The Effect of Sleep Deprivation on Creative Cognition: A Systematic Review of Experiment-Based Research

ABSTRACT

Tales of working through the night permeate biographical accounts of eminent creative figures. However, empirical research on the association between sleep deprivation and creativity is scarce and inconsistent. Some studies indicate that sleep deprivation impairs creative thinking, while others suggest that sleep deprivation enhances it. The present article provides a systematic review of literature reporting experiments assessing the impact of sleep deprivation on creative cognition. The literature search and screening followed the PRISMA Statement and used the PICOS framework to structure extraction of information from studies, while the NIH quality assessment tools were used to assess risk for flaws in study methods. From an initial pool of 521 studies, eight met inclusion criteria. These studies used diverse methods and measures; therefore, a qualitative review approach to data synthesis was adopted. Results revealed that sleep deprivation tends to impair creative thinking. However, the quality of studies was fair to poor, statistical power was low, and confidence intervals for effect sizes were wide, limiting what can be confidently inferred about the effect of sleep deprivation on creative thinking from the existing literature. This observation underscores the pressing need for additional research, for which this article offers methodological advice and research directions.

The relationship between sleep and creativity is not well understood. On the one hand, greater sleep quantity is associated with better cognitive function (Pesoli et al., 2021). Conversely, anecdotal accounts of sleep-deprived musicians, painters, and poets working late into the night pervade biographical accounts of highly creative individuals (Flaherty, 2004; Iranzo, Stefani, Hogl, & Santamaria, 2018; Iszáj & Demetrovics, 2011; Kingston, 2019; Zeidler, 2021). Survey-based studies have found that more creative persons tend to have more sleep difficulties compared to the general population, and high scores on some measures of creativity are associated with nighttime sleep disturbances (Beaty, Silvia, Nusbaum, & Vartanian, 2013; Healey & Runco, 2006; Vaag, Saksvik-Lehouillier, Bjorngaard, & Bjerkeset, 2016). There is little research experimentally examining causal links between sleep deprivation and creativity, and the little that exists does not seem to reach consistent conclusions. This systematic review integrates and critiques this literature to present the evidence on the relationships between sleep deprivation and creativity, including the direction of causality, and critically evaluates the study designs commonly used in this field to identify avenues for future research.

Conceptualizing creativity

Creativity is seen in such diverse endeavors as sculpture, physics, music, architecture, and engineering, as well as everyday creativity, such as producing a joke or using a paper clip to fix a broken zip (Merrotsy, 2013). How can scientific inquiry capture such a heterogeneous phenomenon? One approach is to limit the scope of creativity to specific cognitive domains, and this is the approach adopted in the present review. Two major theories of so-called creative cognition are influential in this regard: i) Guilford’s (1957) model of productive thinking abilities (divergent and convergent thinking), and ii) Mednick’s (1962) associative theory of creativity.

Guilford (1957) believed that producing ideas requires divergent thinking (generating novel solutions to open-ended problems) and its counterpart convergent thinking (considering potential solutions to a problem and identifying one correct answer). Guilford described divergent thinking as composed of fluency, flexibility, and originality; these qualities providing the most obvious indications of creativity. Fluency, flexibility, originality, and elaboration (the ability to expand on an idea and embellish it with details; Guilford, 1957) are elements of divergent thinking assessed in contemporary creative cognition research (Paek, Abdulla Alabbasi, Acar, & Runco, 2021). These concepts provided the basis for laboratory measures of creativity, the best-known of which are the Alternate Uses Test (AUT; Guilford, Christensen, Merrifield, & Wilson, 1960), which involves generating as many uses for an everyday object (e.g., a brick) as possible, and the Torrance Tests of Creative Thinking (TTCT; Torrance, 1974), which includes both verbal and visual divergent thinking tasks.

