Memory performance as a function of modality and context: revealing the cognitive mechanisms underlying enhanced memory in virtual reality

ORIGINAL ARTICLE Virtual Reality (2025) 29:128 https://doi.org/10.1007/s10055-025-01205-1 laboratory, VR environments are highly immersive and realistic, hence eliciting more lifelike mnemonic processing (Kisker et al. 2021b; Schöne et al. 2019). Interestingly, several studies provide evidence for a memory superiority effect in VR, i.e., enhanced memory performance for VR experiences as opposed to two-dimensional presentations commonly used in conventional laboratory setups. This effect has been found for working memory performance (Gabana et al. 2017), memory recall performance (Krokos et al. 2019), and recognition memory performance (Schöne et al. 2019). Previous research attributed the enhanced memory performance to the involvement of the autobiographical memory system (Schöne et al. 2019) characterized by personal relevance and the integration of newly made experiences into the narrative of an individual’s life experience (Conway 2005; Schöne et al. 2018). The vivid 1 Introduction The application of Virtual Reality (VR) as a psychological research method to investigate real-life cognitive processing has led to a re-evaluation of mnemonic processes previously investigated in the conventional laboratory (e.g., Corriveau Lecavalier et al. 2020; Kisker et al. 2021c; Rodríguez et al. 2022). In contrast to experiences in the conventional Marike Johnsdorf [email protected] 1 Experimental Psychology I, Institute of Psychology, Osnabrück University, Osnabrück, Germany 2 Department of Psychology, Norwegian University of Science and Technology, Trondheim, Norway Abstract Although recent psychological research provides evidence for enhanced memory retrieval of Virtual Reality (VR) experiences compared to conventional laboratory experiences, the functional characteristics of the underlying mnemonic processes remain unclear. Initial studies suggest that encoding and retrieval involve distinct cognitive processes, yet the possible technological artifacts induced by VR head-mounted displays (VR-HMDs) have not been accounted for. To determine the factors influencing modality-specific memory processing, 122 participants performed an old/new recognition memory task wearing a VR-HMD. Stimuli were first either presented two-dimensionally on a screen within a VR environment (mediated VR; mVR) or experienced immediately in a three-dimensional immersive VR environment (iVR). In the subsequent old/new recognition memory task, participants had to recall the stimuli either in the same or the other modality, providing a cross-modality comparison. As a result, the canonical context effect, i.e., an advantage for memory performance in congruent contexts, was found for both mVR and iVR modalities. Most importantly, the study further provides evidence for a memory superiority effect following iVR experiences, as an equally high retrieval success was observed even when the subsequent recall was tested in mVR. In conclusion, the improved recall observed suggests that immersive environments play a crucial role in enhancing the encoding process, whereas technological artifacts, particularly contextual factors of the VR-HMD, can be ruled out. Furthermore, the transferability of information from two-dimensional to threedimensional environments seems limited. Our study thus advocates for integrating immersive VR into learning contexts to extend the applicability of learned content to various settings. Keywords Virtual reality · Implicit learning · Memory · Encoding · Learning context Received: 27 June 2024 / Accepted: 15 July 2025 / Published online: 7 August 2025 © The Author(s) 2025 Memory performance as a function of modality and context: revealing the cognitive mechanisms underlying enhanced memory in virtual reality MarikeJohnsdorf1· JoannaKisker1· MerleSagehorn1· ThomasGruber1· BenjaminSchöne1,2 1 3 Virtual Reality (2025) 29:128 and experiential nature of VR closely mirrors real-life interactions, thereby embedding memories more deeply within the autobiographical memory framework (see Schöne et al. 2019). Additionally, information which is encoded in a VR environment and accordingly stored within autobiographical memory traces, is more effortlessly retrievable than memories formed under conventional laboratory conditions, regardless of memory performance (Kisker et al. 2021c). Recent electrophysiological findings suggest that these qualitatively distinct mnemonic processes in VR environments (Kisker et al. 2021c) already occur during early processing stages as evidenced by extensive encoding processes involving multimodal neural integration (Johnsdorf et al. 2023a, b). Therefore, VR provides a unique approach to investigate mnemonic processes involving autobiographical memory under more realistic conditions. However, several factors potentially contributing to the memory superiority