Table 1.

Summary of the included studies. VE: virtual environment; HMD: head-mounted display; VR: virtual reality; OA: older adults; YA: young adults; aMCI: amnestic mild cognitive impairment; AD: Alzheimer’s disease; HNC: high navigation control; LNC: low navigation control; IC: itinerary control; Exp.: experiment; //: same as above; 1A: first trial apartment; 2A: second trial apartment; B: third trial apartment.

Ref.	Sample (s)	Conditions	Design (Navigation)	Virtual Apparatus	Memory Assessment	Primary Outcomes
[62]	48 YA (age range: 21–38; 24 males)	Effect of active (no decisional level) vs. passive (prerecorded travel) vs. snapshot exploration (static condition) on scene recognition and memory of displacements. Intentional encoding.	Within	Non-immersive (CaTS driving simulator); input device: joystick.	Scene recognition (route snapshots); Pointing toward the origin test using the joystick; drawing test (shape of the path).	Path shape task benefitted from active condition, whereas recognition and pointing task were not affected by the exploring conditions.
[41] Exp. 1	30 YA (mean age = 27.1; 14 males)	Effect of active vs. passive (recorded navigation) navigation on spatial memory. No intentional encoding. Participants could freely navigate the apartment.	Between (yoked)	Non-immersive; input device: joystick; house apartment navigation.	Spatial layout test (spatial layout drawing of the VE); recall test (location and objects name on VE map).	Active group showed better spatial layout scores. No effect on recall test.
[41] Exp. 2	40 YA (mean age = 26; 18 males)	Effect of active vs. passive (recorded navigation) navigation on spatial memory. No intentional encoding. Participants could freely navigate the apartment.	//	Non-immersive; input device: joystick; house apartment navigation.	Spatial layout test (spatial layout drawing of the VE); recognition task (objects); object location test.	Spatial layout recall replicated for active condition. No effect on other tasks.
[63]	30 YA (age range = 18–30) and 30 OA (age range = 58–72)	Role of active vs. passive (pre-recorded video) motor exploration in spatial memory and wayfinding. Intentional encoding. No decision-making.	Between	Non-immersive; input device: joystick; virtual replica of Bordeaux.	Wayfinding task (replication of the path; use of spatial representation, errors and stops were calculated); spatial memory task (map drawing + picture classification; route and survey representations).	Active condition worsened survey knowledge (spatial map) in both groups, led to better wayfinding scores in YA and worsened in OA. Executive functions have a crucial role during active navigation.
[64] Exp. 1	22 undergraduates (14 males)	Exploring the role of motion control vs. passive condition (VE tour) on spatial learning. Intentional encoding (learning phase before test phase). Navigation instructions were given.	Between	Non-immersive; input device: keyboard; VE of a research lab.	Spatial learning test (indicate position and direction of egocentric pictures on a lab map; object location test).	View positioning test was better for active participants, no significant difference between the conditions was observed for object location task. Active navigation contributes partially to survey representation.
[64] Exp. 2	80 undergraduates (49 males)	Exploring the role of optical flow (action with object in active condition) vs. passive navigation vs. static condition on spatial learning between active. Intentional encoding (learning phase before test phase). Navigation instructions were given.	//	Non-immersive; input device: keyboard; VE of a research lab.	Spatial learning test (indicate position and direction of egocentric pictures on a lab map; object location test).	Active participants performed better in object locations task and passive condition performed better than static condition. No difference among the conditions was observed for the first task.
[65] Exp. 1	82 university students (age range: 19–33; 43 males)	Effect of intentional vs. incidental encoding vs. active vs. passive navigation (observing the participant navigating) on spatial memory. Auditory route instructions were given (no decision-making).	Between (yoked)	Non-immersive; input devices: keyboard and mouse; virtual city.	Spatial memory test (landmark recognition task, pointing task and path-sketching, route navigation task; respectively, landmark, survey and route knowledge).	Active navigation led to better landmark and route knowledge performances. No effect on survey knowledge. No effect of encoding.