Mednick (1962) built upon Guilford’s model, focusing on the remote association subcomponent of originality. He defined creative thinking as “the forming of associative elements into new combinations which either meet specific requirements or are in some way useful. The more mutually remote the elements of the new combination, the more creative the process or solution” (Mednick, 1962, p. 221). Based on this definition, he suggested that highly creative individuals have a greater ability to access mutually remote associative elements, which can be used to generate a creative idea. The classic measure used to assess this associative basis of creative thinking is the Remote Associates Test (RAT; Mednick, 1968). RAT trials present three seemingly unrelated words (e.g., paint/doll/cat) and ask participants to produce a fourth word that unites the other three (e.g., house; house paint, dollhouse, and cat house).

Sleep and creative cognition

Sleep has been shown to influence cognitive abilities that support creative thinking. Good sleep supports learning, memory, and executive functions (Pesoli et al., 2021), whereas poor sleep impairs cognitive performance (Krause et al., 2017). Lack of sleep, such as that associated with insomnia, is a common sleep complaint and may have particularly detrimental effects on cognitive performance. Indeed, there is broad consensus that insufficient sleep quantity impairs speed and/or accuracy on many cognitive tasks, particularly those thought to rely on working memory and attention (Gevers et al., 2015; Krause et al., 2017; Kusztor et al., 2019; ; Lim & Dinges, 2008; Pesoli et al., 2021). Both acute sleep deprivation (where individuals are continuously deprived of sleep over a 24–48-h period) and chronic partial sleep restriction (restricted sleep over several consecutive nights) have a dose-dependent effect on task performance. The longer the duration or the higher the total amount of sleep loss, the worse the performance (Alhola & Polo-Kantola, 2007; Krause et al., 2017; Pesoli et al., 2021). These negative effects of sleep deprivation are hard to reconcile with the sleep problems observed among highly creative people who are motivated to produce many creative works.

The present study

The apparent contradiction between the detrimental effect of sleep deprivation on cognitive performance and accounts of reduced sleep in artists highlights the need for a systematic literature review to examine the effects of sleep deprivation on creative thinking. The present study therefore aimed to (a) summarize literature reporting experiments that have examined the effect of sleep deprivation on creative thinking and (b) critically evaluate the methods used in these studies. The present study specifically addressed the following questions:

1. Does sleep deprivation influence creative thinking?

2. Are different aspects of creative thinking enhanced or impaired by sleep deprivation?

3. What is the quality of study designs in the relevant literature?

Method

Information sources and search strategy

The search followed PRISMA Statement guidelines (Higgins et al., 2021; Moher, Liberati, Tetzlaff, & Altman, 2009). Nine databases were searched for peer-reviewed articles and conference proceedings: PsycINFO, Scopus, Cochrane Library Trials, PubMed Central, CINAHL, Embase via Ovid, the British Library, Web of Science, and Medline. Search areas were the title, abstract, and keywords. Search terms were based on two concepts: sleep deprivation (acute and chronic sleep restriction, total or partial sleep deprivation) and creative thinking (divergent, convergent, and associative creative thinking). The search was limited to studies of adults. Search terms for all databases were: adult* AND sustained wakefulness OR sleep loss OR sleep deprivation OR prolonged wakefulness OR sleep restriction AND divergent thinking OR convergent thinking OR creative thinking OR novel idea generation OR creative ideation OR associative thinking OR creative problem solving OR creativ*. The original search was conducted in 2021 and updated in 2024.

Inclusion and exclusion criteria

The review included only peer-reviewed articles or full-text conference proceedings; experimental studies that used lab or field-based sleep deprivation protocols; and studies that included a measure of convergent, divergent, or associative creative thinking as a dependent variable (e.g., AUT, TTCT, or RAT). Sources were excluded if they were unavailable in English or were review articles, book reviews, media reviews, editorials, or case studies.

Article selection process

Search results were processed in Microsoft Excel. One reviewer (ARL) screened the title and abstract of each record and marked articles for full-text assessment based on inclusion/exclusion criteria. Two reviewers (ARL and BB) conducted full-text eligibility assessments independently before proceeding to data extraction.

