Neutral and emotional episodic memory: global impairment after lorazepam or scopolamine
- First Online:
- Cite this article as:
- Kamboj, S.K. & Curran, H.V. Psychopharmacology (2006) 188: 482. doi:10.1007/s00213-006-0552-7
- 128 Views
Benzodiazepines and anticholinergic drugs have repeatedly been shown to impair episodic memory for emotionally neutral material in humans. However, their effect on memory for emotionally laden stimuli has been relatively neglected.
We sought to investigate the effects of the benzodiazepine, lorazepam, and the anticholinergic, scopolamine, on incidental episodic memory for neutral and emotional components of a narrative memory task in humans.
Materials and methods
A double-blind, placebo-controlled independent group design was used with 48 healthy volunteers to examine the effects of these drugs on emotional and neutral episodic memory.
As expected, the emotional memory advantage was retained for recall and recognition memory under placebo conditions. However, lorazepam and scopolamine produced anterograde recognition memory impairments on both the neutral and emotional components of the narrative, although floor effects were obtained for recall memory. Furthermore, compared with placebo, recognition memory for both central (gist) and peripheral (detail) aspects of neutral and emotional elements of the narrative was poorer after either drug.
Benzodiazepine-induced GABAergic enhancement or scopolamine-induced cholinergic hypofunction results in a loss of the enhancing effect of emotional arousal on memory. Furthermore, lorazepam- and scopolamine-induced memory impairment for both gist (which is amygdala dependent) and detail raises the possibility that their effects on emotional memory do not depend only on the amygdala. We discuss the results with reference to potential clinical/forensic implications of processing emotional memories under conditions of globally impaired episodic memory.
KeywordsAcetylcholineScopolamineGABALorazepamEmotionMemoryEmotional memoryAlzheimer’s diseaseAmnesiaPTSD
Transient impairments in episodic memory are a characteristic effect of benzodiazepines like lorazepam and anticholinergics such as scopolamine. A number of studies have shown that despite acting on distinct neurotransmitter systems, lorazepam and scopolamine produce very similar effects on a variety of memory tasks (see Curran 2000 and Curran and Weingartner 2002 for reviews). However, human psychopharmacological studies examining the amnesic properties of these drugs have tended to focus on neutral rather than emotional material. This is despite the fact that such drugs are commonly used as preoperative sedatives, and their clinical utility depends, to a large extent, on their amnesic action during aversive and emotionally arousing events (e.g., surgery).
Emotional occurrences are generally better remembered than neutral events, suggesting that the processing of emotional stimuli is privileged in some way. The neurochemical basis for this emotional memory advantage has recently been investigated in humans. A number of psychopharmacological and neuroimaging studies suggest that enhancement of memory for emotionally arousing events depends on stress–hormone and noradrenergic modulation of the amygdala (see Chamberlain et al. 2006 for a review), although neurochemical/neuroanatomical data suggest that it is likely that the GABAergic, cholinergic, and opioid neurotransmitter systems are also involved in amygdala modulation and emotional memory (see Cahill and McGaugh 1998 for a review). To date, there seems to be only a single study investigating the role of GABAergic modulation of episodic emotional memory in humans. Buchanan et al. (2003) found that the benzodiazepine, triazolam, differentially affected memory for “gist” (especially of unpleasant stimuli) while leaving memory for “detail” relatively intact.
The specific role of the amygdala in focusing information-processing resources on key aspects of complex emotional stimuli (gist) has been examined recently. It seems that during emotional arousal, investment of information-processing resources on the central gist aspects of complex (emotional) scenes occurs at the expense of processing peripheral detail. Thus, patients with damage to the amygdala show impairment in memory for gist but not for detail (Adolphs et al. 2001, 2005). Taken together with data from animal studies (Tomaz et al. 1992), Buchanan et al. (2003) suggested that triazolam’s effect on emotional memory in humans was mediated through GABAergic enhancement within the amygdala.
Although all benzodiazepines share a number of behavioral effects (e.g., anxiolysis, transient anterograde amnesia), lorazepam seems to be a distinct benzodiazepine in that it also results in implicit memory impairments (see Curran 2000 for a review) and prolongs the processing of visual stimuli (Giersch and Herzog 2004). The distinct behavioral effects of lorazepam may be explained by the possibility that different central GABA receptor subunit combinations could result in different binding preferences for the various benzodiazepines (Möhler et al. 2002). Given this possibility, lorazepam may produce effects on emotional processing, which are distinct from effects that other benzodiazepines, such as triazolam or diazepam, produce (e.g., Kamboj and Curran 2006).
