Chain-of-feeling
Synthetic Users use a “chain-of-feeling” approach—combining emotional states with OCEAN personality traits—to produce more human-like, realistic user responses and yield richer UX insights.
Chain-of-feeling
To cut to the chase, a “chain-of-feeling” is essentially your synthetic user’s emotional narrative as they reason through a task. Just as a chain-of-thought ensures a more transparent cognitive process (something OpenAI introduced in O1, a chain-of-feeling ensures a more realistic, human-like emotional process—critical for generating diverse, nuanced insights in UX research.
If you’ve been to one of our webinars or demos you know how important the OCEAN personality model is in generating better Synthetic Users, i.e. Synthetic Users that when compared to organic users score considerably higher than going directly to a GPT.
The Chain-of-feeling method doubles down on a more visceral approach to UX research. As models migrate from a System 1 to a System 2, from instinct to rational explanation, we at Synthetic Users are doubling down on S1. We want our Synthetic Users to be as close or organic users (what you call real humans). Closer implies factoring in all the cognitive impairments and emotional drivers that influence users’ behaviour.
1. What Is a Chain-of-Feeling?
• Chain-of-thought: A stepwise cognitive reasoning sequence.
• Example (simplified):
1. I see a question.
2. I recall relevant data.
3. I reason through possible answers.
4. I synthesize and present a final solution.
• Chain-of-feeling: A parallel or intertwined emotional sequence.
• Example (simplified):
1. I start curious or uncertain.
2. I become slightly anxious if I can’t find a clear solution.
3. I get relief upon finding a direction.
4. I end with contentment once I feel the solution is good.
In an interaction with a Synthetic User, each step captures not only the “what” (chain-of-thought) but also the “how it feels” (chain-of-feeling). This helps produce more realistic and emotionally diverse behaviors that mimic how real users respond to frustration points, excitement, confusion, etc.
2. Why we came up with a Chain-of-Feeling?
1. Better Realism: Organic users’ behaviors are heavily influenced by how they feel at each moment in a workflow (e.g., frustration can cause users to abandon a task even if they could complete it with more time).
2. Personality Integration: Synthetic Users each have an OCEAN-based personality profile (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism). We needed a mechanism to model how those traits alter emotional responses over time. A user high in Neuroticism might escalate frustration quicker, while a highly Agreeable user might maintain calm longer.
3. Better Insights: By incorporating emotional states into the chain of reasoning, we get a more “human-like” variety of responses, often revealing edge cases. This enriches your UX research and highlights potential improvements in product or service design.
3. Modeling the Chain-of-Feeling
1. Defining Emotional States and Dimensions
• We use Plutchik’s wheel of emotions. It’s the one with most scientific backing.
• We attach intensities or valences (e.g., scale of 1–5 for mild to intense).
2. Triggers and Transitions
• Triggers are events in the user’s journey: encountering an error message, discovering a new feature, facing a time constraint, etc.
• Transitions describe how the user moves from one emotional state to the next, informed by the user’s personality traits and the context. For example, a user with low Neuroticism might shift from mild anxiety back to curiosity quickly, while a high-Neuroticism user escalates from mild anxiety to strong frustration.
3. Incorporating Personality (OCEAN)
• Openness: More open users might respond to new designs with curiosity or excitement instead of confusion.
• Conscientiousness: Might drive perseverance in the face of frustration—leading them to stay longer before giving up.
• Extraversion: Could lead to positive affect more quickly, or a tendency to consult help sooner.
• Agreeableness: Might lead to self-blame or calmness instead of anger at the interface.
• Neuroticism: Might intensify negative emotions or cause them to linger.
4. Annotating Steps
• Each “step” the user takes can include a short emotional annotation. For example:
1. Thought: “I need to find the settings tab.”
Feeling: “Mild frustration (2/5) after not seeing it immediately on the screen.”