[65] Exp. 2	88 university students (age range: 18–33; 10 males)	Effect of movement (active navigation vs. passive) vs. instruction control (instructing vs. listening) vs. instruction specificity (landmark information vs. layout information) on spatial memory. Navigation instructions were written (no decision-making).	//	Non-immersive; input devices: keyboard and mouse; apartment with rooms.	Spatial memory test (landmark recognition task, tour integration task, route navigation task).	Landmark knowledge, tour integration and route knowledge benefited from self-contained condition. Effect on performance was mediate by instruction specificity and control in the latest task.
[65] Exp. 3	102 students (age range: 19–41; 21 males)	Effect of active vs. passive navigation vs. decision-making (map) vs. less decision-making (map with suggested path) vs. no-map condition on spatial memory. Participants were asked to find the shortest possible route.	//	Non-immersive; input devices: keyboard and mouse; apartment with rooms.	Spatial memory test (landmark recognition task and tour integration task; route navigation task).	Active navigation led to better landmark recognition performance. Decision-making helped participants in observed movement condition and less decision-making worsened route knowledge.
[66]	24 undergraduates (age range: 18–21; 7 males)	Effect of active free navigation (with decision-making) vs. passive on object-memory. Intentional encoding.	Between (yoked)	Non-immersive; input device: keyboard; virtual city.	Object task (locate objects) and recognition task.	No difference between the two conditions in the tasks.
[53] Exp. 1	72 undergraduates (age range: 18–27; 22 males)	Effect of psychological activity (decision-making vs. no decision-making on directions) and physical activity (motor control vs. no motor control on keyboard) vs. control group on spatial performance. No intentional encoding (explore VE).	Between	Non-immersive; input device: keyboard; virtual city.	Orientation task (direction test + map drawing).	No difference was observed between the conditions manipulated.
[53] Exp. 2	36 undergraduates (age range: 18–42)	Effect of active exploration vs. passive observation of navigation vs. control (no exploration of VE) on wayfinding. No intentional encoding (explore VE).	Between	Non-immersive; input device: keyboard; virtual arena.	Wayfinding task.	No difference was observed between the conditions manipulated.
[67]	18 YA (age range: 20–39; 9 males)	Effect of active (with decision-making) vs. passive dynamic (recorded video) vs. passive static (slide-like scenes) free exploration on spatial layout performances. Intentional encoding.	Within	Non-immersive; input device: joystick; virtual arena with cubes.	Target location test (8 trials; score, time, orientation and verbal or drawing description of the strategies used to reach a given target were calculated).	Active participants performed better than the two passive conditions in the task (scores, time, verbal and layout descriptions but no orientation task). Active motor behaviour with active perception is crucial to extract invariants in the VE.
[68]	30 YA (age range = 18–25)and 30 OA (age range = 60–81)	Active (with decision-making) vs. passive (computer-guided tour) free navigation effects on memory for everyday objects. Intentional encoding.	Between	Non-immersive; input devices: keyboard and mouse; VR-based Human Objects Memories from Everyday Scenes (HOMES).	Free recall and recognition (learning, proactive interference, semantic clustering, recognition hits, and false recognitions). VE 1A followed by free recall task + VE 2A followed by free recall task + VE B followed by recognition task.	Active navigation had a beneficial effect on recognition hits only, in both YA and OA compared with passive mode. Active mode reduced false recognitions in YA but increased these in OA. Active navigation enhanced memory in older adults when to not demanding.
[69]	44 students (mean age = 21.94, SD = 2.13; 21 males)	Active (with decision-making) vs. passive (computer-guided tour) free navigation effects on memory for everyday objects. Intentional encoding.	Between	Non-immersive; input devices: keyboard and mouse; VR-based HOMES.	Free recall and recognition (learning, proactive interference, semantic clustering, recognition hits, and false recognitions). VE 1A followed by free recall task + VE 2A followed by free recall task + VE B followed by recognition task.	Active navigation led to better recognition hits performances compared to passive condition. Active participants had less source-based false recognition compared with passive participants. Active navigation was useful to enrich visuomotor details of episodic memory traces but had no effect on semantic relational processing.
