A sunk cost is money spent on a project that has not provided the desired outcome. For example, a pharmaceutical company spends $25 million in R&D on drug #1. The outcome is not what the company was hoping for. The drug does not cure a particular disease and has negative side effects. The $25 million is already spent and there is no chance of recouping that investment.
The company can spend $10 million more pursuing drug #1 or put the money towards a higher probability outcome for drug #2. The fact that $25 million has already been spent should not factor into the decision. If the company decides to continue with drug #1, in hopes of recouping some of its loss, it will have engaged in what economists call loss aversion.