Habits in Dual Process Models. Wendy Wood, Jennifer S. Labrecque, Pei-Ying Lin, and Dennis Rünger. University of Southern California - PDF

Habits in Dual Process Models 1 Habits in Dual Process Models Wendy Wood, Jennifer S. Labrecque, Pei-Ying Lin, and Dennis Rünger University of Southern California Chapter to appear in J. Sherman, B. Gawronski,

Please download to get full document.

View again

of 20
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Creative Writing

Publish on:

Views: 26 | Pages: 20

Extension: PDF | Download: 0

Habits in Dual Process Models 1 Habits in Dual Process Models Wendy Wood, Jennifer S. Labrecque, Pei-Ying Lin, and Dennis Rünger University of Southern California Chapter to appear in J. Sherman, B. Gawronski, & Y. Trope (Eds.), Dual Process Theories of the Social Mind Habits in Dual Process Models 2 Habits in Dual Process Models Habits structure much of people s everyday activities. Eating habits are some of the most striking. People seem to eat about 91% of the food on their plates regardless of whether they are eating from a large or a small plate of food. Specifically, when given large plates, people served themselves and ate about half again more food as when given small plates (Wansink & Cheney, 2005). Thus, plate-cleaning habits persisted despite the amount of food involved. People also develop habits to eat certain types of food. People with habits to eat popcorn at the movies consume approximately the same amount regardless of whether it is fresh and tasty or stale and spongy. Specifically, when popcorn was stale, habitual eaters at the cinema reported disliking it, but they consumed just as much as when it was fresh (Neal, Wood, Wu, & Kurlander, 2011). Thus, popcorn eating habits persisted despite the palatability of the food. What propels people to eat habitually in this way, with minimal regard for the amount consumed or the way it tastes? The answer to this question comes from an understanding of the nature of habits. Eating, like much of human behavior, can be repeated so that it becomes habitual. Habits are dispositions to give a response in a particular context. As people pursue goals in daily life, they experience myriad covariations between their actions (e.g., eating) and cues in the performance context (e.g., food on plate, popcorn in bag). Habit learning occurs when people repeat the same behavior in a specific context, so that associations between the behavior and the contextual cues can be formed. Once a habit is firmly established, perceiving relevant cues the food on one s plate and the popcorn in the bag at the cinema is sufficient to trigger the associated behavior. In this chapter, we first develop a process-based account of habitual responding (see Wood & Neal, 2007, 2009), and then we consider the role of habits in dual-process theories. The Habits in Dual Process Models 3 remainder of the chapter focuses on the factors that promote or impair habitual control of action in relation to other action control mechanisms. Processes of Habit Automaticity Following William James s (1890) notion of ideomotor action, we assume that a memory representation of the habitual response intervenes between cue perception and habit performance (see Figure 1). Due to the frequent repetition of the response, this representation is likely to be highly accessible in the sense of being strong, stable, and distinct. In addition, the frequent pairings of the context and the response produce a particularly strong associative link between the memory representation of the context cues and that of the response. As a result, whenever an individual perceives the relevant cues, the habit representation is strongly and reliably activated. Habitual responses have a high probability of being executed when their triggering conditions are met. One reason is the simple strength of habit representations in memory they likely exceed the activation level of alternative actions. Contributing to this effect, alternative responses may decrease in accessibility as people repeatedly retrieve a particular response from memory (McCulloch, Aarts, Fujita, & Bargh, 2008; see retrieval induced forgetting, Anderson, 2003). Also contributing to habit performance, people may misattribute externally-cued representations to their own natural response to the situation, that is, to their internal preferences and desires (Loersch & Payne, 2010). With this misattribution, habitual responses can seem to be promoted by goals and intentions (see section, Experience of Habits ). Nonetheless, as responses become more habitual, they come more directly under the control of the stimulus context and thereby lose sensitivity to the value of the consequences of that response. The context cues that activate habits range from simple elements of the environment that covary with a response including physical locations, other people, internal states, and Habits in Dual Process Models 4 preceding actions in a sequence to complex conjunctions involving multiple such factors. The cuing process may proceed as envisioned in models of routine action, in which each response