The significant importance of GABAA receptors in the treatment of central nervous system (CNS) disorders has been known for a long time. However, only in recent years have experimental protein structures been published that can open the door to understanding protein-ligand interactions and may effectively help the rational drug design for the future. In our previous work (Szabó, G. J. Med. Chem. 2022, 65(11), 7876), where a promising selective α5-GABAA negative allosteric modulator (NAM) was developed containing the 3-(4-fluorophenyl)-5-methyl-1,2-oxazole headgroup, we noticed a switch-like effect of a single nitrogen atom for the receptor function in some derivatives having a dihydro-naphthyridinone or dihydro-isoquinolinone moiety. Here, we focused on this chemotype, and a small set of compounds were designed to investigate ligand-receptor interactions experimentally and through computational methods. Elaborated compounds were tested against GABAA α1 and α5 subunit-containing receptors, and binding affinities and functional activities were measured. Starting from the published experimental structure of an engineered, homopentameric, basmisanil-binding GABAA receptor-like construct consisting of modified α5 subunits and an α1-containing GABAA structure, we created a new model of the ligand binding site at the α5/γ2 interface. Using this model, the measured ligand affinities were able to be reproduced well by free energy perturbation (FEP) calculations. In addition, calculations were able to explain the obtained structure-activity relationships, among others, the switch-like effect of the aromatic nitrogen position in the dihydro-naphthyridinone motif for the functional character, and suggest different binding poses for the ligands presenting silent versus negative allosteric effects in this set (SAMs vs. NAMs, respectively). We believe that our results can help design α5 selective GABAA negative allosteric modulators and better understand the GABAA receptor.