Data extraction

The PICOS framework was used to extract the following information from each study: (a) study design (e.g., randomized/nonrandomized group comparison pre-post, single group pre-post); (b) participant characteristics (e.g., sample size, age, gender, recruitment technique, country); (c) manipulation type (e.g., sleep deprivation regime); (d) outcome measures (e.g., type of creative thinking task); and (e) main results (e.g., significance, effect size).

Quality assessment

Two quality assessment tools, one for controlled intervention studies and one for before–after studies with no control group, were used to assess risk for flaws in study methods (National Institutes of Health [NIH], 2021). These tools provide criteria and scoring guidelines to help reviewers evaluate each study’s internal validity. Two authors (ARL and BB) independently rated the quality of the studies using prescribed criteria. Responses were scored to create an overall rating (“good,” “fair,” or “poor”) based on each criterion. Discrepancies between the two raters were resolved through discussion and consensus.

Results

The PRISMA diagram for the review is shown in Figure 1. The initial search returned 521 records, with 511 remaining after duplicates were removed. Title and abstract screening excluded 499 sources; the main reasons for exclusion were that studies did not measure sleep/creativity, were non-experimental designs, had inappropriate samples (children, clinical disorder populations, non-human participants), or were reviews or case studies. The 12 remaining articles underwent full-text screening. Two articles were excluded because sleep deprivation was measured via self-report or observational methods rather than manipulated experimentally (Schonauer et al., 2018; van Heugten-van der Kloet et al., 2015), and a further article was excluded because it did not meet full-text criteria. One final article was excluded as “innovative thinking” was measured by a financial decision-making game that does not fulfill the creative thinking inclusion criteria definition (Harrison & Horne, 1999). Thus, a total of eight articles met the criteria for the review.

Figure 1. Systematic Search According to the PRISMA Statement Methodology.

General study characteristics

The identified articles had such diverse methods and outcomes that a qualitative review approach was adopted. Table 1 summarizes the studies’ characteristics. Three of the eight articles utilized a pre-post intervention with participants randomly allocated to the experimental and control groups (Horne, 1988; Landmann et al., 2016; Wimmer, Hoffmann, Bonato, & Moffitt, 1992). The remaining five studies used a pre-post intervention design without a control group (Glaubman et al., 1978; Lewin & Glaubman, 1975; Nelson, Dell’angela, Jellish, Brown, & Skaredoff, 1995; Vartanian et al., 2014; Webb & Levy, 1982).

Table 1. Characteristics of Included Studies.

Study	N	Country	Sample	Intervention	Control Group	Creativity Task/s	Dependent Variable/s	Primary Outcome/s
Pre-Post Intervention with Control Group Designs
Horne (1988)	24	UK	15 female, 9 male, 19-23 years	TSD for 32 hrs in a lab-based setting (n = 12)	Allowed to sleep normally at home (n = 12)	TTCT Verbal & Figural	Fluency, Flexibility, Originality & Elaboration	SD impaired all aspects of the TTCT
Wimmer et al. (1992)	25	Canada	All male, 17-27 years	TSD for one night in a lab-based setting (n = 12)	Allowed to sleep normally in a lab-based setting (n = 13)	TTCT Figural	Fluency, Flexibility, Originality & Elaboration	SD impaired TTCT Figural Flexibility only
Landmann et al. (2016)	60	Germany	30 female, 30 male, 20-31 years	TSD for one night in a lab-based setting (n = 20)	Two, one normal sleeping group (n = 20) and one daytime wake group (n = 20) in a lab-based setting	CRA	Number correct and solution time for baseline, strengthening, reorganization, and control	SD enhanced reorganisation whereas sleep enhanced strengthening of associations
Pre-Post Intervention Designs (no control group)
Lewin and Glaubman (1975)	12	Israel	All male, 22-25 years	REM deprivation for one night in a lab- based setting	All participants were compared pre-post	AUT	Total fluency, Mean fluency, Spontaneous flexibility, Mean originality, Sample originality	REM deprivation impaired all aspects of performance on the AUT
Glaubman et al. (1978)	10	Israel	8 females, 2 males, 17-18 years	REM deprivation for one night in a lab-based setting	All participants were compared pre-post	Consequences	Total Fluency, Mean Fluency, Flexibility, and Originality	REM deprivation impaired Mean Fluency and Flexibility only
Webb and Levy (1982)	16	USA	All male, 6 x 18-22 years, 10 x 40-49 years	TSD for two consecutive days in a lab-based setting	No control group, young and older participants compared	Object Uses RAT	Number of uses (Fluency) Number correct	SD reduced the number of object uses generated for the older group only
Nelson et al. (1995)	9	USA	Gender and age not reported	PSD over a 24-hr period, approximately 30 minutes sleep	All participants compared pre-post	TTCT Figural TTCT Verbal	Fluency, Flexibility, & Average Fluency, Originality,Flexibility, & Average	PSD impaired Figural Originality and Verbal Fluency, Originality, and Flexibility
Vartanian et al. (2014)	13	Canada	10 males, 3 females, mean age = 32.23 years	TSD for one night in a lab-based setting	All participants compared pre-post	AUT	Number of uses (Fluency)	SD reduced number of alternate uses generated