2. Thought: “I notice a hamburger menu, maybe it’s there.”
Feeling: “Slight relief (1/5) but still uncertain.”
3. Thought: “Yes, the settings are there. Let me open them.”
Feeling: “Neutral, trending to curious.”
4. Using the Chain-of-Feeling in UX Research
1. Scenario Simulation
• We run multiple Synthetic Users through the same concept. We collect the chain-of-feeling logs alongside decisions to see which personalities or emotional triggers lead to task success/failure, early drop-off, or dissatisfaction.
2. Identifying Pain Points and barriers
• We locate patterns where frustration spikes or confusion becomes prevalent. This helps us identify areas in the concept or design that might cause negative user experiences.
3. Concept Iterations
• We tweak the concept based on these identified emotional friction points, then we re-run the Synthetic Users to compare emotional states pre- and post-changes.
• If after a design fix the chain-of-feeling logs show fewer spikes in frustration for the same tasks, we might infer that real users will also have a smoother experience.
5. Summary of Key Points
• A chain-of-feeling mirrors the structure of a chain-of-thought but tracks emotional progressions.
• Each step includes both cognitive reasoning (“What am I thinking/doing?”) and emotional response (“How do I feel about it?”).
• OCEAN personality traits guide how emotions shift and how strongly they manifest.
• This framework yields richer, more human-like Synthetic User, unveiling friction points and user experience issues that purely rational or purely “thought-based” models might miss. As per our earlier point of Super Intelligence vs Human intelligence. Synthetic Users is all about Human Intelligence.
Damasio’s View: Emotions vs. Feelings
Emotions as Bodily States, Feelings as Mental Experiences. Emotions happen in the body first. Feelings are the mind’s interpretation of those emotions.
Applying This Distinction to Synthetic “Chain-of-Feeling”
When we say “chain-of-feeling” in the context of Synthetic Users, we’re primarily talking about the subjective experience side—i.e., feelings in Damasio’s sense. Of course, Synthetic Users do not have real physiological changes (like a racing heart), so we rely on simulated or “fictional” emotional states.
To be precise:
• We are not truly modeling bodily/physiological changes (which would be “emotions” in Damasio’s terms).
• We are modeling the subjective interpretation of states—the synthetic user’s personal, conscious experience of “frustration,” “relief,” “anger,” “curiosity,” etc.
Thus, from Damasio’s perspective, what we’re calling a “chain-of-feeling” is, strictly speaking, a “chain of felt or perceived emotional states.” Each entry in that chain is akin to: “I (the user) perceive I am tense or relieved, so I feel frustration or relief.”
Why This Matters. It makes for Synthetic Users that are closer to Organic Users
Because we’re only modeling subjective feelings (not actual physiological changes), we gain a human-like flavor of emotional diversity without needing to simulate the complexity of real biological processes.
Our novel approach pushes the boundaries of UX research, because understanding perceived frustration or perceived confusion is typically more relevant than the underlying hormonal changes that might occur in real humans.
Releated Articles
More articles for you
Teaching Synthetic Users What Real People Actually Think
Synthetic Users without calibration are individually believable, but collectively wrong. The missing piece is calibration, not better models.
The Lie We Tell Ourselves About Customer Research
Most research asks what people say. The problem is people don't do what they say. This piece breaks down the gap between stated and revealed preference — and why behavioral modeling, not better interviews, is how you close it.
Two ways to run research with Synthetic Users and why the difference matters
Iris, what is the difference of using agents to accelerate research.
Synthetic Users vs digital twins
You don’t need a twin for “a parent in rural Ohio who shops weekly at Walmart, prefers fragrance-free, and has a toddler with eczema.” You sample a parent profile with relevant traits and constraints, add retail and dermatology context, and generate behaviors consistent with both.
Two major papers. One shared direction.
LLM-powered Synthetic Users have crossed from concept to validated method. This proves they can predict human behavior accurately, letting teams run fast, low-cost behavioral experiments without replacing real participants.
