Outlines of a Computer Model of Motivation

Sure, here's a blog-style summary of the paper "Outlines of a Computer Model of Motivation":

TL;DR

This paper proposes a computational framework for human motivation, viewing it as a system of parallel, cooperating goal-processes. The authors introduce the concept of "will" as a resource allocator between these goals and suggest that mental effort can be understood as computational work. This model aims to provide a foundation for AI systems that can interact intelligently with humans, particularly in clinical applications.

What Problem Does This Paper Address?

Understanding and modeling human motivation is a complex challenge in both psychology and artificial intelligence. While existing theories often rely on abstract concepts like "psychic energy," this paper seeks to develop a more concrete, computational model that can be implemented in AI systems. The authors aim to bridge the gap between psychological theories of motivation and the practical requirements of building intelligent agents capable of interacting with humans, especially in areas like clinical psychology where a nuanced understanding of motivation is crucial. They highlight that motivation is a neglected area in AI and a central problem in personality theory.

Methods

The proposed model is based on an information-processing approach. Key elements include:

  • Parallel Cooperating Goal-Processes: Motivation is seen as a system where multiple goals are pursued concurrently. These processes interact and can interrupt each other.
  • Will as a Resource Allocator: The "will" is conceptualized as the mechanism that allocates finite resources (computational work, time, memory) to these competing goal-processes. This is analogous to resource management in modern computer systems.
  • Wishes vs. Wants: A "wish" is defined as the recognition of a desirable situation, while a "want" implies an intention to allocate resources to achieve that goal.
  • Mental Effort as Computational Work: The subjective feeling of mental effort is linked to the amount of computational work required to execute a process.
  • Active and Reactive Motivation: The model accounts for both proactive behavior (driven by regnant goals) and reactive behavior (triggered by unexpected environmental events).
  • Planning and Intentions: The paper outlines processes for planning actions to achieve goals and forming intentions, which involve evaluating alternatives and selecting a course of action.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Operationalizing "Will": Defining the "will" as a resource allocator provides a concrete, computational mechanism for a concept often treated abstractly.
  • Computational Basis for Mental Effort: Linking mental effort to computational work offers a quantifiable way to understand this subjective experience.
  • Distinction Between Wishes and Wants: This distinction clarifies the difference between passive desire and active intention, crucial for modeling goal-directed behavior.
  • Integration of Active and Reactive Behavior: Acknowledging both proactive and reactive aspects of motivation provides a more complete picture of how agents interact with their environment.
  • Motivation as a Foundation for AI Interaction: The authors argue that a robust model of motivation is essential for AI systems to engage meaningfully with humans, especially in sensitive applications like therapy.

Key Results

This paper is primarily theoretical and outlines a framework rather than presenting empirical results. The "results" are the conceptual framework and the proposed computational model itself. The authors demonstrate how their model can conceptually handle aspects like planning, intention formation, and the distinction between active and reactive behavior.

Limitations and Future Work

The authors acknowledge that this is an early-stage project and that the ultimate aim is to implement these concepts as computer programs. They state that the detailed implementation of the planning process and other aspects will require further development. Future work will involve experimentally implementing these ideas over the next few years, with a focus on developing a comprehensive framework. They also note that the analogy of parallel processes in computer science becomes inaccurate when considering the high degree of communication and control sharing assumed in human motivation.

Summary updated

BibTeX
@inproceedings{1973outlines,
  title = {{Outlines of a Computer Model of Motivation}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {1973},
}
Figure