The underwood project: A virtual environment for eliciting ambiguous threat

Abstract

Threatening environments can be unpredictable in many different ways. The nature of threats, their timing, and their locations in a scene can all be uncertain, even when one is acutely aware of being at risk. Prior research demonstrates that both temporal unpredictability and spatial uncertainty of threats elicit a distinctly anxious psychological response. In the paradigm presented here, we further explore other facets of ambiguous threat via an environment in which there are no concrete threats, predictable or otherwise, but which nevertheless elicits a building sense of danger. By incorporating both psychological research and principles of emotional game design, we constructed this world and then tested its effects in three studies. In line with our goals, participants experienced the environment as creepy and unpredictable. Their subjective and physiological response to the world rose and fell in line with the presentation of ambiguously threatening ambient cues. Exploratory analyses further suggest that this ambiguously threatening experience influenced memory for the virtual world and its underlying narrative. Together the data demonstrate that naturalistic virtual worlds can effectively elicit a multifaceted experience of ambiguous threat with subjective and cognitive consequences.

Keywords: Anxiety; Fear; Predictability; Uncertainty; Virtual worlds.

Fig. 1

Ratings of overall subjective experience. The estimated marginal means (black dots), their confidence intervals (purple line), and distributions (violin plots) for ratings of the overall experience of the virtual world. The red arrows are for comparisons between means. Overlapping arrows between two means indicates that the difference is not significant based on the adjusted p-values

Fig. 2

Screenshots from the virtual world. These are a subset of the scenes in Fig. 3

Fig. 3

Scene-by-scene affect and heart rate. The estimated marginal means (black dots) and their confidence intervals (shaded bars) for the scene-by-scene ratings and heart rate levels. The red arrows are for comparisons between means. Overlapping arrows between two means indicates that the difference is not significant based on the adjusted p-values. The order of the scenes (x axis) is chronological

Fig. 4

Relationship between affect and memory. One latent variable (LV) emerged (p < .01) in the partial least squares (PLS) analysis demonstrating an association between memory measures and overall emotions. Ratings of unpredictability and surprise contributed to the LV, which correlated with memory bias for both the scenes and the prelude, and with sensitivity (d′) and “confabulation” about the prelude (see main text)