Effects of antidepressants,anesthetics and sedatives/hypnotics on sleep-associated memory consolidation – A systematic review

Inclusion and exclusion criteria

This research focuses on the influence of pharmaceuticals on memory consolidation. The original studies had to meet specific requirements to ensure their design was suitable for identifying the influence of the pharmaceutical on the consolidation phase. To exclude that any effect on memory consolidation was due to a modulatory effect on learning or retrieval processes the timing of drug application was required to abide by two rules: (1) It must occur after termination of the encoding phase and (2) a sufficient time (retention) interval must pass before the (recall) test to avoid any influence of the drug on the retrieval process. These temporal aspects are reflected in the inclusion and exclusion criteria I3, I7, E9. The latter takes drug half-life into account (Tables 1 and 2).

OPEN IN VIEWER

Table 1.

Inclusion criteria.

Inclusion criteria	Description
I1	Approved drug
I2	Rat, mice, and human interventional studies
I3	Drug must have been administered after training
I4	Healthy subjects
I5	Control condition
I6	Six or more subjects
I7	Sleep has been measured by EEG or else there was a retention phase of at least 5 hours

OPEN IN VIEWER

Table 2.

Exclusion criteria.

Exclusion criteria	description
E1	Application of more than one drug at a time
E2	Case studies, field studies
E3	Human children as study population; mice and rats had to be at least 4 weeks old
E4	Reviews
E5	Studies on reconsolidation
E6	Additional non-pharmacological interventions (i.e., simultaneously)
E7	Studies on extinction consolidation
E8	Articles in other languages than English or German
E9	Repetitive drug application (except if all applications take place before sleep and the drug half-life is short enough to rule out drug effects during recall – single case decision)

Notably, sleep EEG and stages were rarely reported, thus necessitating an inclusion criterion based on a minimal retention time. Figure 1 reveals that despite differences in temporal structure and circadian phase (Soltani et al., 2019), sleep in all three species includes rapid eye movement (REM) and NREM sleep (Bastianini et al., 2014).

OPEN IN VIEWER

Figure 1.

Cross-species NREM and REM sleep composition. Exemplary sleep architecture of an 8-hour sleep period in a human subject (top), and of a 12-hour lights-on period in a rat (middle) and mouse (bottom).

When subject age was not provided, an adult population was assumed. All included studies were grouped by drug group (antidepressants, anesthetics, and sedatives/hypnotics) and further separated according to the individual drug, species, and memory task.

Search methodology

First, key terms were identified. These included not only drug names but also drug group names such as “antihistamines” or “antidepressants.” This ensured that all substances belonging to the searched group were identified. All terms describing the drugs or drug groups were combined with the operator “OR” and the key terms “memory” and “consolidation” were added by using the operator “AND.”

On the chosen databases, PubMed and Web of Science, the final searches were executed on the following dates:

The complete search terms can be found in the Supplemental Material.

Studies on anesthetics and sedatives/hypnotics were searched collectively because of the similar subject area and later separated according to drugs at the level of the full-text screening. Morphine was initially included in the search for anesthetics because it is used intraoperatively. According to the anatomical therapeutic chemical (ATC) classification, however, it does not belong to the general anesthetics and was therefore subsequently removed from the analysis. Some benzodiazepines are not classified under hypnotics and sedatives in the ATC code, but under anxiolytics. Nevertheless, all benzodiazepines were included in the evaluation to represent the entire class of chemical substances. The PubMed MeSH terms function also automatically completes the search queries. The exact search history including each inclusion or exclusion decision is available upon request.

Selection of studies included

For the purpose of this review the definition for antidepressants is any drug that is included in the German ATC code N06A (without N06AP (herbal antidepressants) or N06AH (homeopathic antidepressants)); N05AN01 (Lithium), an antipsychotic, was additionally included because it is used as maintenance therapy not only for bipolar but also for unipolar depression (Abou-Saleh et al., 2017; Coppen, 2000). An anesthetic is defined as any drug included in the German ATC code N01A, and hypnotics and sedatives are defined as any drug included in the German ATC code N05C (without N05CP (herbal hypnotics and sedatives)); N05BA01 (Diazepam) and N05BA06 (Lorazepam) were additionally included. Ketamine potentially fits into two categories. Only (S)-Ketamine is included in the antidepressant category because no preparations for depression containing the racemate or (R)-Ketamine are approved. Preparations with the ketamine racemate are included in the anesthetics category, since the racemate is used in clinical practice. Additional information can be found in the Supplemental Tables 1 and 2.

The program Rayyan, which allows co-working of several editors was used to manage literature during the review process (Ouzzani et al., 2016). For the study selection, first all articles were screened by title and abstract after duplicates were removed. The screening was performed blindly by two independent researchers. The articles were rated: “included,” “excluded,” or “maybe.” Articles that received the rating “included” or “maybe” from at least one researcher were reviewed as a full-text article in a second step. The number of articles excluded in this second step is shown in the Supplemental Material with reasons for their exclusion (Supplemental Figures 2 and 3). All articles receiving discordant ratings from reviewers following full-text screening were reviewed by a third researcher, who made the final decision on study inclusion or exclusion. The third researcher was not blinded to the previous reviewers’ decisions but rather read the given reasoning.

Methods of data extraction

Data was extracted from the articles manually by one researcher using a predefined data extraction sheet. Data were compiled in an excel file.

Risk-of-bias analysis

In animal studies, use of methods to minimize risk of bias is not yet standard practice, hence neither is their reporting. Also, some aspects of randomized controlled trials in humans cannot be accounted for in animal studies. Since this review includes animal and human interventional studies we selected the Syrcle risk-of-bias tool for animal studies that is best appropriate for the purpose of this review (Hooijmans et al., 2014). The following items were chosen: D1: Were baseline characteristics between groups similar? (selection bias); D2: Randomization? (selection bias); D3: Were caregivers/investigators blinded? (performance bias); D4: Were outcome assessors blinded? (detection bias); D5: Were subjects/data excluded and why? (attrition bias). The studies were assessed by one researcher. A random sample of about 33% of the studies was also assessed by a second researcher independently.

The results of the risk-of-bias assessment were then visualized using an online tool (McGuinness and Higgins, 2021).

Synthesis

Studies are summarized in Tables 3 to 5. The study characteristics include author, publication year, investigated drug, study population, their sex and age, time of experiment, retention interval, use of EEG/local field potential (LFP)/polysomnography (PSG) for sleep measurement, memory task, retention performance, effect size, sample size and mechanism of action. Each pharmaceutical category was grouped individually.

Quantitative analysis

Statistical analysis was only conducted when more than two studies were found that were homogeneous in the categories:

Available data (means and standard deviations, or alternatively t-values and number of subjects) were extracted from the studies. Statistical analyses were conducted using Meta-Essentials v1.5 (Suurmond et al., 2017), “workbook 3 – differences between independent groups with continuous data” CC BY (https://creativecommons.org/licenses/by-nc-sa/4.0/). Weights were automatically assigned to the studies. A random effects model was chosen due to different retention intervals. Assuming that consolidation is a time-dependent process, measured effects likely differ if retention intervals vary. Hedges’ g was chosen as the effect measure, because it provides less distorted estimates in small samples than Cohens’ d.

Precision of the results was discussed using the 95% confidence interval (CI). Heterogeneity was discussed using I², Tau, and the prediction interval (PI).

