Memory processes play a major role in many models of decision making. Several fast-and-frugal heuristics assume a sequential search of information in memory (e.g. the take-the-best heuristic). Fast-and-frugal heuristics exploit regularities in the structure of the environment and basic cognitive capacities, such as memory. However, until now only few attempts have been made to relate models of memory and decision making to the structure of information in the environment. The ACT-R architecture provides a quantitative theory about the interplay between the information structure in the environment and the memory system. Based on internet statistics, we use ACT-R to predict people´s recognition and knowledge about objects in the world, as well as the associated retrieval time distributions of respective memories. We show how a corresponding model integrating memory and decision processes within ACT-R allows predictions about the ecological rationality of decision strategies that operate on the accessibility of information in memory.