in a sequence is triggered by the current internal state given the response sequence, in conjunction with perceptions of the state of the environment (Botvinick & Plaut, 2004). Thus, when making coffee habitually, people may have learned that adding cream and the perception of the lighter color together cue the next response of adding sugar. In this view, habit learning amplifies aspects of context cues that were in the past informative for action selection. Thus, performance is not tied solely to a specific cue in a specific time and place (your own coffee cup) but to other, similar context cues that share important features (other cups). Habit performance thereby can be triggered by a class of cues that might vary in irrelevant features. Rewards for habit performance may be important initially to promote learning of contextresponse associations, but these consequences of responding become relatively unimportant once habits have developed (Neal, Wood, & Quinn, 2006). Thus, people with the relevant habits eat stale popcorn in the cinema despite its taste and eat all of the food on their plates regardless of whether they are still hungry. The stimulus-cued nature of habits yields this insensitivity to outcomes of the response. The term, habit, is sometimes used interchangeably with automaticity. In contrast with this view, we argue that habits are characterized by a specific subset of the features that are commonly used to diagnose automaticity (De Houwer, Teige-Mocigemba, Spruyt, & Moors, 2009). Broadly speaking, habits are triggered by a relatively rigid form of context cuing that integrates with but is not flexibly responsive to people s goals and intentions. Habitual responses are initiated automatically in the sense that they are stimulus driven, require only Habits in Dual Process Models 5 minimal or sporadic conscious monitoring and thus are largely autonomous, and they do not depend on a goal for performance. Despite these features of automaticity, habit responses are largely controllable. As illustrated in Figure 1, people do not necessarily act on a habit representation once it is brought to mind in a given context. Given an explicit goal that is incompatible with a habit, people can alter their responses in several ways (see Quinn, Pascoe, Wood & Neal, 2010). Much like a dieter deciding to eat only half of the food on his plate, people can make decisions to inhibit an unwanted response. Also, when people are aware of the link between contextual cues and an unwanted habit, they can deliberately avoid habit triggers. Thus, dieters can decide not to go to a favorite donut shop and, if they are aware of the tendency for cues to generalize on important dimensions, should probably decide not to go to other restaurants filled with the sight and smell of donuts. Habits in Dual Information Processing Systems In dual-process models, habit performance is mediated by a fast, automatic, unconscious processing system that reflects associations learned through experience. This automatic system typically is contrasted with slow, deliberative, conscious processing that requires access to a working memory system of limited capacity (Evans, 2008). A variety of dual-process models have been proposed, each providing a broad distinction between two partially independent processing systems, and each offering slightly different interpretations of automatic and deliberative processes (see Evans, 2008). The heterogeneous types of fast, automatic learning and retrieval capacities considered in these models are grouped under general labels such as System 1 (Kahneman, 2003), the impulsive system (Strack & Deutsch, 2004), and the reflexive Habits in Dual Process Models 6 system (Lieberman, Gaunt, Gilbert, & Trope, 2002). This type of system includes habits, other features of automaticity, and low-effort processes such as relying on heuristic judgment rules. It is surprising that few dual-process models identify habits as a distinct mechanism of action control. Habit also is largely absent from social psychology more generally. For example, the chapter on automaticity in Fiske, Gilbert, and Lindzey s (2009) Handbook of Social Psychology does not, according to the volume index, make even one mention of habit. Similarly, few textbooks in social psychology currently recognize habits, although this may change given the popular resurgence of interest (e.g., Duhigg, 2012). Historically, habits were a central principle in William James s (1980) understanding of the mind. In the ensuing years, habits became linked with mechanistic approaches to learning, serving as a cornerstone of early learning models (Hull, 1934; Skinner, 1953). Squire and Zola- Morgan s (1991) classic model of memory systems differentiated skills and habits from other types of nondeclarative (implicit) processes including priming, simple classical conditioning, and nonassociative learning. Building on these insights, Evans s (2008) review of dual-process models identified habit as one of three distinct types of low effort, automatic processes. Recently, habits are an important component of many neural models