The effectiveness of the loci method as a mnemonic device: Meta-analysis

Abstract

The “loci method” is a popular mnemonic device that involves visualising and recalling items at specific points along a familiar route. The loci method has been used for thousands of years, and by many successful memory athletes; yet there have been relatively few educational and clinical applications, possibly owing to empirical uncertainty. The current meta-analysis of 13 randomised controlled trials (RCTs) mostly based in university settings demonstrated the effectiveness of the loci method as a mnemonic device, with a medium effect size (g = 0.65, 95% confidence interval [CI] = [0.45, 0.85]; I² = 45.5%). The effect size remained at similar levels in further analyses adjusting for publication bias, the impact of removing each study, setting, control conditions, outliers, and number of loci method sessions. High risk of experimental bias was indicated, however, as the vast majority of studies did not report procedures to minimise biases relating to random sequence generation and allocation concealment. Overall, this meta-analysis of predominantly university-based RCTs has provided good initial support for the loci method as a mnemonic device and this may encourage future investigations and applications, particularly in educational settings, where it has the potential to improve recall of information relevant to academic success.

Keywords

Meta-analysis mnemonic memory loci

Introduction

The “loci method” (loci being Latin for “places”) is a popular mnemonic device that involves visualising to-be-remembered items at specific points along a familiar route, and then mentally retracing the visualised items—as they appear on the route—during recall (Maguire et al., 2003). More specifically, five visualisation steps are involved in the loci method: (1) imagining a familiar route/journey (e.g., a commute or a walk around one’s house); (2) selecting several memorable landmarks on the route (e.g., particular buildings, trees, or rooms); (3) creating imagery for each to-be-remembered item (e.g., a person); (4) linking each item to one of the landmarks; (5) for recall, imagining the journey, observing the items at each landmark along the route. It is thought that the loci method relies on navigation-based cognition, and findings from neuroimaging studies demonstrate an association of loci method engagement with increased connectivity of visuospatial brain regions (Caplan et al., 2019; Dresler et al., 2017; Maguire et al., 2003); however, the method also appears to incorporate general mnemonic devices such as effortful attention, memory for emotion, organisation, linking, chunking, and elaboration (Bellezza & Reddy, 1978; Caplan et al., 2019; Carney et al., 1994; McCabe, 2015; Restorff, 1933; Ross & Lawrence, 1968).

The origins of the loci method—also referred to as the memory journey, the mental walk, and the memory palace—can be traced to hunter–gatherer societies (Kelly, 2015); however, the method is commonly attributed to the Greek poet Simonides of Ceos (556–468 BC) who (1) identified the bodies of fellow banquet guests killed by a building collapse using his memory of their seating allocations and (2) later formalised the loci method based on this experience (Yates, 1966). The loci method has been continually used in virtually unchanged format for thousands of years, demonstrating its longevity and utility (Maguire et al., 2003). In recent years, the method has been extensively used by the world’s most successful memory athletes who have regularly demonstrated the ability to memorise hundreds of abstract information units within minutes (Dresler et al., 2017; Maguire et al., 2003). In a particularly striking example, the loci method was employed by Lu Chao who recalled the mathematical constant π to 67,890 decimal places—without error (Hu et al., 2009; Raz et al., 2009).

Aside from the above feats of memory, the loci method has frequently been used in combination with other mnemonic devices in multifaceted memory training programmes for older adults and people with cognitive impairment (Gross et al., 2012, 2014; Hudes et al., 2019; Verhaeghen et al., 1992; Wolgemuth et al., 2008; Yang et al., 2018). However, there have been relatively few direct applications of the loci method within clinical and educational settings (Dalgleish et al., 2013; McCabe, 2015). While age, cognitive functioning, health status, and other demographics may interfere with clinical applications of the loci method (Rebok et al., 2013), these factors are unlikely to be as important in educational settings. It is therefore puzzling that the loci method is not more widely used by students, especially since the expected enhancement of information recall would likely aid exam performance (McCabe, 2015).

A potential reason for the few practical applications of the loci method relates to the lack of high-quality studies directly investigating its effectiveness. Many observational and quasi-experimental studies have been conducted across several decades—with generally favourable results (Cornoldi & De Beni, 1991; Dresler et al., 2017; Groninger, 1971; Gross et al., 2012; Lea, 1975; Maguire et al., 2003; McCabe, 2015; Ross & Lawrence, 1968), and a 1992 meta-analysis of 12 studies of varying design and quality supported the effectiveness of the loci method for older adults, with a large effect size yielded for a within-groups pre-to-post analysis (d = 0.80) (Verhaeghen et al., 1992). However, relatively few of the most rigorous study design for evaluating intervention effectiveness—randomised controlled trials (RCTs)—have been undertaken and findings from RCTs have yet to be synthesised through meta-analysis. Accordingly, the current meta-analysis of RCTs investigates the effectiveness of the loci method as a mnemonic device—with a view to encouraging the consideration of future practical and research applications.

