Learning and Memory Mechanisms Involved in Compulsive Drug Use and Relapse - Drugs of Abuse: Neurological Reviews and Protocols - page 76

has emerged to support this hypothesis, including data from humans (16 , 17)
and nonhuman primates (18) . Some of the clearest demonstrations come from
the work of White, Packard, and colleagues, who trained rats in a set of radial
maze tasks. In the “win-shift” task animals received food rewards for choosing
maze arms that they had not recently visited; in the “win-stay” task they
were rewarded for choosing an illuminated arm, with their recent history of
spatial positions being less important. Lesions (19 , 20) and local drug infusions
(21) demonstrated that the hippocampus is important for acquisition and
performance of the win-shift task (and not win-stay), whereas the (dorsal)
striatum is closely involved in the win-stay task (and not win-shift). The task
components are similar in each case (maze, rewards, etc.) but are combined
under different contingencies. The involvement of distinct neural circuits
thus appears to refl ect their differing “rules of operation” (22) or information
processing “styles” (13) . Among other roles, the hippocampus is essential for
the ability to keep track of the when, where, and what of particular recent
experiences (episodic memory; reviewed in 23 , 24 ). In contrast, the dorsal
striatum appears to be critical for the progressive acquisition of fi xed response
tendencies (habits)—in this case, an approach response to a light stimulus.
In some task situations the relationship between the biologically meaningful
events and other task elements is ambiguous, and “correct” performance can
rely on either of two or more strategies. This has been classically demonstrated
in a four-arm radial maze (“plus maze”), with an animal consistently placed
in one arm (“South”) and allowed to fi nd food consistently in another (“East”)
(25) . The animal learns to retrieve the food rapidly; the central experimental
question is what exactly is learned—a fi xed response habit (e.g., “turn right!”)
or a more cognitive use of spatial information (e.g., “food is in East—go
there!”). A probe test—in this case, starting from the North arm—can help to
resolve this, as the two theories make opposite predictions about the animal's
choice (West if turning right, East if using spatial information). Many experi-
ments on this “place vs response” question (26) established that both strategies
are learned in parallel. Which strategy predominates depends on a variety of
factors such as availability of local and distal cues, and also on the amount of
training received by the animal. Early in training rats tend to make use of a
hippocampus-dependent spatial/cognitive strategy, but with many repetitions
the fi xed response “habit,” dependent on the dorsal striatum, comes to dominate
behavior (27) . In a sense, the strategies implemented by different neural circuits
“compete” for behavioral output. Acquisition of the striatum-dependent win-
stay task is considerably accelerated in animals with a hippocampal lesion
(19) ; in fact, hippocampus-damaged animals tend to act generally as if they
were easily dominated by stimulus-response habits (28 , 29) . Inactivation of the
hippocampus early in plus-maze training leads rats to use the nascent striatum-