Using a laboratory experiment, we investigate whether contagion can emerge between two risky assets despite an absence of correlation in their fundamentals. To guide our experimental design, we use the ‘Two trees’ asset pricing model developed by Cochrane, Longstaff and Santa-Clara (2007). We draw on the model to make predictions regarding changes in the time-series and cross-section of returns in response to fundamental value shocks. We observe positive autocorrelation in the shocked asset, and a positive contemporaneous correlation between assets, as the model predicts. The dividend-price ratio forecasts the returns of risky assets, both in the time series and in the cross-section. There is more support for the model’s predictions in markets in which traders have greater cognitive ability.