Recently, however, some philosophers have begun to pay attention to habits. An example is a series of papers by Bill Pollard starting in the mid-aughts (Pollard, 2006a, 2006b), and more recently Steve Matthews (2017). Pollard tackles some fundamental issues arguing (positively) for habit-based explanations of action as a useful addendum (if not replacement) for folk-psychological accounts (along the lines of previous posts). Here I’d like to focus on Mathews more recent work, which deals with the core characteristics making something a habit.
One useful (implicit) message in this work is that consistent with the modern notion of concepts in cognitive semantics, habits are a radial category. Rather than being a crisp concept with necessary and sufficient conditions of membership, habits are a fuzzy concept, with some “core” or “central” exemplars that share most of the features of habits, and some “peripheral” members that only share some features.
Most anti-habit theorists (with Kant and Kant-inspired theorists such as Parsons being one of the primary examples) equate habit with mindless compulsion and use this equation to expunge habit from the category of action. Critiques of habit theories can thus be arranged on a strength gradient depending on which element of the radial category they decide to focus on. The weakest critiques pick peripheral members, passing them off as “prototypes” for the whole category. Peripheral members of the habit category, such as tics, reflexes, addictions, and compulsions, tend to share few features with action that is experienced as intentional. It is thus easy for these critics to deny habit-based behavior the characteristic that we usually reserve for “action” proper.
Much like American sociological theory post-Parsons (Camic, 1986), habits have been given short shrift in the analytic philosophy of action tradition. As noted in previous posts, one problem is that habit-based explanations, being a form of dispositional account of action, are hard to reconcile with dominant intellectualist approaches to explaining action. The latter, require resort to the usual “psychological” apparatus of reasons, intentions, beliefs, and desires. In habit-based explanations, actions are instead accounted for by referring to their own tendencies to reliably reoccur in specific environments given the agent’s history. This makes them hard to square with typical folk psychological explanations, in which “mental” items are the presumed causal drivers.
Matthews’s argument is that the core or prototypical members of the habit category are what Marcel Mauss called techniques. Skilled ways of being proficient at an action, acquired via an enskillment process requiring training and repetition. These include both “behavioral” skills (e.g. playing the piano, typing, riding a bike) and “cognitive” or “mental” skills, although the latter is less central members of the habit category for most people. In this respect, most bona fide habits are mindful, without necessarily being intentional in the folk psychological sense. They also have five core features, which I discuss next.
Habits are socially shaped.- This might seem obvious. However, there is a tendency in some corners of social theory to think of habit-based accounts as somehow imposing an “individualistic” explanatory scheme. Some people decry while others celebrate (Turner, 1994) the alleged commitment to individualism that comes with habit-based accounts of action. This conception is misguided. Matthews is correct in noting that for core (prototypical) habits are hardly individualistic since they comprise culturally transmitted “techniques” for how to do things (Tomasello, 1999). That each person could have their own way (say of typing or swimming) does not make habits purely individual since they would not be constructed or transmitted if people were Crusoe-like isolates. Instead, most true habits, as revealed by recent sociological “apprenticeship ethnographies” require the embeddedness of the individual in some pedagogical context (for most children this being the family). In this way, most habits are “relational” in a fairly straightforward sense.
Habits are acquired through repetition.- Another one that seems obvious. Nevertheless, I believe this point is more consequential than meets the eye. Recent work emphasizing the root of so-called “dual process” models in theories of learning and memory emphasize that routes to cultural acquisition (ideal-typical “fast” and one shot and “slow” and high repetition) are a key way to partitioning different cultural elements. Namely propositional “beliefs” from non-declarative practices. Habits, having only the slow route of acquisition open to them belong to the latter. Hence the relatively harmless analytic equation of habits and practices. This criterion also serves to demarcate degenerate or borderline examples of the habit radial category such as phobias acquired after a single exposure to a threatening object (e.g., fear of dogs after a dog bite), which depend on analytically and physiologically distinct neural substrates. These we can safely rule out as robust members of the habit category based on the acquisition history criterion.