Antidepressants

For antidepressants, 25 reports of 25 studies were included in the analysis, a quantitative analyses was not performed in this category. The results for all antidepressants are shown in Table 3. Five human studies and 20 animal studies were found. Fourteen different antidepressants were investigated. Memory tasks included passive avoidance tasks, novel object recognition tasks, fear conditioning, finger tapping tasks, mirror tracing tasks, visual discrimination tasks, autoshaping, and word paired-associate tasks (WPA). Three studies did not state subjects’ sex, four studies included female and male subjects, and most studies only included male subjects. Six studies used EEG/LFP/PSG measurements. Across all studies measuring EEG and/or PSG the suppression of REM sleep was most characteristic.

OPEN IN VIEWER

Table 3.

Results antidepressants.

CrRef	Author and year	Drug	Study population, sex, age	Experiments: daytime or nighttime, Retention interval	EEG/LFP/PSG: yes or no	Memory task	Retention performance against control	Main mechanism of action a	Effect size, sample size
§	Diaz-Mataix et al. (2014)	Agomelatine	Rats (strain not stated), not stated, not stated	2 h before dark onset 24 h	No	FC	↔ 40 mg/kg	Melatonergic agonist	ES: not stated, n: not stated
	Goerke et al. (2014)	Amitriptyline	Human, male 26.8 ± 5.6 years	-, 24 h	Yes, PSG: REMS decreased; REMS-lat., N2 increased	VDT; FT; Rey-Osterrieth complex figure test; Rey auditory verbal learning test	↓ 75 mg p.o. VDT; ↔ all other tests	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: Cohen’s d = 0.89, n: 25 (parallel design; n (drug) = 12)
	Gvilia et al. (2003)	Amitriptyline HCl	Rats (strain not stated), male, adult	-, 2 days	No	active avoidance test	↔ 8 mg/kg i.p.	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: unclear
	Everss et al. (1999)	Amitriptyline	OF1 mice, male and female, 42 days	Dark phase, inversed dark/light cycle, 24 h	No	PA	↓ 7.5 mg/kg i.p. (f); ↓ 7.5 mg/kg i.p. (m); ↔ 15 mg/kg i.p. (f);↓ 15 mg/kg i.p. (m);↓ 30 mg/kg i.p. (f); ↓ 30 mg/kg i.p. (m)	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: 124 (n = 15–16 per group)
	Everss et al. (2005)	Amitriptyline	CD1 mice, male and female, 42 days	Dark phase, inversed dark/light cycle, 24 h	No	PA	↓ 30 mg/kg injection	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: 10–12 per group
	Parra et al. (2002)	Amitriptyline	CD1 mice, male and female, 42 days	Dark phase, inversed dark/light cycle, 24 h	No	PA	↓ 7.5 mg/kg i.p. (f); ↓ 7.5 mg/kg i.p. (m); ↔ 15 mg/kg i.p. (f); ↓ 15 mg/kg i.p. (m);↓ 30 mg/kg i.p. (f); ↓ 30 mg/kg i.p. (m)	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: 48 male and 48 female mice (n = 12 per group)
	Kruk et al. (2011)	Bupropion	naive Swiss mice, male, Not stated	Light phase, 24 h	No	Elevated plus maze	↔ 10 mg/kg i.p.; ↔ 20 mg/kg i.p.; ↔ 40 mg/kg i.p.	NARI	ES: not stated
									n: 7–10 per group
+	Feltmann et al. (2015)	Citalopram	Sprague–Dawley rats, male, Not stated	Light phase 24 h	No	NOR	↑ 10 mg/kg	SSRI	ES: not stated, n: 5–8 per group
	Montezinho et al. (2010)	Escitalopram	Sprague–Dawley rats, male, Not stated	Light phase 24 h	No	CFC	↔ 0.5 mg/kg s.c.; ↔ 1 mg/kg s.c.; ↑ 5 mg/kg s.c.	SSRI	ES: not stated, n: 8–10
+	Feltmann et al. (2015)	Desipramine	Sprague–Dawley rats, male, not stated	Light phase, 24 h	no	NOR	↔ 10 mg/kg	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: 5–8 per group
	Watts et al. (2012)	Desipramine	Fisher 344 rats, male, 6 months	Light phase, 24 h (5 days)	Yes, intracranial EEG:REMS, PreREMS decreased	Familiar eight box maze; novel eight box maze; t-maze task	(↓ 10 mg/kg p.o. familiar eight box maze; ↓ 10 mg/kg p.o. novel eight box maze; ↑ 10 mg/kg p.o. t-maze task); Experiments were conducted over several days. It is unclear whether results were significant after the first day.	Monoamine reuptake inhibitor (5-HT and NA) with affinity to bind to α-adrenergic, histaminergic (H1) and muscarinic (M1) receptors	ES: not stated, n: 30 rats (n (placebo) = 15)
	Pereira et al. (2009)	Duloxetine	SR-1 mice, male, 2–3 months	Normal cycle, 24 h	no	PA	↔ 10 mg/kg i.p.; ↔ 20 mg/kg i.p.	SNRI	ES: not stated, n: 10–14 per group
	Introini-Collison et al. (1992)	Fluoxetine	Sprague–Dawley rats, male, not stated	Light phase, normal cycle, 48 h	no	PA	↔ intraamygdala: 1.0; 3.0 and 10.0 µg; ↑ 15 mg/kg i.p.	SSRI	ES: not stated, n: 9–10 per group
	Lee et al. (1992)	Fluoxetine	Sprague–Dawley rats, male, not stated	Light phase, 24 h	no	PA	↔ 3 µg; ↑ 6 µg; ↔ 12 µg; application in all cases into the lateral septum, bilaterally, per side	SSRI	ES: not stated, n: 14–18
	Meneses (2007)	Fluoxetine	Wistar rats, male, 12 weeks	Normal cycle, 24 h	no	Autoshaping	↔ 1 mg/kg i.p.; ↔ 5 mg/kg i.p.; ↑ 10 mg/kg i.p.; ↓ 10 mg/kg i.p. 6 h after learning	SSRI	ES: not stated, n: 8 per group
	Flood and Cherkin (1987)	Fluoxetine	CD1 mice, male, not stated	Light phase, 1 week	no	PA; T-maze test	T-maze: ↔ 2.5 mg/kg s.c. ↑ 5 mg/kg s.c. ↑ 10 mg/kg s.c. ↑ 15 mg/kg s.c.; ↑ 15 mg/kg s.c. 30min after learning; ↑ 15 mg/kg s.c. 60 min after learning; ↔ 15 mg/kg s.c. 90 min after learning; ↔ 4 µg i.c.v. ↑ 8 µg i.c.v. ↑ 16 µg i.c.v. ↑20 µ i.c.v. PA: ↔ 5 mg/kg s.c. ↔ 10 mg/kg s.c.; ↑ 15 mg/kg s.c.; ↔ 20 mg/kg s.c.; ↔ 25 mg/kg s.c.	SSRI	ES: not stated, n: 20 per group
	Monleon et al. (2001)	Fluoxetine	OF1 mice, male and female, not stated	Dark phase, reversed cycle, 4 days	no	PA	↔ 5 mg/kg i.p.; ↔ 10 mg/kg i.p.; ↑ 20 mg/kg i.p.	SSRI	ES: not stated, n: 12 per group
	Pommer et al. (2006)	Fluvoxamine	Human, male, young	-, 36 h	Yes, not specified	FT, MT	↔ 50 mg p.o.	SSRI	ES: not stated, n: 14
+	Rasch et al. (2009)	Fluvoxamine	Human, male, mean age 25.7 ± 0.7 years, range 18–36 years	-, 32 h	Yes, PSG: REMS decreased	FT, MT, WPA	50 mg p.o., ↑ FT (accuracy); ↔ MT; ↔ WPA	SSRI	ES: not stated, n: 16
**	Gais et al. (2008)	(S)-Ketamine	Human, male, 24.9 ± 0.8 years	-, ~10 h	Yes, PSG: no effect	VTD	↓ 0.25 mg/kg BW i.v. over 6 h	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 15
**	Ide et al. (2019)	(S)-Ketamine	C57BL/6 mice, not stated, >18 weeks	-, 24 h	No	NOR	↓ 10 mg/kg i.p.; ↓ 20 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10–11
	Amiri et al. (2020)	Lithium chloride	Wistar rats, male, adult	Light phase, normal cycle, 24 h	No	PA	↔ 10 mg/kg i.p.; ↓ 20 mg/kg i.p.; ↓ 30 mg/kg i.p.; ↓ 40 mg/kg i.p.	Unknown, several proposed	ES: not stated, n: 8 per group
	Zarrindast et al. (2008)	Lithium chloride	Albino MNRI mice, male, adult	Light phase, 24 h	No	PA	↓ 10 mg/kg i.p.	Unknown, several proposed	ES: not stated, n: 10 per group
	Walker and Foley (2010)	Mianserin	Swiss-Webster mice, male, Not stated	-, 24 h	No	autoshaping	↔ 10 mg/kg i.p. after a 1 or 2 h learning session; ↓ 32 mg/kg i.p. after a 1 or 2 h learning session;	Presynaptic α receptor antagonist	ES: not stated, n: 5–12 per group
	Gais et al. (2011)	Reboxetine	Human, male, 25.7 ± 1.3 years, range 18 and 35	-, ~34 h	Yes, PSG: REMS decreased; N2, W increased	Odor recognition task; WPA	2 mg p.o., ↑ Odor recognition; ↔ WPA	NARI	ES: d = 0.7 (odor recognition), n: 16
+	Rasch et al. (2009)	Reboxetine	Human, male, Mean age 25.7 ± 0.7 years, range 18–36 years	-, 32 h	Yes, PSG/EEG: REMS decreased; N2, SpiDens, W increased	FT, MT, WPA	2 mg p.o. ↑ FT (accuracy); ↔ MT; ↔ WPA	NARI	ES: not stated, n: 16
+	Feltmann et al. (2015)	Reboxetine	Sprague–Dawley rats, male, not stated	Light phase, 24 h	No	NOR	↑ 3 mg/kg	NARI	ES: not stated, n: 5–8 per group
	Alvarez-Ricartes et al. (2018)	Sertraline	C57BL/6 mice, male, 3 months	-, The next day	No	FC	↔ 3 mg/kg p.o. 2 h after training	SSRI	ES: not stated, n: 5–8

