Curiosity shapes spatial exploration and cognitive map formation in humans

Active exploration of novel environments is key to building hippocampus-dependent cognitive maps, yet the motivational drivers of exploration absent external reinforcers are less understood. Theories posit curiosity—the desire to seek novel information—as a primary driver of exploration and cognitive map construction. While human curiosity research has largely focused on semantic learning (epistemic curiosity) and perceptual curiosity, the impact of curiosity on spatial exploratory behaviours and spatial memory in humans has not been directly tested. The study investigates whether pre-room curiosity states stimulate spatial exploration in novel virtual environments and whether this exploration enhances the fidelity of cognitive maps, also contrasting effects of pre-information curiosity with post-exploration interest.

Prior work in animals and humans highlights the hippocampus’s role in navigation and cognitive map formation and suggests exploration as a behavioural manifestation of curiosity. Human curiosity research distinguishes epistemic curiosity (knowledge-driven) and perceptual curiosity (stimulus-driven), both linked to exploratory behaviours and memory enhancement chiefly for semantic information. Recent studies show curiosity influences information-seeking (e.g., online searches), anticipatory gaze, and memory via dopaminergic circuits interacting with the hippocampus. However, spatial exploratory behaviours common to motile species and their relationship with spatial memory have been less examined in humans, motivating the current work to bridge curiosity theories with spatial cognition.

Two experiments employed a desktop-based virtual reality task built in Unity 3D (v2019.4.15). Participants from Cardiff University explored 16 novel rooms (with two practice rooms), each 16 m × 16 m, designed with layout-defining objects and occlusions to necessitate movement for full mapping. An outdoor pier with a zigzag pathway led to rooms; trial order was randomized. Apparatus: LCD monitor (1920×1080, 60 Hz), keyboard for forward/back (W/S) and interaction (E to enter, B to exit), mouse for head direction; walking speed fixed at ~3.4 m/s; sideward movement restricted; footsteps via headphones. Procedure: At the pier, participants viewed the room label (e.g., Lounge, Library) and rated pre-room curiosity on a 1–10 Likert scale. After entering and freely exploring without time limits, position (location) and head-direction (field-of-view angles) were recorded at 60 Hz. Upon exit, participants rated post-room interest (1–10). Experiment 2 additionally administered the Five-Dimensional Curiosity Revised (5DCR) scale before familiarisation and included an immediate sketch map memory test after a 5-min break. Sketch map test: for each room, participants sketched layouts on paper with room boundaries marked, instructed to include key spatial elements (furniture, doors, windows). A scoring system (1–5 in 0.5 increments) assessed four dimensions: Object Presence, Spatial Distortion and Rotation, Relative Positioning, and Spatial Proportion. Two independent raters scored maps; inter-rater reliability 0.70–0.78; composite score was the average of the four dimensions; Cronbach’s alpha posterior mean 0.93 (93% HPDI [0.92, 0.94]). Quantification of exploration: Roaming entropy (RE) measured variability of coverage. Path RE computed as Shannon’s entropy over accessible locations in a 32×32 grid (0.5 m cells), normalized by log2(k), where k is accessible cells. Head-direction RE computed similarly over an 18×36 grid spanning horizontal (−180° to 180°) and vertical (−90° to 90°) directions (k=648), normalized by log2(k). Statistical analysis: Bayesian multilevel (and multivariate) models using brms in R, with 4 chains × 4800 iterations, weakly informative normal priors centered at 0. Predictors were centered around individual means to examine intra-individual relationships; grand-mean centered averages captured inter-individual effects. Bivariate models jointly estimated path RE and head-direction RE allowing residual correlation; duration in room included as covariate. Models reported posterior means and 93% HPDIs; posterior predictive checks and sensitivity analyses were conducted. The study was not preregistered. Participants: Experiment 1 final N=28 (3 men, 25 women, ages 18–25; mean=19.79, SD=1.70) after excluding 4 (scale adjustments and incomplete exploration). Experiment 2 N=60 (5 men, 55 women, ages 18–25; mean=19.6, SD=1.28); 5 excluded from memory analysis due to instruction/order changes. Ethical approval was obtained; informed consent given; compensation via payment or course credit.

