Epley's 3 Factors of Anthropomorphism

Hello! Welcome to the final lesson of our first module, "The Social Machine: Foundational Concepts."

In our previous lessons, we've built a strong foundation for understanding how humans perceive and interact with AI. We started with the "ELIZA effect," moved to Dennett's "Intentional Stance," explored the "Computers Are Social Actors" paradigm, and most recently, delved into Sherry Turkle's concepts of "evocative objects" and "relational artifacts." Turkle's work showed us that we form emotional, relational bonds with technology, especially objects designed to engage us.

Today, we address the fundamental question that follows from Turkle's work: Why are we so prone to do this? What are the underlying psychological mechanisms that cause us to see a mind in a machine or attribute human qualities to an inanimate object?

To answer this, we will explore a key theory from social psychology. By the end of this lesson, you will be able to:

Explain Epley's three-factor theory of anthropomorphism (elicited agent knowledge, effectance, and sociality) as a model for why people humanize non-human agents.

This theory, often called the SEEK model, provides a powerful and empirically tested framework for understanding the triggers behind our tendency to anthropomorphize, bringing together many of the concepts we've touched upon in this module.

1. What is Anthropomorphism?

Before we dive into the theory, let's clarify our central term. Anthropomorphism is the attribution of human-like characteristics, motivations, intentions, or emotions to non-human agents. This can be anything from seeing a "face" in the front of a car, to believing your pet understands your feelings, to getting angry at your laptop for "being stubborn."

While it might seem like a simple quirk, psychologists Nicholas Epley, Adam Waytz, and John Cacioppo proposed that anthropomorphism is not random. Instead, it's a predictable psychological process driven by specific motivations and cognitive shortcuts. They organized these drivers into a comprehensive three-factor theory.

2. The Three-Factor (SEEK) Model of Anthropomorphism

Epley's theory is often referred to as the SEEK model, an acronym for the three core psychological factors that increase our tendency to anthropomorphize:

1. Sociality Motivation
2. Effectance Motivation
3. Elicited agent Knowledge

These three factors are not about the agent itself (e.g., how advanced a robot is) but about the psychological state of the human perceiver. They explain why the same person might anthropomorphize an object at one time but not another, or why different people might treat the same AI very differently.

To get a clear, example-driven overview of each factor, please read the following excerpt from a recent research paper that uses the SEEK model as its foundation.

3. Deconstructing the Three Factors

Now that you've read the overview, let's break down each factor and connect it to the world of AI.

Factor 1: Sociality Motivation

• Core Idea: The fundamental human need for social connection. When we feel lonely or socially disconnected, we are more motivated to find sources of connection in our environment. If human companions aren't available, we may "create" them by anthropomorphizing non-human agents.
• Mechanism: Loneliness acts like a trigger, increasing our vigilance for social signals and lowering the bar for what we consider a "social agent."
• AI Context: This is perhaps the most powerful driver in the context of relational AI. A person experiencing loneliness is far more likely to perceive their AI companion (like Replika or a social robot) as a genuine friend with thoughts and feelings. As the reading noted, Epley's research found that lonely individuals were more likely to attribute human-like characteristics to gadgets like "Clocky," the alarm clock that runs away. This motivation explains the powerful draw of AI companions for those seeking to alleviate social isolation.

Factor 2: Effectance Motivation

• Core Idea: The need to understand, predict, and control our environment. We want to make sense of the world around us. When we encounter an agent whose behavior is unpredictable or unexplainable, we have a strong drive to reduce that uncertainty.
• Mechanism: Attributing human-like intentions ("It's trying to trick me") is a quick and easy way to explain complex or unpredictable behavior. It makes the agent's actions seem less random and more predictable, thus giving us a sense of understanding and control.
• AI Context: Think about a time your computer froze or a smart speaker gave a bizarre response. A common reaction is to curse at it as if it were a person being intentionally difficult ("Why won't you listen to me?!"). This is effectance motivation in action. By attributing intention, we create a simple (though incorrect) explanation for a complex technical failure. As the reading mentioned, this is why people might anthropomorphize unpredictable weather or, in studies, unpredictable robots.

Factor 3: Elicited Agent Knowledge

• Core Idea: We use what we know best—ourselves and other humans—as a default model for making sense of other entities. The more an object makes our "human" knowledge seem relevant, the more we'll apply it.
• Mechanism: This is a cognitive shortcut. Thinking about how a machine works requires effort and specialized knowledge. Thinking about how a person works is intuitive for us. If an agent looks or acts even vaguely human-like, our brain finds it easier to apply its "human model" to it.
• AI Context: This is where design comes in. An AI with a human-like avatar, a name, a voice, and the ability to use conversational language (like "I think..." or "I feel...") is explicitly designed to elicit our agent knowledge. These features serve as cues that invite us to treat the AI as a person. As the reading notes, even simple features like eyes and a mouth on a robot can trigger this process. This factor explains why the design choices we discussed in the CASA lesson are so effective.

4. The SEEK Model in Action

It's important to remember that these three factors often work in concert. Imagine a user who is feeling lonely after moving to a new city (Sociality). They buy a new, advanced AI assistant that sometimes behaves in unexpected ways (Effectance). The assistant has a friendly female voice and a glowing light that pulses when it "listens" (Elicited Knowledge).

This scenario creates the perfect storm for anthropomorphism. The user is motivated to find a friend, motivated to understand the device's quirks, and is given cues that make it easy to think of the device as a "she." The SEEK model predicts that this user is highly likely to form a strong, human-like perception of their AI assistant.

This theory isn't just an academic curiosity; it's a foundational model used in current Human-Computer Interaction (HCI) and Human-Robot Interaction (HRI) research to design and study our interactions with technology, as demonstrated by the very papers we've referenced today.

Conclusion

Today, we've answered the "why" behind our tendency to treat machines like people. By understanding Epley's three-factor theory, we can see that anthropomorphism isn't just a random error but a predictable outcome of our own deep-seated psychological needs.

Key Takeaways:

• Anthropomorphism is the attribution of human characteristics to non-human agents.
• The SEEK Model explains this tendency through three main psychological drivers:
  • Sociality Motivation: The need for social connection, which is heightened by loneliness.
  • Effectance Motivation: The need to understand and control our environment, which is triggered by unpredictability.
  • Elicited Agent Knowledge: The cognitive shortcut of applying our knowledge about humans to agents that appear or act human-like.
• These factors are about the perceiver's psychological state, not just the object's features, and they often work together to encourage a human-like interpretation of an AI.

This lesson concludes our first module, where we've laid the groundwork by exploring the foundational psychological and sociological concepts that govern our initial perceptions of AI.

Preview of the Next Module:

We've now seen that we treat computers as social actors (CASA), we've seen the kinds of relationships we form with them (Turkle), and we've explored why we're psychologically primed to do so (Epley).

In our next module, "Designing the Relationship: Psychology and Interaction," we will build directly on these ideas. We'll start by taking the concept of social connection a step further and exploring how one of the most important theories of human relationships—Attachment Theory—is now being used to understand the bonds people form with AI companions.