The role of tacit kno wing in adherence to social norms

This paper aims to contribute to the ongoing discussion on adherence to social norms. It considers insights from multiple research traditions in an effort to explain how individual learning and action are connected to social norms. One strand of philosophical tradition holds that non-representational learning and skillful coping carried out unconsciously are underestimated by both scientific and philosophical traditions. The present research combines this tradition with the literature on the evolution of social norms and suggests that experienced individuals in a society adhere to social norms better than novice agents do. We explain this phenomenon by unconscious and non-representational cognitive processes. This framework is then used to investigate population-level outcomes of individual learning.


Introduction
The paper investigates how implicit learning affects the formation of social norms and rule fo lowing in societies.The role of learning and tacit knowledge has been a consideration in economics since the studies by Hayek (1937Hayek ( , 1945) ) and Nelson and Winter (1982).Herein, the paper posits that their research on price formation in markets and production organization in firms could be extended to a herence to norms and rule-fo lowing behavior in societies.In essence, tacit knowledge and unconscious coping abilities-the ha lmarks of expertise-have significant effects on rule-fo lowing behavior.Moreover, the rational actor model cannot adequately capture these effects when population-level outcomes are considered.
The classical constitutional conundrum, according to Samuel Bowles (2006), is the fundamental pro lem of a l societies.It is about how to harmonize individual preferences and social outcomes.In other words, how can we direct, if indeed necessary, self-intere ed individuals toward a ions for the benefit of society?In answering this question and considering its variations the literature is inconsistent, as a result of differences in assumptions, presumptions, and methods.On the one hand, mechanical and rational answers dominate the agenda, as behavior is reduced to the calculation of costs and benefits, on the other hand, intuition, disposition, and arational patterns are used to explain the formation and evolution of behavior.
On the other hand, disciplines including philosophy, economics, and biology deal with the issue.Philosophy employs the analytical tradition and phenomenology, economics uses neoclassical theory and new institutional theory, and cognitive science uses symbolism and connectionism.A l of these research disciplines maintain the division mentioned above and provide conflicting explanations for the causes and implications of the classical constitutional conundrum.
In the paper, it is attempted to bring together the approaches of these two traditions to explain a herence to norms in societies.The rational actor model and 'pure logic of choice' are not adequate for gaining an understanding of norm a herence; however, criticism alone is not sufficient.Our aim is to use the findings of conflicting ap roaches in a complementary manner to synthesize an explanation for rule-fo lowing behavior.The theoretical and empirical literature on both sides is extensive, but to the best of our knowledge few researchers have brought them together, nota ly Lane et al. (1996), Langlois (1998), Negru (2013); see also Op (2013) and Aydogmus et al. (2015).It is our aim to highlight the possibility of bridging these divergent paths.In this regard, this paper further develops the ar uments in Aydogmus et al. (2015).
The paper begins with a review of the literature on unconscious cognitive processes and its philosophical basis.Next, evolutionary literature on the formation and evolution of social norms is reviewed.Then, a simple game theoretic example is provided on how studies on rule-fo lowing behavior could be improved by taking into account unconscious cognitive processes.Lastly, the implications of our framework and further research needs are discussed.

Unconscious cognitive processes in decision-making: non-representational abilities
When it comes to norm formation at the societal level, individuals' decisions and the evolution of their behavior are important microdeterminants.That is why the way individuals learn and the way their decisions evolve are also important.How does an individual choose to act according to a norm?The dilemma to consider is whether individuals act rationa ly in a complex environment, as Herbert Simon (1996) su gests, or do they use ski ls beyond their conceptual framework, without (consciously) thinking about them, as Hubert Dreyfus andStuart Dreyfus (1980, 1988) insist.The answers to these questions have implications for the formation and persistence of norms.
Maurice Merleau-Ponty's (1996) embodied knowledge, Michael Polanyi's (2012) ski ls, and Dreyfus's (1993) work relating both concepts to modern cognitive science constitute the epistemological background of the present work.They highlight the importance of unconscious cognitive capacity and contend that unconsciously formed ski ls and tacit knowledge play a primary role in how individuals understand and cope with reality.
Learning and a ion cannot be explained by refer ing only to intentional and representational content.For example, an individual does not consciously make complex physical calculations when bicycling, even though it ap ears to be necessary for balancing a bicycle.To safely stop a moving car an individual does not need to calculate the stop ing distance, according to such varia les as the cur ent eed, state of the brake pads, friction between the road and tires, and air resistance.Individuals cope with life with the help of multiple capacities, some of which are not consciously contro led.These capacities seem to lack conscious representational content.