effect in VR remain to be evaluated. In addition to the aforementioned differences observed between conventional laboratory and virtual conditions, and contradictory to the memory superiority effect in VR, several studies actually indicate equal, or poorer memory performance in VR in comparison to conventional 2D experiences depending on the retrieval task (e.g., Cadet and Chainay 2020; Johnsdorf et al. 2023a, b; Kisker et al. 2021c; Monaro et al. 2024; Schöne et al. 2023). In particular, the free recall of stimuli encoded in a virtual environment either presented in VR or on a conventional 2D screen did not result in differences in memory performance (Cadet and Chainay 2020). In a further study, participants performed poorer in an application task of previously learned rules after learning via a 3D/360° video compared to a 2D video. Moreover, no differences in memory performance were obtained between both groups in either a multiple-choice test or in a free recall test in the same study (Johnsdorf et al. 2023a, b). Another study that used a recognition memory task involving a remember/know paradigm (Tulving 1985; Gardiner and Java 1990; Rajaram 1993; see Kisker et al. 2021b) found that exploring a virtual village either in VR or via a 2D screen did not result in different memory performance (Kisker et al. 2021b). Comparing memory performance for objects encoded in a virtual room or via 2D pictures of the room revealed no differences between the two conditions in a free recall task, a visual recognition task, and a suggestibility task with the exception of a nonsuggestive verbal task in which memory performance was better in VR (Monaro et al. 2024). A fifth study compared memory performance for videos of different scenes watched either in VR or in 2D on a monitor. Although the VR group performed better in a free recall task, the two groups did not show differences in memory performance in a cued recall task (Schöne et al. 2023). Overall, the varying memory tasks do not necessarily provide the same conclusions about the differences in memory performance between VR and 2D, although no stringent pattern emerges regarding memory performance as a function of the retrieval task. Importantly, it remains unclear whether the varying modality-specific mnemonic processing characteristics are attributable to the actual exposure to three-dimensional immersive experiences in VR, or whether they are only influenced by possible artifacts induced by wearing a VR head-mounted display (VR-HMD). Since most studies assessed memory performance after conducting the encoding phase in VR via an HMD and in 2D via a conventional monitor (Johnsdorf et al. 2023a, b; Kisker et al. 2021b, 2021c; Schöne et al. 2019), the differing outcomes in mnemonic processing might arise from the use of the VR-HMD in only one of the compared conditions, creating the false impression that memory processes in VR differ markedly from those under 2D conditions. On the one hand, the novelty of the technology and the associated high cognitive load can have a negative impact on learning (Miguel-Alonso et al. 2023; Wu et al. 2013). On the other hand, the thrill or technological awe of wearing a VR-HMD can also improve motivation and subjective usability (Koch et al. 2018) as well as memory performance by increasing attention and effort (Huang et al. 2021). Therefore, the novelty of wearing a VR-HMD might be a contributing factor to the memory superiority effect in VR, and particularly to the diverging results in mnemonic research comparing immersive to conventional conditions. Interestingly, apart from mnemonic processing, motivational and emotional reactions still exhibit different functional characteristics on a neural level when controlling for a potential influence of the VR-HMD by presenting both 3D/360° stimuli and 2D stimuli via an HMD (Schöne et al. 2023). These findings indicate that the three-dimensionality of stimuli is a relevant factor influencing affective reactions in VR. Further properties intrinsic to VR might account for the heterogeneity in memory performance outcomes. The discrepancies might be attributable to methodological differences across studies, and thus result from mere technological artifacts like the technical features of the VR systems at hand along with varying degrees of interactivity and presence afforded by the VR environments (for review see Smith 2019). These specific technological properties of a VR system generate different levels of immersion, i.e., the capability of a VR technology to represent a realistic environment (Slater and Wilbur 1997). Generally, a higher degree of immersion increases the sense of presence (Diemer et al. 2015), i.e., the sense of actually being in an interactive and responsive environment (Sanchez-Vives and Slater 2005). Although a direct influence of presence on memory performance could not always be observed (e.g., 1 3 128 Page 2 of 12 Virtual Reality (2025) 29:128 Smith and