What’s at the center of Claude’s mind?

Name: What’s at the center of Claude’s mind?
Uploaded: 2026-07-06T17:33:34.191Z
Duration: 5 min 27 s
Channel: Anthropic
Description: Anthropic investigates Claude's internal representations by applying global workspace theory, a neuroscience theory about conscious accessibility. They identify mechanisms in Claude's neural activity comparable to how the human brain processes information.

Anthropic6 July 2026Watch on YouTube

Description

Out of everything happening in your brain right now, only a tiny fraction is consciously accessible — thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside our AI model, Claude. Our experiments were inspired by a leading theory in neuroscience: the global workspace theory. It holds that a thought becomes consciously accessible when it enters a shared "workspace" that's broadcast across the brain. We found a set of representations in Claude’s neural activity that play a similar role. Read more about the research here: http://www.anthropic.com/research/global-workspace

What you'll learn

Research reveals Claude contains internal representations similar to the 'global workspace' concept from neuroscience theory
Like the human brain, Claude processes information where only a small portion is consciously accessible
Anthropic identified mechanisms in Claude's neural activity that mirror how the brain processes information

Frequently asked questions

What is global workspace theory and how is it relevant to AI?

Global workspace theory is a neuroscience theory describing how a thought becomes consciously accessible when it enters a shared 'workspace' broadcast across the brain. Anthropic applied this theory to Claude's neural activity to understand how the AI processes information.

What similarity did Anthropic discover between Claude and the human brain?

Anthropic found that Claude, like the human brain, has a division where only a small fraction of information is consciously accessible. They identified representations in Claude's neural activity that play the same role as the global workspace in the brain.

How does this research help in understanding AI models?

This research contributes to mechanistic understanding of how AI models process information and which internal structures determine which thoughts are 'accessible'. This type of insight can improve AI interpretability.

Topics

ai-interpretability research cognitive-science claude mechanistic-understanding

In this video

Claude

Anthropic