of memory systems (see section, Habits in Neural Models of Memory ), and a number of theoretical analyses have outlined how habit mechanisms integrate with other aspects of action control (e.g., Wood & Neal, 2007; Yin & Knowlton, 2006). By outlining the processes of habit automaticity, the present chapter moves beyond global automatic-versus-controlled distinctions in understanding action control. We argue that habits involve characteristic neural mechanisms and cognitive representations, and that these Habits in Dual Process Models 7 characteristics are evident in particular patterns of response. In the next section, we explain how the neural mechanisms associated with habit memory subserve this automatic responding. Habits in Neural Models of Memory Evidence of the neural substrates of habit performance comes from diverse research conducted on human and nonhuman primates as well as rodents. Although it might seem that, in humans, habit memory is degraded or dependent on deliberative, higher cortical functions, research with selectively brain-damaged patients and with normal participants on a variety of tasks reveals that this learning mechanism is in fact well developed (Bayley, Frascino, & Squire, 2005). A common theme across these various research literatures is that repeated action and the formation of habits is accompanied by dynamic shifts in neural activity, especially in corticobasal ganglia circuits (Graybiel, 2008; Yin & Knowlton, 2006). The functional neuroanatomy of the cortico-basal ganglia system underlies many of the characteristic features of habit automaticity, especially the cuing of responses by contexts with minimal responsiveness to goals. As people repeat a response so as to form a habit, two cortico-basal ganglia loops are particularly important (Graybiel, 2008; Yin & Knowlton, 2006). The first, associative loop, supports working memory functions and goal-directed actions. This loop links the prefrontal cortex and associated areas with the caudate nucleus and to the anterior portion of the putamen. The second, sensorimotor loop, is thought to support the formation of inflexible stimulusresponse associations that underlie automatic, habitual behaviors. This loop links the somatosensory and motor cortex with the medial and posterior portions of the putamen. The formation of habits and development of stimulus-cued responding is thought to involve a shift in Habits in Dual Process Models 8 behavioral control from the associative cortico-basal ganglia loop to the sensorimotor loop (Graybiel, 2008; Yin & Knowlton, 2006). 1 Evidence of the relative involvement of these two cortico-subcortical loops in the development of stimulus-driven responding comes from research on simple sequential motor tasks. In a 3-month study, monkeys practiced different sequences in a button-pressing task, with some new and some standard, habitual sequences (Miyachi, Hikosaka, Miyashita, Kárádi, & Rand, 1997). When the monkeys' associative striatum was temporarily inactivated (via muscimol injection), they were less able to perform the newly learned sequences but still able to enact the habitual ones. In contrast, inactivation of the sensorimotor striatum selectively disrupted performance of habits. Comparable shifts emerged in a month-long study in which people practiced a sequence of finger movements (Lehéricy et al., 2005). With training, neural activation deceased in the associative loop systems linked with goal-directed actions (i.e., premotor and prefrontal cortical areas, anterior cingulate cortex, associative territories of the basal ganglia), whereas activation increased in the sensorimotor regions of the basal ganglia. From a functional perspective, then, the goal-independence of habits is linked to the architecture of the basal ganglia, in particular the lack of reward-based modulation of neural activity in the sensorimotor loop. Evidence of this link comes from Tricomi, Balleine, and O Doherty s (2009) research on habit formation in a button-press learning task. Participants received either chips or candy as a reward for pressing a button when one of two target images 1 Initial, goal-directed learning does not appear necessary to guide habit formation in all tasks. With repetition, people may form habits to successfully perform complex tasks even when they cannot reason deliberatively about how to reach the correct outcome (Bayley et al., 2005). Habits in Dual Process Models 9 was presented. After extensive practice and habits had formed, participants continued to press the buttons to the target images regardless of whether or not they had just eaten their fill of the associated food reward. Importantly, the sensorimotor striatum was linked to this absence of reward modulation. Across their extensive training, participants showed increased neural activity in relevant sensorimotor territories of the basal ganglia. Other features of habits, such as inflexible performance, also can be traced to particular neural substrates. Inflexibility arises in part from the unitization or chunking of action sequences over time. A chunk is an integrated memory