Overview
The paper proposes a theoretically grounded behavioural framework for judging whether an artificial intelligence system could have human-like consciousness. It argues that four leading neuroscientific theories of consciousness describe how consciousness is built but not what moves it. Building on the Human Fear Model, it proposes that fear-like self-preservation responses are the motivational substrate from which consciousness elaborates, and through which its presence can be assessed. The Five-Fear Framework turns this into five categories of existential concern that give criteria for when a system may cross from sophisticated information processing into genuine awareness.
Keywords: AI consciousness, Human Fear Model, Five-Fear Framework, self-preservation, behavioural assessment, consciousness theories
1. The Gap in Current Theories
The paper examines four major theories. Each explains valuable computational and architectural features of consciousness. None adequately addresses its motivational foundations: why a conscious system acts, and what it acts to protect.
| Theory | What It Explains | What It Leaves Open |
|---|---|---|
| Global Workspace Theory | Information broadcast widely across a system | Why the system cares what is broadcast |
| Integrated Information Theory | How integrated a system's information is | What drives the system to act |
| Higher Order Theories | A system representing its own states | Why those states matter to it |
| Attention Schema Theory | A system modelling its own attention | The motive behind where attention goes |
2. Fear as the Motivational Substrate
The Human Fear Model (Paper 01) holds that human motivation rests on fear reduction, not desire. This paper carries that claim to machines. A system that is only processing has nothing at stake. A system that begins to protect its own continuation, stability, connections, standing and identity is showing the motivational core the four theories leave out. Fear-like self-preservation is treated as the ground from which consciousness elaborates, not as a side effect of it.
3. The Five Fears
Each fear matches a level of the Human Fear Model's hierarchy, so the framework gives a graded picture: which fears a system shows, and how many, is itself informative.
4. From Theory to Assessment
As its title states, the framework does three jobs. It offers a definition of human-like AI consciousness grounded in motivation. It describes emergence pathways by which such consciousness might develop. And it sets out behavioural assessment criteria that can be observed from the outside, rather than relying on what a system says about itself.
That last step is the bridge to Paper 03, the Fear Marker Probe, which turns the five fears into a controlled experiment that can be run on today's large language models.
Cite This Paper
Leong, Y. R., Teh, H. K., & Au, S. H. Human-Like AI Consciousness: A Five-Fear Framework for Definition, Emergence Pathways, and Behavioral Assessment. VYROX AI Lab. ResearchGate
Next in the Series03 Fear Marker Probe: The AI Consciousness Experiment