Local, error-driven and associative, biologically realistic algorithm (LEABRA) is a widely used framework to design neurocomputational models for cognitive processes. The complex structure of brain layers and interconnected neuronal units form a pattern to store specific information. In an object the information content is high at edges, corners and angles formed in between two planes. It is quoted in various research journals that the neuronal weight computation is based the high information content parts than the less variation in colour in the image. In this work we have proposed a neurocomputational model to store and retrieve the information of an object. After training the model is tested on various similar objects and it can recognise the object with some error. The model can also recognise the objects having similar in terms of number of sides and number of angles.