New Anthropic research: Signs of introspection in LLMs. Can language models recognize their own internal thoughts? Or do they just make up plausible answers when asked about them? We found evidence for genuine—though limited—introspective capabilities in Claude.

We developed a method to distinguish true introspection from made-up answers: inject known concepts into a model's “brain,” then see how these injections affect the model’s self-reported internal states. Read the post: anthropic.com/research/intro…

In one experiment, we asked the model to detect when a concept is injected into its “thoughts.” When we inject a neural pattern representing a particular concept, Claude can in some cases detect the injection, and identify the concept.

However, it doesn’t always work. In fact, most of the time, models fail to exhibit awareness of injected concepts, even when they are clearly influenced by the injection.

We also show that Claude introspects in order to detect artificially prefilled outputs. Normally, Claude apologizes for such outputs. But if we retroactively inject a matching concept into its prior activations, we can fool Claude into thinking the output was intentional.

This reveals a mechanism that checks consistency between intention and execution. The model appears to compare "what did I plan to say?" against "what actually came out?"—a form of introspective monitoring happening in natural circumstances.

We also found evidence for cognitive control, where models deliberately "think about" something. For instance, when we instruct a model to think about "aquariums” in an unrelated context, we measure higher aquarium-related neural activity than if we instruct it not to.

In general, Claude Opus 4 and 4.1, the most capable models we tested, performed best in our tests of introspection (this research was done before Sonnet 4.5). Results are shown below for the initial “injected thought” experiment.