Methods

Eligibility criteria for study selection

Only peer-reviewed and published RCTs investigating the effectiveness of the loci method as a mnemonic device were included. RCTs minimise error and bias, offering the most rigorous method of determining whether a cause–effect relation exists between treatment and outcome (Sibbald & Roland, 1998). Excluded were studies with different designs, RCTs investigating other mnemonic devices, RCTs which examined the effectiveness of the loci method in combination with other mnemonic devices, and RCTs with the loci method incorporated as part of all control conditions. No limits were set according to study outcome, status, or language (studies not written in English were considered).

Literature search and data extraction

Search terms relating to the loci method (i.e., loci method OR loci system OR journey* OR palace) were combined with words relating to memory (i.e., memory* OR remember* OR retrieval OR recall OR mnemo* OR mind) and RCTs (i.e., random* OR control* OR experimental OR RCT). It is worth noting that the trialled search term “method of loci” did not return additional search results (owing to its similarity to the term “loci method”) and it was therefore not included. PsycINFO, ERIC, MEDLINE, and Academic Search Premier comprised the databases, last searched on 22 September 2020. Additional records were identified through hand-searching of reference lists of included studies. The first author screened all abstracts and the second author independently screened 50% of abstracts. When we disagreed regarding the screening outcome of an abstract, it was included in screening at “full-text” level, which was subsequently conducted by the first author. Data were managed using EndNote X7 (Thomson Reuters Corp.) and word processing software. Extracted data covered setting, participant characteristics, loci training, control conditions, outcome measures, data collection timepoints, dropout, and risk of bias.

Statistical analysis

The meta-analysis was conducted using a “random effects” model which assumes that the variance in observed effects reflects both sampling variability and real differences resulting from heterogeneity in study populations, follow-up length, and other factors (Riley et al., 2011). All statistical analyses were performed using Comprehensive Meta-Analysis (version 2.0, Biostat Inc. see supplementary material). Pooled mean effect sizes (Hedges’ g) with 95% confidence intervals (CI) were calculated; effect sizes of 0.2, 0.5, and 0.8 refer to small, moderate, and large effect sizes respectively (Cohen, 1988). All effect sizes were automatically calculated using Comprehensive Meta-Analysis; means (and standard deviations) and sample sizes for the intervention and control groups comprised the requisite data for the automated effect size calculations, and these data were independently extracted by the authors, before being cross-checked. Higgin’s I² percentages interpreted the heterogeneity of effect sizes; scores of 25%, 50%, and 75% indicate low, moderate, and high heterogeneity, respectively (Higgins & Thompson, 2002). Higgin’s I² has less precision in relatively small meta-analysis such as the current one; therefore, two additional heterogeneity statistics were calculated: Cochran’s Q (Cochran, 1950) and τ² (Higgins, 2008). The former provides a more conservative estimate of heterogeneity based on statistical significance, the latter represents the among-study variance of the true effect sizes. Data from the final post-intervention data collection point were analysed, reflecting the practical need for mnemonic devices to be effective for a period of time (Higbee et al., 1990). These data were extracted and cross-checked by the study’s two authors, with any discrepancies resolved through discussion. To avoid double-counting (Senn, 2009), the effects of different intervention arms representing the loci method included in a single study were averaged and entered once in the analysis; similar averaging was conducted for certain outcomes (Table 1). Three publication bias analyses were conducted: (1) the “Trim and Fill” procedure which corrects for funnel plot asymmetry (Duval & Tweedie, 2000); (2) the rank correlation test which examines the associations of effect sizes with their corresponding sampling variances (Begg & Mazumdar, 1994); and (3) Egger’s regression test which regresses standardised effect sizes on their precisions and assumes that the regression intercept is zero in the absence of publication bias (Egger et al., 1997). Three sensitivity analyses were conducted: (1) a one-study-removed analysis (Comprehensive Meta-Analysis (version 2.0, Biostat Inc.) computed pooled effect sizes when each study was removed from the meta-analysis, therefore illustrating the impact of each study on the primary pooled effect size; (2) a meta-analysis wherein the only two studies with non-student samples were excluded; and (3) a meta-analysis that removed outliers identified through visual inspection of the forest plot. Finally, two-moderator analyses were conducted: (1) a comparison of pooled effect sizes from studies with “active” and “no instruction” control conditions (a more nuanced control condition comparison would have been at the expense of limited available statistical power) and (2) a meta-regression with the number of loci method training sessions as a predictor of the effect.

Table 1.