By introducing the concept of "tacit knowledge" Polanyi (2012Polanyi ( [1958]]) makes a significant contribution to the philosophy of mind and a ion.Tacit knowledge, unlike explicit or codified knowledge, is extremely difficult or impossi le to codify (express with lan uage) and transfer to others.Knowledge of riding a bicycle, playing tennis, or driving a car are examples of knowledge with tacit content.In the remainder of this paper these adaptive behaviors based on unconscious ski ls and/or tacit knowledge wi l be refer ed to as ski lful coping, according to Dreyfus (1993).To acquire such knowledge or ski ls an individual needs to go through an ap ropriate sequence of experiences relevant to the ski l being acquired.According to Polanyi (2012Polanyi ( [1958]]), ski ls are performed and assessed via subsidiary awareness.Such knowledge, ski ls and tools associated with them become internalized via experience.Polanyi su gests that they become integral to us in the same sense that our limbs are part of us.As such, a racket becomes an integral part of a professional tennis player and a wa king stick becomes an integral part of a lind man.
In both cases, unlike a novice, an expert does not need to pay attention to the use of his/her tool; it is as if they are using their own limbs.
According to Dreyfus's theory, which is based on Merleau-Ponty's (1996) Phenomenology of Pe ce tion, learning is not independent of the body.At least some knowledge becomes part of one's unconscious decision-making pattern and becomes a ski l.Tacit knowing and ski ls of this type become endogenous to the decision-making process, without representational content in the mind of the individual (Dreyfus and Dreyfus, 2004;Dreyfus, 1993, p. 24).In essence, ski l acquisition reduces the costly burden of rational calculation without compromising the outcome.
Hubert Dreyfus and Stuart Dreyfus (1988) describe five discrete stages of learning, from novice to expert: Novice; advanced beginner; competence; proficiency; expert.Throughout the process of advancing from novice to expert, individuals acquire new knowledge and behavioral patterns.While in the first few stages a learner acts in a rule-driven manner, in the later stages experience is assimilated in such a way that intuitive rea ions replace reasoned responses.On the road to expertise an individual recognizes increasingly more patterns and behaves according to newly acquired ski ls.While moving through these five stages an individual becomes less procedure-driven, less analytical, and more intuitive (Dreyfus and Dreyfus, 2004, 1988, 1980).
Merleau-Ponty, Polanyi, and Dreyfus agree that unconscious processes play a pivotal role in our behaviors and discredit a l attempts to analyze cognition without considering these processes.They also sup ort the notion of a holistic approach to learning and the importance of human intera ion with the environment.It is our conclusion that unconscious cognitive processes could be classified as fo lows: (a) Non-representationality: Ski lful coping is not accompanied by representational ideas and analytical processes.Instead, it is integral to the body.(b) Holistic nature: Ski lful coping is holistic.(c) Experience dependence: Ski lful coping improves with experience.(d) Stability: They are less sensitive to conscious cognitive changes and this makes them more sta le.That is to say, they do not change a l of a su den.Instead, they change gradua ly.To i lustrate, both learning and unlearning bicycling happen gradua ly and continuously in contrast to learning how to multiply two integers or learning the capital city of Austria.
According to Hubert Dreyfus and Stuart Dreyfus (1988), modern philosophy of mind and artificial inte ligence research underestimate the importance of non-representational learning, namely, learning that is not mediated by consciously accessi le representations.Similarly, in economics it is also highlighted that the a ions of individuals are only partia ly explaina le by conscious processes.In these ap roaches, which are critical to the mainstream idea that representational ski ls and knowledge are the only determinants of decision making, it is ar ued that unconscious cognitive processes are the ha lmark of expertise (Lane et al., 1996, p. 52).As more fee back accumulates, individuals progress from novice to expert.In complex environ-ments a novice simply relies on analytical computations and rationality, with limited cognitive resources, whereas experts employ a holistic, non-representational ap roach.Expert individuals "experience and understand their worlds only through their intera ions with other agents" (Lane et al., 1996, p. 75).The rational actor model in economics cannot adequately explain expertise of this sort (Lane et al., 1996).
In general, unconscious cognitive abilities are associated with expertise in a task, e.g.hitting a tennis ba l with top spin.However, these abilities and tendencies are not only confined to these kinds of tasks but also are related to norm-fo lowing behavior in societies.Just like in the case of expert and novice tennis players in which the former unconsciously does more tasks compared to the latter, in norm-related contexts experts unconsciously perform more norm-related tasks than novice individuals.In other words, we enlarge the extent of "expertise" to include norm-related decisions and ski ls such that individuals who have been members of their re ective societies for longer periods of time, accumulate and internalize experience and use them unconsciously are qualified as experts.Moreover, the chara eristics of task-related unconscious cognitive processes, i.e. non-representationality, holistic nature, experience dependence and stability are shared by norm-related unconscious cognitive processes which are the ha lmarks of expertise in social contexts.

