Introduction

Rationale

Human episodic memories are intrinsically defined in space and time1. The spatial memory and navigation system of the mammalian brain is now a relatively well-understood neurocognitive domain, and it was the subject of the 2014 Nobel Prize in Physiology or Medicine. Spatial memory can be defined as a “brain function responsible for recognizing, codifying, storing and recovering spatial information about the arrangement of objects or specific routes”2. A common assumption in psychological science is that cognitive abilities are domain-general. Often, implicit assumptions are made when generalizing cognitive findings from the sensory system in which it was researched (e.g., making conclusions about “working memory” from studying only visual working memory). However, there is in fact scarce evidence on whether human spatial memory interacts similarly with different sensory modalities, and whether findings related to spatial memory can be generalized across all senses in humans. In research on human participants, the literature is dominated by vision3, even though research on rodents often involves the sense of smell. Although one should always be mindful of the profound differences between species when discussing any physiological or sensory/cognitive function, the olfactory system in rodents shares many similarities with the human olfactory system4and often provides models for human cognitive abilities e.g4,5,6,7.,.

Recent evidence indicates that olfaction is underrepresented in research on human spatial navigation and memory[8](https://www.nature.com/articles/s41598-025-25503-5#ref-CR8 “Schwarz, M., Yang, A. L. & Hamburger, K. The human sense of smell in Spatial orientation: A state-of-the-art review. Psychol. Conscious. Theory Res. Pract. https://doi.org/10.1037/cns0000395

(2024).“), but spatial processes might be uniquely influenced by olfaction. Human olfaction stands out by having a uniquely direct (unmediated by a thalamic relay) and strong neuroanatomical connection to higher processing centers9. Olfaction operates through a cortical network extending from the olfactory bulb to the primary olfactory regions in the medial temporal lobe9, which are in turn strongly connected to the core spatial memory and navigation system, including the hippocampus (HC) and entorhinal cortex (EC)9,10,11. Moreover, there is an apparent neuroanatomical integration of perceptual and cognitive functions in olfaction, where cognitive areas aid perceptual tasks, and perceptual areas engage in cognitive tasks. Indeed, the HC and EC contribute to olfactory tasks such as odor identification (i.e., matching familiar odors to corresponding labels12,13. Previous studies showed that human visual spatial abilities correlate with olfactory identification performance, and volume of the HC predicts both olfactory functions and spatial memory functions14,15,16. Likewise, the primary olfactory areas engage in smell-related cognitive tasks, such as mental navigation in conceptual olfactory space17 or odor-cued spatial navigation tasks18. Furthermore, cognitive maps in the EC-HC network, which may provide a foundation for spatial memory and navigation ability, were recently recorded in humans using olfactory landmarks in navigation tasks18, and amnesic patients with HC damage are unable to associate specific smells with their correct locations in an odor-place association task19. Finally, some argue that the ability to remember locations of smells might be evolutionarily beneficial because it helps locating and navigating to food sources20,21. It has even been suggested that olfaction might have evolved primarily to aid spatial navigation, so that spatial orientation is its primary function22,[23](https://www.nature.com/articles/s41598-025-25503-5#ref-CR23 “Jacobs, L. F. The PROUST hypothesis: the embodiment of olfactory cognition. Anim. Cogn. https://doi.org/10.1007/s10071-022-01734-1

(2022).“) and that olfaction should indeed be regarded as a spatial sense[24](https://www.nature.com/articles/s41598-025-25503-5#ref-CR24 “Barwich, A. S. Olfaction is a Spatial sense. Rev. Philos. Psychol. 1–29. https://doi.org/10.1007/s13164-024-00764-7

(2025).“). Notably, there is a growing interest in studying olfactory spatial memory, shown by a steady increase in the number of publications related to that topic over the last 20 years (Supplementary Fig. 1). Therefore, it is of interest to summarize the current state of knowledge about human olfactory spatial memory, to compare whether human olfactory spatial memory is as good as other sensory modalities, or maybe even better.

We systematically reviewed the evidence regarding human abilities to memorize the arrangement of objects or routes based on olfactory cues, and we compared olfactory spatial memory performance with that of other senses qualitatively and in a meta-analysis. Notably, although the individual studies might focus on many different aspects of human olfaction or memory, we emphasize aspects that are of particular relevance to our theoretical viewpoint, as outlined above.

Objectives

To systematically evaluate human ability to memorize locations of smells and use them in spatial navigation tasks, and to test whether olfactory spatial memory abilities differ from those of other senses.