[9] Exp. 3	41 participants (age range: 18–34; 12 males)	Active vs. passive (recorded actions; no motor response) selection affects memory for object. Intentional encoding. Instructions.	Between	Non-immersive; input device: keyboard; WWW (what–where–when) variation built with Second Life arena.	Object name cued recall (full episodic recall: what + where + when; non-episodic: where + what or when + what or what only).	Active condition reduced distractor encoding compared to the passive viewing of the action of the avatar.
[43]	72 Psychology students (mean age = 22.23, SD = 3.94; 36 males)	Interaction condition (motor trace in memory, no decision on itinerary) vs. planning condition (no control of the vehicle; decisional level) vs. passive (recorded video). Intentional encoding.	Between	Non-immersive; input devices: steering wheel and pedals; virtual city.	Free recall of elements; visuospatial memory test (draw map + locate elements); visuospatial cued recall (locate elements on a prepared map); recognition test (elements, locations and navigation directions after seeing the elements).	Interaction enhanced memory recall, in particular spatial memory test (no effect on influence on visuospatial cued recall or recognition); however, interaction worsened elements recognition compared with passive condition; planning condition boosted visuospatial recalls. Both interaction and planning had an effect on episodic memory.
[70]	21 healthy OA (4 males), 15 aMCI (7 males) and 15 AD (2 males)	Active vs. passive (recorded video) encoding influences episodic memory. Intentional encoding. Predetermined route.	Between	Non-immersive; input devices: steering wheel and pedals; two virtual cities.	Immediate free recall (what, details, when, egocentric where, allocentric where, binding); recognition (elements, spatial and temporal relations between elements; remember/know paradigm); delayed free recall (same as immediate free recall).	Active exploration led in OA, aMCI, and AD groups to better recall of elements, allocentric spatial information and binding. Procedural skills and self-involvement may be crucial for episodic performances in aMCI and AD patients.
[71]	113 psychology students (mean age = 21.57, SD = 2.99) and 45 OA	Effect of active vs. passive navigation and intentional vs. incidental encoding on episodic memory. Predetermined route.	Between	Non-immersive; input devices: steering wheel and pedals; virtual city.	Free recall (what, verbal where, visuospatial where, when, details); recognition test (elements).	Encoding conditions affect differently episodic features in YA and OA. However, any effect due to sensorimotor implication emerged in the study.
[8]	64 YA and 64 OA (32 males)	HNC (real-life driving conditions) vs. LNC (only pedals; no enactment associated with direction) vs. IC (verbal instructions without driving; decisional level only) vs. passive (no driving no decision) effect on episodic memory performance. Intentional encoding.	Between	Non-immersive; input devices: steering wheel and pedals; virtual city.	Immediate free recall (what and details; binding: what + where + when; remember/know paradigm); visuospatial recall test (what, where, when on real map); delayed free recall test (what, details, where, and when); recognition test (elements).	Binding, regardless of age-groups, was enhanced by LNC and IC; HNC and passive conditions did not help episodic memory performance in both groups. Interestingly, Remember responses were boosted in older adults by IC condition. Active condition may be helpful when do not overload cognitive resources.
[72]	90 students (average age of 20; 45 males)	Passive VE (recorded route) vs. active VE vs. real environment (navigate the environment with instructions) and immediate vs. 48-h recall. Predetermined route.	Between	Non-immersive; input device: joystick; virtual replica of the Bordeaux area.	Immediate or 48-h recall task: real world wayfinding (replication of the real route), freehand sketch (directional changes) and photograph classification (picture in chronological order).	Transfer and sketch task are efficient after 48 h of retention and it is efficient for the two paper-pencil tasks. Active navigation led to benefits in wayfinding task, irrespective of the delay retention.