Habits modify people in durable ways.- As Mike and I have noted (Lizardo & Strand, 2010), this criterion serves to demarcates “strong” habit or practice theories from theories who purport to pay attention to practices but from which embodied agents with their own inertia and history of habituation seem to be absent. Commitment to habit as an explanatory category entails commitment to persons as causally powerful particulars who have been modified by habit in a durable way. This makes durable habits a disposition to behave in such-and-such ways under certain circumstances. Durable modification also entails making conceptual room for the fact that, once acquired, habits are hard to get rid of. So it is usually easier to “refunctionalize” a habit (e.g. take an old habit and put to use for new purposes) than to completely retool.
Since habits operate according to a Hebbian “use or lose it” rule, it is possible for habits to atrophy and decay. However, this decay is relatively graceful and gradual, not fast and sudden. In addition, the previous acquisition of a habit entails faster re-acquisition even when that habit has been weakened or partially lost. This is behind the folk idea that many things are “like riding a bike,” so they can come back easier when you try them again even after a period of disuse.
Considering the “second nature” created by habit means we need to differentiate the temporality of habit (acquisition, use, rehearsal or decay) from the temporality of “macro” social life, as these may not always be in sync; habits will try to persevere even under changing or adverse conditions (Strand & Lizardo, 2017). Durable modification also links nicely to classic sociological notions on the power of “cohorts” to enact social change as history is “encoded” in individuals (Bourdieu, 1990; Ryder, 1965; Vaisey & Lizardo, 2016).
Habits are activated by environmental cues and triggers.- This is one of the better documented empirical regularities in the psychology of action (Ouellette & Wood, 1998). Yet, its meager representation in sociological action theory as an explanatory tool is telling, despite sociologists obvious preference for environmental over attribute-based explanations. Perhaps part of the problem if conceptual; thinking of the environment as a “trigger” may bring fears of removing voluntarism (or as we call it today “agency”) out of the equation thus producing a unidimensional theory of action that reduces action to “conditions” (Parsons, 1937). Yet this fear is unfounded.
First, most people can prospectively plan to enter an environment they know will trigger a habit. For instance, we may set up our work space in the office in a way that facilitates the evocation of the “writing” habit. Second, agents can actively perceive that certain situations have certain “moods” or affordances and they welcome that these trigger reliable (usually pleasant) habits. For instance, a social butterfly can actively perceive that a cocktail party will be good for triggering the complex of habits making up their “outgoing” personality. These have “negative” versions; we avoid certain environments precisely because we know that they’ll trigger a habit we may want to atrophy or decay. There’s no reason to think of the triggering function of environments in purely mechanical ways.
Second, that habits are automatically triggered by environmental cues does not impugn their link to goal-oriented action. In fact, habits can be thought of as a way to facilitate the pursuit and attainment of goals. It is a Parsonian prejudice to presume that the only way to pursue goals is to “picture” them reflectively before the action is initiated and then deploy “effort” to get moving. In fact, this effortful control of action may be subject to more disruption (and thus failure in the attainment of goals) than when agents “offload” the control of action to the environment via habit. In the latter case, goals can be pursued efficiently in a way that is more robust to environmental disruption and entropy.
Habits partake of certain conditions of “automaticity”.- That habits are “automatic” also seems self-evident. However, this can also be conceptually tricky. The problem is that automacity is not a molar concept; instead, it decomposes into a variety of features, some of which can vary independently (Moors & De Houwer, 2006). This can lead to semantic ambiguity because different theorists may actually emphasize different aspects of habitual action when they use the term “automatic” to refer to it.
As already intimated earlier, for prototypical habits, the automaticity feature that most people have in mind is efficiency. After acquiring a habit via lots of repetition people gain proficiency in performing the action. This means that the action can be performed faster and more reliably. Another feature of efficiency is that we no longer have to monitor each step of the action; instead, the action can be performed while our attention resources can be freed to do something else. For instance, experienced knitters can become so efficient at knitting they can do that while reading a book or watching TV.
