Resumen

The standard assumption in social learning environments is that agents can only learn from others through choice outcomes. We argue that in many settings, agents can also infer information from others’ response times (RT). This can occur any time there is information about the rate of transactions. If RTs reveal agents’ private beliefs in a reliable way, this can increase the provision of information. To investigate this, we conduct a standard information-cascade experiment where subjects make publicly observable decisions. We find that RTs contain information that is not contained in choice outcomes and we identify a simple log-linear relationship between beliefs and RTs. Moreover, in two conditions we manipulate subjects’ ability to observe others’ RTs and find that subjects do incorporate information from others’ RTs into their decisions. Our results suggest that in environments where RTs are publicly available, the information structure may be richer than previously thought.