In contrast to the previous study investigating the effect of a benzodiazepine on episodic emotional memory in humans (Buchanan et al. 2003), there appear to be no such studies of the anticholinergic drug scopolamine. This is despite the fact that scopolamine—like benzodiazepines—produces a robust impairing effect on neutral episodic memory. Furthermore, delineating the role of the cholinergic system in emotional memory may improve our understanding of the neuropsychological effects of cholinergic hypofunction in Alzheimer’s dementia, a disease associated with emotional blunting and impaired emotional memory (Budson et al. 2004; Hamann et al. 2000; Kensinger et al. 2002, 2004; but see Kazui et al. 2000 and Moayeri et al. 2000 for reviews). Simulating cholinergic hypofunction using scopolamine in healthy human volunteers results in patterns of mnemonic performance which are similar to some of the memory deficits seen in Alzheimer’s (Curran 2000; Curran and Weingartner 2002). In particular, the anterograde (but not the retrograde) episodic memory impairments of Alzheimer’s disease are mimicked by scopolamine. As such, scopolamine has been considered to be a partial pharmacological model of Alzheimer’s, the effects of which are reversed by commonly used cholinesterase inhibiting antidementia drugs.
Furthermore, the limited but emerging literature on the effects of benzodiazepines and other anxiolytic drugs on emotional processing tasks may provide insights into the mechanisms of action of these drugs in mood disorders (e.g., Blair and Curran 1999; Harmer et al. 2003a,b; Zangara et al. 2002).
The aim of the current study was to compare the effects of lorazepam and scopolamine on human emotional memory because animal studies suggest a key role for GABAergic and cholinergic modulation of the amygdala (see Cahill and McGaugh 1998 for a review). We decided to compare the effects of these two particular drugs for a number of reasons. Firstly, they have both been studied independently and repeatedly shown to safely and reliably produce robust memory impairment (for neutral material). Secondly, comparison studies have established equivalent doses for amnesic and sedative effects (e.g., Curran et al. 1998). Finally, the current report extends our previous findings comparing the effects of lorazepam and scopolamine on emotional processing (in particular, emotion recognition; Kamboj and Curran 2006).
On the basis of previous research on benzodiazepines and neutral episodic memory, one prediction was that lorazepam would result in a reduction in emotional and neutral memory performance, but that the pattern of performance (i.e., the presence of an emotional memory advantage) would be retained (see Curran 2000 for a review). Alternatively, lorazepam may result in impairments in both neutral and emotional memory and result in a reduction in emotional memory enhancement (i.e., have a disproportionate effect on emotional memory). Furthermore, given parallels between the effects of scopolamine on episodic memory in healthy volunteers, it might be expected that scopolamine would produce a similar pattern of impairment to lorazepam. In addition, it was of interest to investigate effects on gist and detail memory with these two drugs. A greater impairing effect on central gist vs peripheral detail would indirectly suggest a preferential action on the amygdala.
Materials and methods
Forty-eight healthy volunteers (24 men) took part in this study and were the same sample of participants from a previous study (Kamboj and Curran 2006). Their mean age was 23±4 years (range: 18–35; there was no age difference between groups, p > 0.5). None was receiving any medication except for oral contraceptive (n = 3), and five were smokers ( ≤ 10/day). All had been screened via a psychiatric history to exclude any with significant current or past mental health or substance misuse problems. The study received ethical approval from the University College London/University College Hospital ethics committee. All participants were fluent in English, and all gave written informed consent and were paid for participating.
Procedure and design
A double-blind, “double-dummy” independent group design was used. Except for balancing groups by gender (eight men and eight women in each), the 48 participants were randomly allocated to one of the three conditions: placebo, 2 mg oral lorazepam, or 0.6 mg subcutaneous scopolamine. These doses (using these modes of administration) have previously been matched for effects on arousal, attention, and memory (Curran et al. 1998). Within each treatment group, we collapsed the data across sexes because previous studies on the effect of sex on emotional memory have produced inconsistent results. Furthermore, a recent study suggests that participants’ “sex-related traits” may be a more important factor in influencing emotional memory than biological sex (Cahill et al. 2004).
