ABSTRACT



The concept of psychological safety has attracted sustained empirical attention since Edmondson's foundational work in 1999, establishing its relationship with team learning behaviour and performance. Despite this extensive literature, most organisations continue to treat psychological safety as one environmental variable among many — important, but additive rather than foundational. This study challenges that assumption through a mixed methods investigation examining whether psychological safety functions as a threshold moderating variable: one whose presence or absence determines whether other organisational investments in learning and development generate measurable performance returns at all — a conceptual reframing with significant implications for how capability investment is sequenced and evaluated.

Drawing on survey data from 342 frontline professionals across six geographically distributed operational contexts, combined with 24 in-depth interviews with senior learning and development leaders, the study employs an Explanatory Sequential Mixed Methods design grounded in an integrated theoretical framework synthesising three streams: psychological safety theory (Edmondson, 1999; Frazier et al., 2017), learning transfer theory (Baldwin & Ford, 1988; Blume et al., 2010), and applied neuroscience (Arnsten, 1998, 2015; Rock, 2008). This integration is deliberate and theoretically consequential: neuroscience provides the biological mechanism, transfer theory provides the outcome framework, and psychological safety provides the environmental moderating condition — each stream illuminating what the others alone cannot explain.

The study finds that high-quality learning design has a statistically significant positive effect on commercial performance outcomes in high-psychological-safety environments (β = .31, p < .001), but a non-significant effect in low-psychological-safety environments (β = .09, p = .14), with the moderation interaction accounting for a meaningful increment in explained variance (ΔR² = .028, p = .003). Qualitative thematic analysis surfaces the construct of the Safety Ceiling — operationally defined as the environmental threshold of perceived interpersonal safety, measurable via team-level Psychological Safety Climate scores, below which cortisol-mediated cognitive suppression prevents the transfer of acquired learning to live performance behaviour regardless of learning design quality. This construct has identifiable neurological architecture and provides mechanistic explanation for the quantitative moderation finding.

Statistical limitations are explicitly acknowledged, including potential social desirability bias in self-reported performance ratings, the absence of inter-rater agreement statistics for team-level safety climate aggregation, and the risk of shared method variance inflating effect size estimates. Generalisability is appropriately bounded: findings are most directly applicable to frontline, multi-cultural, high-performance-visibility service sector environments, and replication in knowledge-work, remote, and monocultural contexts is recommended before broader claims are made. Practical implications are strong and directly actionable for both managers and learning and development professionals, with specific guidance on safety diagnostic sequencing, the compliance-activation distinction as a real-time team climate indicator, and the incorporation of safety climate as a baseline moderating variable in programme evaluation. Future research priorities include longitudinal panel designs to establish causal direction, experimental or quasi-experimental studies targeting safety-building managerial interventions, and replication across diverse organisational forms.

Keywords: psychological safety; learning transfer; performance architecture; safety ceiling; action research; mixed methods; neuroscience; L&D effectiveness; frontline performance; threshold moderation