1 Disputatio N ecessaria D e G ustubus : Why economics should mind human wants Steven J. Bosworth , University of Reading, s.j.bosworth@reading.ac.uk 25 June 202 6 Introduction In their seminal 197 7 article De Gustibus Non Est Disputandum , George Stigler and Gary Becker argue that revealed preference alon e lacks sufficient structure to explain people ’ s choices and economic behaviour. They instead advocate that economists assume behaviour to be motivated by rational self - interest because such a view yields sharp predictions across a wide domain of economic life. Economics can then be construed as the study of human activities where the homo economicus assumption they advocate makes accurate predictions. In the intervening decades that approach has been found unsatisfactory to a number of economists for varied reasons Alternative approaches which expand either beyond self - interest or strict rationality , collectively referred to as behavioural economics , have gained traction because they expand the scope of what economics is able to deal with (Camerer and Loewenstein, 200 4 ; Truc, 2002 ) The behavioural economics research programme has not been able to escape the criticism that its success in explaining aspects of human choice that the homo economicus paradigm could not relies on ad hoc complications to a model of human decision making which cannot make useful predictions outside of the purpose for which they were developed however (Gigerenzer and Berg, 2010) This chapter will argue that the critique is well founded, but that behavioural economics has stronger foundations than its practitioners may explicitly invoke , in the form of (embodied) 1 cognitive psychology. Most theories about preferences in behavioural economics can be justified with reference to scientific investigation of the human mind and reasonable postulates about human beings’ material and social needs. The human mind provides just enough structure for a theory of behaviour grounded in basic human objectives. We will first revisit the Stigler and Becker argument against unrestricted revealed preference. We will then argue, by analogy to their examples, that while structure is necessary to generate hypotheses about economic behaviour, assuming naively self - interested rationality can lead us astray just as easily. We conclude by arguing that a more general point is relevant: people’s wants must themselves have explanatory principles. In some cases we can deduce a narrowly self - interested homo economicus from these principles. However in other domains we deduce social preferences from these same principles A sound science of human behaviour must continu ously reconcile deducti ve reasoning about preferences with observed choice. 1 The embodied cognition paradigm in cognitive psychology rejects that cognition can be reduced to a set of computational algorithms for processing external stimuli and furthermore rejects that cognition is located solely in the brain (Shapiro and Spaulding 2025). I argue below (and elsewhere) that preferences are not located in individuals but instead arise from social proce sses and are subject to environmental influence Wh ere in the body we may locate cognition is not relevant for my purposes. 2 Background Stigler and Becker promoted a particular form of methodological orthodoxy in explaining economic choices. Their basic approach remains wholly compatible with the revealed preference framework developed by Samuelson (1948) , among others , in that individuals choose in such a way as to realise a stable preference ordering over outcomes given particular constraints (instrumental rationality) . Stigler and Becker wanted to elevate the explanatory role played by the constraints, namely prices and costs, within this framework over that played by tastes or preferences. The analyst should not only treat tastes as stable and well defined for the agent doing the choosing, but also ideally across the various agents being analysed. Stigler and Becker first and foremost are discouraging explanations of differences between people as arising from differences in preferences. Though it has proved obsolete by this point, t here is still much to recommend such an approach. The theory of revealed preference identifies choice with taste by axiom: if an alternative is selected it must be preferred to everything else in the choice set. The analyst however has limited information about the alternatives in the choice set and substantially less information about the preference ordering over those alternatives not in the choice set While one can explain the choice of A by person 1 and the choice of B by person 2 as arising from either persons 1 and 2 having different preference orderings or from having different choice sets , the latter explanation is much easier to falsify than the former. This is precisely what Stigler and Becker meant when they titled their article , that since tastes cannot be observed or “disputed,” they do not usefully explain human choices. Though revealed preference axioms are indeed sometimes violated , these violations are usually small in cost ( for example Dean and Martin, 2016 test this in a consumer grocery expenditure panel ). For concrete contextual environments, assuming fixed tastes is a reasonable simplifying assumption. W e should be more concerned that in Dean and Martin’s context revealed preference alone does not have much more predictive power than random choice! Owing to the strong influence of logical positivism on economics in the latter half of the 20th century via Friedman ( 1953 ) and others , a methodological imperative to search for explanations which refer to people’s circumstances