PURPOSE: To provide an associative module learning method and device capable of properly learning a new pattern to be stored in an associative neural net of a module structure.

CONSTITUTION: An associative module learning device is prepared in an associative neural net consisting of plural modules that can change the threshold value. Then this module learning device is provided with an associative executing means 5 which obtains the final output of each associative module that is acquired through the convergence calculation of a binary pattern image to be inputted to each associative module, an output sorting means 7 which calculates the similarity between the final output acquired by the means 5 and the supplied binary image pattern, and a storage control means 9 which identifies the associative module that controls the storage of the supplied image pattern based on the calculation result of the means 7 and performs the storage control of the supplied image pattern for the associative module.