Note. REM = Rapid Eye Movement sleep. TSD = total sleep deprivation. PSD = partial sleep deprivation. TTCT = Torrance Tests of Creative Thinking. CRA = Compound Remote Associates. RAT = Remote Associates Test. AUT = Alternate Uses Test. Reorganization in the Landmann et al. (2016) study refers to newly solved primed items, whereas strengthening refers to the re-solving of problems presented at baseline. The Lewin and Glaubman (1975) and Glaubman et al. (1978) studies deprived participants of REM sleep only in the experimental condition, and non-REM sleep deprivation served as the control or baseline condition.

The sample size of studies employing a control group ranged from 24 to 60 participants (M = 36.33, SD = 20.50), while that of studies without a control group ranged from 9 to 16 participants (M = 12, SD = 2.74). Horne (1988) and Landmann et al. (2016) had equal sized experimental and control groups, while Wimmer, Hoffmann, Bonato, and Moffitt (1992) had unequal group sizes. Studies were conducted in five countries: United Kingdom (1), Canada (2), Germany (1), Israel (2), and the United States of America (2).

Participant characteristics

Most studies used convenience sampling, and the majority of participants were university students or staff recruited on campus. Webb and Levy (1982) compared the performance of younger (18–22 years; recruited via a student magazine advertisement) and older persons (40–49 years; recruited via a letter addressed to faculty members). The majority of participants across all studies were men (104 men, 56 women); three studies included only men, and one study did not report participant gender. For the six studies that reported age range, participants ranged from 17 to 49 years of age.

Sleep manipulation characteristics

All studies attempted to monitor or control for sleep patterns prior to testing. Horne (1988) and Wimmer, Hoffmann, Bonato, and Moffitt (1992) instructed participants to go to bed no later than 23:30 and 23:00, respectively, for four consecutive nights preceding testing. Nelson, Dell’angela, Jellish, Brown, and Skaredoff (1995) used a self-report measure of sleep prior to testing. In contrast, Landmann et al. (2016), Vartanian et al. (2014), and Webb and Levy (1982) objectively measured sleep patterns prior to testing. Landmann et al. monitored sleep patterns for two weeks prior to testing using actigraphy, Vartanian et al. did the same for six nights. Lewin and Glaubman (1975), Glaubman et al. (1978), and Webb and Levy monitored participants as they slept in the lab before the experimental manipulation. No participants were excluded from the main analyzes based on wrist activity or sleep monitoring data.