↓: significantly impaired consolidation; ↑: significantly enhanced consolidation; ↔: no significant effect on consolidation; CrRef, Cross references; PSG: polysomnography; LFP: local field potential; PA: passive avoidance task; NOR: novel object recognition task; FC: fear conditioning; CFC: contextual fear conditioning; FT: finger tapping task; MT: mirror tracing task; VTD: visual discrimination task; WPA: word paired-associate task; 5-HT: serotonin; NA: noradrenaline; SSRI: selective serotonin reuptake inhibitor; NARI: noradrenaline reuptake inhibitors; SNRI: serotonin-noradrenaline reuptake inhibitors; p.o.: per os; s.c.: subcutaneous; i.p.: intraperitoneal; i.c.v.: intracerebroventricular; ES: effect size; n: sample size; SWS: slow-wave sleep; N2: sleep stage 2 (S2 or N2); SpiDens: Spindle density; REM: rapid eye movement; REMS: REM sleep; lat.: latency; PreREMS: pre REM sleep **: also, in the results for anesthetics; §: study was funded by Servier laboratories; +: study was shown more than once, since more than one drug was investigated; -: Information not stated.

a

Data compiled based on information from Geisslinger et al. (2020) and Goodman (2005).

Dividing the antidepressants into groups, with one containing second-generation (selective serotonin reuptake inhibitor (SSRI), noradrenaline reuptake inhibitor (NARI), serotonin-noradrenaline reuptake inhibitor (SNRI)), and one containing antidepressants developed longer ago (tricyclic and tetracyclic antidepressants), shows a tendency for improved consolidation with second-generation antidepressants and a tendency toward impairment in the other group.

Anesthetics

For anesthetics, 24 reports of 23 studies were included in the analysis. Quantitative analyses were not performed in this group. The results for anesthetics are shown in Table 4. Four human studies and 19 animal studies were found. Six different anesthetics were investigated, and the memory tasks included WPA, finger tapping tasks, 2D face location tasks, visual texture discrimination tasks, novel object recognition tasks, Morris water maze and passive avoidance tasks. One study did not state subjects’ sex, four studies included female and male subjects and most studies only included males. Six studies used EEG/LFP/PSG measurements. Sodium oxybate was investigated in four of these, increased SWS and/or decreased spindle density were most characteristic.

OPEN IN VIEWER

Table 4.

Results anesthetics.

