Learning and memory studies have been performed for more than two decades using the crab Chasmagnathus in our laboratory. Here, our research was aimed at disclosing some instances of learning in field conditions. Three experiments were performed non-simultaneously, all with a 22.5-min pre-training preceding the first visual-danger-stimulus, an opaque rectangle passing overhead. In Experiment 1, crabs received a single stimulus followed by 22.5-min testing without stimulation, where the re-emerging latency was considered the basic latency response. In Experiment 2, training consisted of 15 stimulus 3-min apart, followed by 22.5-min testing without stimulation. Throughout training crabs were underground but re-emerged at testing with latencies longer than the basic latency response. Both at pre-training and testing the usual strategy of exploring was the short-near excursions. In Experiment 3, training included three stimulus 22.5-min apart, followed by 22.5-min testing. Crabs left their burrows before the end of each inter-trial, showing a mean latency like the basic latency response, but a sensitization to the stimulus and a preponderance of the fast-far excursions over the usual slow-near. In brief: through 15-3 training, crabs learn that the stimulus is iteratively presented; through 3-22.5 training, crabs acquire sensitization to the stimulus and a different strategy of exploration.