Toward an~Integrated Cognitive--Ergonomic Architecture for~Human--Machine Interaction: Combining Cognitive Models with~Human Factors Ergonomics

1 Introduction

• HMI in safety-critical and knowledge-intensive domains requires models accounting for cognition, action, context, and adaptation.Cognitive architectures formalize reasoning and learning, while HFE examines conditions shaping activity in real work situations.Source: §1 · pp. 1–2
• Cognitive modeling and ergonomics have largely evolved in parallel despite shared objectives.Source: §1 · p. 2
• The paper applies its approach to welding expertise and outlines subsequent reviews of cognitive architectures, industrial robotics, and critical interface design.It also describes future perspectives after discussing these applications.Source: §1 · p. 2
• The proposed architecture combines symbolic and subsymbolic cognitive models with activity regulation, situation awareness, and competence.It aims to bridge internal cognitive mechanisms and observable activity in complex systems.Source: §1 · p. 2

2.1 Cognition, Activity, and Competence

• Competence is an evolving structure for adaptive behavior, while activity is regulated between internal states, task constraints, and environmental resources.Figure 1 depicts double regulation linking task constraints and internal regulation processes.Source: §2.1 · p. 2, Figure 1 · p. 3
• Ethnographic data support welding schemes for GTAW and GMAW and a superordinate training scheme, with gesture rules differing by process.The filler-metal manipulation subscheme varies between the welding techniques.Source: §2.1 · p. 2
• Future welder validation is intended to test scheme robustness, identify applicable situation classes, and assess consistency across task repetitions.The model may be challenged by highly variable activity, incomplete subschemes, or unfamiliar situations requiring substantial adaptation, potentially infeasible in robotics.Source: §2.1 · pp. 2–3

2.2 Memory, Knowledge, and Cognitive Resources

• Working-memory limits and long-term knowledge structures help explain skilled performance, error, learning, and expertise, bridging cognitive architecture and ergonomic analyses.Source: §2.2 · p. 3
• Industrial robotics addresses skilled-worker shortages and health risks, while complex variable environments still require human skill and cognitive flexibility.Welding is cited as an example amid a shortage of skilled operators in France.Source: §2.2 · p. 4

3.1 SOAR

• SOAR models problem spaces and production rules symbolically, with chunking compiling experience into procedural knowledge.Source: §3.1 · p. 4

3.2 ACT-R

• ACT-R combines symbolic rules with subsymbolic memory activation and action timing, using modular components to model perception, cognition, and motor interaction.Source: §3.2 · p. 4

3.3 LIDA

• LIDA models perception-action cycles and attention for adaptive cognition, including real-time regulation and feedback control.Source: §3.3 · p. 4

3.4 COCOM

• COCOM distinguishes strategic, tactical, opportunistic, and scrambled control modes for behavior in dynamic, time-critical settings.Source: §3.4 · p. 4

3.5 Application in Welding Operations

• The welding application combines scheme theory with cognitive architectures to characterize manual-operator skills and graphically represent cognitive and gestural processes.Schemes and competencies are treated as equivalent; they may be individual or shared and identified by confronting expert knowledge with operators’ practical know-how.Source: §3.5 · p. 5
• Modeling human skill is proposed to increase robotic robustness and improve operator training, while robotization can facilitate communication among roboticists, operators, and engineers.The selection scheme can support graph-based or probabilistic representations, but the goal is decision orientation rather than an exact current-setting equation; full robotization is not described.Source: §3.5 · p. 5

4 Human Factors Ergonomics Perspective

• HFE analyzes real activity rather than prescribed tasks and interprets performance variability as adaptive regulation.This perspective is relevant to dynamic work requiring adaptation to changing conditions and constraints.Source: §4 · p. 5

4.1 Operator Expertise and Normative Knowledge

• Qualification standards and safety protocols regulate expertise, making normative knowledge central to welding execution, evaluation, and skill transmission.Source: §4.1 · p. 6

4.2 Participant Observation

• Participant observations in engineering work and welding initiation training documented ergonomic, sensory, and psychological dimensions such as posture, visibility, fatigue, and stress.Logbooks and experience feedback organized transcripts, sensory impressions, and reflections to support later qualitative inquiry; observation also enabled access to embodied knowledge and empathy.Source: §4.2 · pp. 6–7

4.3 A Multiphysics System Model

• Welding is modeled as a multiphysics process in which continuous sensory feedback guides gesture adjustments, including torch angle and travel speed, to maintain bead uniformity.The conceptual architecture integrates sensory processing, gesture regulation, and cognitive decision-making in real-time control.Source: §4.3 · p. 7, Figure 3 · p. 7

• The architecture combines cognitive-science foundations with ergonomic insights to address operator expertise, task regulation, and adaptive behavior in complex systems.The authors present it as bridging cognitive modeling and Human Factors Ergonomics in human-machine interaction.Source: §5 · p. 8
• The framework aligns with Vergnaud’s scheme of activity and integrates feedback mechanisms to support real-time decision-making and skill acquisition.Figure 4 presents the skill model as an integration of Vergnaud’s scheme and cognitive architectures in a feedback scheme.Source: §5 · p. 8, Figure 4 · p. 8
• In welding and other real-world scenarios, the architecture has potential to improve operator performance and system design by formalizing normative knowledge and operator expertise.The authors describe this as a foundation for more intuitive and adaptive human-machine interfaces.Source: §5 · p. 8
• Future work will validate the architecture across diverse industrial contexts and refine its computational models for a broader range of cognitive and ergonomic factors.The authors also propose exploring machine learning to potentially enhance system adaptability and predictive capabilities.Source: §5 · p. 8
• The research contributes to human-centered technologies aligned with users’ cognitive and ergonomic needs.Source: §5 · p. 9