All participants received two treatments: an oral capsule containing a placebo (lactose) or lorazepam (formulated in identical red capsules), then followed 1 h later by a subcutaneous injection containing scopolamine or placebo (physiological saline). The injection and capsule were prepared and administered by a third person to maintain double-blind conditions. Participants were asked to refrain from alcohol and other psychoactive substances (including alcohol, caffeine, and tobacco) for 24 h before the test session and from alcohol for 24 h after the session. Before receiving the capsule, participants completed a 16-item visual analogue subjective rating scale measuring alertness–sedation, calmness–anxiety, and contentedness–discontentedness (Bond and Lader 1974). An estimate of verbal IQ was obtained using the NART (Nelson and Willison 1991).
Posttreatment testing began 1 h after the injection to allow peak concentrations of active drugs to be reached (Curran et al. 1998). To determine changes in mood and arousal, a second mood rating scale was completed. Other tasks completed before presentation of the emotional narrative included recall of a story from the Rivermead Behavioral Memory Test (RBMT; Wilson et al. 1985) and an emotion recognition task (Kamboj and Curran 2006).
Emotional narrative encoding and testing
The story used in this experiment was based on the material used by Cahill et al. (1994) and consisted of visual and audio elements (presented on a computer as an 11-picture “slide show”). The story consisted of three phases (P 1, P 2, and P 3). Slides 1–4 made up the first “emotionally neutral” phase (P 1); slides 5–8 constituted the emotional phase (P 2), and slides 9–11 made up the second neutral phase (P 3). The original version of this task (Cahill et al. 1994) contains a number of culturally specific references. Therefore, the task was adapted slightly to replace American words with corresponding words used more commonly in Britain (e.g., “football” instead of “soccer ball”, and “telephone box” instead of “telephone booth”). Immediately after the story presentation, participants were asked to rate how emotionally arousing they found the story on a scale from 0 (not emotional at all) to 10 (extremely emotional).
Participants returned 1 week after the initial test day (i.e., on day 7) and were given an unexpected memory test for the story. Recall was tested by asking participants to write down as much as they could remember about the story. If a participant could not recall anything after 5 min, recall testing was abandoned. Scoring for the recall task involved giving a single point for each key audio or visual idea presented in the story. The emotional phase contained 20 such ideas, and the neutral phases contained 24 idea units (P 1) and 20 idea units (P 3). Recognition memory testing followed the procedure of Cahill et al. (1994) and involved 80 forced choice questions (four options per question), with 6–9 questions per slide. Following Cahill et al. (2004), the questions were divided into those which tapped peripheral detail (61 questions) and central gist (19 questions). In our analysis of gist and detail, we only used data from P 1 and P 2 of the narrative as these contained very similar numbers of questions related to detail (22 or 23) and gist (six each).
Repeated measures ANOVA was used with treatment group as the between-participant factor and story phase as a within-participant factor. Post hoc comparisons were Bonferonni-corrected. Within groups, analyses used two-tailed t tests. Unless indicated, all data were presented as means ± SD.
There was no statistical difference between the groups in terms of estimated verbal IQ (placebo = 108.9 ± 7.2, scopolamine = 110.3 ± 5.2, lorazepam = 111.3 ± 7.4; p > 0.5).
Emotional story recall and recognition
Participants in the three groups did not differ on their rating of emotionality of the story (placebo: 5.2 ± 2.5; scopolamine: 5.3 ± 1.7; lorazepam: 5.3 ± 2.9; F(2,45) = 0.014, p > 0.5).
As expected, in the free recall test, the placebo group showed a significant memory advantage for the emotional story phase (P 2; 3.9 ± 1.9 idea units recalled) compared with neutral P 1 and P 3 (a mean of 2.8 ± 1.2 idea units for both phases; t(15) = 2.91, p < 0.02). Both the scopolamine and lorazepam groups showed floor effects for all story phases (mean of < 0.5 idea units recalled). Therefore, recall data were not analyzed further.
Central gist and peripheral detail memory
A similar pattern emerged with memory for peripheral detail of the narrative. Thus, detail memory was impaired in the lorazepam and scopolamine groups compared with placebo (F(2,45) = 14.713, p < 0.001), but there was no main effect of story phase (F(1,45) = 1.47, p = 0.232) nor a phase × drug group interaction (F(2,45) = 2.24, p = 0.118).