The period participants were deprived of sleep differed among the studies. Table 1 summarizes the different sleep deprivation periods used in each experiment. One study partially deprived participants of sleep overnight, with participants napping approximately 30 min over a 24-h period (Nelson, Dell’angela, Jellish, Brown, & Skaredoff, 1995). Two studies deprived participants of REM sleep only in the experimental condition compared to a control condition, where participants were deprived of non-REM sleep only (Glaubman et al., 1978; Lewin & Glaubman, 1975). The remaining studies used total sleep deprivation interventions of 12–37 consecutive hours of forced wakefulness.

Dependent variables: creativity tasks

Creativity tasks used in each study are shown in Table 1. Six of the eight utilized divergent thinking tasks only, one study used associative creative thinking tasks only, and one study used both divergent thinking and associative tasks.

The studies that measured divergent thinking used Consequences, Object Uses (OU), the AUT, and the TTCT. Consequences administered by Glaubman et al. (1978) was developed by Guilford (1967) and asks participants to list the consequences of certain hypothetical situations (e.g., the Earth was covered in water). The OU administered by Webb and Levy (1982) asks participants to report all the possible uses they could think of for a given object in 2 min, for six different objects. The OU is based on the Unusual Uses Test developed by Wilson et al. (1953) that preceded the development of more commonly used AUT and is scored using only the number of uses generated (fluency). Lewin and Glaubman (1975) administered the traditional AUT, whereas Vartanian et al. (2014) administered a modified version for use in brain imaging studies; objects were presented during the generation phase (12 s per object), then in the response phase participants were given 3 s to enter the number of uses generated. Horne (1988) and Wimmer et al. (1992) used the TTCT and followed the administration and scoring instructions that accompany the test package (see Torrance, 1974). Nelson et al. (1995) also used the TTCT, but the publisher scored the tests.

Two different versions of the RAT were used in the studies that assessed the associative basis of creative thinking. Webb and Levy (1982) used the original RAT (see Mednick, 1962) and administered 25 triads with 1 min allocated to solve each triad. Landmann et al. (2016) used the German-language version of the Compound Remote Associate triads (CRA; Landmann et al., 2014). The CRA triads are based on traditional RAT triads (e.g., same/tennis/head; same = match, tennis match, and matchhead), but CRA solutions are always a compound word (e.g., matchhead), whereas the RAT solutions include compound words, but are not limited to them (e.g., same = match). The English version of the CRA was developed by Bowden and Jung-Beeman (2003) to provide a large pool of items where the solution word was always related to the triad in forming a compound word (e.g., age/mile/sand; stone age, milestone, and sandstone). Landmann et al. presented 80 CRA triads in total. At baseline, 60 CRA (30 easy, 30 difficult) were randomly presented for a maximum of 10 s. At re-test, the 60 items were randomly presented again for a maximum of 30 s each, with the addition of 20 (10 easy, 10 difficult) novel control items to control for state-dependent effects following the experimental conditions.

Webb and Levy (1982) scored their RAT as participants’ number of correct solutions. Landmann et al. (2016) recorded both number of correct CRA solutions and solution time, with the addition of strengthening and reorganization scores. Strengthening represented solidification of newly encoded memories and was scored by the percentage of solutions re-solved and percentage improvement in solution time on the second test. Reorganization represented newly solved items post-intervention that were presented but unsolved at baseline, operationalized as the number of newly solved items and the solution time for those items.

The effect of sleep deprivation on creativity tasks

Both Horne (1988) and Lewin and Glaubman (1975) found that total sleep deprivation significantly impacted all creativity outcomes. The six remaining studies reported at least one significant outcome. Only Vartanian et al. (2014) and Landmann et al. (2016) reported effect sizes. Effect sizes for the remaining studies were computed using information from the published manuscripts and described as small, medium, and large based on the benchmarks suggested by Cohen (1988). See supplementary material for statistical summaries of the main results for each study, as well as calculated effect sizes, CIs for effect sizes, and power sensitivity analyzes.