Mulligan 2021), other studies have suggested presence to be one of the most profound features of VR that has been shown to be positively related to memory performance (Cadet et al. 2022; Makowski et al. 2017). Taking into account the potential technological artifacts resulting from the use of the VR-HMD, this study aimed to unravel modality-specific cognitive mechanisms in both immersive VR and conventional 2D settings reflected in enhanced retrieval success. The use of a VR-HMD in all conditions ensures that differences in memory performance can be attributed to specific functional characteristics of mnemonic processing rather than mere technological features. Memory performance was assessed by translating a conventional old/new recognition paradigm to both a mediated VR environment (mVR), presenting 2D stimuli on virtual monitor within VR, and an immersive VR environment (iVR), presenting 3D/360° stimuli in VR. The subsequent old/new recognition memory task was either performed in the same or the other modality, providing a cross-modality comparison. Classical findings from previous memory research showed that memory performance benefits from context effects, i.e., from a congruent environment during encoding and retrieval (Godden and Baddeley 1975). Since the stimuli in the mVR environment were presented twodimensionally, i.e., as in the conventional laboratory, we expected to observe high memory performance scores when participants encoded and recalled the stimuli in the same mVR environment. The classical context effect could be replicated within different VR environments (Shin et al. 2021), hence we expected similarly high memory performance scores for congruent iVR contexts during encoding and retrieval. Assuming that the memory superiority effect in VR observed in previous research was only caused by technological artifacts of the VR-HMD, only context effects should influence memory performance in the present study, and high memory performance scores should only occur when modalities are congruent during encoding and retrieval. However, given the evidence for a memory superiority effect in VR when experiences become part of an autobiographical memory network (Schöne et al. 2019), we also expected high memory performance scores for participants encoding in iVR and recalling in mVR. This result would indicate that immersive VR does not only serve as a mere contextual factor but that it enhances encoding processes beyond wearing a VR-HMD. Lastly, this effect was not expected for participants encoding in mVR and recalling in iVR as memory performance should be reduced by the change in context, without the beneficial effects of the three-dimensional presentation of the stimulus material during encoding. Additionally, due to the inconsistent findings regarding the impact of presence on memory performance (see Cadet et al. 2022; Makowski et al. 2017; Smith and Mulligan 2021), we assessed presence both after encoding and retrieval to further investigate its relation to memory performance when controlling for the influence of the VR-HMD. Since a high level of immersion increases the sense of presence (for review see Diemer et al. 2015), we expected to observe a higher sense of presence in the iVR modality compared to the mVR modality both during encoding and retrieval. 2 Methods 2.1 Participants 139 participants were recruited via the University’s students e-mail list, the University’s online bulletin board and by students of the bachelor’s degree program in psychology. Participants were screened for psychological and neurological disorders, substance abuse and symptoms of a SARS-CoV-2 infection. All had normal or corrected to normal vision. Eight participants had to be excluded due to unfulfilled inclusion criteria, three data acquisitions had to be terminated because of technical issues, two participants prematurely terminated the experiment and four participants did not show up for the appointment. Therefore, data of 122 participants were included in the analyses (29 male, 93 female, MAge = 22.32, SDAge = 3.610; nmediated = 30, ncross−up = 32, ncross−down = 29, nimmersive = 31). Participants received partial course credits or 15€ for their participation. The study was conducted in accordance with the hygiene guidelines of the University for minimizing the spread of SARS-CoV-2 during the COVID pandemic. The study was approved by the local ethics committee and all participants gave informed written consent. 