rather than their tastes ma de a great deal of sense. Postulating a representative utility function provides enough structure to allow the analyst to focus on these observable circumstantial factors. E conomics has advanced markedly in its use of non - choice data in the last decades (Manksi, 2004 ; Camerer, 2007; Layard, 2010; Cooper et al., 2019 ) , making tests of preference - based explanations less tautological, the structure imposed by having to assume that human beings share largely the same preferences allows the discipline of economics to maintain a strong link between its theoretical and empirical practices. What then, did Stiger and Becker’s programme suggest as the common basis for human tastes ? Naturally, this was to be self - interested expected utility , possibly augmented with a paternalistic concern for close family members (Becker, 1981) , and invariant to social forces other than prices. Though Becker held a joint appointment in the Sociology department at Chicago, h is work carries through it the theme of explaining sociological phenomena exclusively with respect to economic variables. The hesitance with which sociologists have adopted economic models cannot have been helped by the pioneer of that approach ignoring people’s basic sociality and need for belonging Stigler and Becker were also following the much earlier established programme of cleansing economics of its psychological roots in utilitarianism ( Bruni and Sugden, 2007 ) and scepticism of psychology more generally (Gul and Pesendorfer, 2008 ) 3 Stigler and Becker propos e that economists restrict their attention to phenomena which can be explained with preferences of this very particular nature, what Mill termed homo economicus . In so doing we are able to better elaborate those explanatory factors which have to do with people’s constraints. This has proved fruitful. However there are more fruits to be harvested To point out that human beings are not homo economicus is not terribly insightful. Mill himself admitted as such (Persky , 1995 ) . But the ways in which humans deviate from this caricature have economic consequences and are amenable to economic analysis . By this I mean the se behaviours can be modelled as arising from constrained optimisation ; and exhibit predictable comparative statics with respect to thos e constraints. We should of course remain careful not to open ourselves up to explanations which have exp ansive degrees of freedom on the preference side, as these remain difficult to observe. Behavioural economics has come under criticism, rightly so, for its proliferation of heuristics and biases which qualify preferences. Behavioural economics is connected with the ‘ Heuristics and Biases ’ programme pioneered by Amos Tversky and Daniel Kahneman in the psychology of decisions Gigerenzer ( 1991 , pp. 102 ) provides an early critique of that agenda : “ Several of the explanations using heuristics are hardly more than redescriptions of the phenomena reported. ” The core of the critique is that heuristics and biases are in many cases inferred from the behaviour they seek to explain as ex - post rationalisations rather than deriving from a priori reasoning. I agree with Gigerenzer and others that such circular reasoning should be avoided on scientific grounds Instead I am proposing that economists, and social scientists more generally, should think very carefully about preferences. We should endeavour to explain more than what homo economicus would do, but nevertheless adopt disciplining structure on preferences which comes from outside the particulars of the phenomenon being explained. More specifically I propose that this structure come from postulates about human needs that can be supported by evolutionary arguments. My thinking about preferences is very much in line with what Cosmides and Tooby (1989) advocate in explaining human psychology. The approach following them is termed “evolutionary psychology.” That term has unfortunately also come to encompass exactly the sort of reasoning which a good social scientist, in the broadest spirit of Stigler and Becker, ought to avoid – namely ad hoc justification. Many bad arguments have been developed by observing differences between people and then “retconning” an evolutionary argument which must have produced it (see Confer et al., 2010 for a review of critiques) I rather mean that all human beings are social animals who must live in groups and make fitness - promoting choices that make economical use of information in an uncertain and changing environment. We must seek food, shelter, and other comforts, conserve our energy, and so on ; but also stay in each other's good grace s, combine to advance our mutual interests, and navigate complex forms of social organisation. Fortunately as economists, much of th e careful thinking about preferences has been done for us P sychology, sociology, anthropology, and political science have a leg on us in elaborating a coherent view of human cognition. As an example, the psychology of motivation identifies a few fundamental drivers of purposeful behaviour in wanting, care for others, the need for achievement, the need for affiliation, anger at material and social insult, and fear of uncertain harm (Heckhausen and Heckhausen , 2018 ) I and others have advocated doing economics being mindful of these motives and the circumstances which might elicit them (Bosworth et al. , 2016 ; Chierchia et al., 2021 ). More work in this vein can shed light on behaviour that otherwise 4 appears to elude rational explanation such as the crowding - in/ - out of intrinsic motivation by material incentives (Bowles and Polanía - Reyes , 