Encoding of others' beliefs without overt instruction

Under what conditions do people automatically encode and track the mental states of others? A recent investigation showed that when subjects are instructed to track the location of an object but are not instructed to track a belief about that location in a non-verbal false-belief task, they respond more slowly to questions about an agent's belief, suggesting that belief information was not encoded or tracked automatically [Apperly, I. A., Riggs, K. J., Simpson, A., Samson, D., & Chiavarino, C. (2006). Is belief reasoning automatic? Psychological Science, 17, 841-844]. In the current experiments, we show that if belief probes occur closer in time to the events that signal the content of the agent's false belief, responses to those probes are faster than responses to probes about reality, and as fast as responses to probes about belief when instructed to track them, suggesting (i) beliefs may get encoded automatically in response to certain cues and (ii) that belief information rapidly decays unless it is maintained via 'top-down' instructions.