• Pre-room curiosity vs. spatial exploration: In Experiment 1 (N=28), path RE positively associated with pre-room curiosity (βcuriosity=0.0059, 93% HPDI [0.0014, 0.0104]); tentative negative association with post-room interest (βinterest=−0.0040, 93% HPDI [−0.0089, 0.0012]). Head-direction RE positively associated with post-room interest (βinterest=0.0048, 93% HPDI [0.0016, 0.0080]); indeterminate relation with pre-room curiosity (βcuriosity=−0.0005, 93% HPDI [−0.0030, 0.0020]). Residual correlation between path and head-direction RE after predictors: 0.18 (93% HPDI [0.10, 0.27]). • Replication in Experiment 2 (N=60): Path RE positively associated with pre-room curiosity (βcuriosity=0.0039, 93% HPDI [0.0003, 0.0073]); tentative negative association with post-room interest (βinterest=−0.0023, 93% HPDI [−0.0062, 0.0016]). Head-direction RE positively associated with post-room interest (βinterest=0.0058, 93% HPDI [0.0034, 0.0082]); indeterminate relation with pre-room curiosity (βcuriosity=−0.0009, 93% HPDI [−0.0028, 0.0011]). Residual correlation: 0.34 (93% HPDI [0.28, 0.39]). Controlling for room type preserved the interest–head-direction RE link. Differences in associations between path vs. head-direction RE: pre-room curiosity difference posterior mean = 0.0064 (93% HPDI [0.0017, 0.0110]) in Exp 1; 0.0048 (93% HPDI [0.0013, 0.0083]) in Exp 2. Post-room interest difference posterior mean = −0.0088 (93% HPDI [−0.014, −0.0030]) in Exp 1; −0.0081 (93% HPDI [−0.0120, −0.0041]) in Exp 2, evidencing a double dissociation. • Trait curiosity moderation (Experiment 2): Stress Tolerance strengthened the positive curiosity–path RE relationship (interaction weight posterior mean = 0.0041, 93% HPDI [0.0012, 0.0068]); Deprivation Sensitivity showed a potential positive effect (0.0025, 93% HPDI [−0.0003, 0.0054]); Joyous Exploration and Thrill Seeking showed no clear interaction. • Cognitive map precision (Experiment 2 memory test): Intra-individual level—composite map score positively associated with pre-room curiosity (βcuriosity=0.066, 93% HPDI [0.027, 0.10]); tentative negative association with post-room interest (βinterest=−0.023, 93% HPDI [−0.060, 0.014]). Path RE positively associated with cognitive map precision (βpath=0.80, 93% HPDI [0.0076, 1.59]); head-direction RE showed no intra-individual evidence (βheadRE=−0.075, 93% HPDI [−1.22, 1.12]). Inter-individual: weak evidence for average head-direction RE (βheadRE=2.22, 93% HPDI [−0.47, 4.80]). • Mediation analysis: Pre-room curiosity’s effect on cognitive map precision may be modestly mediated by path RE (mediation effect posterior mean = 0.0028, 93% HPDI [−0.00035, 0.0077]). • Descriptive: Duration spent per room covariate—Exp 1 M=30.35 s (SD=19.19); Exp 2 M=33.10 s (SD=22.82). Positive correlations observed between path and head-direction RE, and between curiosity and interest ratings. Inter-rater reliability for map dimensions ranged 0.70–0.78; composite reliability high (alpha ~0.93).

Findings support the hypothesis that curiosity drives spatial exploration in humans and enhances cognitive map formation, aligning with theories that curiosity acts as a cognitive enhancer and extends its role beyond semantic memory to spatial-relational memory. A double dissociation emerged: pre-room curiosity predicted spatial coverage (path RE), whereas post-room interest predicted visual scanning (head-direction RE), suggesting distinct mechanisms—curiosity motivating movement through space to reduce uncertainty, and interest focusing attention on specific features. Individual differences in Stress Tolerance strengthened the curiosity–exploration link, indicating the importance of coping with uncertainty for translating curiosity into exploratory action. The positive association between path RE and map precision underscores the role of active spatial coverage in encoding spatial relationships, potentially via dopaminergic circuits interacting with the hippocampus that are implicated in curiosity-induced learning. The potential negative relation of post-room interest to map precision may reflect selective attention to features at the expense of broader spatial layout encoding, highlighting differential impacts of curiosity and interest on types of memory in spatial contexts.

The study provides the first direct behavioural evidence in humans that curiosity states enhance spatial exploration and improve cognitive map formation. It demonstrates a double dissociation between curiosity and interest effects on spatial versus visual exploration, and shows that trait Stress Tolerance amplifies curiosity-driven exploration. These insights have practical implications for designing spaces and experiences—architecture, urban planning, museums, and games—to harness curiosity for better exploration and memory in real or virtual environments.

The sketch map task, while validated, captures only one aspect of spatial memory; complementary navigation/wayfinding measures could provide a fuller assessment. Path roaming entropy primarily indexes spatial exploration but may partially reflect visual exploration; head-direction roaming entropy chiefly captures visual exploration yet can be influenced by spatial layout, indicating some overlap. Samples were heavily skewed toward women, limiting generalisability; future work should ensure balanced gender representation. A detailed analysis of environmental features and their relationship to interest and exploration was not conducted and presents an avenue for future research.