Periodic Reporting for period 3 - ActionContraThreat (Action selection under threat: the complex control of human defense)
Reporting period: 2022-04-01 to 2023-09-30
Humans seem have access to an extraordinary repertoire of actions to avoid or escape threat. Sometimes these actions must be selected with high precision and under extreme time pressure. This poses a significant challenge to any biological or artifical system. The current project addresses the cognitive mechanisms by which the brain decides on a course of action under threat, implements it in behaviour, monitors its success, and updates the plan. This is likely to be of importance to people who have to cope with physical hazards in their daily work, and also to clinical conditions in which people seem to over-avoid or escape harmless objects or situations.
Thus, the project will investigate: (a) what exactly are the actions and movements that humans can take to avoid, escape, or defend themselves, and (b) how are these actions planned and coordinated. Since ethical considerations prohibit exposing humans to actual threat, we build a virtual reality platform in which people can freely move and encounter various virtual threats. We record their full-body movements in different ways, and analyse the paths of different body parts in these situations. Virtual reality enables us to create a natural environment, but also to manipulate the environment in ways that are physically or biologically impossible. This allows probing the limits of the cognitive mechanisms and thus expose their inner workings. For example, we can ask to what extent can a person suppress a spontaneous action when it leads to negative consequences.
Ultimately, the knowledge garnered in this project could help improving the training that professionals receive in order to select workplace-optimal protective actions, and that patients receive in order to suppress harmful over-avoidance.
Thus, the project will investigate: (a) what exactly are the actions and movements that humans can take to avoid, escape, or defend themselves, and (b) how are these actions planned and coordinated. Since ethical considerations prohibit exposing humans to actual threat, we build a virtual reality platform in which people can freely move and encounter various virtual threats. We record their full-body movements in different ways, and analyse the paths of different body parts in these situations. Virtual reality enables us to create a natural environment, but also to manipulate the environment in ways that are physically or biologically impossible. This allows probing the limits of the cognitive mechanisms and thus expose their inner workings. For example, we can ask to what extent can a person suppress a spontaneous action when it leads to negative consequences.
Ultimately, the knowledge garnered in this project could help improving the training that professionals receive in order to select workplace-optimal protective actions, and that patients receive in order to suppress harmful over-avoidance.
We engineered a software toolkit that implements a virtual reality environment in which various threats can occur. We selected threats by their relevance to human survival and well-being throughout history. Thus, we included environmental threats, other humans, apex predators, self-defending animals such as large herbivores, domestic animals, and snakes, and small animals that can inflict injury or carry diseases.
Through series of experiments in virtual reality, we found that humans behave in sophisticated ways that are closely related to the type of threat, its time-to-impact, and existing escape options. We found that humans have almost no difficulties adapting their behaviour to abrupt changes in the environment, and that they can rapidly learn and integrate completely novel actions. We also found that some selected behaviours are more resistant to change.
Through series of experiments in virtual reality, we found that humans behave in sophisticated ways that are closely related to the type of threat, its time-to-impact, and existing escape options. We found that humans have almost no difficulties adapting their behaviour to abrupt changes in the environment, and that they can rapidly learn and integrate completely novel actions. We also found that some selected behaviours are more resistant to change.
Our VR platform goes beyond current state-of-the-art in terms of complexity, breadth, and affordances for future experiments. At the end of the project, we expect to be able and identify individual threat-related actions from video-observations, analyse their temporal structure, and highlight novel and currently unknown action patterns. We expect to have detail knowledge on cognitive mechanisms of threat avoidance, as well as on implementation principles in the brain. Furthermore, we hope to be able and characterise actions that are hard to change by instruction or training, and investigate suitable ways of overcoming this resistance.