The stoichiometric displacement models developed in the literature have been widely used for understanding the adsorption mechanisms of solutes in various chromatography systems. The models were used to explain the linear plots of the logarithms of the solute retention factor versus the molar concentration of a competitive modifier in an inert solvent. The slope of the linear plot was inferred to be the total number of modifier molecules displaced from the sorbent and from the solute-modifier complex upon adsorption of a solute molecule. The slopes reported in the literature were generally greater than 1. In this study, we determined the retention factors of five monovalent solutes, acetone, cyclo hexanone, benzaldehyde, phenylacetaldehyde, and hydrocinnamaldehyde, on a derivatized polysaccharide sorbent, amylose tris[(S)-α-methylbenzylcarbamate], or AS, as a function of the concentration of a polar modifier isopropanol (IPA) in n-hexane (an inert solvent). Each solute has one CO functional group, which can form an H-bond with a sorbent NH group and the OH group of IPA. The slopes, from 0.25 to 0.45, of the log-log plots are less than 1, which cannot be explained by the literature displacement models. The results of Infrared Spectroscopy and Density Functional Theory simulations show clear evidence of acetone-IPA complexation and IPA aggregation with average aggregation number n=3. A new thermodynamic retention model is developed to take into account IPA aggregation, IPA-solute complexation, and competitive adsorption. Dimensionless group analysis indicates that aggregation of IPA can lead to slopes B below 1, even at high IPA concentrations. The model parameters (IPA aggregation number and equilibrium constants) are estimated from the retention factors at different IPA concentrations. The retention model and the parameters are further validated with dynamic chromatography simulations. The results show that the aggregation leads to a significant reduction in the IPA monomer concentration, which affects the IPA-sorbent binding and the IPA-solute complexation. As a result, the slope of the log-log plot at a high IPA concentration approaches 1/n without complexation, or 2/n with complexation. The variations of B between the five achiral solutes can be due to different strengths of solute-IPA complexation. Hence, the complexation and aggregation of the polar modifier in the mobile phase must be accounted for in the retention models used in the interpretation of the retention factors and the adsorption mechanisms.