2012). Behavioural economics is no monolith, but it is characterizable by a few key themes. The most important of these would be a rejection of Homo Economicus as an admissible model of human behaviour. While that model may fit some behaviours, the boundary between behavioural and mainstream or orthodox economics is that the behaviourists concern themselves with behaviour which is not. Another fundamental feature distinguishing behavioural economics from other heterodox approaches is a belief that behaviour which cannot be explained by Homo Economicus is ultimately reconcilable with most of the methods or assumptions of the orthodox model. Methodological individualism remains firmly in place, as does constrained optimisation. It only the objective function or constraints which need amending according to behavioural economics The desire to keep all mainstream economics’ insights and add to them has led to a proliferation of minimum feasible extensions to the standard model – perhaps a cognitive constraint in the form of non - equilibrium beliefs 2 ( Crawford et al., 2013 ) or a simple linear preference over inequality of distribution (Fehr and Schmidt, 1999). What is most efficient for the purposes of a single explanation may lead however to inefficiency when we want to apply the newly formed models to out - of - context behaviour. Which among the several existing social preference models should an experimenter apply to newly collected data? Should it be inequality - averse distributional preferences like those of Fehr and Schmidt or Bolton and Ockenfels (2000) . Or intentions - based reciprocity (Rabin , 1993 ; Dufwenberg and Kirchsteiger, 2004) or more tractable type /action - dependent reciprocity (Levine, 1998; Cox et al. 2007 ) ? In practice the one which fits best will be used, and so the researcher has as many degrees of freedom as there are parameters in all the models which have been published. Well perhaps not quite. There is consensus about general features of social preferences ( Fehr and Charness, 2025 ) just as there is – to greater and lesser degrees – in the fields of non - expected utility risk preferences ( Starmer, 2000 ; Fehr - Duda and Epper, 2012 ), non - exponential discounting ( Ericson and Laibson, 2018 ), non - equilibrium beliefs ( Crawford et al., 2013 ) and so on. The problem isnot that Homo Economicus is a model suited to a particular domain. Its extensions are likewise suitable for a different range of domains. The compatibility of behavioural economic models with the orthodox Homo Economicus model, oftentimes up to a single e xtra parameter, lends the enterprise an opportunity for an overarching and encompassing theory that ties those parameters to specific strategic or contextual features of the environment where it is to be applied. The research programme I advocate here and in ( Bosworth et al., 2016; Chierchia et al., 2021 ) is to connect the literature on the evolutionary origin of preferences (Samuelson, 2001 ) with that on the context - specificity of choice orderings ( Gilboa and Schmeidler, 2001 ). Concretely, a concern for the wellbeing of others is highly adaptive in situations where collaborative effort increases the size of the pie and ongoing relationships between motivationally aligned actors prevail . This is because good will elicits good will in kind in these situations. Conversely it would be very disadvantageous to take into account others’ welfare when bargaining over a fixed pie, or when needing to deter potential aggression. In such situations generosity invites exploitation. Expectations are also specific to the strategic context. Indeed it is most advantageous for one’s 2 The purpose of this chapter is the degrees of freedom afforded by preferences but the same principle applies to bounds on rationality. 5 career if colleagues expect you to pull your weight on team projects, but also expect you to fight for your share when negotiating salary. Insisting that this person’s behaviour be modelled as always selfish or always altruistic does not then put any desirable or disciplining structure on the explanation if it does not fit when both behaviours are observed. At least one must be an exception to the ‘simple’ rule. It is also no less parsimonious to respond “it depends” to the question of whether people are altruistic or self - interested if one has a parsimonious theory of the “it depends”. This framework can furthermore explain – and not just accommodate – differences in choices by people who face the same incentives or choice sets. In Bosworth, Singer, and Snower (2016) we show that between - person differences in the distribution of expected social settings induces these different people to adopt different context - contingent preferences over actions via two channels. Directly because some are simply more likely to encounter situations where selfishness is more adaptive (complementarity in motives) ; and indirectly via the actions of the other people whom they are likely to meet in those situations (complementarity in types ). This explanation creates an ancillary testable prediction as well – that frequency and type of environment are associated with cross - sectional differences in preferences. The reader will note that I have used the word advantageous somewhat loosely above. We wish to expand from Homo Economicus to a wider range of context - contingent preferences, but avoid the pitfall of overfitting inherent in the revealed preference approach as criticised by Stigler and Becker. So we must tie down those preferences by requiring them to satisfy an evolutionary advantageousness criterion . Why then does this not reduce to the same? Put differently, why