CrRef	Author and year	Drug	Study population, sex, age	Experiments: daytime or nighttime, Retention interval	EEG/LFP/PSG: yes or no	Memory task	Retention performance against control	Main mechanism of action a	Effect size, sample size
	Yang et al. (2021)	Isoflurane	C57BL/6J mice, male and female, 3–9 months	-, 24 h	Yes, LFP, dCA1: Power >4 Hz reduced; power <0.5 Hz increased	MWM	↔ 2% for 1 h; applied 30 min after acquisition	Several, including positive allosteric modulation of GABAA	ES: not stated, n: 10–14 per group (from Supplemental Material)
	Goulart et al. (2010)	Ketamine	Wistar rats, male, 70–95 days	Light phase, normal cycle, 24 h	no	NOR	↔ 4 mg/kg i.p.; ↓ 8 mg/kg i.p.; ↓ 20 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10–15 per group
** +	Ide et al. (2019)	Ketamine	C57BL/6 mice, Not stated, >18 weeks	-, 24 h	no	NOR	↔ 10 mg/kg i.p.; ↓ 20 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10–11 per group
	Liu et al. (2014)	Ketamine	Sprague–Dawley rats, male, adult	Light phase, 24 h	no	MWM	↔ 25 mg/kg i.p.; ↔ 50 mg/kg i.p.; ↓ 100 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10 per group
	Liy-Salmeron and Meneses (2008)	Ketamine	Wistar rats, male, 12 weeks	Normal cycle, 24 h	no	Autoshaping	↔ 10 mg/kg s.c.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
	Liy-Salmeron and Meneses (2007)	Ketamine	Wistar rats, male, 12 weeks	Normal cycle, 24 h	no	Autoshaping	↔ 10 mg/kg s.c.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
	Moosavi et al. (2012)	Ketamine	Sprague–Dawley rats, male, Not stated	Light phase, 24 h	no	MWM	↔ 15 mg/kg i.p.; ↓ 100 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
*	Morena et al. (2017)	Ketamine	Sprague–Dawley rats, male, Not stated	Light phase, normal cycle, 48 h	no	PA	↔ 100 mg/kg i.p.; ↑ 125 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 12–13
	Morena et al. (2021)	Ketamine	Sprague–Dawley rats, male, Not stated	Light phase, normal cycle, 48 h	no	PA	↑ 125 mg/kg i.p.; ↔ 125 mg/kg i.p. 3 h after learning	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
	Pitsikas and Carli (2020)	Ketamine	Wistar rats, male, 3 months	Light phase, normal cycle, 24 h	no	NOR	↓ 8 mg/kg i.p.; ↓ 20 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10 per group
	Radford et al. (2018)	Ketamine	Sprague–Dawley rats, male, adult	Dark phase, inversed cycle, 48 h	no	AFC	↔ 0.5 mg/kg bolus plus 2 mg/kg i.v. over 2 h; ↑ 2 mg/kg bolus plus 10 mg/kg i.v. over 2 h; ↔ 2 mg/kg bolus plus 20 mg/kg i.v. over 2 h; ↔ 10 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 7–11 per group
	Khanegheini et al. (2015)	Ketamine	NMRI mice, male, Not stated	Light phase, normal cycle, 24 h	no	novelty recognition	↔ 0.005 mg/kg i.p.; ↔ 0.01 mg/kg i.p.; ↓ 0.1 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
	Saha et al. (1990)	Ketamine	Swiss strain albino mice, male, Not stated	-, 24 h	no	PA	↔ 3 mg/kg i.p.; ↓ 10 mg/kg i.p.; ↓ 30 mg/kg i.p.	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 8 per group
**	Gais et al. (2008)	(S)-Ketamine	Human, male, 24.9 ± 0.8 years	-, ~10 h	yes, PSG: no effect	VTD	↓ 0.25 mg/kg BW i.v. over 6 h	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 15
** +	Ide et al. (2019)	(S)-Ketamine	C57BL/6 mice, Not stated, >18 weeks	-, 24 h	no	NOR	↓ 10 mg/kg i.p.; ↓ 20 mg/kg i.p.;	Several, including glutamatergic (NMDA) antagonism and HCN1 blockade	ES: not stated, n: 10–11 per group
*	Hauer et al. (2011)	Propofol	Sprague–Dawley rats, male, not stated	Light phase, normal cycle, 48 h	no	PA	↔ 250 mg/kg i.p.; ↑ 300 mg/kg i.p.; ↑ 350 mg/kg i.p.; ↑ 300 mg/kg i.p. 30 min after learning; ↔ 300 mg/kg i.p. 90 min after learning; ↔ 300 mg/kg i.p. 180 min after learning	Unclear, GABAA involvement postulated	ES: not stated, n: 9–13 per group
	O’Gorman et al. (1998)	Propofol	Wistar rats, male, 80 days	-, 12 h	no	PA	↔ 100 mg/kg i.p. application 3 h after training, ↓ 150 mg/kg i.p. application 0, 1, 2 and 3 h after training	GABAA receptor modulator, postulated NMDA antagonism	ES: not stated, n: 6 per group
	Zhang et al. (2013)	Propofol	Sprague–Dawley rats, male, not stated	Light phase, 24 h	No	MWM	↔ 75 mg/kg i.p.; ↓ 150 mg/kg i.p.	GABAA receptor modulator, postulated NMDA antagonism	ES: not stated, n: 12 per group
	Li et al. (2011)	Sevoflurane	Sprague–Dawley rats, male, adult	Light phase, normal cycle, 24 h	No	Continuous multiple trial inhibitory avoidance	↓ 2% for 2 h	Several, including positive allosteric modulation of GABAA	ES: not stated, n: 10–11 per group
	Liu et al. (2010)	Sevoflurane	Sprague–Dawley rats, male, adult	Light phase, normal cycle, 24 h	No	Continuous multiple trial inhibitory avoidance	↔ 0.5% for 2 h; ↔ 1% for 2 h; ↓ 2% for 2 h	Several, including positive allosteric modulation of GABAA	ES: not stated, n: 10 per group
	Nakamura et al. (2016)	Sevoflurane	C57BL/6 mice, male, 4 weeks	Light phase, normal cycle, 24 h	No	PA	↑ 2.5% for 3 h	Several, including positive allosteric modulation of GABAA	ES: not stated, n: 16
*	Kaestner et al. (2013)	Sodium oxybate	Human, male and female, 18–39 years	-, 9 h	Yes, PSG/EEG: SWS increased; N1, SpiDens decreased	emotional recognition memory	↔ 2.5 g	Several, including GABAB agonism	ES: not stated isolated for sodium oxybate, n: 28 (15 women)
*#1	Mednick et al. (2013)	Sodium oxybate	Human, male and female, 22 ± 3 years	-, 9 h	Yes, PSG/EEG: SWS increased; N1, SpiDens decreased	WPA, VTD, FT	↔ 2.5 g all tasks	Several, including GABAB agonism	ES: not stated for the memory tests, n: 30 (15 females)
*#1	Niknazar et al. (2015)	Sodium oxybate	Human, male and female, 22 ± 3 years	-, 9 h	yes, PSG	WPA	↔ 2.5 g	Several, including GABAB agonism	ES: not stated for the memory test, n: 28 (14 females)
	Vienne et al. (2012)	Sodium oxybate	Human, male, 23.5 ± 1.6 years	-, ~5 h	Yes, PSG: TST, SWS, REMS increased; REMS-lat., N1 decreased	WPA, 2DFL, FT	↔ 30 mg/kg p.o. all tasks	Several, including GABAB agonism	ES: not stated, n: 13 (no females)

↓: significantly impaired consolidation; ↑: significantly enhanced consolidation; ↔: no significant effect on consolidation; i.v.: intravenous; i.p.: intraperitoneal; s.c.: subcutaneous; p.o.: per os; CrRef: Cross references; PSG: polysomnography; LFP: local field potential; AFC: auditory fear conditioning; WPA: word paired-associate task; FT: finger tapping task; 2DFL: 2D face location task; VTD: visual texture discrimination task; NOR: novel object recognition task; MWM: Morris water maze; PA: passive avoidance task; ES: effect size; n: sample size; SWS: slow-wave sleep; N2: sleep stage 2 (S2/N2); SpiDens: Spindle density; REMS: REM sleep; lat.: latency; TST, Total sleep time; dCA1: dorsal CA1 region**: also in the results for antidepressants *: also in the results of hypnotics and sedatives; #1: reports of the same study; +: paper shown twice; -: Information not stated.

a

Data compiled based on information from Geisslinger et al. (2020) and Goodman (2005).

Sedatives/hypnotics

For hypnotics and sedatives, 24 reports of 22 studies were included in the analysis and five reports of three studies were subjected to quantitative analysis. The results for all sedatives/hypnotics are shown below in Table 5. Eleven human drug studies and 11 animal drug studies were found. Ten different sedatives/hypnotics were investigated, and the memory tasks included WPA, passive avoidance tasks, auditory fear conditioning, contextual fear conditioning, visual texture discrimination task and finger tapping tasks. One study did not state the subjects’ sex, eight studies included females and males and most studies only included males. Nine studies used EEG/LFP/PSG measurements. Across all human studies measuring EEG/PSG an increment in SWS, N2 or spindle density often concurrent with a decrease in REM sleep or an increase in REM sleep latency were most frequently reported.

OPEN IN VIEWER

Table 5.

Results sedatives/hypnotics.

