Subtopic Deep Dive

Computational Models of Emotion in AI
Research Guide

What is Computational Models of Emotion in AI?

Computational models of emotion in AI formalize appraisal theories and dimensional models to synthesize and recognize emotions in human-robot interaction.

Researchers implement affective modules that influence decision-making in social robotics (Hudlická, 2011; 143 citations). Key architectures include FLAME, EMA, FAtiMA, and SEAI based on Damasio’s theory (Kowalczuk & Czubenko, 2016; 64 citations; Cominelli et al., 2018; 56 citations). Over 20 models reviewed span cybernetic and psychological approaches across 250M+ papers.

15
Curated Papers
3
Key Challenges

Why It Matters

Emotional AI enables natural interactions in healthcare, where socially adaptable robots detect user affective states to enhance engagement in dementia care (Tanevska et al., 2020; 55 citations; Perugia et al., 2018; 41 citations). In companionship robotics, models like SEAI improve human-like perception and response (Cominelli et al., 2018). Ethical taxonomies guide emotion proxy data use in AI systems (Stark & Hoey, 2021; 142 citations).

Key Research Challenges

Integrating Emotions in Architectures

Linking emotion models to cognitive architectures remains complex, as seen in LIDA's global workspace lacking full affective integration (Ramamurthy et al., 2006; 47 citations). Hudlická outlines design guidelines but notes variability in appraisal mechanisms (Hudlická, 2011; 143 citations).

Ethical Emotion Representation

Proxy data for emotions risks bias in AI systems, with taxonomies revealing impacts on decision-making (Stark & Hoey, 2021; 142 citations). Duffy identifies fundamental issues in making social machines affective without anthropomorphic pitfalls (Duffy, 2008; 35 citations).

Real-time Human-Robot Adaptation

Achieving continuous affective awareness in interactions challenges current frameworks (Tanevska et al., 2020; 55 citations). Bionic models struggle with human-like perception speed and accuracy (Velik, 2009; 41 citations).

Essential Papers

1.

Guidelines for Designing Computational Models of Emotions

Eva Hudlická · 2011 · International Journal of Synthetic Emotions · 143 citations

Rapid growth in computational modeling of emotion and cognitive-affective architectures occurred over the past 15 years. Emotion models and architectures are built to elucidate the mechanisms of em...

2.

The Ethics of Emotion in Artificial Intelligence Systems

Luke Stark, Jesse Hoey · 2021 · 142 citations

In this paper, we develop a taxonomy of conceptual models and proxy data used for digital analysis of human emotional expression and outline how the combinations and permutations of these models an...

3.

Computational Approaches to Modeling Artificial Emotion – An Overview of the Proposed Solutions

Zdzisław Kowalczuk, Michał Czubenko · 2016 · Frontiers in Robotics and AI · 64 citations

Cybernetic approach to modeling artificial emotion through the use of different theories of psychology is considered in this paper, presenting a review of twelve proposed solutions: ActAffAct, FLAM...

4.

SEAI: Social Emotional Artificial Intelligence Based on Damasio’s Theory of Mind

Lorenzo Cominelli, Daniele Mazzei, Danilo Emilio De Rossi · 2018 · Frontiers in Robotics and AI · 56 citations

With the increase in device integration level and the growth in complexity of Integrated circuits, small delay and low power dissipation become important parameters as these increases performance a...

5.

A Socially Adaptable Framework for Human-Robot Interaction

Ana Tanevska, Francesco Rea, Giulio Sandini et al. · 2020 · Frontiers in Robotics and AI · 55 citations

In our everyday lives we regularly engage in complex, personalized, and adaptive interactions with our peers. To recreate the same kind of rich, human-like interactions, a social robot should be aw...

6.

LIDA: A Working Model of Cognition

Uma Ramamurthy, Bernard J. Baars, Sidney K. D’Mello et al. · 2006 · Human Pathology · 47 citations

The object of this study was the investigation of carbohydrate antigen expression in malignant epithelial cells and benign mesothelial cells in serous effusions from patients diagnosed with epithel...

7.

Understanding Engagement in Dementia Through Behavior. The Ethographic and Laban-Inspired Coding System of Engagement (ELICSE) and the Evidence-Based Model of Engagement-Related Behavior (EMODEB)

Giulia Perugia, Roos van Berkel, Marta Díaz Boladeras et al. · 2018 · Frontiers in Psychology · 41 citations

Engagement in activities is of crucial importance for people with dementia. State of the art assessment techniques rely exclusively on behavior observation to measure engagement in dementia. These ...

