Looking beyond the boundaries: time to put landmarks back on the cognitive map?

Since the proposal of Tolman (1948) that mammals form maplike representations of familiar environments, cognitive map theory has been at the core of debates on the fundamental mechanisms of animal learning and memory. Traditional formulations of cognitive map theory emphasize relations between landmarks and between landmarks and goal locations as the basis of the map. More recently, several models of spatial coding have taken the boundaries of an environment as the basis of the cognitive map, with landmark relations being processed through alternative, operant learning mechanisms. In this review, the evidence for this proposed dichotomy is analyzed. It is suggested that 2 factors repeatedly confound efforts to compare spatial coding based on landmark arrays, formed by 2 or more landmarks, and that based on the boundaries of an environment. The factors are the perceived stability of the landmark arrays and their placement relative to the larger environment. Although the effects of landmark stability and of placement on spatial navigation have been studied extensively, the implications of this work for debates concerning the role of boundaries in cognitive map formation have not been fully realized. It is argued that when these 2 factors are equated between landmark arrays and bounded environments, current evidence supports a commonality of spatial coding mechanism rather than a dichotomy. The analysis places further doubt on the existence of a dedicated geometric module for reorientation and is consistent with models of navigation containing mapping and operant learning components, both taking as input local views (Sheynikhovich et al., 2009).