CrRef	Author and year	Drug	Study population, Sex, Age	Experiments: daytime or nighttime, Retention interval	EEG/LFP/PSG: yes or no	Memory task	Retention performance against control	Main mechanism of action a	Effect size, sample size
	Chamadia et al. (2020)	Dexmedetomidine	Human, male and female, not stated	-, 10 h	Yes, PSG: N2 increased; REMS decreased	Motor sequence task	↔ 700 µg p.o.	Selective alpha-2 receptor agonist	ES: not stated, n: 15
*	Morena et al. (2017)	Dexmedetomidine	Sprague–Dawley rats, male, not stated	Light phase, normal cycle, 48 h	No	PA	↓ 0.3 mg/kg i.p.; ↓ 0.4 mg/kg i.p.	Selective alpha-2 receptor agonist	ES: not stated, n: 12–13 per group
	Asth et al. (2012)	Diazepam	Swiss mice, male, not stated	Normal cycle, 24 h	No	Elevated T-maze	↔ 2 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 7–10 per group
	Both et al. (2005)	Diazepam	CF1 mice, male, 2 months	-, 24 h	No	PA	↓ 0.85 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 20 per group
+	Jensen et al. (1979)	Diazepam	Swiss-Webster mice, male, 50–60 days	Light phase, normal cycle, 3 days	No	PA	↔ 0.01 mg/kg i.p.; ↔ 0.1 mg/kg i.p.; ↔ 1 mg/kg i.p.; ↔ 10.0 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 21–26 per group
	Molinengo et al. (1999)	Diphenhydramine	Albino mice, male, not stated	-, 24 h	No	PA	↔ 2 mg/kg i.p.; ↔ 10 mg/kg i.p.	H1-histamine antagonist	ES: not stated, n: 10 per group
	Nonaka et al. (2013)	Diphenhydramine	C57BL/6J mice, male, 8–18 weeks	Light phase, normal cycle, 24 h	No	Fear conditioning	↔ 10 mg/kg p.o.; ↓ 30 mg/kg p.o.	H1-histamine antagonist	ES: not stated, n: 11–13
	Mylonas et al. (2020)	Eszopiclone	Human, male and female, 30 ± 6 years	-, 13 h	YES, PSG	Motor sequence task	↔ 3 mg	GABAA receptor modulator with great α1 receptor selectivity	ES: No effect size stated for the healthy participants, n: 29
+	Jensen et al. (1979)	Flurazepam	Swiss-Webster mice, male, 50–60 days	Light phase, normal cycle, 3 days	No	PA	↔ 0.025 mg/kg i.p.; ↔ 0.25 mg/kg i.p.; ↓ 1 mg/kg i.p.; ↔ 2.5 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 21–26
+	Jensen et al. (1979)	Lorazepam	Swiss-Webster mice, male, 50–60 days	Light phase, normal cycle, 3 days	No	PA	↔ 0.2 mg/kg i.p.; ↔ 2 mg/kg i.p.; ↔ 5 mg/kg i.p.; ↔ 20 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 21–26
	Bustos et al. (2006)	Midazolam	Wistar rats, male, adult	Light phase, normal cycle, 48 h	No	CFC	↔ 1 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 10–12 per group
	Gafford et al. (2005)	Midazolam	Long Evans rats, male, not stated	Light phase, 24 h	No	AFC and CFC	CFC: ↓ 1 µg per side intrahippocampal; ↔ 1 µg per side intrahippocampal (application 3 h after training), AFC: ↔ 1 µg per side intrahippocampal (immediate and 3 h after training)	GABAA receptor modulator	ES: not stated, n: 14 per group
*	Hauer et al. (2011)	Midazolam	Sprague–Dawley rats, male, not stated	Light phase, normal cycle, 48 h	No	PA	↔ 30 mg/kg i.p.; ↔ 50 mg/kg i.p.; ↔ 70 mg/kg i.p.	GABAA receptor modulator	ES: not stated, n: 9–10 per group
	Kroon and Carobez (2009)	Midazolam	Wistar rats, male, 12–16 weeks	-, 24 h	No	Fear conditioning	↓ 0.375 mg/kg i.p.; (head-out time), ↓ 0.5 mg/kg i.p. (head-out time and hide time)	GABAA receptor modulator	ES: not stated, n: 10 per midazolam group and 20 in the saline group
+	Zanin et al. (2013)	Midazolam	Swiss EPM-M1 mice, male, 3 months old	Normal cycle, 10 days	No	PM-DAT	↔ 2 mg/kg i.p.; ↓ 2 mg/kg i.p. (1 h after training); ↔ 2 mg/kg i.p. (2 h after training)	GABAA receptor modulator	ES: not stated, n: 10 per group
+	Meléndez et al. (2005)	Triazolam	Human, male and female, 19.8 ± 0.35 years	-, 11 h	No	word and non-word list learning	↔ 0.25 mg p.o.	GABAA receptor modulator	ES: not stated, n: 22
+	Morgan et al. (2010)	Triazolam	Human, male, 39 ± 3 years	-, ~10 h	Yes, PSG: TST, N2, REMS-lat. increased	Motor sequence task	↓ 0.375 mg p.o.	GABAA receptor modulator	ES: not stated, n: 12
+	Hall-Porter et al. (2014)	Zaleplon	Human, male and female, 29.4 ± 6.7 years	-, 10.5 h	Yes, PSG: SWS increased	WPA, Finger tapping task	↔ 10 mg (both tasks, dosing 3.5 h after bedtime)	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 22 (FTT) and 20 (WPA)
	Carbone et al. (2021)	Zolpidem	Human, male, 25.0 ± 0.5 years	-, ~10 h	Yes, PSG	word-sound pair recognition	↑ 10 mg	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 22
#2	Chen et al. (2021)	Zolpidem	Human, male and female, 20.85 ± 2.97 years	-, 13.5/26.5 h	Yes, PSG: SWS increased; REM decreased (26.5 h)	WPA	↑ 10 mg p.o. 26.5 h; ↔ 10 mg p.o. 13.5 h	GABAA receptor modulator with great α1 receptor selectivity	ES: Calculated to be (Hedge’s g): 0.44 for 13.5 h retention interval and 0.50 for 26.5 h retention interval, n: 38 (19 females)
+	Hall-Porter et al. (2014)	Zolpidem ER	Human, male and female, 29.4 ± 6.7 years	-, 10.5 h	Yes, PSG: SWS, N2, TST increased; Slplat., N1 decreased	WPA, Finger tapping task	↓ 12.5 mg (both tasks)	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 22 (FTT) and 20 (WPA)
*	Kaestner et al. (2013)	Zolpidem	Human, male and female, 18–39 years	-, 9 h	Yes, PSG/EEG: SWS, SpiDens increased, REMS decreased	emotional recognition memory	↑ 10 mg (negative and high-arousal stimuli)	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated isolated for zolpidem, n: 28 (15 women)
*#1	Mednick et al. (2013)	Zolpidem	Human, male and female, 22 ± 3 years	-, 9 h	Yes, PSG/EEG: SWS, SpiDens increased; REMS, decreased	WPA, visual texture discrimination task, motor sequence task	↑ 10 mg WPA; ↓ 10 mg VTD; ↔ 10 mg motor sequence task	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated for the memory tests, n: 30 (15 females)
+	Meléndez et al. (2005)	Zolpidem	Human, male and female, 19.8 ± 0.35 years	-, 11 h	No	Word and non-word list learning	↔ 10 mg p.o.	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 22
+	Morgan et al. (2010)	Zolpidem	Human, male, 39 ± 3 years	-, ~10 h	Yes, PSG: REMS-lat. increased	Motor sequence task	↔ 10 mg p.o.	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 12
*#1	Niknazar et al. (2015)	Zolpidem	Human, male and female, 22 ± 3 years	-, 9 h	Yes, PSG	WPA	↑ 10 mg	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated for the memory test, n: 28 (14 females)
	Simon et al. (2020) b	Zolpidem	Human, male and female, 18–30 (mean: 20.8) years	-, 24 h	Yes, PSG/EEG: SWS, fast SpiPow increased; REMS decreased (data from Simon et al. (2022)	Negative/neutral picture recognition	↑ 10 mg (negative pictures)	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 25 (Behavior), 28 (PSG/EEEG)
#2	Zhang, Yetton et al. (2020)	Zolpidem	Human, male and female, 21.00 ± 2.97 years	-, 13.5/26.5 h	Yes, PSG: SWS increased, REMS decreased	WPA	↑ 10 mg both 13.5 and 26.5 h	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 26 (12 females)
	Zhang and Mednick (2020)	Zolpidem	Human, male and female, college-aged	-, 24 h or overnight	Yes, EEG	WPA	↑ (Dose not stated) 24 h; ↔ (dose not stated) overnight	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 28 (13 females)
+	Zanin et al. (2013)	Zolpidem	Swiss EPM-M1 mice, Male, 3 months old	Normal cycle, 10 days	No	PM-DAT	↔ 2 mg/kg i.p.; ↔ 5 mg/kg i.p.; ↔ 10 mg/kg i.p.; ↔ 10 mg/kg i.p. (1 h after training); ↓ 10 mg/kg i.p. (2 h after training); ↓ 10 mg/kg i.p. (3 h after training); ↔ 10 mg/kg i.p. (6 h after training)	GABAA receptor modulator with great α1 receptor selectivity	ES: not stated, n: 10–12 per group