would Homo Economicus not be the most fit among possible preference types as it would maximise an agent’s expected material resources? One response is that implementing fully rational and selfish choices is cognitively costly (Kahnema n, 2011; Székely and Michael, 2021 ) or that attempting to maximise makes inefficient use of limited information ( Simon, 1956; the Gigerenzer and Selten , 2001 ). More relevant to the social domain, most human institutions and endeavours are coordination problems ( Bernard, 1965 ; Camerer and Knez , 1997 ; Adsera and Ray, 1998; Marengo and Dosi, 2005 ; Young, 2015 ). By coordination problems I mean joint projects whose success hinges on costly individual effort, but is much more efficient than autar kic production. Because coordination problems have multiple equilibria, they can support altruism as an evolutionary stable strategy (Boyd and Richerson , 2010 ). Humans as a social animal primarily engaged in group endeavours should therefore be particularly shaped by group selection forces (Boyd and Richerson, 2009) Another feature of most human interactions is that they have an ongoing or dynamic character. What are ex - post one shot or finitely repeated interactions often have some ex - ante uncertain ty about the game horizon ( Salmi, 202 4 ) or at the very least about players’ confidence in each other’s back ward induction ( Alao ui and Penta, 2016 ). Finally, the very notion of a long - lived self demands a modicum of internal coherence. It is simply less costly, in terms of cognitive dissonance (Festinger , 1957 ) to reconcile one’s need to maintain long - run relationships of mutual benefit with consistent (if conditional) morality than with ruthless opportunism. Such a need for coherence gives rise to self - signalling ( Bénabou and Tirole, 2004 ) and social identity formation (Akerlof and Kranton, 2000). What then, is the cost of failing to appreciate the contours of human preferences and their origins? The straightforward answer is that models based on Homo Economicus will often fail to 6 make accurate or useful predictions in novel contexts . If we believe, as Stigler and Becker did, that human behaviour responds to material incentives in predictable ways, we should want to push that explanation as far as it will go. Thinking carefully in advance about what preferences it is reasonable to assume of the agents in our models will ensure that the explanation we arrive at is more robust to having mis - specified other elements of the explanatory structure, such as for example the production function(s) or noise structure . Following Stigler and Becker’s five - fold examples of applying price theory in the fields of family economics, addiction and taste adaptation, traditions and customs , advertising, and fashions, I will show how each can be enriched , and by Stigler and Becker’s own criteria sharpened , by assuming “non - standard” preferences. Economics of the Household One of the topics that Stigler and Becker use to motivate their insistence that economics focus on costs , opportunities , etc ., is in the economics of the family Becker (1981) is most famously associated with the idea that the rational explanation for why there i s a gender division of market work and household work is because the spouse who is ‘ more productive ’ in earning a market income (i.e. has a higher wage) specialises in that activity and that spouse who earns a lower wage specialises in home production 3 Apart from any essentiali s ing postulates about what men and women are good at , this theory has an interesting set of predictions about what is going to happen when the distribution of market income within the household changes Namely if the explanation is solely about cost and benefits we should see for th at increasing minority of heterosexual households where a woman earns more than a man the man should bear the majority of household production. This is not borne out in the data Husbands with lower wages than their wives do not substantially reduce their hours worked and do not pick up greater shares of housework (Bertrand et al., 2015) . The asymmetry in the distribution of household and market labour when the gender identity of the person earning a higher wage effectively falsifies the Stigler and Becker explanation. Hook (2016) furthermore shows that relative housework distribution does not differ between weekdays and weekends, suggesting men’s opportunity costs are not relevant to the division of labour T he distribution of housework instead seems to conform with societal norms prescribing these tasks to women (Lichard, et al., 2021 ; Cavapozzi et al., 2024 ) T he literature on the economics of the household has been in the vanguard among economists (e.g. Lundberg and Pollak, 1993 ) for recognising that social norms provide a functional structuring role in human societies. These social norms have regular properties of stability which we can analyse using the tools of game theory because social norms involve wanting to do particular things conditional on what other people do 4 and what other people expect one to do 5 (Bicchieri, 2005) . That lends itself very well to the modelling of psychological game theory (Gavrilets et al., 2024) Using the social norms lens there have been a number of puzzles that are difficult to understand from the perspective of just costs and benefits or production functions, but which make sense 3 This is the most prominent, but certainly not sole, application of Becker’s Theory of Social Interactions (1974), which is reiterated in De Gustibus 4 This is the descriptive aspect of social norms. 5 This is the injunctive aspect of social norms. 