Definition 1 refers to 'habit' irrespective of whether it is innate ("a priori), acquired, or induced by disease or drugs. Definition 2 restricts 'habit' to a type of "learned behaviour; it is a special case of definition 1, but is stated separately because this is what 'habit' commonly means in the contemporary learning literature. In the early learning literature, 'habit' was also occasionally used to denote a learned act in general, but this is now unacceptable. Habits comprise a substantial chunk of our behavioural repertoire, often to a larger degree than we tend to concede. Driving a car becomes a habit, but so is also eating junk food, or delivering the same seminar for the ninth time. Nowhere is the dominance of habit in our lives better epitomized than in James' "classic Principles of psychology (1890). Dubbing organisms as 'bundles of habits' (referring to the generic definition 1), James considered habits not only as an inescapable manifestation of the "plasticity of organic material, but also as driving force of the daily operation of individuals and "cultures: '... Habit a second nature! Habit is ten times nature. Habit is. the enormous fly-wheel of society... it keeps the fisherman and the deck-hand at sea through the winter; it holds the miner in its darkness, and nails the countryman to his log-cabin.'1 Many a reader nowadays will clearly disagree with James' social conservatism, but not with his appreciation of the role in our life of automatic or semiautomatic procedures, many of which are acquired via repetitive "stimulus-response conditioning (Thorndike 1911; "instrumental conditioning). Habits could be sensorimotor, emotional, or cognitive, or all of the above elements combined in a single assembled routine. They could be highly useful and beneficial, as in vocational training, because they spare "attentional and cognitive resources, perfect performance, and prevent superfluous response; but the same brain circuits that bring all these benefits could also go astray and generate viciously disturbing habits (Long and Miltenberger 1998; Robbins and Everitt 1999; Leckman and Riddle 2000).
Being independent of conscious awareness, habit is by definition a non*declarative (implicit) type of memory (Hirsh 1974; Mishkin et al. 1984; Squire and Zola 1996; "skill, "taxonomy). As opposed to "declarative memory, which is subserved by "cortico"limbic circuits, habits are subserved by corticostriatal circuits, i.e. reciprocal connections between the "cerebral cortex and the basal ganglia (Mishkin et al. 1984; Salmon and Butters 1995; Knowlton et al. 1996; Jog et al. 1999). Two selected studies, one in the "rat, the other in "Homo sapiens, will serve to illustrate the conserved role of the corticostriatal circuits in the formation and expression of habit in mammals.
Let's start with the rat first. McDonald and White (1993) subjected rats to three types of tasks in an eight-arm radial "maze: a win-shift task (Olton and Schlosberg 1978), a conditioned *cue preference task, and a win-stay task (ibid.). In the win-shift task, all the eight arms of the maze are baited with food. The rat is placed on the central platform of the maze, and allowed on each trial to choose an arm. The performance is scored as the number of revisits (i.e. errors) to arms from which the rat has already obtained food within the first eight choices. Success in this type of task demands the rat to identify, classify, and remember multiple spatial locations, i.e. use explicit memory.2 In the conditioned cue preference task, two arms are used and the rest blocked. One arm is cued with light at the entrance to the arm. In the training trials, the rat is conditioned by confining it with food to either the 'light' or the 'dark' arm. In the test, both arms are left unbaited, and the amount of time spent in each arm is recorded. This type of task taps into the ability of the rat to associate a stimulus (light) with a food reward. In the win-stay task, four randomly selected arms are baited with food and lit. The rat is placed on to the centre platform of the maze. Immediately after the rat leaves a lit arm having eaten the bait, the arm is rebaited. When the rat retrieves the second pellet for any arm, the light is turned off and no further food is placed there. Overall, the rat is required to visit each of the four lit arms twice, earning eight pellets within a trial. The trial is terminated after a fixed time or after the eight food pellets have been retrieved. Entries into unlit arms are scored as errors. This task involves learning an approach response to a specific sensory cue (light), irrespective of other environmental cues,3 and repeat the choices to each lit arm within a trial. This is construed to involve the formation of a stimulus-response habit (Packard et al. 1989; McDonald and White 1993).