Evolution of and adherence to norms in societies
Is expertise that results from social intera ions a factor associated with a herence to social norms?Put in another way, how does learning or expertise affect a herence to social norms?When individuals make decisions, they usua ly do not consider the effects their decisions have on society (Hardin, 1968); however, such consideration is of utmost importance to evolutionary social theory.As Gintis (2007) points out, the beliefs, constraints, and preferences of individuals in the social sphere matter.As such, individual learning makes sense in a framework in which reconciliation of individual and social outcomes is properly a dressed (Bowles, 2006).Therefore, the evolutionary ap roach emphasizes that individual learning occurs in an environment chara erized by strategic intera ions.
As discussed in the previous section, there is a vast literature on the complexity of decision making and coping with it.Bounded rationality, according to Simon (1996), highlights our limited capacity to engage in complex issues.We economize the use of our limited cognitive capacities by a hering to evolved rules of thumb.Other mechanisms that ena le individuals and society to cope with complexity include unconscious cognitive processes and evolved social norms.These mechanisms are foundational to how we think and behave (Young, 2015).In other words, unconscious cognitive processes and a herence to social norms reduce the cognitive cost of acquiring ski ls (Boyd and Richerson, 1985).
Rather than deciding which social norms to fo low and which not to, individuals simply inherit surviving social norms and learn through intera ion with others.In this regard, expert agents have the ability to a here to social norms and exhibit ap ropriate behavior in novel situations.It is sugge ed that expertise is related to the agent's level of understanding of the social context in which he/she is intera ing, which is acquired via multiple intera ions over time (Lane et al., 1996).Thus, according to evolutionary game theory it is reasona le to assume that the behavior of expert agents may "lead to novel situations that the participants construct together in anticipation of mutual benefits that they cannot clearly foresee […]" (Lane et al., 1996, p. 61).
In this framework social norms evolve through dynamic learning and these norms form the basis of social and economic order (Young, 2015).In other words, a herence to social norms provides the basis of social order.Here we a d to this framework the notion that expert agents a here to social norms in greater number than novice agents, resulting in social order.The inclusion of expert agents in a game-theoretical framework can significantly improve our understanding of a herence to social norms.
Evolutionary game theory usua ly treats the evolution of social norms as a non-cooperative common interest game (Bowles, 2006).Common interest is crucial to the formation of social norms, as the percentage of a population that wants a particular social outcome must reach a threshold level for the behavior associated with the desired outcome to be accepted as a social norm.In this case, a herence to the social norm provides great benefit to both individuals and society.In contrast, social norm violation is beneficial to free riders, as norm enforcement is endogenous, i.e. it is individual social agents that decide for themselves whether or not to a here to norms.The development and evolution of social norms has been a focal point of researchers (Hamilton, 1964;Trivers, 1971).Expert agents genera ly a here to social norms, which is beneficial to society.The fo lowing section provides an example that i lustrates how novice and expert social agents a here to traffic rules, i.e. a social norm.

Formation of traffic rules
Consider cooperation in traffic as discussed in Aydogmus et al. (2015).The evolution of cooperation in traffic can simply be represented as a prisoner's dilemma (PD) game, as fo lows.For the sake of simplicity, consider two drivers trying to reach their destinations.If there is no traffic both drivers reach their destination in 10 minutes and if there is traffic it takes 20 minutes, provided each driver in the traffic drives nicely; however, if one of the drivers (D1) does not a low the other driver (D2) to merge into traffic safely, i.e.D1 does not drive nicely, D2 may be forced onto another street.Consequently, reaching the destination for D2 takes 30 minutes, whereas for D1 it takes 15 minutes due to his a gressive driving behavior.If D1 and D2 drive a gressively, there is bound to be a co lision elsewhere and then traffic wi l eventua ly cause the average commute to be 25 minutes.These four scenarios can be summarized as fo lows:  D1 drives nicely (20 minutes to commute); D2 drives nicely (20 minutes to commute). D1 drives a gressively (15 minutes to commute); D2 drives nicely (30 minutes to commute). D1 drives nicely (30 minutes to commute); D2 drives a gressively (15 minutes to commute). D1 drives a gressively (25 minutes to commute); D2 drives a gressively (25 minutes to commute).