Methods

The present study was conducted in line with the PRISMA statement that outlines recommendations and guidelines for systematic reviews and meta-analyses25. The study was not preregistered. All data for this project, along with supplementary material, relevant code, and codebook with metainfo, have been made publicly available and can be accessed on our OSF repository (https://osf.io/5y9vj/)26.

Information sources and search strategy

Three databases (PubMed, Web of Science, Scopus) were searched on the 10th of September 2024 to identify the relevant studies. Search terms were selected in line with the PICO (Population, Intervention, Comparison, Outcome) framework. Search terms were partially identified by utilizing ChatGPT (GPT-3) to provide synonyms of phrases that could be used in relation to odor, olfaction, and spatial memory. Example search terms are: Adult population, healthy adults, odor, olfaction, olfactory sensation, scent, cognitive mapping, location memory, object-place association (for all search terms see Supplement Table 1 in “Supplement_1_tables_and_figures.pdf”). Additionally, a filter limiting the language of studies to English was applied. Search strategies for the three databases are available in Supplement Table 2.

Eligibility criteria

Spatial memory may be treated either as implicit or explicit in nature. Some cases in real life where spatial memory guides behavior does not require conscious recall, but often spatial memories are evoked by explicit questions that require conscious recall (e.g., “Where did I leave my phone?”). Notably, the concept of spatial memory in this review does not include studies that tested memories of fully implicit nature, where participants were neither asked to memorize location of smells, nor tested on their memory specifically e.g27,28.,. Instead, we included studies that allow for a direct investigation of olfactory spatial memory through conscious memory recall. More inclusion and exclusion criteria are presented in Table 1.

Study selection

Our search strategy resulted in identification of 899 articles, out of which 272 were duplicates. After abstract (n = 627 articles) and full-text (n = 37 articles) screening, 15 articles were selected for the systematic review. During full-text screening of the relevant 15 articles and manuscript preparation, additional eight relevant articles were discovered by reference tracking (for identification of articles discovered with search and reference tracking, see Supplement Table 3). These articles were screened and added to the systematic review. Also, one additional peer-reviewed empirical article, which was published during preparation of this manuscript, was added to the systematic review. Together, 24 articles are included in the present review (Fig. 1). List of articles excluded during full-text screening and reason for exclusions is available in Supplement Table 4.

Fig. 1

PRISMA 2020 flow diagram for new systematic reviews which included searches of databases, registers and other sources25.

Selection process

All studies were screened at two stages (title and abstract, and full text screening) by one reviewer (MS). Uncertainties were consulted with the other reviewer (NC) leading to a consensus decision. During the title and abstract screening process, articles were assessed manually and, in parallel, utilizing artificial intelligence with the ASReview application, a free-source tool, created to make the article screening process faster and more efficient[29](https://www.nature.com/articles/s41598-025-25503-5#ref-CR29 “van de Schoot, R. et al. ASReview: active learning for systematic reviews. Zenodo https://doi.org/10.5281/zenodo.4158671

(2020).“),30. For more details and analytic insights from the automated screening process, see “Supplement_2_ASReview_details.pdf” on OSF26.

Data collection process

We found 15 studies that compared performance in the spatial memory task between sensory modalities, including olfaction (either between- or within-group design). Thirteen studies compared performance in olfaction with that in vision, four with hearing, one with touch, one with a combination of olfaction and semantics (i.e., participants first heard the list of odor names, then were presented with a sequence of odors at specific locations, and during recall were tasked with replicating that sequence while identifying the odors using previously heard list), two with a combination of olfaction and vision, one with partial sensory deprivation (no hearing, vision, or olfaction), and one with multimodal perception (vision, olfaction, and taste). One reviewer (MS) extracted data from all these studies. Besides the olfactory-visual comparison, only qualitative syntheses could be derived from the retrieved data in other modalities, because the studies were too few to perform meta-analyses. However, we retrieved enough data regarding the differences between olfaction and vision to perform a meta-analysis of how olfactory spatial memory may differ from visual spatial memory. Note that not all studies provided raw data or group means and standard deviations (SD) in text or table for the relevant variables, but some provided supplementary files with data or showed data on the plots. If no other source of data was available, the reviewer (MS) used a screen measuring ruler application (Page Ruler add-on to Google Chrome browser) and proportion calculations to estimate relevant values from the plots. For studies that reported standard error of the mean (SEM) or 95% confidence interval (95% CI), SD was calculated using appropriate formulas (see script “Supplement_metaanalysis.R”26).