Gartner says we lead. That's kind of them.
Gartner’s latest report on AI-powered synthetic user research cites Synthetic Users as a leader.
Introducing Shuffle v2
Shuffle v2 is a feature that intelligently shuffles between multiple large language models via a routing agent to produce more realistic, diverse Synthetic Users with better organic parity.
Chain-of-feeling
Synthetic Users use a “chain-of-feeling” approach—combining emotional states with OCEAN personality traits—to produce more human-like, realistic user responses and yield richer UX insights.
Generative Agent Simulations of 1,000 People
A paper that thoroughly executes a parity study between Synthetic and Organic users.
21 Peer reviewed papers that support the Synthetic Users thesis
Here is a compilation of all the papers that help make a case for Synthetic Users.
Why we shuffle between models — to ensure both parity and diversity!
Synthetic Users balances aligned and unaligned models to maintain diversity and authenticity in simulated users while ensuring ethical standards and user expectations are met.
Latest press articles for Synthetic Users
Synthetic Users and AI are transforming research methodologies, offering innovative, cost-effective alternatives to traditional human subject studies.
Comparison studies. The opportunity lies in the deviation.
When we compare different studies, especially looking at what synthetic (artificial interviews) and organic (real-world interviews) data tell us, we often find they mostly talk about the same things but there's also a bit where they don't match up. This gap is super interesting because it's like finding hidden treasure in what we thought we knew versus what we might have missed.
How we deal with bias
Harnessing the power of AI in our Synthetic Users, we strive for a balance between reflecting reality and ethical responsibility, ensuring diversity and fairness while maintaining realism.
The transition to Continuous Insight
The transition towards Continuous Insight™ aligns research activities more closely with the dynamic needs of the business and ensures that product development is continuously informed by up-to-date user insights.
The Art of the Vibes Engine
Large language models (LLMs) like GPT-4 serve as powerful "vibes engines," empathizing with diverse groups and generating contextually relevant content. Their applications span market research, customer support, user experience design, and mental health support, offering invaluable insights and personalized experiences. While not infallible sources of truth, LLMs enable creativity, personalization, and connection within the realm of human language.
There is a faster and more accurate way to do research. Use Synthetic Users.
How Synthetic Users is changing the research process.
The wisdom of the silicon crowd
In the light of an ancient parable, we explore a new paper that dives into how ensembles of large language models match the prediction accuracy of human crowds. It reveals that combining machine predictions with human insights leads to the most robust forecasting results.
Three research papers that helped us build ❤️ Synthetic Users
For the sceptics amongst us who need more tangible research in order to engage with this brave new world. Full disclosure: we are part of the sceptics.
What is RAG and why it’s important for Synthetic Research
Ahead of our RAG launch we explain Retrieval-Augmented Generation (RAG) and how it enhances Synthetic Users by providing increased realism, contextual depth, and adaptive learning, with profound implications for market research, user experience testing, training, education, and innovative product development.
Synthetic Users system architecture (the simplified version).
Foundation models underpin Synthetic Users with advanced capabilities, enhanced by synthetic data and RAG layers for realism and business alignment, all within a collaborative multi-agent framework for richer interactions.
Saturation score. How do we know how many interviews to run?
Determine your interview target for achieving topic saturation using our efficient approach, leveraging the historical wisdom of research pioneers. This method ensures deep insights with theoretical sampling at its core.
How Synthetic Users are gaining depth
Synthetic Users are evolving to address criticism about their generalist nature by incorporating representative data sets and personal narratives.
How we compare interviews to ensure we improve our Synthetic Organic Parity — 85 to 92%
How do we know we are right? How do we know our Synthetic Users are as real as organic users? We compare.
Synthetic Users: Merging Qualitative and Quantitative Research, in seconds.
At Synthetic users we are blurring the lines between qualitative and quantitative research. Here's how we are going about this transformative approach.