Of guns and snakes: testing a modern threat superiority effect^*

* The research was conducted at the Psychological Institute, Laboratoire de Psychologie Sociale: Menaces et Société, Université Paris Descartes.

ABSTRACT

Previous studies suggest that ancient (i.e. evolutionary-based) threats capture attention because human beings possess an inborn module shaped by evolution and dedicated to their detection. An alternative account proposes that a key feature predicting whether a stimulus will capture attention is its relevance rather than its ontology (i.e. phylogenetic or ontogenetic threat). Within this framework, the present research deals with the attentional capture by threats commonly encountered in our urban environment. In two experiments, we investigate the attentional capture by modern threats (i.e. weapons). In Experiment 1, participants responded to a target preceded by a cue, which was a weapon or a non-threatening stimulus. We found a larger cuing effect (faster reaction times to valid vs. invalid trials) with weapons as compared with non-threatening cues. In Experiment 2, modern (e.g. weapons) and ancient threats (e.g. snakes) were pitted against one another as cues to determine which ones preferentially capture attention. Crucially, participants were faster to detect a target preceded by a modern as opposed to an ancient threat, providing initial evidence for a superiority of modern threat. Overall, the present findings appear more consistent with a relevance-based explanation rather than an evolutionary-based explanation of threat detection.

KEYWORDS:

• Attentional capture
• threat
• relevance

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

* The research was conducted at the Psychological Institute, Laboratoire de Psychologie Sociale: Menaces et Société, Université Paris Descartes.

1. We used the 90% confidence intervals for eta-squared, because eta-squared cannot be negative and, therefore, 90% confidence intervals for eta-squared correspond to 95% confidence intervals for other indexes.

2. It is worth mentioning that mixed models often come with approximation of degrees of freedom (in our case Satterthwaite approximations). This explains why these degrees of freedom contain decimals. In addition, there is still no real consensus about which effect size to use with such models. Accordingly, we refrain from presenting such estimates.