Study characteristics.

Study	Setting (country)	Participants	Recruitment procedure	N	%f	M age (SD)	Loci method training	Control	Outcome	Final PI time	Dropout (%)^a		Quality
Study	Setting (country)	Participants	Recruitment procedure	N	%f	M age (SD)	Loci method training	Control	Outcome	Final PI time	Loci	Ctrl	R	A	C
Bass & Oswald (2014)	University (US)	Students	Online and course credit	94	64	19.0 (1.2)	Instructions and video	No instructions	Word recall	Same day	0	0	–	–	+
Bellezza & Reddy (1978) ^b	University (US)	Students	Course credit	72	NS	NS	Instructions	Visualisation	Word recall	Same day	0	0	–	–	+
Crovitz (1971) ^b	University (US)	Students	Course requirement	70	NS	NS	Instructions	No instructions	Word recall	Same day	0	0	–	–	+
Dalgleish et al. (2013) ^c	Community (UK)	Individuals with depression	Newspaper and health centre ads	42	64	45.6 (10.7)	Instructions and practice	Chunking and rehearsal	Memory recall	2 weeks	16	0	–	–	+
De Beni et al. (1997) ^b	University (Italy)	Students	NS	125	50	NS	Three training sessions	Rehearsal	Word recall	1 week	0	0	–	–	+
De Beni et al. (1997) ^d	University (Italy)	Students	NS	34	82	20.0 (NS)	Three training sessions	Rehearsal	Word recall	1 week	0	0	–	–	+
De Beni et al. (1997) ^c,d	University (Italy)	Students	NS	32	NS	20.0 (NS)	Four training sessions	Rehearsal	Word recall	2 weeks	0	0	–	–	+
Engvig et al. (2010)	Community (Norway)	Middle-aged and older adults	Newspaper ad and interview	45	51	60.8 (9.3)	8-week programme	No instructions	Source memory	9 weeks	0	9	–	–	+
Hill et al. (1991) ^c	Community (USA)	Adults over the age of 60	Newspaper and TV ads	71	61	70.4 (6.2)	Instructions	Location memory tips	Word recall	3 days	0	0	–	–	+
Kroneisen & Makerud (2017) ^e	University (Germany)	Students	Course credit	48	93	20.3 (1.9)	Instructions	Visualisation	Word recall	Same day	0	0	–	–	+
Kroneisen & Makerud (2017) ^f	University (Germany)	Students	Course credit	52	81	21.0 (3.3)	Instructions	Visualisation	Word recall	Same day	0	0	–	–	+
Legge et al. (2012) ^g	University (US)	Students	Online and course credit	142	62	19.1 (1.6)	Instructions and practice	No instructions	Word recall	Same day	0	0	–	–	+
Massen & Vaterrodt-Plunnecke (2006) ^h	University (Germany)	Students and unspecified others	NS	132	38	NS	Instructions	Rehearsal	Word recall	Same day	0	0	–	–	+
Massen & Vaterrodt-Plunnecke (2006) ^h	University (Germany)	Students and unspecified others	NS	108	64	NS	Instructions and practice	Rehearsal	Word recall	Same day	0	0	–	–	+
Qureshi et al. (2014)	University (Pakistan)	Students	Part of class	78	NS	NS	Four training sessions	Worksheets	MCQ Test	Same day	0	0	–	–	+
Weinstein et al. (1981) ⁱ	University (US)	Students	Course requirement	100	NS	NS	Instructions or practice	No instructions	Word recall	Same day	0	0	–	–	+

%f: % female; Ctr: control; MCQ: Multiple Choice Questionnaire; NS: not stated in paper; PI: post-intervention data collection point.

Dropout (%) from study at post-intervention.

This study was subsequently excluded from the meta-analysis because the requisite data were not provided in the manuscript, and the corresponding author did not provide another record of this data.

Data from the final post-intervention data collection point were analysed.

Word recall scores for both written text and orally presented stimuli were averaged in analysis.

Word recall scores for both low and high “imageability” stimuli were averaged in analysis.

Word recall scores for four “imageability”/“survival” scenarios were averaged in analysis.

Word recall scores (from the “lenient scoring” category) for two loci conditions were averaged in analysis.

Word recall scores (across three lists) for two loci conditions and two rehearsal conditions were averaged in analysis.

Word recall scores for four loci conditions were averaged in analysis. Quality: A = allocation concealment; C = completeness of data; R = random sequence generation. ± = procedure to minimise bias reported/not reported. In the case of “completeness of data,” data were considered complete if no dropouts were present or if appropriate statistical procedures accounted for missing data.

