One area of real omission in attitude research is how attitudes towards novel items may be acquired as a result of one's experience with them. Studies conducted using the B~anFest .paradigm (Fazio, Eiser & Shook, 2004; Fazio, Eiser, Stafford & Prescott, 2003) have begun to shed light upon these processes, however this particular paradigm is restricted in the dimensions that it can research. The Treasure Island paradigm was developed to build upon the discoveries of BeanFest and to consider the process of attitude formation toward spatial location within a more realistic environment. Within the paradigm, participants enter a computerised environment (i.e. island) that they must search for resources. Search responses can yield either positive (i.e. treasure) or negative (i.e. pirates) outcomes dependent on location. Participants must then recall and make predictions about the locations of the treasure and pirates. The paradigm enables the study of how direct valenced experience impacts upon future sampling decisions within the environment, how readily and accurately people form evaluative associations towards different regions of the environment (i.e. spatial attitudes) and whether experiences at visited locations might generalise to those that have not been visited (i.e. generalisation). This thesis outlines the development of the paradigm and considers participants' performance within versions of the game aimed at researching the impacts of two manipulations of environmental predictability (i.e. valence inconsistency and resource depletion). With respect to the valence consistency manipulation, it was found that participants in the consistent condition showed more accurate spatial attitude formation compared to participants in the inconsistent condition. Further analysis of the responses in the inconsistent condition revealed that whilst some participants had managed to form relatively accurate regional attitudes, others were employing a more simplistic 'west is best' heuristic to assist their sampling and recall. With respect to the resource depletion manipulation, it . was discovered that, regardless of differential beliefs about the availability of resources (i.e. treasure) within the game (i.e. limited vs unlimited), participants were able to form similarly accurate spatial attitudes. This research offers pioneering insight into spatial attitude formation within novel environments in which feedback (and learning) is contingent upon personal exploration.