↓: Significantly impaired consolidation; ↑: significantly enhanced consolidation; ↔: no significant effect on consolidation; i.p.: intraperitoneal; p.o.: per os; CrRef, Cross references; ER: extended release; PSG: polysomnography; LFP: local field potential; WPA: word paired- associate task; PA: passive avoidance task; AFC: auditory fear conditioning; CFC: contextual fear conditioning; PM-DAT: plus maze discriminative avoidance task; VTD: visual texture discrimination task; ES: effect size; n: sample size; SWS: slow-wave sleep; N2: sleep stage 2 (S2/N2); SpiDens: spindle density; SpiPow: spindle power; REMS: REM sleep; lat.: latency; Slplat.: sleep onset latency; TST: total sleep time; +: study was shown more than once; *: also in the results of anesthetics; #1 and #2: reports of the same study; -: Information not stated.

a

Data compiled based on information from Geisslinger et al. (2020) to Goodman (2005).

b

The data was updated with information from Simon et al., (2022), which had not been published at the time of our search.

Results of quantitative analysis

Five reports on effects of zolpidem on young adults (Chen et al., 2021; Mednick et al., 2013; Niknazar et al., 2015; Zhang and Mednick, 2020; Zhang et al., 2020) were combined in a meta-analysis. Mednick et al. (2013) and Niknazar et al. (2015) as well as Chen et al. (2021) and Zhang et al. (2020) are reports of the same study and combined as one in the statistical analysis (Table 6). One otherwise eligible study was not included in the quantitative analysis because it used zolpidem extended release (ER; Hall-Porter et al., 2014).

OPEN IN VIEWER

Table 6.

Characteristics of studies included in the statistical analysis.

CrRef	Author and year	Drug	Study population	Sex	Age	Experiments: daytime or nighttime	Retention interval	EEG yes or no	Memory task
#2	Chen et al. (2021) and Zhang et al. (2020)	Zolpidem	Human	Male and female	20.85 ± 2.97 years/21.00 ± 2.97 years	–	13.5/26.5 h	Yes	WPA
#1	Mednick et al. (2013) and Niknazar et al. (2015)	Zolpidem	Human	Male and female	22 ± 3 years	–	9 h	Yes	WPA
	Zhang and Mednick (2020)	Zolpidem	Human	Male and female	College-aged	–	24 h or overnight	Yes	WPA

#1 and #2: reports of the same study.

CrRef: Cross references; WPA: word paired-associate task.

The combined effect size estimate shows an effect of zolpidem on consolidation of the WPA of 0.57 (95% CI 0.04–1.10; Figure 2). The CI is wide due to the relatively small overall sample size. Yet, even at the low end of the 95% CI, the effect size still hints at a beneficial effect of zolpidem on memory consolidation compared to placebo, even if small.

OPEN IN VIEWER

Figure 2.

Statistical analysis. Forest plot of the effect sizes of the studies on zolpidem. Two of the studies (Chen et al., 2021; Zhang and Mednick, 2020) had two separate retention intervals as indicated by the number of hours of the retention interval or the time of day of testing. The same participant groups were used to establish the different test time points in Zhang and Mednick (2020) and in Chen et al. (2021). Zhang et al. (2020) report on the same data as Chen et al. (2021). We chose the data from Chen et al. (2021) for the meta-analysis since they included a larger number of participants in their statistics.

The heterogeneity was I² = 58.97% and Tau = 0.10. The PI (–0.45 to 1.60) indicates substantial between-study heterogeneity, suggesting that future studies may find effects ranging from negligible or slightly negative to very large. This might reflect the different retention intervals (from 9 to 26.5 h), different sleep periods (one nap, three nocturnal) and potentially different doses (once not reported).

Antidepressants

Within the drug group antidepressants, a tendency emerged for second-generation antidepressants to enhance memory consolidation as compared to placebo. These included SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, sertraline), NARIs (bupropion, reboxetine) and the SNRI duloxetine (lithium, agomelatine, and esketamine do not belong in this pharmacological class). In contrast, first-generation antidepressants, namely tri- and tetracyclic antidepressants (amitriptyline, desipramine, mianserin) either had no effect or impaired memory consolidation, with only one exception.

A benefit of second versus first-generation antidepressants presents an important finding, since a symptom of major depressive disorder is memory impairment (Austin et al., 2001; Beblo et al., 2020). Moreover, for young adults, for whom depression is one of the most common mental health problems and whose brain is still maturing, choice of medication should take potential effects on mnemonic processes and maybe additionally on related structures into account.

The differential effects of first- and second-generation antidepressants, likely reflect the critical difference between these drug generations regarding their affinity to receptors outside of their main mechanism of action. In addition to acting as inhibitors on noradrenergic and serotonergic reuptake transporters they bind to muscarinic, histaminergic, and alpha-adrenergic receptors (Geisslinger et al., 2020; Goodman, 2005) possibly explaining the adverse effects on memory consolidation. For instance, H₁-histamine receptor antagonists impair memory consolidation (da Silveira et al., 2013; Nonaka et al., 2013) and histamine receptors in the ventral hippocampus, a relevant hippocampal region for sleep-associated memory consolidation, were found to mediate this impairment (Alvarez and Banzan, 2008; Binder et al., 2019). In addition to acting as a histaminergic antagonist Goerke et al. (2014) argued that impaired consolidation of the procedural task, with amitriptyline in humans was due to the reduced cholinergic tone reflected in reduced REM sleep. In rodents, doses of amitriptyline as used in this study (5 mg/kg i.p.) suffice to strongly suppress REM sleep (Obál et al., 1985). REM sleep reduction was likewise reported here for desipramine (Watts et al., 2012). Whereas REM sleep generation relies significantly on cholinergic activity of the brain stem laterodorsal/ pedunculopontine tegmental nucleus (Wang et al., 2021), norepinephrine and serotonin suppress REM sleep onset and maintenance (Mallick et al., 2008; Monti et al., 2008a). Interestingly, within the included first-generation drugs only consolidation performance of a striatum-dependent task increased (Colavito et al., 2013; Porsolt et al., 1993; Watts et al., 2012).

