This paper explores a novel approach for developing artificial general intelligence (AGI), called Synthetic Cognition. We point out that the existing Transformer architecture is the state-of-the-art in generating context-aware responsive actions but lacks inference capabilities, and thus we conduct research on developing immediate responsive actions using Synthetic Cognition. In particular, we add a sequence processing mechanism to the existing implementation of Synthetic Cognition and apply it to a DNA sequence classification task, and conduct comparative experiments with DNA-based foundation models. Our experimental results show that the proposed method outperforms the DNA foundation models and achieves state-of-the-art scores on several benchmark tasks, thereby claiming that it extends Synthetic Cognition's sequence processing and outperforms the Transformer architecture in sequence classification tasks.