Risk of bias assessment

In line with previous meta-analyses conducted by the first author and colleagues (Twomey et al, 2020), three criteria from the Cochrane Collaboration’s tool for assessing risk of bias (Higgins & Green, 2011) were deployed: random sequence generation, allocation concealment, and completeness of outcome data. For random sequence generation, a low risk of bias is present when a random component in the process of generating the order of participant allocations to experimental or control conditions is reported (e.g., using a computerised random number generator); a high risk of bias is present when a non-random component in the sequence generation process is reported (e.g., allocation by case record number or date of birth). For allocation concealment, a low risk of bias is present when the assignment of participants or experimental and control conditions cannot be foreseen by researchers (e.g., using sequentially numbered, opaque, sealed envelopes); a high risk of bias is present when researchers can foresee and possibly influence the allocation of participants, indicating a selection bias. For completeness of outcome data, the risk of bias is related to the percentage of missing data (<20% is considered a low risk in RCTs with a short-term follow-up) and whether or not this is related to the outcome, and balanced across experimental conditions. When procedures in relation to minimising the above potential biases are not reported in the published manuscript of a given study, the risk level can be adjudged “unclear”; however, a high risk of bias is also commonly adjudged in these cases, to promote greater caution in relation to the interpretation of findings (Twomey et al., 2020). The latter stance was taken in the current study: if procedures to minimise a bias were not reported in a study’s article, a high risk of bias was marked. Separate risk of bias assessments were conducted by this study’s authors; discrepancies were discussed and resolved.

Results

Study selection and characteristics

A total of 1,326 records were screened and 16 studies were selected for review, with both authors agreeing that 82.3% of independently screened abstracts should be subsequently examined at “full-text” level (note that all of the disputed selections—the 17.7% remaining abstracts—were also subsequently examined at “full-text” level). However, only 13 studies could be included in the meta-analysis because requisite data were unavailable for 3 reviewed studies, and their authors who were subsequently contacted could not provide it (Figure 1). It is also worth noting here that the requisite data for one study was provided by that study’s lead author (Kroneisen & Makerud, 2017). Table 1 outlines the characteristics of the selected studies (Bass & Oswald, 2014; Bellezza & Reddy, 1978; Crovitz, 1971; Dalgleish et al., 2013; De Beni et al., 1997; Engvig et al., 2010; Hill et al., 1991; Kroneisen & Makerud, 2017; Legge et al., 2012; Massen & Vaterrodt-Plunnecke, 2006 ; Qureshi et al., 2014; Weinstein et al., 1981). Publication dates ranged from 1971 to 2016, the majority of studies were set in US or European universities, and there were a variety of recruitment procedures. Sample sizes ranged from 34 to 142 (combined N = 1,244), the majority of studies had mostly female participants, and mean ages ranged from 19 to 70. Simple instructions were used for loci method training in most studies, though training sessions (of varying number) were also provided in some studies. Word recall comprised the relevant outcome in all-but-two studies; outcomes were mostly assessed on the same day as the RCT experiment. Dropout was zero in all-but-two studies: (1) a community-based study with a 2-week follow-up period: 16% of loci method participants dropped out (Dalgleish et al., 2013); (2) another community-based study with a 9-week follow-up period: 9% of controls dropped out (Engvig et al., 2010) . Finally, as indicated in Table 1, risk of bias studies was high: although dropout was invariably minimal, the vast majority of studies did not document procedures to minimise biases relating to random sequence generation and allocation concealment. It is worth noting that, based on independent evaluations, the authors agreed on 46 out of the 48 risk of bias ratings across the 16 studies (95.8% agreement rate), and upon further discussion they agreed upon the final two ratings.

Figure 1.

Literature search flow.

Inter-rater agreement levels for data used for effect size calculations

Independent extraction of the data required for Comprehensive Meta-Analysis effect size calculations had the following inter-rater agreement levels: intervention means = 100%; intervention standard deviations = 100%; intervention sample sizes = 92.3%; control means = 100.0%; control standard deviations = 100%; and control sample sizes = 100%. The one discrepancy in intervention sample sizes was resolved through subsequent author discussion.

