Social media platforms today generate vast amounts of user-generated data, leading to increased interest in cognitive-based sentiment analysis, particularly in the automated identification of user behaviors such as personality traits through social media text. This research utilizes the Myers-Briggs personality type (MBTI) dataset available on Kaggle, which classifies human personalities into 16 distinct types based on four key dimensions: sensing versus intuition, introversion versus extraversion, thinking versus feeling, and judging versus perceiving. Our research involves applying various machine learning algorithms to the MBTI dataset and performing a comparative analysis with our proposed model, which combines a CNN + BiLSTM architecture with a CPSO optimizer. The CPSO optimizer, inspired by animal social behavior, is designed to optimize nonlinear continuous functions, based on swarm intelligence seen in groups such as flocks and shoals. Compared to other advanced methods, the CNN + BiLSTM model with the CPSO optimizer demonstrated superior performance, achieving 93% precision, 93% recall, 93.85% accuracy, and an F1-score of 89.99%.