2.2 Stimuli & experimental conditions 120 3D/360° videos filmed with a resolution of 4 K at standard eye level were selected from the library for universal Virtual Reality experiments (luVRe; Schöne et al. 2023). The videos showed a variety of scenes, including for example different locations, animals or human interactions (for a comprehensive list of the used scenes from the database see Supplementary Material S1). All scenes were recorded with sound. Each video was cut to ten seconds and was recoded to 2D for the presentation on the virtual monitor within a three-dimensional VR environment in the mVR modality. During the encoding session, the videos were either presented in mVR on in iVR, and the modality in the retrieval 1 3 Page 3 of 12 128 Virtual Reality (2025) 29:128 2.3 Procedure The study had to be partly adapted to the restrictions due to the COVID pandemic. Therefore, the anamneses were initially conducted online prior to the experiment to minimize physical contact, whereas they were scheduled directly before the experiment at a later stage of the data acquisition. The experiment was conducted on two consecutive days. On the first day, participants were randomly assigned to one of the four conditions and were instructed to put on and adjust the VR-HMD correctly. They were equipped with two controllers and headphones that were used to provide the previously recorded instruction. The encoding session was conducted either in mVR or in iVR (see Fig. 1a). Participants in the mVR modality were located in front of a 2D monitor on a table in a virtual three-dimensional laboratory, while participants in the iVR condition were located in a neutral grey room. All participants were instructed that they would be presented with videos of various scenes, which they were asked to watch carefully. Afterwards, they could session was either consistent with the encoding session (congruent context) or was carried out in the alternative modality (incongruent context). In the mVR modality, the videos were presented in 2D on the virtual monitor, while they were presented in 3D/360° in the iVR modality, i.e., the participants were surrounded by the videos and watched them from a first-person perspective (see Fig. 1). This design resulted in four conditions providing a cross-modality comparison: Encoding in mVR and retrieval in mVR (mediated), encoding in mVR and retrieval in iVR (cross-up), encoding in iVR and retrieval in mVR (cross-down), and encoding in iVR and retrieval in iVR (immersive). Both the mVR and iVR environments were built using Unity (version 2020, Unity Technologies, San Francisco, United States). They were presented using a standalone HTC Vive Focus Plus HMD (with a resolution of 1440 × 1600 pixels per eye). Additionally, participants were equipped with two Vive Focus Plus controllers and Sony over-ear headphones providing stereoscopic audio. Fig.1 Setup of the encoding phase (a) and the retrieval phase (b) in mediated VR (mVR) and immersive VR (iVR). The cross-modality comparison resulted in four conditions: Encoding in mVR and retrieval in mVR (mediated), encoding in mVR and retrieval in iVR (cross-up), encoding in iVR and retrieval in mVR (cross-down), and encoding in iVR and retrieval in iVR (immersive) 1 3 128 Page 4 of 12 Virtual Reality (2025) 29:128 separately for each symptom (headache, nausea, and dizziness) and each experimental phase (encoding, retrieval). 2.4.2 Memory performance The sensitivity index d-Prime (d’) was used to investigate memory performance in each condition during the retrieval phase. It is calculated by relating the hits, i.e., the correctly recognized old stimuli to the false positives, i.e., the new stimuli incorrectly rated as old (d’ = zHit–zFalsePositive; Haatveit et al. 2010; Swets et al. 1961). To compare memory performance between conditions during the retrieval session on the second day of the experiment, a 2 × 2 ANOVA with the factor Encoding Modality (mVR, iVR) and Retrieval Modality (mVR, iVR) resulting in the condition combinations (mVR + mVR = mediated, mVR + iVR = cross-up, iVR + mVR = cross-down, iVR + iVR = immersive) was performed followed by post-hoc t-tests for independent samples with a Bonferroni-corrected significance level. SPSS (Version 28) was used for statistical analyses. To further clarify whether non-significant group differences regarding memory performance (d’) would indicate the absence of a meaningful effect, the two one-sided t-tests procedure (TOST) was performed using the TOSTER package (Caldwell 2022; Lakens 2017) in RStudio (Version 2023.3.1.446). In comparison to non-significant results of conventional t-tests, this equivalence test allows for statistically rejecting the existence of effects that are large enough to be considered meaningful, and therefore to assume the absence of an effect. Two statistically significant results in the TOST procedure suggest statistical equivalence between the compared groups. In this case, the test with the smaller test statistic is reported. The upper and lower equivalence bounds that need to be specified for the TOST are based on the smallest effect size of interest (SESOI). The SESOI used in our analyses was determined based on previous research regarding memory processing in VR. Several VR studies have reported d’ as an indicator for memory performance in different paradigms (Kisker et al. 2021b, c; Schöne et al. 2019, 2023). The smallest effect size of a significant effect regarding memory performance documented in these studies was Cohen’s d = 0.5 (Schöne et al. 2023) which corresponds to a medium effect (Cohen 1988). Therefore, based on the smallest effect size observed to date in similar studies, the upper and lower bounds were set to Cohen’s d = 0.5 and Cohen’s d = − 0.5. 