[73]	59 YA (age range: 19–29; 19 males)	Effect of active vs. passive free exploration on object recognition in VEs.	Between	Immersive; input device: keyboard; virtual rooms with objects.	Recognition task (objects).	Active navigation led to higher hit and lower miss responses than the passive condition. Active navigation has an important role in landmark recognition.
[74] Exp. 1	32 YA (age range: 18–34; 16 males)	Active vs. passive free navigation with four trials with different virtual maze.	Between	Non-immersive with static navigation; input devices: keyboard and mouse; virtual maze.	“Active” test: number of moves and time.	Navigational knowledge is represented regardless the kind of exploration condition.
[58]	20 male students (age range: 20–26)	Active (self-governed) vs. passive (avatar-guided) free exploration. Four exploring sessions. Intentional encoding.	Between	Immersive (HMD); input device: joystick; virtual school.	Wayfinding task (short route to starting point); pointing task (orienting to the starting point); sketch-map (local accuracy or survey-type organization).	Self-governed explorers were better in completing the wayfinding task. Sketch-map accuracy was similar in both groups, whereas self-governed group had better survey-type organization. No differences were shown in pointing task. Self-governed participants organize their knowledge in survey mode.
[75]	34 YA (age range: 18–38; 7 males)	Active vs. passive (passenger condition) condition. Participants were before divided in driver and non-drivers. Intentional encoding and decision-making.	Between	Non-immersive: input devices: steering wheel and pedals; virtual city.	Survey knowledge: pointing error scores (street-level view) and map placement error scores (bird’s eye view); route knowledge: route scores (shortest route).	Driver had better route scores during active navigation compared with drivers in passive and non-driver in active conditions. Drivers showed better map scores (no condition effect). Active navigators do not learn more spatial layout knowledge and dual task effect may affect scores in non-drivers.
[59]	54 students (9 males)	Active vs. passive (passenger condition). Three exposures (3, 10, or 15 times). No intentional encoding. Predetermined path.	Between (yoked)	Non-immersive: input devices: steering wheel and pedals; virtual city.	Survey knowledge (map sketch drawing and map rates’ score), route knowledge (travel directions) and landmark knowledge (landmarks recall).	Passengers recalled more landmarks across exposure conditions. Survey errors reduced between 5 and 15 times in both conditions. Exposure led to better map reliability especially for the passive condition. Attentional resources could have led to worst performance in active drivers.
[76] Exp. 3	41 undergraduates (age range: 18–24; 20 males)	VE active vs. VE passive (watch experimenter navigation) vs. VE + line (active with path to follow; no free exploration) vs. control (no VE training; real -world wayfinding).	Between	Non-immersive. Input device: keyboard; virtual replica of an office.	Real world transfer task (balloons wayfinding times and errors from virtual to real places); training task (wayfinding time and errors in VE conditions).	Times for active condition were lower compared with control condition and active and VE + line led to fewer errors than control condition. Virtual real transfer occurs thanks to virtual interaction.
[60]	64 students (average age of 20; 32 males)	Ground vs. areal point of view and active vs. passive navigation. Predetermined route	Between	Non-immersive; input device: joystick; virtual replica of the Bordeaux area.	Real world wayfinding task (replication of the real route; error scores), sketch-drawing task (directional changes; errors and omissions scores) and scene-sorting task (errors).	Active navigation boosted sketch-mapping task and worsened wayfinding and picture-sorting scores. Grounded-level condition improved performance in wayfinding and picture-sorting tasks, whereas aerial-level in sketch-mapping task. Active navigation and grounded-level interaction had a positive effect in the wayfinding and picture-sorting tasks, whereas passive and aerial-level condition improved sketch-mapping scores. Egocentric information and motor information create a correct perception-action coupling.
[77]	64 students (average age of 20; 32 males)	Detailed vs. undetailed visual fidelity and active vs. passive navigation. Predetermined route.	Between	Non-immersive; input device: joystick; virtual replica of the Bordeaux area.	Real-world wayfinding task (replication of the real route; errors and hesitations scores), sketch-mapping task (directional changes; errors and omissions scores), and scene-sorting task (errors).	Results highlighted better performance for each spatial task in both active and detailed condition. Interaction effect (active and detailed) led to better scores for sketch task and active condition combined with undetailed VE worsened scene-sorting task. Perceptual-motor information is crucial in spatial knowledge. Visual fidelity has positive effect for allocentric representation but not for route knowledge.