7 in light of norms. One very recent example of this of a paper of my own together with Almudena Sevilla and Nacho Gimenez - Nadal (Bosworth et al., 2025) analysing the time allocation of mothers towards childcare stratified by educational status. We try to explain a phenomenon that is pervasive across rich countries, which is that there is an education gap in parenting time. Parents who have greater levels of education spend more time with their children than less educated mothers and the gap between the two has been growing (Dotti Sani and Treas, 2016) The total amount of time invested has also been growing (Doepke and Zilibotti, 2021) T h e pattern we document does n o t really make sense from the perspective of opportunity costs C ertainly people with more education have a strong association with higher market wages and so therefore the y have higher opportunity costs for each additional hour spent with their children th an would someone who has a lower market wage ( associate d with people of l ower education al attainment). An other thing which is puzzling about this pattern is that it cannot be rationalised with respect to the production function for human capital 6 The human capital investment literature suggests that each additional unit of time invested in increasing children's human capital is no more productively spent by mothers with higher levels of education than it is by mothers with lower levels of education 7 I t is mothers with the lowest socioeconomic status whose children would benefit the most from them spending additional time with them! (Agostinelli and Wiswall, 2025) A final confounding piece of this puzzle is contributed by our work using data from the American Time Use Survey suggesting that during these childcare episodes , hedonic well - being (at least if we assume that self - reported happiness measures it) is lowest for the most educated mothers. Why would people who have the highest opportunity costs, lowest marginal product, and least hedonic enjoyment do more of any parti cular activity? W e resolve this puzzle by not ing that people take pride in how other people view their children W anting to appear as a high human capital person is something that confers its own set of benefits ; and that motive combined with the inherent scrutiny on one’s child that comes with socialising with other parents leads strategically to a fragmentation of social groups along human capital lines. T he ‘ entry costs ’ of socialising with high human capital families is that one’s children need to be able to withstand the scrutiny of highly educated parents and comparison with other well - heeled children. I t is only those mothers with the highest levels of human capital who se children are able to 8  S o from a very simple motive to impress others within one's one social circle we get a resolution of this puzzle , which if we were purely looking at a social costs and benefits would not make sense I ncorporat ing human beings ’ social motives rescues the price theory framework of economics of the household. Addiction and Taste Adaptation The model of rational addiction put forward by Stigler and Becker is one in which careful reasons for increasing marginal utility of consumption are postulated and then matched to stylised patterns of consumption that we would characterise as addiction. In its most basic form, the argument is that by consuming particular products these products then suggest to the consumer only after having consumed the product that buying an additional unit would be more 6 See Ramey and Ramey (2010) for an argument of this type. 7 Cunha et al., (2010); Gayle et al. (2018); Caucutt et al. (202 3). 8 Such scrutiny can also differentially affect fertility by education level (Kim et al., 2024). 8 satisfying than buying the first unit. This might be rationalised by informational problems - people cannot foresee what the experience of consuming the good will be until they consume it. It could also be something like what Stigler and Becker call consum ption capital, where one must develop the capacity for enjoying a particular good. The example they use is appreciation of classical music B ut a good model for thinking about how to appreciate classical music does not pas s the sniff test if we wish to model what is properly understood to be addictive consumption. Furthermore, I think that these sorts of models, developed from Stigler and Becker's work most famously by Becker and Murphy (1988) , suffer from the exact critique that Stigler and Becker themselves were making in De Gustibus The economist observe s a pattern of choice and then preferences which rationalise those choices are created post hoc to justify th em. There is no argument made for why these preferences arise n or why they apply to particular goods, nor why they apply to particular people and not other people. Again, this is a reference to preferences differing across people which cannot be directly observe d beyond choice. E conomists following this theory have missed the reasons that people seek out and enact addictive behaviours , despite those reasons potentially interfacing with the broader economy! Instead, we should reason from the premise that there must be some need that that people who engage in addictive behaviours have which those behaviours satisfy . Th ese need s should be consonant with what we generally know about the human mind, basic human needs, human psychology, and that more basic framework should be able to explain differences across people and across classes of goods. O ne decision - theoretic explanation that has been put forward for addictive behaviours is that they exhibit the kinds of patterns which we