The vertebrate self-organizes feature detcting neural cells in the neural net through an adaption to his environment. The feature detecting cells play an important role in neural information processing. Therefore, it seems to us that the feature detecting cells take part in the brain memory in some form or other. In the paper, we give an adaptive associative memory model based on the formation of the feature detecting neural cells. We explain the association mechanism of the model and consider the ability of the recall through computer-simulated experiments. As the result, it is shown that the model works well under the general environment.