, during non-REM sleep, emotional memories require coordinated reactivations between the hippocampus and amygdala (Girardeau et al., 2017). Considering that sedatives dampen amygdalar activations (Del-Ben et al., 2012; Gilman et al., 2008; Paulus et al., 2005), negative memories may reactivate less often during the first night of sleep. In the days that follow, subsequent instances of remembering and reactivations of emotional memories may become less likely compared to neutral memories, thereby resulting in preferential enhancements of neutral memories over longer delays. Moreover, because environmental information is typically neutral, longer periods between encoding and retrieval would be expected to contain more interfering neutral information, allowing more opportunity for previously encoded neutral memories to benefit from a manipulation that initially prevents interference and/or facilitates memory stabilization. Together, these explanations would suggest that retrograde facilitation involves both a facilitation of memory stabilization processes and a reduction in interference.

Regarding retrieval effects of different drugs, the striatum may be common to drug-induced false memories. The striatum is increasingly being recognized for its role in memory retrieval (King et al., 2018; Scimeca & Badre, 2012) including false recognition (Abe et al., 2008), and all psychoactive drugs seemingly increase striatal processing (Di Chiara & Imperato, 1988; Yager et al., 2015), including drugs not typically abused like scopolamine (Antonova et al., 2011; Chapman et al., 1996) and psychedelics (Vollenweider, 1999). Finally, the striatum may especially be involved in retrieving positive memories (Speer et al., 2014), and false memory effects under drugs tended to be larger for positive memory. One possibility is that mood-congruent information is more often spontaneously retrieved resulting in greater familiarity or fluency even for previously non-processed stimuli, resulting in more false memories for these stimuli. This conjecture is supported by the finding that THC, which produces more mixed affective states compared to amphetamine and MDMA, increased both positive and negative false memories.

Although some drugs with minimal psychoactive effects can also preferentially modulate emotional memories (e.g., the β-adrenergic antagonist propranolol at encoding, Cahill et al., 1994), these effects do not always mirror those found here. For example, pre-encoding naltrexone (a μ-opioid antagonist) impaired neutral memory but enhanced negative memory (KatzenPerez et al., 2001), and pre-encoding reboxetine (a norepinephrine reuptake inhibitor) enhanced memory more selectively for positive compared to negative memory (Harmer et al., 2003, 2009). In contrast to these biases in either negative or positive memory, most psychoactive drugs at encoding had larger effects for both negative and positive memory. Furthermore, stress manipulations at encoding or consolidation, which impair and enhance memory, respectively, do not appear to be more selective for emotional memories (Shields et al., 2017), whereas most of the consolidation enhancements from sedatives were larger for emotional memories. Stress manipulations at retrieval, however, more selectively modulate emotional memories, but they appear to impair true memory rather than increase false memory (Shields et al., 2017).

There are still many gaps in our knowledge regarding different classes of psychoactive drugs and their effects at different phases of emotional episodic memory processing. Some drugs such as opioids and psychedelics have yet to be tested on objective emotional episodic memory tasks, and most drugs have yet to have their effects isolated to the consolidation and retrieval phases of an emotional episodic memory task. Additionally, all drug manipulations in the studies reviewed here were within the clinical and/or moderate recreational range, but there could be non-linearities at lower or higher doses. Below, we discuss clinical implications of the work reviewed here and novel directions for future work.

Clinical Implications

Drug effects on different phases of emotional episodic memory have important consequences for understanding their long-term effects relevant to affective disorders and addiction. For example, drugs that impair memory encoding may bias the kind of events that are stored in memory, with more emotionally laden ones suppressed. If one is particularly prone to remembering negative events (because of an emotional disorder or a maladaptive environment), remembering fewer negative experiences could shape their reality, thereby influencing subsequent use. However, the accompanying impairment of positive memories may explain why affect does not improve and can worsen (Brady & Sinha, 2005; Davis et al., 2008). In contrast, enhancing the encoding of both negative and positive memories with stimulants may have the opposite effects. That is, enhancing positive memories may drive subsequent use, but the simultaneous enhancement of negative memories may worsen affect, especially once sober, thereby driving further use and contributing to more incremental mechanisms of addiction (e.g., developing/avoiding withdrawals, conditioned/procedural memories).

How a drug impacts consolidation may also have unintended consequences. The enhancements of emotional memory consolidation with sedatives suggest that one should “forget drinking to forget” (cf. Bruce & Pihl, 1997). Although one study alludes to the possibility of time-dependent decays of such emotional memory enhancements (Weafer et al., 2016), this study was the exception. Furthermore, considering that studies in humans (Wießner et al., 2022) and animals (Zhang et al., 2013) have recently found psychedelics at consolidation can enhance memory, how they impact emotional memory will be an important factor regarding their use in psychotherapy. For example, it might be best to avoid psychedelic therapy immediately after a negative event or even after cuing a negative memory.