The effectiveness of the loci method as a mnemonic device

As per the forest plot (Figure 2), comparisons from 13 studies demonstrated the effectiveness of the loci method as a mnemonic device, with a medium effect size (g = 0.65, 95% CI = [0.45, 0.85] and moderate heterogeneity that was statistically significant (I² = 45.5%; τ² = 0.06, SE = 0.056; Cochrane’s Q = 22.03; p < .05). Publication bias was indicated by asymmetry in the funnel plot (Figure 3), a moderate correlation in the rank correlation test (τ = 0.34; p < .05) and a significant Egger’s regression intercept (B0 = 3.73, p < .05). The publication bias was handled using the “Trim and Fill” procedure (Duval & Tweedie, 2000), which resulted in a slight reduction in the effect size (g = 0.55, 95% CI = [0.31, 0.78]). The results from the one-study-removed analysis are displayed in Figure 4; adjusted pooled effect sizes remained in the medium range (ranging from 0.58 to 0.69), indicating that the overall pooled effect size is only minimally impacted upon by the removal of any individual study. The sole inclusion of university-based studies in the meta-analysis also produced a medium effect size (g = 0.64, 95% CI = [0.39, 0.90]). When two outliers identified in the forest plot (the first study from Kroneisen & Makerud, 2017 and the final study from De Beni et al., 1997) were removed the effect size decreased slightly but remained in the medium range (g = 0.53, 95% CI = [0.38, 0.68]) and heterogeneity was eliminated (I² = 0%; τ² = 0.00, SE = 0.030; Cochrane’s Q = 8.92; p > .05). Although moderator analyses were restricted by the relatively small number of included studies, it is worth noting that the 9 studies with “active” control conditions produced a slightly (and non-significantly) larger effect size (g = 0.69; 95% CI = [0.42, 0.97]; I² = 49.5%) than the 4 studies with “no instruction” control conditions (g = 0.58; 95% CI = [0.26, 0.91]; I² = 47.9%). The meta-regression indicated that the number of loci method training sessions was not a significant predictor of the effect size (Q = 1.34; p > .05).

Figure 2.

Meta-analysis forest plot: effectiveness of the loci method as a mnemonic device.

Figure 3.

Funnel plot and “fill and trim” method. High asymmetry indicates publication bias; in this plot there is some asymmetry. In the “trim and fill” method, the shaded dots represent the imputed studies, and the shaded diamond represents the adjusted effect size.

Figure 4.

Meta-analysis forest plot for one-study removed analysis.

Discussion

This meta-analysis of 13 RCTs demonstrated the effectiveness of the loci method as a mnemonic device, with a medium effect size (g = 0.65, 95% CI = [0.45, 0.85]) that remained at similar levels in further analyses adjusting for publication bias, the impact of removing each study, setting, outliers, control conditions, and number of loci method sessions. Interpretive caution is warranted, as the findings are based on relatively small number of predominantly university-based studies. It is worth noting that in the sensitivity analysis that adjusted for (two) outliers with very large effect sizes, the effect size remained in the medium range and (previously moderate) heterogeneity was eliminated; the reasons for the very large effect sizes in the outliers are unclear as several of their characteristics—sample sizes, demographics, cognitive ability levels, loci training conditions, follow-up periods, control conditions, and risk of bias levels—are similar to the other studies; moreover, the larger effect sizes could merely be a consequence of sampling error.

In other reasons for interpretative caution, most reviewed RCTs had a “same-day” follow-up, which limits the generalisability of the findings, and is incongruent with the practical need for mnemonic devices to be effective for a period of time (Higbee et al., 1990). Furthermore, a cumulative picture of high risk of bias was indicated across the included studies and this should be taken into account. More specifically, while there was good completeness of data across all of the included studies, the vast majority of studies did not report procedures to minimise biases relating to random sequence generation and allocation concealment. For all studies, therefore, there is a considerable risk that the order in which participants were allocated to loci and control conditions may have been influenced by non-random methodological procedures, limiting the validity of the findings. It is worth noting, however, that the risk of bias was predominantly adjudged to be high due to the absence of reported safeguarding measures, rather than the reporting of inappropriate experimental procedures.

The yielded support for the loci method is in line with findings from a 1992 meta-analysis of studies of varying design supporting the effectiveness of the loci method for older adults, with a large effect size yielded for a within-groups pre-to-post analysis (d = 0.80) (Verhaeghen et al., 1992). The slightly lower effect size in the current study may be attributable to its sole inclusion of RCTs—which provide a more stringent examination of cause-and-effect relationships (Sibbald & Roland, 1998)—and its focus on a between-groups analysis that directly compared the loci method with control conditions. The yielded support for the loci method is also in line with its documented use in feats of memory (Dresler et al., 2017; Hu et al., 2009; Maguire et al., 2003; Raz et al., 2009), findings from meta-analyses demonstrating the effectiveness of heterogeneous mnemonic devices for older adults (Gross et al., 2012; Hudes et al., 2019; Verhaeghen et al., 1992) and people with cognitive disorders (Yang et al., 2018), and findings from a systematic review supporting the effectiveness of heterogeneous mnemonic devices for youth with learning disabilities (Wolgemuth et al., 2008). What sets the current meta-analysis apart from these previous reviews is its sole inclusion of RCTs, its direct focus on the loci method which offers greater precision, and—owing to the sample compositions of included studies—the ability to generalise the findings to university settings, to those in younger adulthood, and to those without significant cognitive impairments.

