Experiments were carried out to test the ability of L. pholis to learn and remember the position of a food reward in a hierarchical maze during daily trials. This task was completed in 9 days in the presence of LegoTM landmarks. Learning did not take place in the presence of a white screen clue or in the absence of any obvious visual clues after 15 days. An attempt was made to identify the effect of a change in conditions on L. pholis who had already learned to successfully navigate the hierarchical maze, and to identify the clues utilised in learning, by altering the intra and extramaze clues. Geomagnetic, olfactory / gustatory and current direction clues were eliminated as the source of spatial information, as was the use of a cognitive map (Tolman 1948, O'Keefe & Nadel 1978). Rather, evidence suggested that the fish were learning a specific route through the hierarchical maze using the LegoTM towers as beacons, and as a prompt as to which direction to turn at important stages in the journey. This memory was retained for a period of at least 30 days. A hierarchy of clue use was suggested by the fact that the subjects used the direction of entry to the experimental arena, or the direction by which the experimenter left the arena just prior to a daily test, as a directional clue to the position of the reward box in the absence of Lego towers. These experiments suggested that in the wild L. pholis uses the position of local landmarks, such as rocks and clumps of algae, to direct movements towards feeding patches. A study of behaviour of L. pholis placed in a novel artificial habitat was also carried out. L. pholis moved along the edges of the objects placed in the arena, followed regularly used paths between refuges, and explored the arena from a series of "base" refuges. In the more active individuals, each refuge was investigated until the subject took up residence in a preferred refuge. This adopted shelter was often centrally placed and commanded a good view of a large area of the arena. Activity was concentrated in the more complex half of the arena and experienced fish directed their movement towards this area 24 hours after 6 hours exploration of the novel habitat. There was also evidence to suggest that certain individuals learned the position of a specific preferred refuge after the 6 hour exploratory period. Finally, the ability of L. pholis to remember the position of a refuge was tested in an artificial habitat under the influence of different clues. L. pholis learned the position of the refuge in the presence of an A4 sized black screen clue only. They responded to this clue by moving towards it and pressing themselves up against it while LegoTM towers and a white screen clue did not provoke such a response. L. pholis continued to respond to the black screen in this way even when it was moved to another location further from the refuge. After 12 days L. pholis learned to use the black screen as an indirect clue and navigate to the refuge directly without first touching it. These results suggested that when placed in a novel habitat the immediate reaction of L. pholis is to move quickly towards the first dark area they see. Later, they systematically explore all available shelters and choose a preferred one according to complexity of the surrounding habitat. With experience, they can use the position of objects around them to navigate quickly and efficiently to their preferred refuge or the nearest suitable refuge depending on the severity of the threat.