Spatial Patterns and Memory for Locations

Article excerpt

Rats obtained food from the tops of vertical poles in a 5 × 5 matrix of locations. On each trial, the baited locations formed one of the two possible exemplars of a checkerboard spatial pattern. During training, locations that had been visited earlier in the trial were indicated by a visual cue. Following training, performance with and without the visual cues was compared. Spatial choices were controlled by the checkerboard spatial pattern. The visual cues enhanced the ability of rats to avoid revisits of locations. However, the visual cues did not enhance control by the spatial pattern, as would be expected if the same spatial memories were involved in avoidance of revisits and coding the location of baited locations.

A series of experiments in our laboratory has shown that spatial choices in rats can be controlled by the spatial relations among discrete locations containing hidden food items (Brown, DiGello, Milewski, Wilson, & Kozak, 2000; Brown & Terrinoni, 1996; Brown, Yang, & DiGian, 2002; Brown, Zeiler, & John, 2001; DiGello, Brown, & Affuso, 2002; Lebowitz & Brown, 1999). In these experiments, rats search for single food pellets hidden on top of vertical poles in a matrix of poles. No visual or other perceptual cues are available to indicate the location of the poles that are baited on a particular trial, and the location of the baited poles changes unpredictably over trials. However, there are consistent spatial relations among the baited poles, and the critical finding is that such spatial patterns come to control the choices made by the rats, thereby increasing the efficiency with which they locate the baited poles. We have found evidence for control by square (e.g., Brown & Terrinoni, 1996), linear (e.g., DiGello et al., 2002), and checkerboard (Brown et al., 2001) patterns of baited poles. Dallai and Meek (1990) reported evidence for similar control of choices by the spatial configuration of baited arms in the radial maze (but see Olthof, Sutton, Slumskie, D'Addetta, & Roberts, 1999).

Much of the spatial learning and memory literature concerns the mechanisms involved in using perceptual cues as beacons or landmarks to guide behavior to a goal location. In contrast, a critical feature of the spatial pattern learning that has been investigated using the pole box paradigm is that the content of the learned spatial pattern cannot be in terms of, or anchored to, perceptual cues, because the placement of the baited locations with respect to such cues changes unpredictably from trial to trial. Thus, for the spatial pattern to control choices, it must be abstracted from the visual or other perceptual cues that specify the location of particular poles.

Our previous investigations have focused on the conditions that produce behavioral control by spatial patterns. That work was motivated by the hypothesis that exposure to consistent spatial relations among the baited poles results in an abstract representation of those spatial relations. However, a different set of questions involves the means by which rats use the learned spatial pattern to locate baited locations. Once a representation of the spatial pattern has been acquired, it must affect performance by guiding choices to poles that are relatively more likely to be baited. This analysis indicates that two distinct sets of processes are involved in the control of choices by spatial patterns: First, acquisition of a representation of the pattern occurs over trials. The resulting representation is independent of perceptual cues and the location of particular poles. Second, within each trial the rat must keep track of visited locations and use information about their locations and bait statuses, combined with the acquired representation of the abstracted spatial pattern, to guide choices to the remaining baited poles. Consider the case of a 2 × 2 square pattern of baited poles within a 5 × 5 matrix, for example. Choosing 1, 2, or 3 of the 4 baited poles can allow the rat to determine the location of the square within the larger matrix (i. …