SSRIs lead to an acute increase in extracellular serotonin levels. Investigating effects of serotonin on declarative memory formation in their recent systematic review (Coray and Quednow, 2022) emphasize discordant effects on consolidation, apparently depending not only on serotonin receptor subtypes, but also on the specific memory task. Cognitive effects were reported to also reveal a bell-shaped dose response curve. Regarding the consolidation step of memory formation, serotonin is reported to support particularly protein synthesis-dependent late phase processes with effects possibly mediated by glutamatergic, GABAergic and/or cholinergic neurotransmission (Seyedabadi et al., 2014). The discrepant findings, that acutely administered SSRIs are frequently reported to impair performance on declarative tasks (Coray and Quednow, 2022), yet in the present review at worst no effect on memory was found, likely reflect brain-state dependent effects of manipulating brain neurochemistry, for example, brain-state dependent effects of increased acetylcholine levels (Gais and Born, 2004; Hasselmo, 1999), and/or on brain electrophysiology that reflect differential network interactions. Improved consolidation of a procedural skill memory task was likely linked to the increased spindle density after both the SSRI fluvoxamine and SNRI reboxetine administrations, found despite REM sleep suppression (Rasch et al., 2009).

Beyond serotonergic mechanisms, activity of the LC has emerged as relevant for sleep-associated processes. In the present study, we reported that the sedative drug dexmedetomidine, a selective alpha-2-receptor agonist, that suppresses noradrenergic activity in the LC (Chiu et al., 1995; Morena et al., 2017; Song et al., 2017), impaired memory consolidation in rats in a passive avoidance task. In humans, oral administration of this drug, also failed to facilitate consolidation of motor task as was found for the placebo condition, although NREM stage N2, and spindle density were enhanced (Chamadia et al., 2020). Pharmacological suppression of noradrenergic transmission by intraperitoneal injection of clonidine and propranolol, decreased not only the occurrence of sleep spindles and slow oscillations, but also their temporal coupling, a relevant electrophysiological signature of memory consolidation during NREM sleep (Durán et al., 2021; Eschenko et al., 2012). Indeed, patterns of LC noradrenergic activity reveal both macro- and micro scale brain-state-associated dynamics that are crystalizing as functionally relevant. In addition to its distinctive NREM and REM sleep-dependent activity, the LC reveals heightened neuronal firing dependent upon phase of the sleep slow oscillation. Omission of LC quiescence during REM sleep as well as prior to spindles through optogenetic stimulation disturbed spindle generation, delta and theta power, as well as the consolidation of spatial memories (Swift et al., 2018). More recently, infraslow oscillations of noradrenergic activity (~0.2 Hz) during NREM sleep were shown relevant for memory consolidation (Kjaerby et al., 2022).

Taken together, the beneficial effect on sleep-associated memory consolidation of antidepressants with mechanisms of action that selectively inhibit reuptake of monoamines, may be attributable to enhanced noradrenergic and/or serotonergic signaling.

Anesthetics

The results in the anesthetics group were mixed for each drug, except for sodium oxybate (n = 4), and isoflurane (n = 1), both of which did not influence memory consolidation in any of the present studies. Although GABA_B receptor-mediated activity, through which sodium oxybate mainly exerts it actions, may be required for synaptic plasticity (Heaney and Kinney, 2016; Sanchez-Vives et al., 2021), the present results suggest that acute administration of sodium oxybate at the given systemic concentrations does not affect memory consolidation in young to middle aged adults. To note, EEG recordings during sleep following administration of sodium oxybate, but also of the GABA reuptake inhibitor tiagabine, were likely to induce slow rhythms, but physiological properties of such rhythms, for example, homeostatic SWS response or spindle-related activity appear to diverge from normal sleep (Feld et al., 2013; Kaestner et al., 2013; Mednick et al., 2013; Vienne et al., 2012).

Table 4 reveals that despite different main mechanisms of drug action the five rodent studies revealing increased retention performance over control, employed an aversive task (passive avoidance or acoustic fear conditioning) as well as a relatively long retention interval for their drug (48 hours for ketamine and propofol; 24 hours for sevoflurane). The other listed studies employed neutral tasks, such as spatial tasks or conditioning with neutral stimuli. Together, these findings reflect the relevance of memory content and task for drug effects. Specifically, stimulus salience and relevance of receptors distributed in the amygdala, but also in the hippocampus or nucleus accumbens, that is., structures on which passive avoidance and fear conditioning tasks strongly rely, influence the mnemonic effect (McGaugh, 2000; Meiering et al., 2024; Moustafa et al., 2013).

Dosages and nature of study are furthermore important to consider. The ketamine studies revealing enhanced consolidation used anesthetic-level doses. For subanesthetic doses of ketamine, as used by many rodent studies in Table 4 (⩽30 mg/kg i.p. ketamine; El Iskandrani et al., 2015; Wu et al., 2020) consolidation performance after spatial learning decreased. This applies also to the consolidation of a visual-procedural task in humans. ln this human study, the subanesthetic ketamine administration was initiated at sleep onset and did not modify sleep stage composition (Gais et al., 2008). The contrasting findings on consolidation using subanesthetic versus anesthetic ketamine levels reflect differential dose-dependent effects at neuronal and neural network levels of the non-competitive NMDA-receptor antagonist and are an ongoing subject of investigation (Obert et al., 2024; Zhou and Duan, 2024). Differential effects of ketamine may however, also extend to the influence on other systems, such as the endocanabinoid system, that has been investigated in relation to ketamine’s antidepressant effects (Lang and Borgwardt, 2013; Sharafi et al., 2022; Short et al., 2018).

For propofol, combined findings of the three studies indicate that lower dosages (150 mg/kg i.p.) of propofol in rats as compared to higher dosages inhibit memory consolidation (Hauer et al., 2011; O’Gorman et al., 1998; Zhang et al., 2013), whereby the very lowest dose in each study remained without an effect on consolidation. In mice, dynamic changes in thalamic-cortical rhythmic activity were reported dependent upon anesthesia depth. Studies show that slow waves of natural NREM sleep are not a unitary phenomenon, and that composition of slow-wave types differs with NREM sleep depth (Kim et al., 2019; Seok et al., 2022). It could therefore be that the thalamo-cortical slow-wave activity of propofol-induced deep anesthesia (corresponding in rats to 300 mg/kg BW i.p.; Guang et al., 2025; Murphy et al., 2011) differs qualitatively regarding memory consolidation from the activity of lighter anesthesia depth. Furthermore, temporal factors such as the relationship between retention time to terminal elimination half-life may play a role. Impaired memory retrieval was found at subhypnotic dosages (Liu et al., 2016), and theta activity was suppressed on emergence from anesthesia, possibly affecting retrieval performance (Flores et al., 2017). Yet most importantly Hauer et al. (2011) concluded from further investigations in their study that the memory enhancement was attributed to the elimination by propofol of an endocannabinoid-degrading enzyme (fatty acid amide hydrolase).

