Programmed Instruction: Unit 2:,,Emotion Modeling Architectures in Affective Computing

🧠 Unit 2: Emotion Modeling Architectures in Affective Computing

Goal: You will understand and apply major computational models used to represent and reason about human emotion in software systems. These models provide structured frameworks for classifying, predicting, and generating emotional states in AI.

🔹 Frame 1: Why Emotion Models?

Instruction:
Raw emotional signals (e.g., facial expressions, voice tone) are ambiguous. Emotion modeling systems provide a conceptual bridge between input and interpretation by mapping signals onto structured representations.

Three widely used models:

OCC Model (Ortony, Clore & Collins)
PAD Model (Pleasure, Arousal, Dominance)
EMA Model (Emotion and Adaptation, Marsella & Gratch)

Each has different strengths depending on use case (e.g., storytelling, virtual agents, robotics, mental health).

Prompt:
What is the purpose of using emotion models in affective computing?

A. To avoid machine learning altogether
B. To organize and reason about emotion beyond raw data
C. To enforce emotional uniformity in users
D. To teach machines facial recognition

📝 Your answer:

✅ Exactly right!
Emotion models help systems organize and reason about emotion beyond the raw input—providing structured logic for interpreting, predicting, or simulating emotional responses.

They don’t replace perception—they translate it into meaning.

🔹 Frame 2: The OCC Model

Instruction:
The OCC Model (Ortony, Clore, Collins) is a cognitive appraisal theory of emotion. It says emotions arise from how we evaluate events, agents, or objects in terms of goals, standards, and attitudes.

It defines emotions as reactions to:

Events (e.g., fear of danger, joy from success)
Agents’ actions (e.g., pride, shame)
Objects (e.g., love for art, dislike of a sound)

Each emotion is derived from appraisal variables: Was the event desirable? Was the agent praiseworthy? Do I like the object?

Prompt:
According to the OCC model, what is an emotion like pride a reaction to?

A. A beautiful object
B. A desirable future event
C. One’s own praiseworthy action
D. An unexpected loud noise

📝 Your answer:

✅ Correct!
Pride, in the OCC model, is a reaction to one’s own praiseworthy action—an example of an agent-based appraisal. Similarly, admiration might arise from someone else’s praiseworthy action.

This model lets systems reason about who did what, why it matters, and how to feel about it.

🔹 Frame 3: The PAD Model

Instruction:
The PAD Model defines emotion as a position in a 3D space made of:

Pleasure (positive ↔ negative feeling)
Arousal (high ↔ low energy)
Dominance (sense of control or submission)

Example mappings:

Joy: High Pleasure, High Arousal, Moderate Dominance
Fear: Low Pleasure, High Arousal, Low Dominance
Boredom: Low Pleasure, Low Arousal, Neutral Dominance

This model is great for emotional state tracking in media, games, avatars, and design.

Prompt:
Which PAD coordinates would best describe fear?

A. High pleasure, high arousal, high dominance
B. Low pleasure, high arousal, low dominance
C. High pleasure, low arousal, high dominance
D. Neutral pleasure, low arousal, low dominance

📝 Your answer:

✅ Perfect!
Fear is typically modeled in PAD space as:

Low Pleasure (it's unpleasant)
High Arousal (intensely activating)
Low Dominance (feeling out of control)

This dimensional approach is elegant for continuous tracking and smooth transitions between emotional states—useful in animation, VR, and character design.

🔹 Frame 4: The EMA Model

Instruction:
The EMA (Emotion and Adaptation) model builds on appraisal theory like OCC, but emphasizes dynamic emotional change over time. It focuses on:

Appraisal Framing: interpreting an event through goals, beliefs, and context
Re-appraisal: updating emotions as new information arrives
Coping Strategies: emotional adaptation through behavior or rethinking

EMA is ideal for real-time interactive agents that adapt emotionally, like in therapy bots or narrative characters.

Prompt:
What makes the EMA model especially useful for real-time affective agents?

A. It uses dimensional models only
B. It maps facial expressions directly to actions
C. It updates emotions dynamically as situations change
D. It simplifies emotions to three basic types

📝 Your answer:

✅ Exactly right!
The EMA model’s strength is in dynamic updating—it continuously re-appraises the situation, allowing agents to adapt their emotional state in real time as new information or context arrives.

This makes it ideal for interactive environments where emotions evolve naturally.

🌟 Mini-Summary: Emotion Modeling Architectures

Model	Key Feature	Ideal Use Case
OCC	Cognitive appraisal of events, agents, objects	Reasoning about cause of emotions, narrative AI
PAD	Continuous 3D space (pleasure, arousal, dominance)	Smooth emotion tracking, games, avatars
EMA	Dynamic re-appraisal and coping over time	Interactive, adaptive agents and therapy bots

Would you like to:

🔄 Continue with practice frames for Unit 2?
📘 Receive a formatted summary PDF of Unit 2?
🚀 Move on to Unit 3: Designing Emotionally Intelligent Systems?
✍️ Apply these models in designing a specific affective computing use case?

Your call, Len.

Do you have publication dates for these models

Certainly, Len! Here are the approximate publication dates and origins for the three major models we discussed:

Model	Key Publication / Origin	Year
OCC Model	Ortony, Clore, & Collins, The Cognitive Structure of Emotions (book)	1988
PAD Model	Mehrabian & Russell, An Approach to Environmental Psychology (book)	1974
EMA Model	Marsella & Gratch, EMA: A Process Model of Appraisal Dynamics (paper)	2009

Additional context:

OCC is foundational in appraisal theory for computational emotion, first presented in the late 1980s.
PAD stems from psychological research in environmental psychology in the 1970s, widely applied in affective computing later.
EMA is a relatively recent model, explicitly designed for dynamic, real-time emotion modeling in AI agents.

Search This Blog

An AI Artist in Process