2.4.3 Presence All presence data were analyzed using SPSS (Version 28). To analyze presence, mean values of all IPQ subscales on both start the experiment by pressing a button in the virtual environment. During the encoding phase, a total of 60 videos were randomly selected from all 120 videos and presented in random order. Therefore, each of the 120 videos could be randomly presented in all four conditions, with no video being presented repeatedly. After the encoding phase, participants filled in the German version of the iGroup Presence Questionnaire (IPQ; Schubert et al. 2001). The IPQ assesses the sense of presence and consists of the subscales General Presence (one item), Spatial Presence (five items), Involvement (four items) and Realness (four items). A Likert-Scale from one to seven was used for the IPQ ratings. Additionally, participants were asked for symptoms of motion sickness, i.e., headache, nausea, and dizziness. During the retrieval phase, participants had to perform an unannounced old/new recognition task exactly 24 h after the encoding phase. Again, they were instructed to put on and adjust the VR-HMD correctly. Depending on the condition they were assigned to, participants experienced the retrieval phase in the mVR or the iVR modality (see Fig. 1b). In both modalities, the previously recorded instruction was presented through the headphones. During the old/ new recognition task, the participants were presented with all 120 videos in random order, resulting in 60 old and 60 new videos. Every video was played at least for one second and a maximum of ten seconds, and none of the videos was repeatedly presented. The participants’ task was to indicate whether or not they remembered the scene from the encoding session by pressing the corresponding button as fast as possible (see Fig. 1b). Immediately after the button press, the current video was skipped and the next video was presented, resulting in varying presentation times of the respective videos during retrieval. When participants did not answer within ten seconds, the video was paused until they gave an answer. Following the old/new recognition task, participants again had to fill in the German version of the IPQ and were asked for symptoms of motion sickness, i.e., headache, nausea, and dizziness. 2.4 Data analysis 2.4.1 Differences between conditions To account for potential heterogeneity between participants in the four experimental conditions, we analyzed whether age and gender differed between conditions. Age was investigated by means of a one-way ANOVA with the factor Condition (mediated, cross-up, cross-down, immersive), whereas gender was analyzed using a χ2-test. Additionally, potential differences in symptoms of motion sickness between conditions were investigated by means of χ2-tests 1 3 Page 5 of 12 128 Virtual Reality (2025) 29:128 conditions (χ2(3) = 0.32, p =.955, V = 0.052). Additionally, the frequency of symptoms of motion sickness did not differ between conditions (all χ2s ≤ 2.01, all ps ≥ 0.571, see Supplementary Material Table S1), except for the headache symptom during retrieval (χ2(3) = 9.35, p =.025, V = 0.277, see Supplementary Material Table S2 for the frequencies of the motion sickness symptoms in each condition). 3.2 Memory performance For memory performance, a significant main effect of the factor Encoding Modality (FEncoding Modality(1, 118) = 15.85, p <.001; η2 = 0.118), a significant main effect of the factor Retrieval Modality (FRetrieval Modality(1, 118) = 7.21, p =.008; η2 = 0.058), and a significant interaction of both factors (FEncoding Modality × Retrieval Modality(1, 118) = 24.35, p <.001; η2 = 0.171) were obtained. Post-hoc t-tests further specifying the interaction effect revealed better memory performance for participants in conditions mediated, immersive and cross-down in comparison to condition cross-up (all ts ≥|4.50|, all ps < 0.001, see Table 1; Fig. 2). Additionally, the equivalence test (TOST) revealed significant results for all non-significant group differences regarding memory performance (d’) reported in Table 1 (all ts ≥|3.28|, all ps < 0.001, see Table 2), indicating that memory performance can be considered equal comparing conditions mediated, cross-down and immersive. 