We have reported in detail the results of the control (neutral) memory task (RBMT) and the subjective effects of lorazepam and scopolamine in our previous study (Kamboj and Curran 2006). Briefly, as expected, there was a treatment × time interaction for the “alert-sedated” mood factor indicating significant sedation after lorazepam or scopolamine compared with placebo (F(2,45) = 13.9, p < 0.001). Furthermore, neutral episodic memory (RBMT prose task) was impaired for immediate (F(2,45) = 14.4, p < 0.001) and delayed recall (F(2,45) = 20.3, p < 0.001) as expected.
Correlational analysis (post hoc corrected) showed no significant association between subjective effects and memory performance in any of the treatment groups.
The current study demonstrates a “blunting” of emotional memory in healthy human volunteers after an acute dose of either scopolamine or lorazepam. Neither drug had an effect on the perceived emotionality of the narrative stimuli used in this study, suggesting that a general “emotional blunting” did not occur after drug administration. This is also supported by our previous finding that lorazepam does not produce impairment in emotion recognition of facial expressions, and scopolamine produces specific rather than general impairments (Kamboj and Curran 2006). However, it should be noted that our current findings relate only to recognition memory, as the recall task showed floor effects after either drug.
While numerous studies have compared the effects of lorazepam and scopolamine on various cognitive tasks and consistently shown transient impairments in neutral episodic memory (which was also found in this study, i.e., impaired recognition memory for story phases 1 and 3), this is the first study that we are aware of which compares the effects of these drugs on emotional episodic memory in humans. Our results complement those of Buchanan et al. (2003)—which also suggested a role for GABA-ergic modulation of emotional memory— but also support a role for cholinergic involvement in emotional memory. These findings may have implications for our understanding of the neurochemical basis for emotional memory impairments found in “organic amnesias”. Of particular relevance to the current study are the findings from Alzheimer’s disease patients, which have shown emotional memory impairments consistent with those found in this study with scopolamine (e.g., Budson et al. 2004; Hamann et al. 2000; Kensinger et al. 2002, 2004). Indeed, Alzheimer’s may be associated with severe degeneration of cholinergic neurons in medial temporal lobe structures including the amygdala and hippocampal regions (Emre et al. 1993). Interaction between these structures is crucial for memory enhancement of emotional events (see Phelps 2004 for review).
The “cholinergic hypothesis” of Alzheimer’s disease has been highly influential, and treatment approaches for Alzheimer’s continue to focus on enhancing cholinergic activity using cholinesterase inhibitors (see Curran and Weingartner 2002 and Mesulam 2004 for reviews). As an antagonist of central muscarinic receptors, scopolamine produces some of the amnesic features of Alzheimer’s disease in healthy volunteers (Curran 2000; Curran and Weingartner 2002). The current study demonstrates a wider range of episodic memory impairments produced by scopolamine in humans, encompassing emotional episodic memory, thus potentially providing a more general pharmacological model of the episodic memory impairments of Alzheimer’s disease. It would be of clinical interest to determine whether the amelioration of cognitive decline seen with cholinesterase inhibitors in Alzheimer’s extends to emotional memory. This would further support a role for cholinergic modulation of emotional memory.
Our findings with lorazepam were similar to those with scopolamine. However, unlike Buchanan et al. (2003), who found that the benzodiazepine triazolam resulted in a relatively specific impairment of gist memory for unpleasant stimuli (thus indirectly supporting a preferential effect of triazolam on the amygdala), our data show that both central gist and peripheral detail memory are impaired after lorazepam or scopolamine. This raises the interesting possibility that while these drugs may interfere with processing within the amygdala (as suggested by impairment of gist aspects of the narrative), this is not a preferential effect, and other neuroanatomical pathways or structures involved in emotional memory (especially those involved in the processing and recollection of detail aspects of complex scenes) may also be affected. We found this global effect (on gist and detail) despite using an experimental protocol which favored a “gist memory advantage” (i.e., incidental encoding, and a relatively long delay between encoding and recollection; see Adolphs et al. 2001). On the other hand, Buchanan et al. (2003) gave participants explicit instructions to remember the stimuli (nonincidental encoding) and used a relatively short delay interval (48 h) but nonetheless found a specific effect for gist memory. It is possible that the discrepancy between our results and those of Buchanan et al. (2003) could be explained by experimental differences or by a relative lack of power in the Buchanan et al. (2003) study, resulting in a failure to detect a difference between the placebo and triazolam groups on detail memory. Further work is needed to distinguish between these possibilities, and the possibility that real differences exist between lorazepam and triazolam in terms of their effects on emotional memory. Furthermore, our interpretation of the gist and detail data must remain speculative. It would be interesting if future studies supplement a pharmacological approach with a neuroimaging methodology to examine the affected neuroanatomical substrates more directly (e.g., Bentley et al. 2003).