also associate with hyperbolic temporal discounting (Frederick et al., 2002) H yperbolic discount ers do not just valu e the future less than the present , but in addition underappreciate in the present how they will value the far future once they get to an intermediate point in the near future. In doing so they draw up plans which are not dynamically consistent. As applied to addiction (Ainslie and Monterosso, 2003) we can for example say that a person who smokes cigarettes and is addicted to smoking cigarettes might feel that they eventually would like to quit because they know that there are adverse health consequences which are coming far into the future but they feel very strongly that they don't want to quit in the immediate present and will therefore often make plans to quit at some point in the future such as when the last cigarette in the packet is expired. Bernheim and Rangel (2004) study how dynamic inconsistency might be cue - triggered, with a particular application to addictive behaviours. A more general treatment may be found in Harris and Laibson (2013). There has been a lot of work on the cognitive underpinnings of temporal discounting or behaviours which exhibit temporal discounting like phenomena. David Laibson’s entire body of work goes in this direction and a lot of it has to do with cognitive limitat ions. In Gabaix and Laibson (2022) people try to project the how they will feel about future experiences but everything which is in the future is speculative and the intensity of how one will feel is very difficult to predict, whereas the intensity of how we feel now is very clear. So our inability to look ahead and anticipate how we will feel about stopping addictive behaviours in the future clearly fits within this explanation for hyperbolic discounting. It is a straightforward application of hyperbolic discounting to behaviours whic h have very high short - term benefits and rather delayed long - term costs. One other thing economists have missed about addictive behaviours in trying to rationalise them is that the people who tend to be drawn into addictive behaviours have characteristics in 9 common which the things that they get addicted to don’t necessarily share. People can be addicted to gambling; people can be addicted to substances; people can be addicted to sex; but in all of those cases psychologists say that the addictive personality l acks some baseline level of satisfaction with their life (Shanahan and Copeland, 2021) . They feel that they are in an injured position and that their reference point of unhappiness - that is their current psychic state is far worse than where they would like to be - makes them prone to risk seeking or receptive to experiences which will soo th their psychic injury. This explanation has been put forward to explain why impoverished people pursue behaviours that perpetuate their poverty and disadvantage , w here it seems rationally they should be avoiding the vices they seem to have a preference for (Karelis, 2007) 9 Th e conceptual backbone of this argument is that the curvature of people’s utility for both sure and risky outcomes exhibits reference dependence of the sort in Kahneman and Tversky (1979): increasing marginal utility (and consequent risk seeking) below the reference level and diminishing marginal utility (and consequent risk aversion) above it. Rigoli and Pezzulo (2023) apply the predictions of reference - dependent evaluation models to addiction specifically, and argue that they can explain key regularities such as withdrawal, tolerance, craving and so on. Relative economic deprivation can form one such reference point. Pickard et al. (2024) find support that upward social comparisons place people in a loss frame and increase risk taking. The literatures on relative deprivation, risk seeking, and addictive behaviours paint a picture, consistent with behavioural economics, which is ultimately more coherent than the ad hoc assumptions about increasing marginal utility for particular goods and consumption capital put forward by the “rational” price theory literature. I assert this understanding is ultimately more parsimonious and grounded in what we know about how the human mind works. The Persistence of Traditions and Customs Stigler and Becker’s third application is to the supposed stickiness or stability of preferences. By this they mean the tendency to prefer courses of action which are seemingly disadvantageous under the present circumstances (alternatives, constraints, prices, etc.) but which have been used previously. They rationalise a seeming preference for stability or conservatism by appealing to the cost of acquiring information a bout where the best options are to be had. Decisions which appear irrational under perfect information are rationalised by the addition of this information constraint. Stigler and Becker argue that this framework is preferable to simply assuming a taste for consistency or conservatism because it makes predictions about what happens when the price of acquiring new information about the best available options changes. We s hould see more conservatism in the presence of worse uncertainty and less conservatism where information becomes more precise and so on. There is certainly value in trying to explain persistent choices with respect to some economising of information, and I think without admitting it Stigler and Becker are recreating in a different guise the satisficing approach of Herbert Simon (1956) . Where acquiring information about the 9 A similar explanation by Mullainathan and Shafir (2013) argues that dealing with poverty ’s acute material scarcity consumes cognitive resources which become unavailable for an optimal consideration of choice options. 