Generic properties of a computational task predict human effort and performance

It has been shown that computational hardness of cognitive tasks affects people’s effort and ability to solve problems reliably. However, prior empirical studies lack generality. They quantify based on particular algorithms or for specific problems. Here, we propose a set measures individual instances task in way is independent any algorithm model can be generalized other Specifically, introduce two measures, typical-case complexity (TCC), measure average random ensemble instances, instance (IC), an instance-specific metric. Both are related structural properties instances. We then test the effect those human behavior by asking participants variants 0-1 knapsack problem, canonical ubiquitous NP-hard problem. find spent more time with higher TCC IC, but decision quality was lower study mathematical contribute development computationally plausible accounts decision-making, just like stochastic have proven critical our understanding decisions probabilistic tasks. • Computational varies substantially. Our metrics task-independent.