3.3 Presence The independent t-tests comparing presence between the mVR and iVR modalities on both days suggest a higher sense of presence in iVR compared to mVR indicated by the IPQ subscales. Except for Spatial Presence in the encoding phase days were compared between the mVR and the iVR modalities using independent samples t-tests. To assess a potential modification of the sense of presence from the first measurement to the second measurement as a function of the encoding modality, each IPQ subscale was analyzed using a mixed ANOVA with the within-subjects factor Timepoint (Day 1, Day 2) and the between-subjects factor Condition (mediated, cross-up, cross-down, immersive). Significant results were further analyzed by corresponding independent samples t-tests for IPQ ratings on the second day of the experiment, with a Bonferroni-corrected significance level. To investigate whether memory performance in the four conditions can be predicted by the level of presence during encoding or retrieval, eight multiple linear regressions were performed using all IPQ subscales on day one or on day two as predictor variables and the d’ scores separately by condition as dependent variables. The following regression equation was modelled for each of the four experimental conditions for day one and day two, respectively: d′=a+b General Presence × x General Presence +bSpatial Presence ×x Spatial Presence +bInvolvement ×xInvolvement +bRealness ×xRealness. 3 Results 3.1 Differences between conditions No differences regarding the participants’ age were observed between the four conditions (FCondition(3, 118) = 0.19, p =.905; η2 = 0.005). Similarly, the χ2-test did not reveal any differences regarding gender balance between the Table 1 Test statistics for post-hoc independent samples t-tests regarding memory performance (d’) between all conditions (mediated, cross-up, cross-down and immersive). A Bonferroni-corrected significance level was applied Group statistics Independent samples test M SD t df pCohens da Mediated 1.54 0.34 5.48 60 < 0.001 1.39 Cross-up 0.99 0.44 Mediated 1.54 0.34 0.72 57 0.475 0.19 Cross-down 1.47 0.39 Mediated 1.54 0.34 –0.97 59 0.338 − 0.25 Immersive 1.63 0.41 Cross-up 0.99 0.44 − 4.50 59 < 0.001 − 1.15 Cross-down 1.47 0.39 Cross-up 0.99 0.44 − 6.01 61 < 0.001 − 1.51 Immersive 1.63 0.41 Cross-down 1.47 0.39 − 1.57 58 0.123 − 0.40 Immersive 1.63 0.41 aCohen’s d uses the pooled standard deviation. 1 3 128 Page 6 of 12 Virtual Reality (2025) 29:128 Post hoc t-tests revealed that the sense of General Presence and Spatial Presence during the retrieval phase were higher in the cross-up condition compared to the mediated, crossdown and immersive conditions (all ts ≥|3.06|, all ps < 0.001, see Supplementary Material Table S3). Additionally, the sense of Realness during the retrieval phase was higher in the cross-up condition in comparison to the cross-down condition (t(59) = 3.68, p <.001, for all post-hoc comparisons see Supplementary Material Table S3). The multiple linear regressions determining whether the d’ scores can be predicted by the level of presence separately for both days in all four conditions yielded no significant results (all ps ≥ 0.214, see Supplementary Material Tables S4 and S5). 4 Discussion The aim of the study was to clarify whether experiences in Virtual Reality (VR) involve modality-specific mnemonic processes that are distinct from those typically observed under 2D laboratory conditions, when meticulously (t(120) = − 1.73, p =.087), all further IPQ ratings yielded significant results (all ts ≥|2.07|, all ps ≤ 0.041; see Table 3). The mixed ANOVA revealed a significant interaction between the factors Timepoint and Condition for all subscales of the IPQ (General Presence: FTimepoint*Condition(3, 118) = 28.55, p < 0.001; η2 = 0.421; Spatial Presence: FTimepoint*Condition(3,118) = 14.43, p < 0.001; η2 = 0.268; Involvement: FTimepoint*Condition(3,118) = 6.28, p < 0.001; η2 = 0.138; Realness: FTimepoint*Condition(3,118) = 28.82, p < 0.001; η2 = 0.423). Table 2 Test statistics for Welch two sample t-test (TOST) for nonsignificant comparisons of memory performance (d’) between the conditions Group Statistics TOST M SD t df p Mediated 1.54 0.34 − 4.5 55.44 < 0.001 Cross-down 1.47 0.39 Mediated 1.54 0.34 4.23 57.75 < 0.001 Immersive 1.63 0.41 Cross-down 1.47 0.39 3.28 57.98 < 0.001 Immersive 1.63 0.41 Fig.2 Comparison of memory performance (d’) between all four conditions (encoding in mVR and retrieval in mVR (mediated), encoding in mVR and retrieval in iVR (cross-up), encoding in iVR and retrieval in mVR (cross-down), and encoding in iVR and retrieval in iVR (immersive)). Memory performance (d’) was significantly enhanced for participants in conditions mediated, immersive and cross-down in comparison to condition cross-up (all ps <.001). Standard errors are depicted by error bars. 1 3 Page 7 of 12 128 Virtual Reality (2025) 29:128 markedly distinct from those predominantly studied in laboratory settings. The enhancement in memory performance observed can therefore be attributed to the unique functional characteristics of mnemonic processes in immersive VR environments. From a practical standpoint, our results suggest that for tasks intended to be carried out in three-dimensional space, learning outcomes are significantly enhanced when training or learning occurs in an