Horne’s (1988) between-group analyzes found total sleep deprivation impaired all aspects of TTCT performance, with very large effect sizes for each aspect of the figural and verbal forms. Horne analyzed the first verbal task of the TTCT and found large performance improvements from baseline to post-intervention for both groups, but the control group improved more than the sleep deprivation group. Nelson et al. (1995) found that fluency, originality, and flexibility scores on the TTCT verbal form deteriorated after partial sleep deprivation, with effect sizes ranging from small to large. For the TTCT figural form, Nelson et al. found a decrement only in originality scores after partial sleep deprivation, and Wimmer et al. (1992) found a decrease in only flexibility following total sleep deprivation. Both effects were large.

Landmann et al. (2016) found that total sleep deprivation was associated with a large reduction in CRA strengthening scores. The main effect of experimental condition for re-solved items was medium to large, and large when the control and experimental groups were directly compared. The main effect of experimental condition for percentage improvement in solution time of re-solved items was large, and large when the control and experimental groups were directly compared. In contrast to strengthening, sleep deprivation enhanced reorganization via speed of solution. The main effect of experimental condition for solution time of newly solved primed items was medium to large, and medium to large when the control and experimental groups were directly compared. No significant differences in absolute number of newly solved primed items, baseline performance, or control items at re-test were found.

Lewin and Glaubman (1975) found that REM deprivation, compared to non-REM deprivation, impaired all aspects of AUT performance and effect sizes ranged from medium to large. In comparison, Glaubman et al. (1978) found that REM deprivation impaired only the flexibility and mean fluency aspects of consequences, and effect sizes were medium and large, respectively.

Webb and Levy (1982) found that total sleep deprivation impaired OU performance for the older age group only, and the effect was very large. However, sleep deprivation did not impact RAT performance for either age group. Vartanian et al. (2014) found that sleep deprivation impaired AUT fluency, with a medium-sized effect.

Effect size precision and study power

The CIs for effect sizes (see Supplementary Table 3) are very wide, indicating low precision of effect size estimates. Power sensitivity analyzes indicated that sample sizes were sufficient to detect only large effect sizes, with the exception of Horne’s (1988) between-within ANOVAs for the first verbal task of the TTCT, which was sufficiently powered to detect a medium effect size.

Quality assessment

Methodological quality of studies was assessed using the National Institutes of Health (2021) tools for controlled intervention studies and before–after studies with no control group. As shown in Table 2 and Table 3, the quality of studies was rated “poor” to “fair.” A poor rating indicates that the study was assessed as having significant risk of bias for reasons such as providing no demographic information for participants (e.g., age and gender) and/or group differences at baseline that may partially explain some of the reported effects. A fair rating indicates that the study is susceptible to bias but at a level deemed insufficient to invalidate results. No reviewed study was assessed as having a “good” methodological quality rating.

Table 2. Methodological Quality of Pre-Post with Control Group Studies.

Study	1	2	3	4	5	6	7	8	9	10	11	12	13	14	Rating
Horne (1988)	Y	NR	NR	N	Y	Y	Y	Y	Y	CD	Y	N	N	NA	Fair
Wimmer et al. (1992)	Y	NR	Y	N	NR	NR	Y	Y	Y	Y	Y	N	Y	NA	Fair
Landmann et al. (2016)	Y	NR	NR	NR	NR	Y	Y	Y	Y	Y	Y	N	Y	NA	Fair

Note. A fair rating indicates that the study is susceptible to some bias but deemed insufficient to invalidate results. See the Quality Assessment Tool for Controlled Intervention Studies for a description of the 14 numbered criteria (National Institutes of Health, 2021). Y = yes, criterion met. N = no, criterion not met. NR = criterion not reported. NA = criterion not applicable. CD = cannot determine.