The support for the loci method as mnemonic device could be used to justify practical applications, particularly in educational settings where it is arguably under-used given its potential to improve recall of information relevant to exam performance and academic success (McCabe, 2015). Barriers such as the lack of awareness of mnemonic devices among students (McCabe et al., 2013) and the considerable effort involved in mastering the loci method (Gross et al., 2014), would also need to be overcome for widespread educational use. The findings offer less support for the clinical application of the loci method, as only one included study had a clinical sample: individuals with depression (Dalgleish et al., 2013). However, the justification of the loci method’s use in this study—to construct an accessible mental repository for positive, self-affirming memories to counteract downturns in negative affect—is intriguing and worthy of future investigation. Similarly, only one included study had a sample consisting solely of older adults (Hill et al., 1991), and this should be addressed in future research.

Given the longevity of the loci method in popular culture (Maguire et al., 2003), it is surprising that relatively few studies were identified in this paper’s systematic literature research, and furthermore, that the review did not identify more RCTs with a low risk of bias. Although there exist several quasi-experimental and observational investigations of the loci method (Cornoldi & De Beni, 1991; Dresler et al., 2017; Groninger, 1971; Gross et al., 2012; Lea, 1975; Maguire et al., 2003; McCabe, 2015; Ross & Lawrence, 1968), there is a pressing need for more methodologically rigorous RCTs. Such RCTs would particularly benefit from longer follow-up periods: as mentioned above most reviewed RCTs had a “same-day” follow-up period. It was beyond the scope of the current meta-analysis to investigate the loci method’s potential mechanisms of action (e.g., visuospatial mechanisms). A systematic review of these mechanisms would be useful, and it appears that many relevant studies are available for review (Bellezza & Reddy, 1978; Caplan et al., 2019; Carney et al., 1994; Dresler et al., 2017; Maguire et al., 2003; McCabe, 2015; Restorff, 1933; Ross & Lawrence, 1968).

In conclusion, this meta-analysis of a relatively small number of predominantly university-based RCTs has provided good initial support for the loci method as mnemonic device and this may encourage future investigations and practical applications, particularly in educational settings. The loci method appears to have been useful to humans since hunter–gatherer times (Kelly, 2015), yet there is still much to be explored and learned for its utility to be maximised in today’s society.

Research Data

sj-cma-1-qjp-10.1177_1747021821993457 – Supplemental material for The effectiveness of the loci method as a mnemonic device: Meta-analysis

sj-cma-1-qjp-10.1177_1747021821993457 for The effectiveness of the loci method as a mnemonic device: Meta-analysis by Conal Twomey and Meike Kroneisen in Quarterly Journal of Experimental Psychology

This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).

Footnotes

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The second author was also an author of two of the studies (within one paper) included in this meta-analysis.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Conal Twomey

Meike Kroneisen

Supplementary material

The supplementary material is available at .

References

Bass

W. S.

Oswald

K. M.

(2014). Proactive control of proactive interference using the method of loci. Advances in Cognitive Psychology, 10(2), 49–58. https://doi.org/10.5709/acp-0156-3

Begg

C. B.

Mazumdar

(1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 50(4), 1088–1101. https://doi.org/10.2307/2533446

Bellezza

F. S.

Reddy

B. G.

(1978). Mnemonic devices and natural memory. Bulletin of the Psychonomic Society, 11(5), 277–280. https://doi.org/10.3758/BF03336829

Caplan

J. B.

Legge

E. L.

Cheng

Madan

C. R.

(2019). Effectiveness of the method of loci is only minimally related to factors that should influence imagined navigation. Quarterly Journal of Experimental Psychology, 72(10), 2541–2553. https://doi.org/10.1177/1747021819858041

Carney

R. N.

Levin

J. R.

Levin

M. E.

(1994). Enhancing the psychology of memory by enhancing memory of psychology. Teaching of Psychology, 21(3), 171–174. https://doi.org/10.1207/s15328023top2103_12

Cochran

W. G.

(1950). The comparison of percentages in matched samples. Biometrika, 37(3/4), 256–266. https://doi.org/10.2307/2332378

Cohen

(1988). Statistical power analysis for behavioral sciences (2nd ed.). Academic Press.

Cornoldi

De Beni

(1991). Memory for discourse: Loci mnemonics and the oral presentation effect. Applied Cognitive Psychology, 5(6), 511–518. https://doi.org/10.1002/acp.2350050606

Crovitz

H. F.

(1971). The capacity of memory loci in artificial memory. Psychonomic Science, 24(4), 187–188. https://doi.org/10.3758/BF03335561

10.

Dalgleish

Navrady

Bird

Hill

Dunn

B. D.

Golden

A.-M.