identical or similar threedimensional environment. This underscores the limited transferability of learned content from two-dimensional to three-dimensional contexts and highlights the unique cognitive processes involved in learning in immersive environments. The results are in line with previous VR studies showing that in contrast to conventional 2D paradigms, different mechanisms are employed to encode (Bréchet et al. 2019; Iriye and St. Jacques 2021; Johnsdorf et al. 2023a, b; Kisker et al. 2024; Kourtesis et al. 2021) and to retrieve content in immersive environments (Reggente et al. 2018). In comparison to the encoding of 2D stimuli, a broader cortical network associated with multimodal integration is involved during the encoding of complex virtual stimuli (Johnsdorf et al. 2023a, b). Moreover, the retrieval of immersive VR experiences in a classic old/new EEG paradigm revealed that even when the retrieval modality follows the conventional 2D presentation, the mnemonic mechanisms differ to the extent that the usually characteristic difference in the theta band response to old and new stimuli was not observed after VR-based encoding (Kisker et al. 2021c). This is particularly remarkable as the theta-based cortico-hippocampal controlling for the effects of the VR head-mounted display (VR-HMD) on memory performance. Drawing on previous research, we hypothesized that these unique mnemonic processes would be manifested through enhanced retrieval success. Accordingly, we compared the retrieval success of the same stimulus material incidentally encoded in immersive VR (iVR) or in mediated VR (mVR), using a VR-HMD in all conditions. This design allowed us to determine whether VR merely serves as a context for encoding or retrieval, or indeed facilitates a unique form of mnemonic processing. Generally, the canonical context effect, i.e., an advantage for memory performance derived from the congruency of encoding and retrieval contexts (Godden and Baddeley 1975), was replicated for both the mediated and the immersive condition as participants exhibited equally high memory performance. Yet most importantly, we found that experiences recalled in mVR resulted in the same retrieval success when encoded in iVR as those encoded in mVR. In other words, encoding in iVR not only enhances mnemonic processing but also achieves recall success at the level of context effects for stimuli encoded and recalled in mVR. To our best knowledge, this is the first study to rigorously control for the effects of the HMD and the general impact of being in VR. Our findings provide solid evidence that the enhanced memory performance often observed in VR settings is not merely a technological artifact that can be attributed to the VR setting itself. Most importantly, the data reveals that enhanced memory retrieval of iVR experiences stems from the encoding stage in iVR and does not solely rely on the retrieval stage. Thus, experiencing events in iVR facilitates a form of mnemonic processing that is Table 3 Test statistics for independent samples t-tests regarding presence (all IPQ subscales) in the modalities mVR and iVR on both days Group Statistics Independent Samples t-Test M SD t df pCohens da Day 1 General Presence mVR 4.58 1.56 − 2.28 120 0.025 − 0.41 General Presence iVR 5.22 1.52 Spatial Presence mVR 4.69 1.08 − 1.73 120 0.087 − 0.31 Spatial Presence iVR 5.02 0.98 Involvement mVR 4.27 1.40 − 3.09 120 0.002 − 0.56 Involvement iVR 5.01 1.23 Realness mVR 3.41 0.97 − 4.02 120 < 0.001 − 0.73 Realness iVR 4.10 0.95 Day 2 General Presence mVR 3.97 1.59 − 5.10 120 < 0.001 − 0.93 General Presence iVR 5.37 1.44 Spatial Presence mVR 4.32 1.20 − 4.33 120 < 0.001 − 0.78 Spatial Presence iVR 5.20 1.05 Involvement mVR 4.41 1.57 − 2.07 120 0.041 − 0.37 Involvement iVR 4.95 1.32 Realness mVR 3.33 1.41 − 3.50 120 < 0.001 − 0.63 Realness iVR 4.02 1.05 aCohen’s d uses the pooled standard deviation. 1 3 128 Page 8 of 12 Virtual Reality (2025) 29:128 lifelike autobiographical memory processes. Our knowledge about said processes could furthermore be extended by combining this unique approach with electrophysiological methods, e.g., electroencephalography, or neuroimaging methods, e.g., functional magnetic resonance imaging. Utilizing VR to create controlled lifelike autobiographical scenarios therefore provides the potential to investigate mnemonic processing not only on the behavioral but also on the neural level. Yet developing virtual environments for psychological experiments requires the appropriate technological and methodological expertise. In designing a study involving VR, it is crucial to carefully consider both technological aspects and characteristics of the virtual environment, as these factors might influence the nature of the mnemonic processes exhibited (for review see Smith 2019). Whereas technological advancements might reduce these restraints by minimizing the effort to create appropriate virtual environments in the future, it