For the volatile anesthetics isoflurane and sevoflurane memory effects are mainly attributed to GABA_A agonistic activity, most likely mediated by the alpha5 GABA_AR subunit that shows a high density in the hippocampus. Differential findings regarding sevoflurane could be due to the heightened aversion associated with the electric shocks of the hippocampus-dependent continuous multiple trial inhibitory avoidance task (Li et al., 2011; Liu et al., 2010). To note, since differences in the minimum alveolar concentration between mice and rats as well as with age are indicated (Cesarovic et al., 2010; Kashimoto et al., 1997) the effects induced by sevoflurane may not be identical between the studies.

Taken together, despite our mixed findings of anesthetics, our results fall in line with others in supporting the need to look beyond classical neurochemicals to explain effects on cognitive performance (Kim et al., 2019; Seok et al., 2022; Zhang et al., 2024).

Sedatives/Hypnotics

Except for dexmedetomidine and diphenylhydramine, all drugs investigated in the included studies act primarily through modulation of the GABA_A receptor. Given the central role of GABAergic neurotransmission in the generation of sleep spindles (Gottesmann, 2002; Sanchez-Vives et al., 2021), it could be expected that sedative drugs influence memory consolidation. Consistent with this assumption, optogenetic silencing of GABAergic neurons in the medial septum albeit during REM sleep has been shown to impair contextual memory consolidation, likely through a concomitant reduction in theta oscillations (Boyce et al., 2016). An enhancement in consolidation was however, only found for zolpidem, as revealed in the small meta-analyses. Benzodiazepines either had no effect or impaired memory consolidation across a range of tasks, including passive avoidance, skill learning, and hippocampus-dependent paradigms (corresponding refs from Table 5).

In agreement with above findings for antidepressants the H₁-receptor antagonist in this drug group, diphenhydramine, impaired memory consolidation, at least in higher doses. Plausible mechanisms could be disturbed interregional brain communication, as deducted from decreased coupling between brain rhythms of NREM sleep, and/or a suppression of REM sleep-associated long-term potentiation (Masuoka et al., 2019). In Chamadia et al. (2020) time spent in REM sleep was decreased.

Zolpidem belongs to a group of drugs called z-drugs, encompassing also zaleplon, zopiclone and eszopiclone. Unlike benzodiazepines, these compounds exhibit greater selectivity for the α1 subunit of the GABA_A receptor. Zaleplon and eszopiclone were each examined in one study included in this review, but neither was associated with improved memory consolidation. In the case of zaleplon, the tasks employed were comparable to those used in the zolpidem study; however, chronopharmacological differences complicate comparisons. Specifically, zaleplon was administered 3.5 hours after bedtime, whereas zolpidem was given 30 minutes before bedtime in an ER formulation (Hall-Porter et al., 2014). In that study, zolpidem-ER was found to impair memory consolidation relative to both placebo and zaleplon, while no difference was observed between zaleplon and placebo. Notably, to equate experimental conditions, participants receiving zolpidem-ER were also awakened 3.5 hours after bedtime, a manipulation that may itself have disrupted memory consolidation compared to undisturbed sleep. The study using the ER formula (Hall-Porter et al., 2014) was therefore excluded from the meta-analysis.

Benzodiazepines, the other major group of GABA_A modulators, had either no effect, or impaired memory consolidation for diverse tasks involving passive avoidance, skill learning and hippocampus-dependent tasks (Table 5). Similar to findings of the one human study administering a benzodiazepine here in which light NREM sleep was enhanced (Morgan et al., 2010), Leong et al. (2022) report in their review enhanced spindle activity across most studies with benzodiazepines. In part, benzodiazepines even revealed greater enhancements of spindle activity than z-drugs (Leong et al., 2022). These findings show that spindle activity of light NREM sleep is not always a reliable correlate for enhanced memory consolidation.

Note that both benzodiazepines and z-drugs have clinically been associated with cognitive impairment in long-term use (Atkin et al., 2018). But as far as the authors know, there is no information on spindle activity with long-term usage of these drugs. More recently, zolpidem was reported to enhance the temporal coordination between slow oscillations and sleep spindles, as well as hippocampal–prefrontal temporal coupling -electrophysiological processes that are closely linked to memory consolidation (Kersanté et al., 2023). Future research will be required to determine whether these effects, relative to benzodiazepines, arise from the higher receptor subunit specificity of zolpidem, differences in regional receptor distribution, or alternative mechanisms.

Despite its beneficial effect on memory consolidation, zolpidem was recently reported to suppress the vasomotor activity relevant for glymphatic clearance by suppressing the noradrenergic infraslow oscillation and associated micro-arousals (Hauglund et al., 2025). It is yet to be investigated, whether the likelihood of neurodegenerative disorders including memory disorders such as Alzheimer’s disease are heightened by (long-term) use of medications (Heneka et al., 2010; Yu et al., 2025).

In summary, the results underscore the concept, that (pharmacologically induced) increased spindle activity generated by increased GABAergic signaling alone does not facilitate memory consolidation, but rather that the interaction with other sleep rhythms such as the sleep slow oscillation is essential (Kersanté et al., 2023; Staresina, 2024).

Limitations

Mechanisms of sleep-associated memory consolidation are still a matter of intense research with studies showing a significant dependence on temporal aspects, such as time between learning and sleep duration of post-learning retention interval of sleep the composition of ongoing brain rhythms in sleep (Deng et al., 2025; Diekelmann and Born, 2010; Eschenko et al., 2006; Poe et al., 2010; Rasch and Born, 2013), time of day per se and neuroendocrine interactions (Feld and Born, 2020; Lazar et al., 2015; Thorn et al., 2004; van der Zwet et al., 2025), and their interactions with the type of memory task (Marshall et al., 2020; Myhrer, 2003). On the other hand, most studies on the investigated pharmacological drug classes and memory tasks have not focused on these aspects. Due to the paucity of corresponding studies, neither polysomnography nor brain electrical activity could be used as inclusion criteria. For the same reason, but also due to the diversity of drugs, the review as a whole did not discriminate between types of memory, as hippocampus-dependent versus non-hippocampus-dependent. Only for a few drugs did literature allow for a discussion on concordant brain electrical activity and underlying mechanisms. In the present review, task-dependent effects for ketamine were apparent: Two studies reported an impairment in memory consolidation at the highest dose using the Morris water maze (Liu et al., 2014; Moosavi et al., 2012), while an improvement in consolidation of a passive avoidance task similarly using a regimen with the highest ketamine dose (Morena et al., 2017). Studies have also shown that, in addition to species-specific differences in drug responses, strain-dependent differences in memory consolidation occur, for example, for dopaminergic drugs (Gasbarri et al., 1997). These aspects further hamper any a priori generalization of the present findings. Importantly, new and distinct mechanisms of action for the investigated drug classes are still being disclosed, suggesting the necessity of more drug-specific investigations of their cognitive impacts during both wakefulness and sleep.

A further limitation is the relatively low number of participants, only reaching 13 in some human studies. Moreover, the participants are exclusively healthy, with no representation from those suffering from disease. Also, age ranges are not strictly limited to young adults. Consequently, the findings of the study are unable to provide any insights into the effects of the drugs in patients. As indicated above, the number of studies eligible for inclusion in the meta-analysis was suboptimal, leading to a wide CI. A placebo-controlled design was part of the inclusion criteria. All but one study with human subjects reported blinding of investigators. However, the reviewed animal studies present a contrasting picture; blinding is scarcely reported. Since the latter, especially during analyses is relevant to ensure unbiased results, mention in publications should become the norm.