By using groups of rats with circumscribed brain lesions, McDonald and White (1993) found that damage to the "hippocampal formation impaired acquisition of the win-shift task but not of the conditioned cue preference or the win-stay task; damage to the lateral "amygdala impaired acquisition of the conditioned cue preference task but not of the win-shift or win-stay tasks; and damage to the dorsal striatum impaired acquisition of the win-stay task but not of the win-shift or the conditioned cue preference task. Hence, these authors showed, by inferring function from dysfunction ("method), that the formation of a stimulus-response habit depends on the striatum, and that the neural system that subserves habit could be dissociated from the neural systems that subserve explicit and affective memory. Additional evidence for the involvement of the striatum in the formation of the "internal representations of habits was found in correlative studies, in which neuronal activity was recorded from the striatum of the behaving rat while engaged in maze routines (Jog et al. 1999).
Whereas McDonald and White (1993) used a 'triple dissociation' approach (three tasks, three lesions), Knowlton et al. (1996) used 'double dissociation' to identify the brain substrates of habit and dissociate it from the brain substrates of declarative memory in humans.4 The two tasks used in this study were a probabilistic classification task and a multiple-choice questionnaire task. In the probabilistic classification task, the "subject is requested to forecast the weather, whether sunny or rainy, on the basis of four visual patterns, each linked to the weather conditions with a prefixed probability, unknown to the subject in advance. On each trial, one, two, or three of these cues are presented side by side on the computer screen, and the response on each trial is "reinforced by visual and auditory feedback. This task involves gradual learning shaped by repetitive stimulus-response-reinforcement loops over multiple trials, and does not require conscious awareness of the information accumulated over trials; these task attributes are characteristic of habit learning. The probabilistic classification task can hence be used to tap into the ability to form a habit.5 The multiple-choice questionnaire referred to declarative information about the procedures and episodes of the habit task.
Knowlton et al. (1996) used the two tasks to test three groups of subjects: "amnesic patients with bilateral hippocampal or diencephalic damage; non"demented patients with Parkinson's disease, which involves striatal damage; and healthy matched "controls. The amnesic patients learned the habit as well as the controls, but failed on the declarative task. The parkin-sonian patients, in contrast, performed well on the declarative task, but could not learn the habit. The conclusion: formation of habit depends on intact striatal circuits, but not on intact mediotemporal circuits, whereas the opposite is true for the declarative task.
To the causal observer, the maze habit and the probabilistic classification habit may look very different. But they do share a lot in common. This is evident first, from the analysis of the behavioural task, which in both cases depends on the formation, via repetitive stimulus-response-reinforcement cycles, of a stimulus-response routine; and second, from the identity of the brain circuits that are essential for the behaviour in both cases. As the striatum subserves different types of habits, which involve different types of information, it is likely to execute generic computations, which are required for the 'syntactic' assembly of pieces of action repertoires into routines (Graybiel 1998). The circuits that encode the internal representations specific to the habit are hence expected to involve additional brain areas, including modality-specific and association cortex. If this is the case, then use-dependent changes in striatal circuits will show larger "transfer than those in other brain areas that encode information about a specific habit. This is why damage to the striatum leads to a 'global' deterioration of habits. This relates to the differentiation between use-dependent changes in particular internal representations, and those that occur in generic computations made over these representations. But this is already a global issue, which exceeds the discussion of habit.
Selected associations: Instrumental conditioning, Learning set, Maze, Skill, Taxonomy
1Not surprisingly, many authors considered the formation as well as the reversal of habits a key to the success of education at large, e.g. Rousseau (1762), Radstock (1886),and Rowe (1909). 2This task also exploits the innate tendency of the rat to shift its search for new locations. When given equal opportunity, the rat will prefer the win-shift over the win-stay strategy (Olton and Schlosberg 1978).This is hence an example of the interaction of an innate predisposition (*a priori) with learning (for another example, see the delayed nonmatching to sample task in the *monkey; *delay task).
3Actually, because the rat is requested to enter the lit arm regardless of the identity and location of the arm, the use of cues other than the light, and of the relationship among cues, becomes counterproductive (McDonald and White 1993). This illustrates that the formation of habit involves not only augmentation of a response, but also what Glaser (1910) calls the '...often painful suppression of irrelevant actions, and the survival of only those that count'.
4For the logic and 'algorithms of 'multiple dissociations' experiments, see 'control.
5The habit in this laboratory task forms much faster than must habits in *real-life.
Was this article helpful?