As is we l known, the Nash equilibrium for this game is that D1 and D2 drive a gressively, whereas in reality some drivers are a gressive and some are not.As mentioned earlier, expert agents tend to a here to a social norm (e.g.driving nicely) unconsciously, whereas novice agents tend to make a more conscious effort in an attempt to gain as much an advantage as possi le.To put it differently, a population of novice agents that are worse at a hering to social norms may exist in a state of defection more easily, i.e. everybody drives a gressively.Be aware that the op osite scenario, in which driving a gressively is the social norm, is as plausi le as the above given scenario.
We have said that the ha lmark of expert agents is adherence to social norms.A herence to social norms by expert agents leads to stabilized outcomes.In other words, expert agents facilitate maintenance of incumbent social norms.Hence, once a norm is formed, the population a heres to it if there is a sufficient number of expert agents.This is to say that for some societies driving nicely becomes the norm, whereas in others defection, i.e. driving a gressively, becomes the norm.There are other examples that model intera ion in traffic by using a PD game (see Levine, 2012, p. 25).
Experts fo low the majority under certain conditions in line with Boyd and Richerson (1985) and Henrich and Boyd (2001).This type of behavior is a shortcut to acquiring several adaptive rules of behavior (Henrich and Boyd, 2001, p. 81).For example, fo lowing the majority may lead to cooperation under certain conditions (Henrich andBoyd, 1998, 2001;Mengel, 2009;Smith and Be l, 1994).Evolutionary game theory, as used by Maynard Smith (1982) in biology and Robert Sugden (1989) in economics, tends to overlook the fact that learning social agents cope ski lfu ly.The formation and evolution of social norms is not only affected by rational processes, but also by unconscious processes.In this regard, the paper contributes to the literature by pointing out that the intera ion of novice and expert agents can improve our understanding of the process of norm a herence.

Discussion: tacit knowing, expertise and social norms
In order to fu ly understand the formation of and adherence to social norms, it is necessary to move beyond the boundaries of the rational actor model.There are several ways to learn and transmit social norms.According to Bowles (2006) and Young (2015), factors such as expected future gains, fear of punishment, and fo lowing the tendency of society may explain why individuals a here to social norms.For example, how punishment affects norm formation is studied by Yu et al. (2015).Here we focused on two a ditional factors that explain a herence to social norms.First, implicit learning that results from persistent intera ions with other agents causes novice agents to gradua ly develop non-representational ski ls and tacit knowledge over time and to eventua ly become expert agents.Second, a herence to social norms, which is made possi le by unconscious ski ls and non-representational cognitive processes, helps to sustain incumbent norms in a society.
The existence of differences in behavior between expert and novice agents could be confirmed using evolutionary game theory and a heterogeneous group of intera ing social agents.By incorporating the role of expert agents into the evolutionary game theory framework, the discussion of how implicit learning is related to adaptation in social environments can be improved.For such an attempt, see Aydogmus et al. (2015).They ar ue that relating the behavior of expert agents to the internalization of social norms and behaving accordingly without consciously represented rules in the mind show that the percentage of expert agents in any society is important for the maintenance of incumbent norms.Accordingly, once a social norm is formed a society a heres to it if the number of expert agents in that society reaches the necessary threshold.
According to the literature, the formation of social norms is considered a non-cooperative common interest game.The relevant theoretical game models rely on the assumption that there are no exogenous rules regarding the acceptance of social norms (Bowles, 2006).Moreover, the evolution of social norms is a common interest game, as a threshold number of norm-fo lowing agents is necessary for a norm to be accepted, and if the threshold is reached, a high-level of expected benefit is provided to society (Bowles, 2006).Nonetheless, as social norms are self-reinforcing, social agents want to fo low them when they think others are doing the same (Young, 2015).In other words, although a social norm may provide a basis for beneficial social intera ions, op ortunistic social agents may benefit by violating it, i.e. the we l-known free rider pro lem.The spread of norms that are socia ly beneficial, e.g.cooperation, is explained by group selection theory, which posits that a herence to certain social norms is beneficial to those groups that a here to them (Boyd and Richerson, 1985;Cava li-Sforza and Feldman, 1981).
A further research need is to determine whether an e a lished social norm is resistant to threats by non-rule fo lowers, which is most likely associated with the number of novice and expert agents that threaten the existing social norm.Hence, research should be conducted to determine the percentage of novice and/or expert social agents in a given society that is required to displace an incumbent social norm (for example, see Lozano et al., 2008, for a stochastic model placed upon a network structure).For reference purposes, there is an extensive literature on the enforcement of cooperative norms (Henrich and Boyd, 2001;Fehr et al., 2002).The present paper contributes to the literature as it is the first one to put together such disparate avenues of research on social norms and learning (novice and expert social agents) in order to examine the role of tacit knowing in a herence to social norms.The framework presented here could be generalized to the study of how several social norms emerge in the presence of heterogeneous agents.