Olton ( 1978) model viewed spatial memory in the rat as a limited number of slots, with information about the cues seen at the end of each arm on the radial maze stored in each slot. When a trial had ended (all rewards had been collected), the list of cues could be flushed, by resetting working memory. My review of issues raised by this model suggests that it may not accurately depict the mechanisms of spatial memory in animals. While it cannot be proved that spatial memory does not have a limited capacity, recent discoveries of good spatial memory for an increasingly larger number of places indicates that the discovery of such a limit may not be in the offing. The notion that rats reset working memory at the end of a trial seems unlikely now, since recent findings indicate that rats are susceptible to proactive interference within a sequence of massed trials and that memories of final choices on trial n - 1 lead to avoidance of those choices on the initial selections of trial n. Finally, a variety of evidence seems to have cumulated, which suggests that spatial memory in animals has the characteristics of a map and not those of a list.
While no formal alternative theory of spatial memory will be presented here, some features of an alternative theory can be mentioned. Such a theory might hold that animals form map-like representations of familiar environments and that these representations ore hold in a long-term or reference memory ( Roithlat, 1982). Places visited within such a mapped environment would be noted or marked on the map as a recently visited place. There would be no limit on the number of places on the map that could be so marked; however, performance may begin to suffer when nearby places are marked as visited, since spatial discrimination of similar areas may become difficult.
Proactive and retroactive interference were found to arise under similar circumstances, when rats were required to make trips to identical points in space. In proactive interference experiments, rats showed forgetting when required to return to alloys entered on immediately preceding trials, and retroactive interference was found when rats were placed directly on the correct alloys before a retention test. When mazes were placed a short distance apart, no retroactive interference was found. If rats record a trip to an arm as a marker placed on that part of the map, exploratory tendencies and past learning may lead the animal to avoid returning to that place. However, when repeated trips are made to the same places, more than one marker will be placed at the some position on the reference map. Coding along a temporal dimension would seem to be required here in order to deal with repeated visits ( Dale & Staddon, 1982).
In a proactive interference experiment, the rat must be able to discriminate recent markers from earlier ones. This becomes a problem in temporal discrimination, and if a rat confuses entry markers on trial n and trial n - 1, unentered alloys will be avoided and entered alloys will be reentered. Similarly, in the RI experiment, direct placements may establish entry markers which are difficult to discriminate temporally from those established by the initial forced entrances. It should be emphasized that rats performed well above chance level in both proactive interference and retroactive interference experiments. Temporal discrimination between