Cognitive Language Learning with Molecular Associative Memory

Ji-Hoon Lee and Byoung-Tak Zhang


In recent years the rising emphasis on brain-like computing has made many interdisciplinary studies in science. However, the attempt to design a model of cognitive language learning using biomolecules is a challenge we still face today. Here, we investigate the cognitive language learning model that uses a molecular parallel associative memory. The specifically designed DNA molecules were used to recognize natural language sentences from a molecular learning process. This process is the execution of wet-lab experiments in molecular biology laboratory protocol that involves DNA computing operations such as hybridization, selection and amplification. Each hybridized molecules makes three-dimensional loops which indicate the distance between the DNA sequences. The most matched DNA sentences were selected and amplified for making a more efficient molecular associative memory. We used 24 English sentences and random DNA sentences for the pilot wet-lab experiment. From the results, the each learning step was verified and thus support that our model can be used for cognitive learning and memory.