(2013). Method-of-loci as a mnemonic device to facilitate access to self-affirming personal memories for individuals with depression. Clinical Psychological Science, 1(2), 156–162. https://doi.org/10.1177/2167702612468111

11.

De Beni

Mò

Cornoldi

. (1997). Learning from texts or lectures: Loci mnemonics can interfere with reading but not with listening. European Journal of Cognitive Psychology, 9(4), 401–415. https://doi.org/10.1080/713752565

12.

Dresler

Shirer

W. R.

Konrad

B. N.

Müller

N. C. J.

Wagner

I. C.

Fernández

. . .Greicius

M. D.

(2017). Mnemonic training reshapes brain networks to support superior memory. Neuron, 93(5), 1227–1235.E6. https://doi.org/10.1016/j.neuron.2017.02.003

13.

Duval

Tweedie

(2000). Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56(2), 455–463. https://doi.org/10.1111/j.0006-341X.2000.00455.x

14.

Egger

Davey Smith

Schneider

Minder

(1997). Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315, Article 629. https://doi.org/10.1136/bmj.315.7109.629

15.

Engvig

Fjell

A. M.

Westlye

L. T.

Moberget

Sundseth

Ø.

Larsen

V. A.

Walhovd

K. B.

(2010). Effects of memory training on cortical thickness in the elderly. Neuroimage, 52(4), 1667–1676. https://doi.org/10.1016/j.neuroimage.2010.05.041

16.

Groninger

L. D.

(1971). Mnemonic imagery and forgetting. Psychonomic Science, 23(2), 161–163. https://doi.org/10.3758/BF03336056

17.

Gross

A. L.

Brandt

Bandeen-Roche

Carlson

M. C.

Stuart

E. A.

Marsiske

Rebok

G. W.

(2014). Do older adults use the method of loci? Results from the ACTIVE study. Experimental Aging Research, 40(2), 140–163. https://doi.org/10.1080/0361073x.2014.882204

18.

Gross

A. L.

Parisi

J. M.

Spira

A. P.

Kueider

A. M.

J. Y.

Saczynski

J. S.

. . .Rebok

G. W.

(2012). Memory training interventions for older adults: A meta-analysis. Aging and Mental Health, 16(6), 722–734. https://doi.org/10.1080/13607863.2012.667783

19.

Higbee

K. L.

Clawson

Delano

Campbell

(1990). Using the link mnemonic to remember errands. The Psychological Record, 40(3), 429–436. https://doi.org/10.1007/BF03399551

20.

Higgins

(2008). Commentary: Heterogeneity in meta-analysis should be expected and appropriately quantified. International Journal of Epidemiology, 37(5), 1158–1160. https://doi.org/10.1093/ije/dyn204

21.

Higgins

Green

(2011, March). Cochrane handbook for systematic reviews of interventions (Version 5.1.0). https://handbook-5-1.cochrane.org/

22.

Higgins

Thompson

S. G.

(2002). Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21(11), 1539–1558. https://doi.org/10.1002/sim.1186

23.

Hill

R. D.

Allen

McWhorter

(1991). Stories as a mnemonic aid for older learners. Psychology and Aging, 6(3), 484–486. https://doi.org/10.1037//0882-7974.6.3.484

24.

Ericsson

K. A.

Yang

(2009). Superior self-paced memorization of digits in spite of a normal digit span: The structure of a memorist’s skill. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(6), 1426–1442. https://doi.org/10.1037/a0017395

25.

Hudes

Rich

J. B.

Troyer

A. K.

Yusupov

Vandermorris

(2019). The impact of memory-strategy training interventions on participant-reported outcomes in healthy older adults: A systematic review and meta-analysis. Psychology and Aging, 34(4), 587–597. https://doi.org/10.1037/pag0000340

26.

Kelly

(2015). Knowledge and power in prehistoric societies. Cambridge University Press.

27.

Kroneisen

Makerud

S. E.

(2017). The effects of item material on encoding strategies: Survival processing compared to the method of loci. The Quarterly Journal of Experimental Psychology, 70(9), 1824–1836. https://doi.org/10.1080/17470218.2016.1209533

28.

Lea

(1975). Chronometric analysis of the method of loci. Journal of Experimental Psychology: Human Perception and Performance, 1(2), 95–104. https://doi.org/10.1037/0096-1523.1.2.95

29.

Legge

E. L. G.

Madan

C. R.

E. T.

Caplan

J. B.

(2012). Building a memory palace in minutes: Equivalent memory performance using virtual versus conventional environments with the Method of Loci. Acta Psychologica, 141(3), 380–390. https://doi.org/10.1016/j.actpsy.2012.09.002

30.

