<p>Recent advances in encryption have increasingly used neural networks and DNA-based models to strengthen security through complex computational frameworks. Chaos-based encryptions, in particular, have gained prominence for their ability to enhance confusion and resist cryptanalytic attacks. However, the effectiveness of these techniques largely depends on the quality of the chaotic systems with limited dependence on input variations. A lack of dynamic sensitivity to changes in input can undermine security, even when standard cryptographic tests are passed. Existing methods tend to suffer from limitations such as overreliance on chaotic signal integrity, short memory in propagating input changes, localized impact at the block level, and static computational behavior. This paper proposes a novel encryption technique that addresses these issues by integrating concepts from color theory in physics and cognitive memory loss models. The approach uses dynamic color space transformations and memory-aware operations to track input variations and adapt the encryption process accordingly. Experimental results on both medical and standard images demonstrate that the proposed method achieves superior performance compared to current state-of-the-art techniques, with near-ideal entropy (7.9998), minimal correlation (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(4\times 10^{-5}\)</EquationSource> </InlineEquation>), and near-optimal NPCR and UACI values (99.6083 and 33.4611, respectively), while also providing high adaptability and computational efficiency.</p>