Brain Mechanisms of Actions vs. Habits
S~^R associations yield habits, behaviors that are elicited by stimuli independendy of expectations about their effects. In appetitive situations (food-reinforced behavior), there is evidence that a transition from action to habit is mediated by a change in the underlying brain circuit controlling behavior.
One approach involves brain lesions of specific areas combined with the reinforcerdevaluation procedure. A lesion in a brain area controlling actions should render the instrumental response insensitive to reinforcer devaluation even after minimal training. By contrast, a lesion in an area controlling habits should leave the instrumental behavior sensitive to reinforcer devaluation even after extended training. Two areas of the rat’s dorsal striatum fulfill these functions. Early in training, the dorsomedial striatum controls the formation of lever pressing actions, such that a lesion in this area converts the action into a devaluation-resistant habit (Yin, Ostlund, Knowlton, & Balleine, 2005). Conversely, lesions of the dorsolateral striatum disrupt the conversion of an action into a habit, rendering the response vulnerable to reinforcer devaluation even after extensive training (Yin, Knowlton, & Balleine, 2004).
Another approach is to look for evidence of the transition from action to habit in spatial learning situations. A spatial problem can in principle be solved in one of two ways (Restle, 1957). Consider a rat in a T-shaped maze receiving a reward for turning right. This problem can be solved either by learning to approach a location or stimulus based on S~^S associations (action) or by performing a particular response based on S^R associations (habit). In one experiment (Packard & McGaugh, 1996), rats were trained in a T-shaped maze with constant start and goal boxes and tested after either 8 or 16 training sessions by releasing them from a new location (Figure 6.5). After 8 training sessions, rats showed place learning, thus approaching the location where they were reinforced before—an action. However, after 16 sessions, rats made a left turn, the same response they were trained to perform, ending in the wrong arm of the maze—a habit. During such tests, either the hippocampus or the caudate nucleus (dorsal striatum) was inactivated with lidocaine (a local anesthetic that disables neurons for a short time). After 8 sessions, hippocampal inactivation impaired test performance, whereas caudate inactivation did not. However, after 16 sessions, the opposite effects were observed by inactivation. Thus, the transition from expectancy-guided actions to habits involves a shift in brain control from limbic (hippocampus) to striatal (caudate) areas (for brain areas, see Chapter 5). Interestingly, the transition from action to habit is accelerated by negative emotions, including sources of emotional stress and physical pain (Gasbarri,
Figure 6.5 The neural basis of expectancy-guided actions and automatic habits. Rats were trained to run from the start to the goal in a plus maze. During two key tests after either 8 or 16 training sessions, rats were released from the test position. After 8 training sessions, rats approached the goal, thus making the opposite response turn relative to training turns (action learning). However, after 16 trials they turned left, thus ending in the incorrect arm (habit learning). Interestingly, reversible inactivation of the hippocampus (Hip) with lidocaine infusions (Lid) selectively disrupted action learning relative to saline infusions (Sal), whereas similar inactivation of the caudate nucleus (CN) selectively disrupted habit learning.
Source: From Packard and McGaugh (1996); reproduced with permission from Elsevier.
Pompili, Packard, & Tomaz, 2014). Thus, encountering aversive conditions moves organisms into an automatic mode of responding.