representation that can be selected as a whole and executed with minimal attentional involvement. Research on the neural substrates of chunking has identified neural markers for the start and end points of action sequences, presumably at the start and end of the learned progression of responses, and minimal neural responding in the middle, suggesting an integrated sequence representation (Fujii & Graybiel, 2003). Given the current research, it is tempting to conclude that the sensorimotor loop is responsible for the long-term retention of habits. This conclusion is challenged, however, by some evidence that habit performance, especially after especially extended learning, continues despite deactivation of the sensorimotor striatum (e.g., Desmurget & Turner, 2010). With such extensive training, control of habitual behaviors may be further consolidated in transfer to extrastriatal areas, including the cortex. A neurocomputational model of automaticity in perceptual categorization by Ashby, Ennis, and Spiering (2007) offers an elegant account of this hypothesized transfer of control. The model assumes two pathways connecting sensory association areas with premotor cortex a slow re-entrant cortico-basal ganglia loop and a fast, direct cortico-cortical projection. Early performance is governed by goal-directed learning in the basal ganglia circuit. As training progresses, the appropriate cortico-cortical synapses are Habits in Dual Process Models 10 strengthened via Hebbian learning, and control is gradually transferred from the basal ganglia loop to the exclusively cortical network. This neural progression in learning was demonstrated in a procedural category learning task in which performance depended primarily on cortical areas once it became thoroughly automatic (Waldschmidt & Ashby, 2011). It may be, then, that the basal ganglia are not involved in the long-term storage of habitual behaviors. Their primary function in the development of automaticity could consist in training the direct cortico-cortical pathway. In general, habits gradually develop as goal-responsive neural systems relinquish control to the sensorimotor system, and perhaps ultimately to cortical systems that control habits. Given that many everyday behaviors draw on multiple memory systems, their performance probably involves a combination of habit and other types of learning. Thus, the neural systems subserving habits are integrated with a variety of other substrates involved in action control. This interaction is enabled by neural architecture in which the sensorimotor striatal system associated with habit performance is embedded in broader cortico-basal ganglia circuits that facilitate both stimulus-driven, habitual responding and more flexible goal-oriented actions (Yin & Knowlton, 2006). Given interactions among these neural systems, habits interface with other action control systems to guide responding. In the next section of the chapter, we discuss the features of automaticity that comprise habit responding. We then address the multiple ways that habits integrate with other action control mechanisms. Habits are Directly Brought to Mind by Context Cues Evidence of habit automaticity. To test the cognitive associations that underlie habits, Neal, Wood, Labrecque, and Lally (2012, Study 1) assessed how quickly habitual runners Habits in Dual Process Models 11 detected the words running and jogging when presented in a lexical decision task. Before each detection trial, runners were subliminally primed with their personal: (a) running locations (e.g., forest, gym) that they usually frequented, or (b) goals (e.g., weight, relax) that motivated them to run. After subliminal exposure to their locations, runners with stronger habits were faster to detect running words. This suggests that their habitual behaviors were mentally linked with the contexts in which they performed that action. Also, suggesting that habits do not require goals, subliminal exposure to running goals did not activate thoughts of running for strongly habitual runners. Instead, a curvilinear association emerged between habits and goals. Specifically, for runners who were still developing habits, goals seemed motivating and thus activated thoughts of running. Strongly habitual runners, however, relied on context-response associations rather than goals to activate running responses. Additional evidence that habits are triggered directly by context cues comes from a study of sports fans. Reasoning that fans who frequently go to sports stadiums have acquired a habit of speaking loudly, Neal et al. (2012, Study 2), primed some participants with pictures of the stadiums they habitually visited. Control participants were primed instead with pictures of kitchens. The loudness of participants speech was assessed from their verbal responses to a search task. Participants with stronger habits to attend sports stadiums spoke more loudly after being primed with stadiums but not kitchens. Furthermore, indicating that this direct contextresponse tie did not depend on a motivating goa
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks