Table 1.
Dependent Measures in Experiments.
All dependent measures were given for all 29 threat scenarios (cf. Table 2).
Table 2.
Example Threat Scenarios Presented to Subjects.
Each scenario is assigned a brief descriptor and label, used throughout the paper. N = Natural; A = Animal; P = Physical; S = Psychological. Full set of 29 scenarios presented in S1 Table. All Physical scenarios taken from [2].
Fig 1.
Subjects (n = 31) in the approach-avoid experiment viewed these instructions, which made explicit that approach-avoidance ratings related to taxis relative to the source of threat.
Table 3.
Comparison of male and female top response options in scenarios for which their first responses differed. When applicable, these differences are compared to prior results [2] in the comments column. While the top response option in scenario P10 did not differ between males and females, the scenario is reported since Blanchard [2] had observed a gender effect. * Denotes a tie between first-choice response options.
Table 4.
Columns demonstrate the range of raw ratings by factor, with lowest and highest rated scenarios listed. Histograms (S1 Fig) show the number of scenarios that received an average rating corresponding to a score of 1 (low) to 5 (high).
Table 5.
Normalized factor ratings (Mean ± SE) by scenario category.
Fig 2.
Factor-Response Option Correlations.
Heatmap of correlation coefficients from Pearson’s correlations between mean factor ratings and specific defensive behaviors (a,b) or approach/freeze/avoid (c,d) for physical (left) and psychological (right) threat scenarios. Row-wise factors organized along an approximate low to high imminence continuum. Column-wise response options organized along an approximate approach-avoidance continuum. Original distance scores were reversed to far-to-near to in accordance with the imminence continuum.
Fig 3.
Decision Tree for Defensive Behaviors to Threatening Scenarios.
Panels a, b. Decision tree predicting the defensive behavior chosen by the majority of subjects based on characteristics of that threat scenario. The tree consists of four main branches, with primary approach/avoid responses predicted by the 2x2 interaction of danger (high, moderate) and threat type (psychological, physical threat). Appraisal of factors along further nodes predicts specific defensive responses for each scenario, denoted by the scenario labels used in Table 2. Where appropriate, gender differences are noted. The tree successfully predicts the group majority decisions of both original subjects (n = 85) and separate replication study subjects (n = 22) for all scenarios. Panel c. The average proportion of original and replication study subjects’ first responses correctly predicted for all scenarios (n = 29), physical scenarios (n = 20), and psychological scenarios (n = 9). Male and female performance reported separately. Dashed line around 0.12 (All: 0.128; Physical: 0.130; Psychological: 0.123) represents chance performance.
Fig 4.
Clustering of Scenario Categories Based on Factor Ratings.
Panel a. Heatmap of scenario factor ratings dissimilarity matrix. Dissimilarity scores (legend right of heatmap) represent the distance between pairs of scenarios, calculated as one minus the sample correlation between the ten factor ratings for each pairwise scenario comparison. Scenario labels indicated on top and left of heatmap, with individual scenarios denoted by each row/column (i.e., 4 “Natural” scenarios: N1, N2, N3, N4). Black diagonal indicates scenarios are minimally dissimilar to themselves; dark clusters indicate within-category scenarios are most similar according to factor ratings. Psychological scenarios are most distinct from the other categories. Within category similarity exceptions exist, e.g. P1, P9, P11 and S3. Panel b. Two-dimensional multidimensional scaling (MDS) of Euclidean distance between scenarios based on factor-rating dissimilarity scores. Human psychological threat scenarios (red stars) mostly clustered separately from physical threats (blue circles: animal and natural threats; black triangles: human physical). A scree plot of stress by MDS dimensions justified the use of 2 dimensions, which had a stress of 0.140. While the primary value of our MDS analysis is as a visualization of the similarity space of scenario factor ratings, we cautiously suggest that the first MDS dimension, positively related to ability to mitigate (r = 0.68) and to communicate (r = 0.90) with the threat, while inversely related to dangerousness (r = -0.88) and immediacy (r = -0.63), related to “social power” or the threatened individual’s ability to communicate with and influence the threat. Meanwhile, the second MDS dimension, inversely related to immediacy (r = -0.68), and positively related to distance (r = 0.74) and the presence of a hiding place (r = 0.75) captured the threatened individual’s ability to thwart the threat and “control” the situation.
Table 6.
Comparison of correlations coefficients between defensive behaviors and scenario characteristics obtained in 4 studies.
Comparison between Blanchard [2], Perkins and Corr [22], and Shuhama [21] reproduced from Blanchard [4]. The first 3 studies (3 leftmost columns) used Blanchard’s original 12 physically threatening scenarios and report male (top) and female (bottom) correlation values separately. In the present study (4 rightmost columns), the 20 physical scenarios included 11 of the original physically threatening scenarios, along with 4 natural disaster and 5 animal scenarios. For the 9 psychological scenarios, one to two comparable defensive response options are reported. V.C. = verbal confrontation. *p<0.05, **p<0.01, ***p<0.001, n.s. = not significant with p<0.05; p-values not reported in Shuhama et al. [21].