Table 3. Methodological Quality of Pre-Post Intervention Studies

Study	1	2	3	4	5	6	7	8	9	10	11	12	Rating
Lewin and Glaubman (1975)	Y	N	N	NR	N	Y	Y	NR	Y	Y	N	NA	Fair
Glaubman et al. (1978)	Y	N	N	NR	N	Y	Y	Y	Y	Y	N	NA	Fair
Webb and Levy (1982)	N	N	N	NR	N	Y	Y	NR	Y	Y	N	NA	Poor
Nelson et al. (1995)	Y	Y	NR	NR	N	N	N	N	Y	Y	N	NA	Poor
Vartanian et al. (2014)	Y	N	Y	NR	N	Y	Y	NR	Y	Y	N	NA	Fair

Note. A fair rating indicates that the study is susceptible to some bias but deemed insufficient to invalidate results, and a poor rating indicates a significant risk of bias. See the Quality Assessment Tool for Before–After Studies With No Control Group for a description of the 12 numbered criteria (National Institutes of Health, 2021). Y = yes, criterion met. N = no, criterion not met. NR = criterion not reported. NA = criterion not applicable. CD = cannot determine.

Discussion

Overall, results indicate that sleep deprivation (including REM deprivation) tends to impair creative cognition. However, the nature of the impact differed across studies, and the heterogeneity of study designs means that the generalization of effects is necessarily limited in scope.

Different aspects of divergent thinking were either impaired or unaffected by sleep deprivation, and the affected components were not consistent across studies. REM deprivation, versus non-REM deprivation, also impaired aspects of divergent thinking, and the affected components also differed across studies. Various aspects of RAT-type task performance were enhanced, impaired, or unaffected by sleep deprivation. REM sleep may be particularly important for creative thinking. Using a nap paradigm, Cai et al. (2009) found that a brief nap including REM improved participants’ RAT performance compared to a non-REM nap and quiet rest condition (Cai et al., 2009). In this case, the REM nap was thought to enhance creativity by integrating new information into existing networks of associations, and thus improve the ability to generate the correct solution to previously primed RAT items.

Verbal divergent thinking was also more negatively impacted by sleep deprivation than figural performance. This difference in the effect of sleep deprivation on verbal versus visual divergent thinking aligns with correlational research that indicates higher visual creativity is associated with fewer hours of sleep and poorer sleep quality (Ram-Vlasov et al., 2016; van Heugten-Van der Kloet et al., 2015). It has been argued that psychobiological mechanisms regulating arousal underlie both visual creativity and poorer sleep quality (Ram-Vlasov, Tzischinsky, Green, & Shochat, 2016). Poor sleep quality is one of the defining characteristics of insomnia (American Psychiatric Association [APA], 2022), and eminent creatives in the visual arts appear to be posthumously diagnosed with insomnia more often than what may be expected in the general population (e.g., Henri Matisse; Zeidler, 2021).

The fair-to-poor rated quality of research limits the inferences that can be drawn about the effect of sleep deprivation on the associative basis of creativity or divergent thinking. The primary limitation of all studies was low statistical power, with the majority of studies having samples capable of detecting only very, and unrealistically, large effects. There is, therefore, uncertainty about which creative processes are impacted by sleep deprivation. Small samples also yield imprecise effect estimates; thus, the size of any effects is uncertain (Higgins et al., 2021). There was also a significant gender bias, with the majority of participants being male. Sleep deprivation has been shown to have a more detrimental effect on women’s cognition compared to men (Corsi-Cabrera, 2003; Rangtell et al. 2019; & Kingston, 2019), marking this as an important feature of the reviewed literature. Finally, experimental designs were limited in their control of practice effects and sleep protocols. These limitations were likely compromises necessitated by the practical difficulties of carrying out such studies and ethical constraints placed on sleep deprivation research (Barber, 2017).

Recommendations for future research

First and foremost, future studies need larger sample sizes to be able to detect realistic effect sizes. This is critical for detecting true effects and avoiding the abandonment of useful research that does not show significant effects owing to low power (Funder & Ozer, 2019; Lakens, 2017, 2022). Future studies should also pay closer attention to the gender profile of samples. It is noteworthy that the only study with an equal gender balance was also the only study to find a beneficial impact of sleep deprivation on creative cognition (Landmann et al., 2016).

