Human Failure Events (HFEs) are complex, multi-layer events culminating with a human-machine team's failure to complete a plant objective. HFEs can be further described by Crew Failure Modes (CFMs) which document specific ways the objective tasks may be unsuccessfully performed. In turn, these CFMs are affected by Performance Influencing Factors (PIFs), some of which exert a more direct influence than others. However, in current Human Reliability Analysis (HRA) methods, the multitudes of causal relationships between PIFs, CFMs, and HFEs are not explicitly modeled. This work seeks to fill that gap by developing structured causal models that document direct and indirect pathways from PIFs, through CFMs, and into HFEs. This work is intended to expand the current application of causal-based HRA modeling beyond control room environments to external environments under natural hazard scenarios.A Bayesian network of information-gathering operator activities in response to a system demand is developed by following the causal mapping methodology defined in Zwirglmaier et al. (2017). The relationships in this structure are substantiated with existing psychological and organizational literature, thereby allowing for the identification of the main causal pathways leading to a particular CFM, and therefore an HFE. The work draws upon proximate causes of failure from the NRC's NUREG-2114, CFMs in the Phoenix HRA method, and PIFs from Groth's 2012 hierarchy. Capturing these causal pathways provides the foundation for an improved causal basis of HRA, which represents a promising strategy for enhancing the accuracy and technical basis of HRA. Future efforts will include validation of the structures, constructing similar models for decisionmaking and action HFEs, and quantification of the Bayesian network structures.