Navigational processes and visual memory
Spatial memory plays a persistent role in many daily activities, perhaps most commonly in our daily navigation—from the bedroom to the kitchen, from home to work, from the office to the vending machine. Navigation itself can also be broken down into the different types of processes we hope to accomplish as we move through space (e.g., Golledge, 1999). At present, there is no unifying theory of the different types of tasks and processes that might engage human spatial memory, but the contribution of visual memory to navigation can be characterized by considering its potential role in these different proposed processes. In the following sections, we discuss some of the known and proposed processes and attempt to draw some preliminary conclusions about the role for visual memory.
Place and response learning
One of the fundamental distinctions in the processes that guide spatial behavior has been the difference between place- and response-learning mechanisms in rats (e.g., Packard & McGaugh, 1996; Restle, 1957;Tolman, Ritchie, & Kalish, 1946, 1947). In their classic studies, Tolman and colleagues demonstrated this dichotomy using a T-maze learning paradigm. Rats were placed in a maze like the one shown
FIGURE 2.5 Schematics of a typical T-maze setup (e.g., Packard & McGaugh, 1996). Black bar shows a blockade, and shapes represent distal cues. A. During training, the same response (left turn) is required repeatedly to reach the goal. B. During test, the rat enters from the opposite direction. The solid arrow shows the place-learning behavior (turn toward the triangle), and the dashed arrow shows the response-learning behavior (turn left)
in Figure 2.5A. During training, the rat was placed at the same starting position and the reward was always in the same place. After training, the critical test was conducted by changing the configuration and starting position (Figure 2.5B). From this new position, there are two “correct” responses depending on what the rat has learned. If the rat has learned to use the cues in the environment, it will turn toward the environmental cue, demonstrating place learning. However, if the rat has learned to make a specific response to the T-maze stimulus, it will turn in the same direction that it has been turning throughout the training, demonstrating response learning.
In rats, place and response learning appear to be occurring in parallel, but several factors determine which will guide behavior (e.g., Cook & Kesner, 1988; Morris, Garrud, Rawlins, & O’Keefe, 1982; O’Keefe & Nadel, 1978; Packard & McGaugh, 1996;Tolman, 1948;Tolman et ah, 1946,1947). First, numerous studies indicate that place learning occurs more rapidly with limited learning and overlearning with variable routes, whereas response learning occurs after extensive training provided that the same route is repeated throughout training. In terms of utility, place learning affords greater flexibility of use, accommodating changes in the environment and the need to find novel routes. However, this flexibility is cognitively demanding. In contrast, response learning lacks flexibility but may allow for accurate performance with limited attention. As such, when attentional resources are limited, it is useful to have a more automated system for navigating familiar environments.
In addition to the behavioral differences, place and response learning have been associated with two different neural systems—hippocampus and caudate, respectively (Cook & Kesner, 1988; Morris et al., 1982; Packard & McGaugh, 1996). For example, using Tolman’s T-maze paradigm, Packard and McGaugh demonstrated that lesions of the caudate resulted in solely place-learning performance whereas lesions of the hippocampus resulted in solely response-learning performance.
In humans, there has been a long-standing assumption that these two systems are also operating (e.g., Burgess et al., 1999), and neuroimaging studies have used the known neural correlates to support this contention (e.g., Hartley et al., 2003; Shelton, Marchette, & Yamamoto, 2007). For example, Shelton et al. (2007) used fMRI to scan participants while learning a fictitious environment by watching a repeated route.The results revealed a negative correlation between activation in the right caudate and the bilateral posterior hippocampus—as a given person showed more caudate activation, he or she showed less hippocampal activation. This difference could be attributed to differences in perspective-taking ability, one indicator of flexible spatial reasoning. Hartley et al. (2003) found a similar task-based difference in these regions. Together, these results have been used to suggest that people may differentially rely on place- and response-learning mechanisms based on individual differences and/or task demands.
In both rats and humans, these relationships have been revealed using largely visual learning conditions. However, there is nothing in the specification of these mechanisms that requires a link to vision (e.g., Hartley, Burgess, Lever, Cacucci, & O’Keefe, 2000). Like the hippocampus, the caudate nucleus receives inputs from multiple modalities, suggesting that stimuli from different modalities may serve as the signal for engaging the learned response.The role of cues in different modalities is discussed more thoroughly in the next section.