Reading Guide

Foundational Papers

Start with Hudlická (2011; 143 citations) for design guidelines, then Duffy (2008; 35 citations) for social machine issues, and Velik (2009; 41 citations) for bionic perception basics.

Recent Advances

Study Stark & Hoey (2021; 142 citations) for ethics taxonomy, Tanevska et al. (2020; 55 citations) for adaptable HRI, and Cominelli et al. (2018; 56 citations) for SEAI.

Core Methods

Appraisal (EMA, FAtiMA), dimensional (FLAME, WASABI), cybernetic (ActAffAct), and theory-based (Damasio in SEAI); LIDA for cognition integration (Kowalczuk & Czubenko, 2016).

How PapersFlow Helps You Research Computational Models of Emotion in AI

Discover & Search

Research Agent uses citationGraph on Hudlická (2011; 143 citations) to map 12+ emotion models like FLAME and EMA, then findSimilarPapers reveals SEAI extensions (Cominelli et al., 2018). exaSearch queries 'appraisal theory robotics' across 250M+ OpenAlex papers for undiscovered affective architectures.

Analyze & Verify

Analysis Agent runs readPaperContent on Kowalczuk & Czubenko (2016) to extract 12 model comparisons, verifies affective integration claims via verifyResponse (CoVe) against LIDA (Ramamurthy et al., 2006), and uses runPythonAnalysis for GRADE grading of engagement metrics in Tanevska et al. (2020) with statistical validation.

Synthesize & Write

Synthesis Agent detects gaps in ethical models post-Stark & Hoey (2021) via contradiction flagging across Duffy (2008), then Writing Agent applies latexSyncCitations and latexCompile for a review paper with exportMermaid diagrams of FLAME-EMA flows.

Use Cases

"Compare statistical performance of FLAME vs EMA in robot emotion synthesis"

Research Agent → searchPapers 'FLAME EMA comparison' → Analysis Agent → runPythonAnalysis (extract metrics from Kowalczuk 2016, NumPy correlation plots) → researcher gets pandas CSV of model accuracies.

"Draft LaTeX section on SEAI Damasio integration for HRI review"

Research Agent → readPaperContent (Cominelli 2018) → Synthesis Agent → gap detection → Writing Agent → latexEditText + latexSyncCitations (Hudlická 2011) + latexCompile → researcher gets compiled PDF section with figures.

"Find GitHub code for affective robot architectures like FAtiMA"

Research Agent → citationGraph (Kowalczuk 2016) → Code Discovery → paperExtractUrls → paperFindGithubRepo → githubRepoInspect → researcher gets runnable FAtiMA emotion module repos.

Automated Workflows

Deep Research workflow scans 50+ papers from Hudlická (2011) citation network, producing structured reports on appraisal models with GRADE scores. DeepScan applies 7-step CoVe to verify ethical claims in Stark & Hoey (2021) against Duffy (2008). Theorizer generates novel hybrid theories from SEAI (Cominelli et al., 2018) and LIDA (Ramamurthy et al., 2006).

Frequently Asked Questions

What defines computational models of emotion in AI?

Models formalize appraisal theories (e.g., EMA, FAtiMA) and dimensional approaches to generate/recognize emotions in AI agents (Hudlická, 2011; Kowalczuk & Czubenko, 2016).

What are key methods in this subtopic?

Cybernetic models (ActAffAct, ParleE), architecture integrations (LIDA, SEAI), and bionic perception (Velik, 2009; Cominelli et al., 2018).

What are the most cited papers?

Hudlická (2011; 143 citations) on design guidelines; Stark & Hoey (2021; 142 citations) on ethics; Kowalczuk & Czubenko (2016; 64 citations) reviewing 12 solutions.

What open problems exist?

Real-time affective adaptation, ethical proxy data biases, and full cognitive-emotion integration beyond partial models like LIDA (Tanevska et al., 2020; Stark & Hoey, 2021).

Research Psychiatry, Mental Health, Neuroscience with AI

PapersFlow provides specialized AI tools for Computer Science researchers. Here are the most relevant for this topic:

See how researchers in Computer Science & AI use PapersFlow

Field-specific workflows, example queries, and use cases.

Computer Science & AI Guide

Start Researching Computational Models of Emotion in AI with AI

Search 474M+ papers, run AI-powered literature reviews, and write with integrated citations — all in one workspace.

See how PapersFlow works for Computer Science researchers