[78] Exp. 1 & 2	28 students (age range: 18–23; 9 males)	Active exploration (experiment 1) vs. passive exploration (experiment 2; video of active exploration). Predetermined route of familiar environment.	Between	Non-immersive; input device: keyboard; university building.	Orientation test trials of external cues from four virtual rooms (internal visited and unvisited, external visited and unvisited).	There is no difference in the two conditions.
[78] Exp. 3	54 visitors (mean age = 17.55, SD = 1.14; 19 males)	Active exploration vs. passive exploration. Predetermined route of unfamiliar environment.	Between	Non-immersive; input device: keyboard; university building.	Orientation test trials of external cues from four virtual rooms (internal visited and unvisited, external visited and unvisited).	No effect of unfamiliarity for active participants. Passive participants had greater error for the internal unvisited room. Active exploration enhances survey knowledge for unfamiliar environments.
[79]	60 adults (mean age = 25.2, SD = 4.5; males 49)	Active exploration vs. passive exploration (video of passive exploration) and immersive vs. computer screen. Predetermined rout of Gowanus canal. Intentional encoding. An allocentric map of the canal was provided in all conditions.	Within	Immersive (Emotiv EPOC headset) and non-immersive; input device: mouse and headset gyroscope; Gowanus Canal.	Elements recognition task	No difference between the two navigation conditions. However, active navigation with the mouse has higher level of engagement.
[80] Exp. 1 and 2		3D active exploration (Exp. 1) vs. 2D passive snapshots presentation (Exp. 2). Free exploration and intentional encoding.	Between	Immersive (nVisor SX111) and non-immersive; input device: Wiimote; virtual apartment.	Search trials of geometric and contextual objects.	Search task improved in both conditions but in the immersive condition initial fixations and time spent in the incorrect rooms and better selection of the correct room indicate higher use of memory.
[81]	28 YA (mean age = 25.6, SD = 5.4; 17 males)	Active navigation vs. passive navigation (video). Immediate (intentional encoding) and 24 h delayed (naïve) recall. Participants could freely navigate the environment.	Between	Non-immersive; input device: keyboard and mouse; virtual city.	Immediate and delayed free recall of semantically linked images of 3D objects placed in the town.	No effect of navigation types on spatial memory.
[57]	14 YA (mean age = 22, SD = 2.08; 7 males)	Full condition (full control over the navigation) vs. medium condition (participants move but do not control pre-recorded navigation) vs. low condition (watch pre-recorded navigation).	Within	Immersive (Oculus Rift DK2); input device: Kinect for legs and arms movement detection; virtual city.	Immediate free recall (what, egocentric where, details, when, binding) and item recognition (source memory, remember/know/guess paradigm), egocentric, allocentric and temporal recognition.	Any significance was found among the conditions. However, the full and medium (virtual embodiment) conditions were more immersive than the passive one.
[82]	16 students (females = 16)	Active navigation vs. passive (watching the navigation of the active participant). Free exploration.	Between (yoked)	Non-immersive (46-inch touchscreen monitor); input device: joystick; virtual rooms	Immediate memory recognition for objects manipulated in each room).	Both passive and active navigation had a significant negative effect on memory of object, with active navigation having a greater effect compared to passive.
[61]	22 YA (mean age = 19.71, SD = 2.19; females = 11) and 22 OA (mean age = 74.55, SD = 7.82; females = 10)	Active navigation vs. passive. Free exploration.	Within (yoked)	Immersive (cardboard) mobile application (input device: button headset and head movements); VE (city, park, mall)	Encoding-Retrieval route overlap accuracy	Active encoding leads to better spatial memory in OA; accuracy is predicted by age, active exploration and visuospatial abilities.