The propensity for drugs during retrieval to distort emotional memories should also be a concern for drug-assisted psychotherapies. On the one hand, biasing memories to be more positive could benefit disorders with a negative memory bias such as depression. However, a troubling concern in the growing interest in the therapeutic use of psychedelics is a disposition toward Freudian frameworks that may attempt to recover supposedly repressed memories (Carhart-Harris & Friston, 2019; Kraehenmann et al., 2017). Memory research has largely shown that memory repression is a questionable concept (Otgaar et al., 2021), and if anything, it is difficult to forget emotional memories (Yonelinas & Ritchey, 2015). As discussed, sedatives can drive false recollections by increasing emotionality and personal significance (Pernot-Marino et al., 2004), and they can also increase suggestibility (Kloft et al., 2021). Psychedelics similarly increase suggestibility (Carhart-Harris et al., 2015), personal significance, and emotionality (Griffiths et al., 2006), as well as mental imagery (Carhart-Harris et al., 2012, 2014; de Araujo et al., 2012; Kraehenmann et al., 2017), which increases the incidence of false memories (Weinstein & Shanks, 2008, 2010). Future work, isolating drug effects to an intervening misinformation phase (e.g., Doss et al., 2019) or both misinformation and retrieval phases may shed light on how drug-induced false memories can be minimized (cf. Kloft et al., 2021).

Future Directions

Although this review was meant to lay the groundwork for how psychoactive drugs impact emotional episodic memory encoding, consolidation, and retrieval, there are several extensions of this research with implications for addiction and therapeutics. One focus of this review has been on which memories are impaired by drugs administered at encoding, but studying those memories that survive an amnestic drug manipulation during encoding may shed light on their influence in future drug-seeking under sober conditions. These “episodic drug memories” (Müller, 2013), especially those with positive emotional content, may be particularly salient in early drug use and result in behaviors such as “chasing the first high” (Bornstein & Pickard, 2020). One prediction might be that they are more reliant on the striatum, considering that psychoactive drugs increase striatal transmission. Although speculative, retrieving episodic memories of a drug experience may even facilitate addiction-dependent striatal learning mechanisms that are typically more incremental.

Work on state-dependency (i.e., boosts in memory for administration of a drug at both encoding and retrieval; Overton, 1991; Radulovic et al., 2017) might also reveal how drug use can transition to abuse. Being better able to retrieve positive memories from prior drug experiences while under the effects of the same drug may provide a reason for further use. Alternatively, state-dependent memory could be an important factor in the context of psychedelic-assisted psychotherapy to maintain continuity between sessions and facilitate the retrieval of emotional content brought up in prior sessions. State-dependency may be particularly important when the encoding of positive memories is impaired by both drug effects and psychiatric conditions like depression, and a goal may be to reappraise memories to be more positive (or less negative). Nevertheless, more recent work has found state-dependent drug effects to be less reliable than earlier work (Schreiber Compo et al., 2017; Weafer et al., 2014).

Re-encoding memories to be more positive also necessitates the study of how psychoactive drugs impact emotional episodic memory reconsolidation (i.e., reactivating a memory to render it labile for subsequent strengthening, weakening, or distorting; Lee et al., 2017). Some work has found the β-adrenergic antagonist propranolol administered during a post-encoding memory reactivation phase to attenuate emotional but not neutral memory on a subsequent test (Schwabe et al., 2012, 2013). Psychedelics and dissociative hallucinogens, specifically, may have interesting effects in such paradigms, as they drive plasticity (Inserra et al., 2021; Olson, 2018) and neural flexibility (Braun et al., 2016; Carhart-Harris & Friston, 2019), factors that may help render emotional memories more labile. In fact, psychedelics can enhance neural flexibility and positive affect for at least one week after dosing (Barrett et al., 2020; Daws et al., 2022; Doss et al., 2021), suggesting that reconsolidation of emotional memories could be facilitated even after the acute effects are over. Additionally, post-acute effects of psychedelics could increase the encoding and/or retrieval of positive relative to negative memories, as is found with traditional antidepressants (Dere et al., 2010; Dillon & Pizzagalli, 2018; Dolcos et al., 2017; Durand et al., 2019).

The idiosyncratic emotional experiences produced by psychoactive drugs are likely important to both their long-term harms and benefits. Episodic memory is one medium through which acute drug effects can transform and persist for better or worse. Whether psychoactive drugs are to be used recreationally or therapeutically, delineating the boundary conditions of how they can shape emotional episodic memories could maximize beneficial drug effects while minimizing more harmful ones.

Highlights.

• Psychoactive drugs preferentially modulate the encoding of emotional memories.

• Sedatives can preferentially enhance the consolidation of emotional memories.

• Psychoactive drugs may preferentially distort the retrieval of emotional memories.