Relational reasoning involves evaluating relations between representations. Spatial relational reasoning problems have long been used in psychology to study deductive inference abilities. Research incorporating such tasks has resulted in findings of a variety of effects and reasons as to why some problems are more difficult than others.Computational accounts of the relational reasoning of spatial information offer valuable insights, such as how individuals may construct a mental model to infer conclusions and why some cognitive strategies might be preferred over others. However, many of these accounts either fail to incorporate or remain general to the impact of other factors affecting the difficulty of these reasoning problems, such as the effects of working memory errors. This thesis aims to investigate how the different types of memory errors of omission and commission may be computationally modelled to provide a theoretically unifying account of qualitatively different cognitive reasoning processes and quantitatively different accuracy rates on experimental measurements of spatial relational reasoning problems. Our models demonstrate that modelling memory errors of omission and commission in Python ACT-R based on PRISM theory produces a similar negative relationship of a decrease in accuracy rates with each increase of premise and dimensionality complexity per question as found in the relational reasoning experimental literature.Our results highlight the need for future modelling to consider individual differences in participant micro-strategy preferences, how reasoning processes may be affected by different memory errors, and how future measures may be constructed to better address raised concerns.