Studying the effect of emotion on memory is complicated by the fact that complex stimuli which produce emotional responses tend to be highly distinctive and semantically cohesive (e.g., see LeBar and Phelps 1998). They may also be more likely to increase attention and be more frequently rehearsed. In the current study, we employed a popular emotional memory task (Cahill et al. 1994), which nonetheless has some limitations. In particular, although the stimuli presented in the neutral and emotional phases are likely to differ in the emotional response they produce, they may also differ in other variables which often covary with emotionality (namely, visual complexity, distinctiveness, and semantic cohesion). This issue of confounding variables has recently been addressed by Adolphs et al. (2005) who tested memory for the same neutral elements of a narrative presented either in a neutral or an emotional context. It would clearly be of interest for future pharmacological studies to use a similar approach to isolate the effects of drugs on emotional arousal per se.
It is possible that the effects of scopolamine and lorazepam on emotional memory were partially due to sedation. The effects of reduced central arousal—and more specifically sedation—on emotional memory are not well established. Although a number of experimental paradigms suggest that the memory and arousal effects of scopolamine and lorazepam can be dissociated (Curran 2000), these have only been applied to neutral episodic memory. In the current study, we found no correlation between memory performance and subjective arousal in any of the treatment groups, indirectly suggesting that the effects of lorazepam and scopolamine on emotional memory were not related to sedation. Future studies may clarify the role of general central arousal (or sedation) by comparing the effects of scopolamine and lorazepam either with an active control for sedation or with conjointly administering compounds which reverse sedative effects. In the current study, we used only single doses of lorazepam or scopolamine (which have previously been established to reliably produce an anterograde amnesic effect; Curran et al. 1998). Future studies may also use lower doses of these drugs not only to reduce the effects of sedation but also to potentially eliminate floor effects on recall and to determine if more specific effects on gist and detail recognition memory might emerge.
In addition to having potential relevance to our understanding of the neurochemical basis of organic amnesias, our findings may have further clinical and forensic significance. For example, our data suggest that there is a neuropsychological rationale for the use of these compounds in clinical settings where aversive memories might be formed (e.g., during minor surgery) above and beyond their well-documented sedative and amnesic effects for neutral material. On the other hand, one might speculate that such emotional memory impairment may have complicating consequences when administration of an amnesic agent precedes an aversive or traumatic incident. For example, episodic memories for traumatic events may be unreliable and fragmented and, hence, potentially susceptible to confabulation (Bryand and Harvey 1998). This may be an important factor to consider in formal (legal) investigation of memory for psychologically traumatic events. For example, surreptitious administration of benzodiazepines has been implicated in cases of “drug-assisted rape” (Anglin et al. 1997) in which the victim can only provide fragmented accounts of the crime. In these cases, benzodiazepines are less likely to affect implicit memory and as such, implicit, “somatic memories” of traumatic incidents may still be formed. A lack of “integration” of memory components (essentially, the implicit and episodic components of memory) for a traumatic event may be one factor underlying the maintenance of posttraumatic stress disorder symptoms (Brewin and Holmes 2003) and may be especially pronounced after a traumatic incident when an amnesic agent such as a benzodiazepine has been administered (e.g., in cases of drug-assisted rape). It is likely therefore that psychological treatment approaches to such trauma may need to be altered to accommodate the paucity of episodic memory and may be more akin to treatments for simple phobias in which there is also, generally, a lack of explicit memory for the onset of fear for the phobic object.
We acknowledge that in the current study, we have used what may be considered “low-level” traumatic imagery in the emotional section of the story. Future research may seek to further elucidate the effects of scopolamine and benzodiazepines by examining more “ecologically valid” stimuli (e.g., Holmes et al. 2004) to determine their effects on the formation of traumatic imagery and memories.
SK was supported by a Wellcome Trust International Prize Travelling Research Fellowship. We thank Dr. Larry Cahill for providing us with a scoring system for central and peripheral elements of the narrative task.