Data items

Data were extracted from the studies that compared spatial memory performance between olfaction and other conditions. Outcome variables of interest were: performance in a spatial memory task in the olfactory condition, and performance in a spatial memory task in the other sensory condition. Performance could be measured in multiple ways, for example number of correctly placed objects, proportion of correct object-place associations, distance error, etc. Any measure of performance was deemed eligible for inclusion. If a study reported results for clinical and control samples, only the control sample was taken into account. If a study reported results divided into groups of participants, and these groups were non-clinical samples (e.g., athletes and non-athletes, younger and older, etc.) combined results for the groups were estimated and included in the meta-analysis. Formulas for pooled means and SDs can be found in the “Supplement_1_tables_and_figures.pdf”.

Study risk of bias assessment

Risk of bias in the included studies was assessed by one reviewer (MS) using a selection of questions from the “Checklist for analytical cross-sectional studies”31, in order to evaluate methodological quality of the studies. Uncertainties were consulted with the other reviewer (NC) leading to a consensus decision. For each study we calculated a score representing a sum of satisfied criteria (max 6 points).

Statistical analysis

Data was prepared, processed, and visualized in R32 and RStudio33 utilizing mostly the tidyverse package (version 2.0.0)34. Meta analysis was performed using the package “metafor” (version 4.2-0.2)35. Scripts are available in the repository26. Because the included studies reported various types of performance measures, the analysis was carried out using the standardized mean difference (i.e., Glass’s estimator obtained with metafor::escalc()) as the outcome measure. A random-effects model was fitted to the data. The amount of heterogeneity (i.e., tau2, was estimated using the Hedges’ estimator. In addition to the estimate of tau2, we report the Q-test for heterogeneity and the l2 statistic. To examine whether studies may be outliers and/or influential in the context of the model, we evaluated the studentized residuals and Cook’s distances, and visually inspected the funnel plot.

Results

Study characteristics

Demographics summary

Figure 2 depicts demographic characteristics of the total population in each study (across all experimental groups) and average across studies. For detailed demographic summary per experimental group within each study, see Supplementary Fig. 1. On average, studies tested 68 participants (SD = 104, median = 40), 57.5% of participants were female, and average age of participants was 26 years old (SD = 5.81). Most of the studies reported mean and SD or standard error of the mean (SEM) for the age, but two only reported participants’ age range36. For the studies that reported demographic information divided into different sample groups but did not provide a global summary, the pooled means and SDs for age shown on Fig. 2 were calculated according to the formulas presented in the supplementary material. Detailed description of demographic characteristics is available in the supplementary material.

Fig. 2

Summary of demographic information for individual studies (top panel) and across the studies (bottom panel). The left column shows the number of participants in individual studies in the top panel, and median (open circle), mean and SD across all studies in the bottom panel. The middle column shows color-coded gender distribution in individual studies (top), and averaged across all studies (bottom). The right column shows the mean, SD (solid line), and range (purple dashed line) of participants’ age in individual studies (top), and a mean and SD across all studies that provided mean age values (bottom). Note that two articles reported results from more than one study36,37, and in those instances, individual studies appear in separate rows in this figure.

Task summary

Figure 3 shows a summary of methods in the selected studies. In this review, we distinguish between odor-place association tasks, where participants were specifically asked to indicate correct odor-place association; and other tasks, which we decided to classify as odor-cued navigation tasks, where participants were asked to make a spatial decision based on an odor encountered on their way (e.g., turn left, move forward, stop, etc.). We adopted this distinction based on previous reviews of spatial memory that also distinguished between these two tasks, suggesting possibly different memory mechanisms for encoding object locations and routes due to additional sequential information present in route encoding2,38. Additionally, recalling routes might be based on associating a given stimulus with a spatial action6 rather than employing a specific spatial map or sequential information, further differentiating the two types of tasks. We acknowledge that our classification is limited, and some studies fit into these schemas better than others. More detailed division could be needed in the future to create more specific tasks categories, for example, those with continuous versus discrete smell sampling, or those using labyrinths versus “open field” navigation. For the purpose of this review, we propose to classify the found studies only into the two task types as a balance between grouping overly different studies together, and splitting them into too many task categories with only one or two studies in each.