10 best options is itself costly, we might thin k of satisficing as the ‘reduced form’ of the Stigler and Becker model. I would simply add that even where information may be abundant the mind presents costs in processing this information ( Falkinger, 2008 ). I want to focus more on what they claim their model can answer, which is this John Stuart Mill quote from A System of Logic (1843): It is not true that the actions even of average rulers are wholly, or anything approaching to wholly, determined by their personal interest, or even by their own opinion of their personal interest.... I insist only on what is true of all rulers, viz., that the character and course of their actions is largely influenced (independently of personal calculations) by the habitual sentiments and feelings, the general modes of thinking and acting, which prevail throughout the community of which they are members; as well as by the feelings, habits, and modes of thought which characterise the particular class in that communit y to which they themselves belong.... They are also much influenced by the maxims and traditions which have descended to them from other rulers, their predecessors; which maxims and traditions have been known to retain an ascendancy during long periods, even in opposition to the private interests of the rulers for the time being. Stigler and Becker completely miss the social aspect of the phenomenon which Mill describes: information about what may be the right option comes from other people. Mill is simply elaborating that some of those other people may be older, and so pass opinions about how things should be done across generations. Social information certainly helps people navigate choice environments with costly information. But specifying how social information is transmitted and how it motivates people’s choices further illuminates why information regimes may transition, sometimes rapidly. W hat M ill is trying to capture with his appeal to an intrinsic preference for tradition is not just that existing ways of doing things might be efficient and hard to discover but also that human social relationships play a role in the transmission of this information . The social information embedded in traditions is often imposed by peers When one do es something against tradition one finds out from other people telling one off. T he ‘pressure’ which pushes information through the social network is therefore the motive for normative conformity. Human beings do n o t just economise on information on an individual level ; they are also adapted as a social animal to economise on information processing at a social level ; and they do this through social network s. S ocial inference is an evolved capacity 10 of human decision making and it allows other people to do much of our information discovery work for us In the economic models of social norm adherence 11 the key driving motivator for behaviour is that people want to take actions which other people normatively expect them to take, generally conditional on w h ether they expect that norm to be generally adhered to by others. T hese are known as respectively the injunctive and descriptive aspects of social norms (Cialdini et al., 1990) . B ecause of the conditionality with which people follow (injunctive) normative prescriptions, they in essence create coordination games. T his is the part of the structure that allows them to efficiently convey information ; because there i s a strong incentive for people to try and infer what it is that other people expect them to do , and to align their expectations in a Nash equilibrium. 10 Axelrod (1986); Henrich and Boyd (1998); Boyd et al. (2011). 11 Sugden (2000); Bicchieri (2005); Battigalli and Dufwenberg (2007); d’Adda et al. (2020). 11 The injunctive and descriptive aspects of social norms are important because they create second - order incentives that allow the threat of relatively minor punishment to affect compliance in a strong way. This may be seen in the models put forward by e.g. Bowles and Gintis (2004) ; Henrich and Boyd (2001) ; Boyd, Gintis and Bowles (2010) where needing to appear to be enforcing the norm on other people dramatically magnifies the transmission effectiveness of the relevant behaviour specified by the social norm. That is if I do not enact a punishment on a norm violator, I should expect that that someone else will punish me and I should expect to be punished for not punishing someone who fails to punish a norm violator, and so on. 12 The key thing to note about this is that these social aspects both allow an economisation of information about the particulars of what the norm might be and also economisation on potential enforcement efforts. One is given very simple objectives: to be see n as correct in the eyes of others, and to want to see others do correct things. Because effective injunctive social norms hardly ever call on punishments to be enacted (because of the surety with which people expect to be punished for deviating) they allow human beings to coordinate actions across a wide range of potential collective action problems. Especially i f we take the cultural evolution perspective of Richardson and Boyd (200 4 ) , the ability for groups of people to coordinate on solutions to collective action problems favours the survival fitness of that group and of the practices which enable their success (through copying); and effectively transforms collective action problems into coordination problems. Making explicit reference to the technology that allows this economising of information to happen also allows us to explain cases where tr