However, other theorists may take efficiency for granted and point to other features of automaticity as definitional of habitual action. The most controversial of these is the link to intention. For some habits are automatic because they are patterns of behavior that, via the environmental trigger condition mentioned above, bypass intention. This leads to a sometimes counterproductive dualism between “intentional action” and “habit.” I believe a better solution is to think of habitual action as having its own form non-representational “intentionality” (Pollard, 2006b). Driving a car, or riding a bike is intentional action with its own feel, the difference from reflexive intentional action being that representing each step in the action is not required (Dreyfus, 2002).
As noted earlier, the feature of automacity that makes the weakest criterion for defining prototypical habits is (lack of) goal dependence. Most habits are not automatic by this criterion since most habitual action is action for something. Habits without goals (e.g., twirling your hair, tapping your fingers) exist, but they are actually fairly peripheral members of the category. In accord with the pragmatist conception, most habits exist because they help the agent accomplish their goals. As mentioned earlier, most goals are reached via habitual action rather than by reflexive contemplation of ends and effortful initiation of action.
Other features of automaticy are even more peripheral for fixing the nature of habit. For instance, the feature that Bargh refers to as “control.” This refers to whether the agent can “stop” an action sequence once it is started. In this sense, prototypical habits (playing the piano, specifying a regression model) are “controlled” not automatic actions (Pollard, 2006b, p. 60). Skills and procedures, especially those that are narratively extended in Matthew’s (2017) sense, are all “stoppable” by the agent so don’t count as automatic by this criterion. Complete incapacity to stop a line of action only applies to peripheral members of the habit category (e.g., reflexes, phobias, etc.) and probably pertain to habitual actions with short temporal windows.
Note that this refers to whether habits are “intentional” as described above. Most habits may fail to be intentional (in the classical sense) because they are triggered by the environment, but they can be controlled because the agent (if they have the capacity) can stop them once triggered. This is why it is useful to keep different features of automacity separate when thinking about the nature of habit.
Nevertheless, the issue of controllability brings up interesting conceptual problems for habit theory. These have been sharply noted in a series of papers by the philosopher Christos Douskos (to be the subject of a future post). The basic issue is that categorizing an action as a “habit” may be separable from its status as “skill.” Basically, we have lots of skills that do not count as habitual (remaining in “abeyance” so to speak), and some habits that are not skillful. Overall, the ascription conditions for calling a pattern of action a habit, may be more holistic, and thus empirically demanding, than pragmatist and practice theories suppose because they do not reduce to features inherent to the action or its particular conditions of acquisition.
How about the feature of the “unconscious” nature of some automatic actions? Only degenerate or peripheral members of the habit category are “unconscious.” This refers to whether the person reflexively knows whether they are performing the action. Once again, for some peripheral members of the category (cracking your knuckles while engrossed in some other activity), this may apply but it is unlikely to apply to prototypical skills and procedures (we are all aware of driving, typing, etc.). Some people point to “mindless” habit-driven actions as having this feature, such as driving to work when we meant to drive to the store. Here, however, it is unlikely that the person was unconscious of performing the action. So the lapse seems to have been one of failure to exercise control (e.g. stopping the habit because it was not the one that was properly linked to the initial goal) rather than lack of consciousness per se.
Other theorists emphasize unconscious cognitive habits, and maybe for these, this feature is more central than for more prototypical behavioral habits and procedures. Even here, however, unconscious cognitive habits may have the potential to become “conscious” (e.g. the person knows of their existence qua habits) without losing the core automaticity features defining their habitual nature (e.g. the fact they are efficient means to the accomplishment of certain cognitive goals). Overall, however, while most habitual action does rely on subpersonal processes embedded in the cognitive unconscious, most habits are performed in a “mindful” manner (without implying reflexive self-consciousness). As such, they are not automatic actions by this criterion.
Dreyfus, H. L. (2002). Intelligence Without Representation–Merleau-Ponty’s critique of mental representation the relevance of phenomenology to scientific explanation. Phenomenology and the Cognitive Sciences, 1(4), 367–383.