Twenty-one studies (from 19 distinct scientific articles) employed an odor-place association task18,19,20,21,36,37,39,40,41,42,43,44,45,46,47,48,49,50,[51](https://www.nature.com/articles/s41598-025-25503-5#ref-CR51 “Szychowska, M., Ersson, K. & Olofsson, J. K. Asymmetric cross-sensory interference between Spatial memories of sounds and smells revealed in a virtual reality environment. J. Exp. Psychol. Learn. Mem. Cogn. https://doi.org/10.1037/xlm0001493

(2025).“) but the specific task designs and outcome measures varied, as shown on Fig. 3; Table 2 in section “Results of individual studies”. Note that on Fig. 3 there are 22 items marked as Odor-place association instead of 21, because in one study participants were tested in two settings, both while walking in the room and combined room and screen42. Furthermore, note that one study used a design that matched both task types18.

Six studies used different paradigms that we considered a type of an odor-cued navigation task18,52,[53](#ref-CR53 “Schwarz, M. & Hamburger, K. Memory effects of visual and olfactory landmark information in human wayfinding. Cogn. Process. https://doi.org/10.1007/s10339-023-01169-7

(2023).“),54,55,56. Specifically, tasks were to: find a way in a virtual labyrinth (on screen) by making turning decisions based on olfactory (or visual, or olfacto-visual) landmarks52,[53](#ref-CR53 “Schwarz, M. & Hamburger, K. Memory effects of visual and olfactory landmark information in human wayfinding. Cogn. Process. https://doi.org/10.1007/s10339-023-01169-7

(2023).“),54,55; find a location in the room where the perceived smell mixture matches previously encoded target, navigating based on the gradient of two odors56; and navigate to a specific odor location (odor-place association) using only olfactory landmarks (odor-cued navigation) in a virtual arena18.

Spatio-contextual environments summary

We identified seven different spatio-contextual environments that were used in the studies. In ten studies participants performed the memory task on a screen18,20,44,45,46,47,48,49,50,55. Notably, it was unclear whether Raithel et al.18 had presented the visual context on a computer screen or through the VR goggles, but because the study also included functional Magnetic Resonance Imaging (fMRI), we assumed the task was on the screen. In four studies, participants performed the task at a table or a board in front of them19,39,40,41; In two studies (described in one article) participants had to memorize odor locations on a board in front of them, and at the same time associate them with a visual context presented on the computer screen37; In three studies, participants were walking inside a room to perform the task21,42,56; In two studies (described in one article) participants were walking between two rooms36; In two studies, the location of objects were encoded inside a physical space, but recalled on a computer screen42,43. Finally, in four studies, participants performed the task in the Virtual Reality environment, wearing a VR headset[51](#ref-CR51 “Szychowska, M., Ersson, K. & Olofsson, J. K. Asymmetric cross-sensory interference between Spatial memories of sounds and smells revealed in a virtual reality environment. J. Exp. Psychol. Learn. Mem. Cogn. https://doi.org/10.1037/xlm0001493

(2025).“),52,[53](#ref-CR53 “Schwarz, M. & Hamburger, K. Memory effects of visual and olfactory landmark information in human wayfinding. Cogn. Process. https://doi.org/10.1007/s10339-023-01169-7

(2023).“),54.

Odor presentation summary

We identified five different ways, in which odorants were presented to the participants, (Fig. 3). Thirteen studies (in 12 articles) used some type of odor container, such as glass jars or bottles19,20,37,41,42,43,52,[53](#ref-CR53 “Schwarz, M. & Hamburger, K. Memory effects of visual and olfactory landmark information in human wayfinding. Cogn. Process. https://doi.org/10.1007/s10339-023-01169-7

(2023).“),54,55, tin-cans40, or custom designed SensaCubes21. Eight studies used an olfactometer18,45,46,47,48,49,50,[51](https://www.nature.com/articles/s41598-025-25503-5#ref-CR51 “Szychowska, M., Ersson, K. & Olofsson, J. K. Asymmetric cross-sensory interference between Spatial memories of sounds and smells revealed in a virtual reality environment. J. Exp. Psychol. Learn. Mem. Cogn. https://doi.org/10.1037/xlm0001493

(2025).“), three studies (described in two articles) used paper strips36,39, one study used Sniffin’ Sticks44, and in one study participants smelled odorants diffused in the air56. List of smells used in different studies is available in the supplementary material (see Supplement Table 6 and Supplement Figs. 4–7 in Supplement_1_tables_and_figures.pdf for odors from studies included in the meta-analysis, and “Supplement_SR_olfactory_spatial_memory.xlsx” for details on all studies).

Fig. 3

Summary of the methods used in the studies for three main variables: Memory task (odor-cued navigation or odor-place association) shown as separate columns, Spatio-contextual environment shown as separate blocks of rows, and Smell delivery marked with color coding. Note that two articles reported results from more than one study36,37, and the individual stud