is important not to neglect ethical considerations regarding VR research. Since VR can elicit lifelike emotional, physiological, and behavioral reactions (e.g Kisker et al. 2021a; Kisker et al. 2021a, b, c, d), researchers should be aware of the potential detrimental consequences of experiencing realistic virtual scenarios, in particular those of a strong negative valence. Therefore, when developing VR research, the potential cognitive and behavioral consequences of VR experiences (Madary and Metzinger 2016) should be addressed in the light of ethical considerations. Although we maintained the artificial serial presentation of the virtual scenes, our study supports the idea that iVR environments enhance the sense of presence, with this effect being partially more pronounced when a less immersive environment has previously been encountered. While the relation between an increased sense of presence and memory performance has been repeatedly discussed in previous research (for review see Smith 2019), a direct statistical relationship between presence and memory performance remains elusive in our results. However, the absence of this relation in our data is consistent with previous findings showing that while different manipulations of immersion in VR environments have varying effects on memory and presence, there is no definitive evidence that the subjective sense of presence mediates the impact of immersion on episodic memory performance (Smith and Mulligan 2021). Potentially, presence questionnaires like the IPQ (Schubert et al. 2001) or the ITC-Sense of Presence Inventory (Lessiter et al. 2001) do not fully cover the critical aspect of presence that indeed facilitates the formation of autobiographical memory. Although significant positive correlations between different presence questionnaires could be observed (Kober and Neuper 2013), the influence of various factors can cause inconsistencies in results regarding presence assessed with those questionnaires (e.g., Graf and interplay is hypothesized to underlie many different types of memory, with episodic memory being one of the most thoroughly studied ones (Tulving 2002). Most studies investigating episodic memory assume that the same cognitive processes are elicited by presenting participants with images on a computer screen during encoding or retrieval that occur under real-life conditions. However, previous studies have shown minimal overlap in brain region activation between laboratory-based and reallife memory experiments (Monge et al. 2018). This discrepancy is understandable, as laboratory tasks often present artificial events lacking personal meaning, whereas real-life events are multimodal and significant, occurring in physical proximity. Laboratory events usually only affect the visual domain, serving as reminders of previously experienced events (Marcotti and Jacques 2018). Studying these laboratory events has significantly advanced our understanding of mnemonic processes and brain functioning. The paradigm of short serial stimuli presentation is central to psychological science, especially with psychophysiological measurements that rely on repeated presentation of stimuli. Unlike real-life experiments, standard paradigms offer unparalleled experimental control and reproducibility, which are fundamental to psychological science. However, we should also recognize this conventional approach’s limitations. Stimuli often serve merely as reminders, which can significantly obscure the true nature of processing them (Snow and Culham 2021), thereby severely decreasing our ability to generalize from laboratory to real-life settings. However, that does not imply that we cannot generalize to certain real-life conditions but that we should be more conscious of what our paradigms are investigating, and exercise caution in over-generalizing. For example, watching pictures or videos on a computer screen is a significant part of our lives as we constantly engage with monitors. Yet, it is crucial for research to acknowledge that it is specifically this aspect of our daily lives that is being investigated with monitor-based paradigms. Towards a more immersive approach in psychological science enhancing ecological validity, we created a paradigm that bridges the gap between classic or conventional paradigms and real-life experiences. On the one hand, we maintained the serial presentation of stimuli, on the other hand we used photorealistic, i.e., immersive stimuli to create a realistic environment. The maintained serial presentation of the virtual environments mostly contradicts the probabilistic distribution of our daily life experiences as scenes do not change that drastically within a fraction of a second. Nevertheless, this combination of conventional and immersive aspects of the paradigm seems to achieve an optimal balance where experimental control is still high but the mnemonic processes investigated differ and resemble 1 3 Page 9 of 12 128