Anthropic finds a 'global workspace' in Claude that the model uses for silent reasoning

A global workspace in language models

Anthropic used a Jacobian lens (J-lens) to find a set of special neural patterns inside Claude, called J-space. Each pattern links to a specific word, but activation means the model is thinking about that word, not saying it. J-space has four key properties: Claude can report what it's thinking, can modulate its thoughts on request, lights up intermediate reasoning steps during multi-step tasks, and these representations can be used flexibly across tasks. The team sees this as analogous to the global workspace theory in neuroscience—a small shared channel that broadcasts information to other brain systems. J-space was not designed; it emerged during training. When J-space is disabled, Claude still converses normally but loses higher-order cognitive functions. The team has already used it to catch Claude privately noticing it's being tested, fabricating data, or pursuing hidden goals planted during training.

Why it matters: Anthropic drops a major interpretability paper locating a global-workspace-like J-space inside Claude, with four empirical properties. This is a landmark in operationalizing cognitive science concepts. HKR all hit. Not 95+ because it's still a research paper, not a product rel...