Studies should include objective measures of sleep to confirm that participants maintain a regular sleep-wake schedule prior to testing. This has become much easier to do, and to do naturalistically, with current technology (i.e., wrist activity monitoring; Dunican et al., 2017). To reduce the impact of practice and fatigue on performance, future studies should include a normal sleep comparison group or counterbalance the order of testing across participants in a crossover design. Changes in performance over time could also be assessed to determine if creativity is enhanced at night compared to the detrimental effects found for 24 h or more of prolonged wakefulness. Emerging research suggests that sleep onset (the first stage of non-REM sleep) may have a beneficial impact on creativity (Lacaux et al., 2021), and such sleep can intrude while awake, particularly in fatigued or sleep-deprived individuals (Poudel et al., 2021). Future lab-based studies should measure such sleep intrusions to determine their impact on creative cognition in fatigued or sleep-deprived participants. Contemporary research also suggests that total sleep deprivation may not be warranted as partial sleep deprivation produces similar effects on performance and may have fewer adverse consequences (Barber, 2017).

Finally, well-validated contemporary measures of creative cognition should be utilized. The RAT in particular has been criticized as it has poor psychometric properties, including low predictive validity, high sensitivity to cultural knowledge, and lack of face validity, as it does not measure remote associations but rather one common mediator linking the three seemingly unrelated words that could be associated in different ways (Bowden & Jung-Beeman, 2003; Jung & Hunter, 2022). New open-ended measures may better capture the associative basis of creative cognition as it occurs in highly creative people (Johnson et al., 2019; Merseal et al., 2023). These measures allow participants to provide any response to a cue word, and the degree of association between two words can be objectively determined using computational tools based on distributional models of semantic distance (Beaty & Johnson, 2020; Merseal et al., 2023). Employing contemporary, well-validated measures of creative cognition will help achieve reliable, reproducible, and robust results.

Strengths and limitations of this review

This review contributes to the creative literature by providing a comprehensive analysis of the effects of sleep deprivation on measures of creative cognition. The review’s focus on experimental studies both strengthens and refines the quality of evidence that can be reported. The evidence presented clarifies what is known and useful recommendations for future studies of the effect of sleep loss on creative cognition are offered.

The findings of the present review should also be considered with respect to its limitations. Primary results were analyzed in each source, but in many cases, required statistics were missing and neither authors nor data could be located; consequently, some statistics had to be inferred from available data and are interpreted with caution. The available body of literature is also small, highlighting how little work has been done on causal processes in this area of research.

Conclusions

This synthesis and critique of the literature provides provisional evidence that sleep deprivation impairs creative thinking on divergent and associative thinking tasks. All studies assessing divergent thinking showed that sleep deprivation impaired at least one aspect of task performance and sleep deprivation either impaired, enhanced, or did not change performance on tasks measuring the associative basis of creative thinking. The apparent association between sleep deprivation and creativity often reported in biographical accounts of creative individuals is thus probably not due to any “benefit” of deprivation, but rather poor sleep quality or insomnia that is disproportionally found in creative individuals (Iranzo et al., 2018; Iszáj & Demetrovics, 2011; Ram-Vlasov et al., 2016; van Heugten-Van der Kloet et al., 2015; Zeidler, 2021). However, the methodological quality of the studies, imprecision of CIs for effect sizes, and heterogeneity of study designs mean that only a cautious conclusion regarding the negative effect of sleep deprivation on creative cognition can be made. This observation highlights the need for additional research utilizing the present review’s design recommendations. Research building on this sparse body of literature could identify the optimum sleep quantity or quality for different aspects of creative thinking. Such knowledge could enhance creativity, a skill that is increasingly recognized as one that drives positive change within society.

Acknowledgments

The authors thank Associate Professor Joseph Ciorciari for providing feedback on the manuscript and the two anonymous reviewers whose suggestions helped improve and clarify it.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10400419.2024.2343508

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This research was supported by an Australian Government Research Training Program Scholarship.