Learning the configuration of a landmark array: I. Touch-screen studies with pigeons and humans.

Abstract

Pigeons and humans searched on a touch-screen monitor for an unmarked goal located relative to an array of landmarks presented in varied screen locations. After training with the goal centered in various square arrays of 4 landmarks, humans, but not pigeons, transferred accurately to arrays with novel elements. Humans searched in the middle of expanded arrays, whereas pigeons preserved the distance and direction to a single landmark. When trained with the goal centered below 2 identical horizontally aligned landmarks, humans responded to horizontal expansions or contractions of the airay by shifting their search vertically, preserving angles from landmarks to goal. Pigeons did not adjust their search vertically. Humans trained with a single landmark adjusted search distance when landmark size was changed. Both pigeons and humans use the configuration of a landmark array, but the underlying processes seem to differ.Experimental work has demonstrated that many animal species can encode the location of an important place in terms of nearby visual landmarks (see Gallistel, 1990, and Collett, 1992, for reviews). In several investigations of landmark use, a goal was located near an array of two or more local landmarks, with a directionally stable frame of reference provided by other cues (e.g.,

Geometric determinants of human spatial memory

“…One paradigm examines the locus of search for an object presented in a location equidistant to 2, 3 or 4 landmarks, how it changes after expansion of the array of landmarks, and how responses vary between different species. Gerbils (Collett, Cartwright, & Smith, 1986) and pigeons (Spetch, Cheng, & MacDonald, 1996;Spetch et al, 1997) tend to search in locations that maintain the distance and angle to individual landmarks consistent with a model in which the vectors from the landmarks to the objects' locations are stored. Experiments probing a location nearer to one of two landmarks show that the vector from the landmark nearer to the target location is more influential than that from the farther landmark.…”

“…However, the exact form of proximity weighting was not clear, and might differ across individuals. When humans search for a central location after expansion of the array of landmarks their searches focus on a location that preserves all angles to the landmarks and thus also preserves the ratios of distances between the landmarks rather than the distance to a given landmark (Spetch et al, 1996;Spetch et al, 1997). Waller, Loomis, Gollege, and Beall (2000) used a similar approach when using virtual reality to test human spatial memory.…”

“…If so we could begin to reconcile neurophysiological and behavioral accounts of spatial memory. Our approach builds on those of Spetch et al (1996Spetch et al ( , 1997 and Waller et al (2000) by probing a variety of locations within a walled arena with a variety of manipulations of environmental shape and size taking place between presentation and testing.…”

Integration of spatial maps in pigeons

“…Pigeons have been shown to acquire separate landmark-goal vectors independently even in situations involving multiple landmarks (Spetch and Mondloch 1993;Spetch et al 1996Spetch et al ,1997. For example, after learning to find a hidden food goal in the presence of an array of landmarks, search locations on probe tests involving transformations of the landmark array (e.g., expansions of landmark locations or removal of individual landmarks) indicate that pigeons had encoded the vector between one or more individual landmarks and the goal.…”

Searching by rules: Pigeons' (Columba livia) landmark-based search according to constant bearing or constant distance.

“…When the landmarks were spread apart on a transformation test, the pigeons did not search in the middle of the expanded array but instead searched in locations that maintained the approximate training distances and directions from individual landmarks. A similar tendency to search on the basis of individual landmarks rather than using the spatial rela-tionship of the goal to the whole array was found for gerbils (Collett, Cartwright, & Smith, 1986) and monkeys (Sutton, Olthof, & Roberts, 2000), whereas adult humans show the opposite tendency (Spetch et al, 1996(Spetch et al, , 1997. Transformation tests like those described above are useful for revealing an animal's typical or preferred encoding strategy, but they do not indicate how flexible that strategy is.…”