Maguire

E. A.

Valentine

E. R.

Wilding

J. M.

Kapur

(2003). Routes to remembering: The brains behind superior memory. Nature Neuroscience, 6(1), 90–95. https://doi.org/10.1038/nn988

31.

Massen

Vaterrodt-Plunnecke

(2006). The role of proactive interference in mnemonic techniques. Memory, 14(2), 189–196. https://doi.org/10.1080/09658210544000042

32.

McCabe

J. A.

(2015). Location, location, location! Demon-strating the mnemonic benefit of the method of loci. Teaching of Psychology, 42(2), 169–173. https://doi.org/10.1177/0098628315573143

33.

McCabe

J. A.

Osha

K. L.

Roche

J. A.

Susser

J. A.

(2013). Psychology students’ knowledge and use of mnemonics. Teaching of Psychology, 40(3), 183–192. https://doi.org/10.1177/0098628313487460

34.

Qureshi

Rizvi

Syed

Shahid

Manzoor

(2014). The method of loci as a mnemonic device to facilitate learning in endocrinology leads to improvement in student performance as measured by assessments. Advances in Physiology Education, 38(2), 140–144. https://doi.org/10.1152/advan.00092.2013

35.

Raz

Packard

M. G.

Alexander

G. M.

Buhle

J. T.

Zhu

Peterson

B. S.

(2009). A slice of PI: An exploratory neuroimaging study of digit encoding and retrieval in a superior memorist. Neurocase, 15(5), 361–372. https://doi.org/10.1080/13554790902776896

36.

Rebok

G. W.

Langbaum

J. B. S.

Jones

R. N.

Gross

A. L.

Parisi

J. M.

Spira

A. P.

. . .Brandt

(2013). Memory training in the ACTIVE study: How much is needed and who benefits? Journal of Aging and Health, 25(8 Suppl.), 21S–42S. https://doi.org/10.1177/0898264312461937

37.

Restorff

H. V.

(1933). Ueber die Wirkung von Bereichsbildungen im Spurenfeld: Analyse von Vorgängen im Spurenfeld. I. Von W. Köhler und H. v. Restorff. [On the effect of field formations in the trace field: Analysis of processes in the trace field. I. By W. Kohler and H. v. Restorff.]. Psychologische Forschung, 18, 299–342. https://doi.org/10.1007/BF02409636

38.

Riley

R. D.

Higgins

J. P. T.

Deeks

J. J.

(2011). Interpretation of random effects meta-analyses. British Medical Journal, 342, Article d549. https://doi.org/10.1136/bmj.d549

39.

Ross

Lawrence

K. A.

(1968). Some observations on memory artifice. Psychonomic Science, 13(2), 107–108. https://doi.org/10.3758/BF03342433

40.

Senn

S. J.

(2009). Overstating the evidence: Double counting in meta-analysis and related problems. BMC Medical Research Methodology, 9(1), Article 10. https://doi.org/10.1186/1471-2288-9-10

41.

Sibbald

Roland

(1998). Understanding controlled trials: Why are randomised controlled trials important? British Medical Journal, 316(7126), 201. https://doi.org/10.1136/bmj.316.7126.201

42.

Twomey

O’Reilly

Bultmann

Meyer

(2020). Effectiveness of a tailored, integrative Internet intervention (deprexis) for depression: Updated meta-analysis. PLOS ONE, 15(1), Article e0228100. https://doi.org/10.1371/journal.pone.0228100

43.

Verhaeghen

Marcoen

Goossens

(1992). Improving memory performance in the aged through mnemonic training: A meta-analytic study. Psychology and Aging, 7(2), 242–251. https://doi.org/10.1037/0882-7974.7.2.242

44.

Weinstein

C. E.

Cubberly

W. E.

Wicker

F. W.

Underwood

V. L.

Roney

L. K.

Duty

D. C.

(1981). Training versus instruction in the acquisition of cognitive learning strategies. Contemporary Educational Psychology, 6(2), 159–166. https://doi.org/10.1016/0361-476X(81)90045-X

45.

Wolgemuth

J. R.

Cobb

R. B.

Alwell

(2008). The effects of mnemonic interventions on academic outcomes for youth with disabilities: A systematic review. Learning Disabilities Research & Practice, 23(1), 1–10. https://doi.org/10.1111/j.1540-5826.2007.00258.x

46.

Yang

H. L.

Chan

P. T.

Chang

P. C.

Chiu

H. L.

Sheen Hsiao

S. T.

Chu

Chou

K. R.

(2018). Memory-focused interventions for people with cognitive disorders: A systematic review and meta-analysis of randomized controlled studies. International Journal of Nursing Studies, 78, 44–51. https://doi.org/10.1016/j.ijnurstu.2017.08.005

47.

Yates

(1966). The art of memory. Pimlico.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB