Understanding the mechanistic drivers behind animal movement is important for advancing ecological theory as well as informing effective management protocols. Recent advances in animal tracking and computational technology have allowed for ecologists to do so in ways that were not previously possible. This has allowed for ecologists to analyze how spatial memory, the cognitive process of remembering the spatial position of specific landmarks in one’s environment, contributes to observed animal movement patterns using animal telemetry data. This subfield of movement ecology has received increasing attention in recent years, but there is still much room for additional research due to the lack of time-dependent memory processes modeled in the literature. Many animals live in environments where resources are temporally heterogeneous, so it may be beneficial to return to food patches at planned and consistent “time lags”. We develop a modeling structure for estimating these time lags in animals and apply it to a population of Arctic grizzly bears (Ursus arctos), which live in a temporally variable habitat with sparse resources. The grizzly bear population displayed a wide array of individual variation concerning their movement behavior, with some bears showing evidence of this complex time-dependent spatial memory and others showing memoryless movement. The bears that did appear to use memory seemed to return to patches of high resource quality approximately one year after their previous visit in most cases, suggesting some sort of annual rhythm in these animals. Our modelling framework has advanced our knowledge of animal cognition and foraging in harsh, Arctic environments, and is applicable to a wide range of other animal taxa.