Andrew Frain pointed me to this article he wrote with Nick Jans, The curse of individualism in organisational thinking, adding:
I wanted to share it with you because I think Nick’s and my concerns align with yours when it comes to the state of the behavioural sciences. I’m specifically thinking of your concerns about (a) the dominance of individualistic perspectives, and (b) the disunity and incoherence of the field (e.g. the piranha problem). The above essay speaks to both, albeit with a particular focus on organisational psychology. I would like to think that the implications for political science, or behavioural science in general, are apparent.
I like the article by Frain and Jans a lot, and it reminds me of a few things we’ve discussed in this space over the years:
The problems with the scientist-as-hero narrative (and lots of examples, including here, here, here, here (giraffes!), here, here, and, most recently here).
Problems with the currently-standard individualistic theories in social science: One issue here is that the individualistic model is so dominant, such a default, that sometimes it seems that people don’t even realize it’s an assumption. I discussed this in the context of how to quit smoking and here in the context of what have been claimed to be transformative experiences. And in this article I wrote:
What we have to do . . . is to go beyond the individualistic approach to survey sampling. Traditionally you ask a person about themselves, maybe about their family, and that’s it, as if each respondent or each household lives in a little bubble. There is such a thing as society, and we can ask people all sorts of things about who they know, what their friends do, and so forth. We can ask whatever we want!
The above-linked article by Frain and Jans goes far beyond my occasional musings in that they engage with all sorts of things in the psychology and management literature that I wasn’t aware of.
That was a really interesting read, but it left me with a big question: what would be the unit of analysis for a psychologist who wanted to study these? Social identity seems very well suited to qualitative methods that someone working in a more statistical paradigm has to ask questions of individuals who have social identities and are situated in groups and cultures. Is the idea that researchers should study the types of practices you described it is still social identities? It seems hard not to get dragged back to an individualistic approach.
Thank you very much to Andrew for shouting us out here, and the positive thoughts! Lovely also to see the discussion.
The unit of analysis question is of course valid (thank you RJB). The key here is to not let an analytical lens, or domain of enquiry, be confused with the structure of the system being studied. So yes, by all means study and measure individual psychology or individual behaviour, but don’t slide into the presumption that the individual is in some way independent from, or more important than, the rest of the system. Or, said much better than me:
“People – persons – may reify “I” and “me”, but psychologists shouldn’t, except as they recognise the causal-functional importance of people’s own reifications” (Smith, 1980).
Hopefully that helps and happy to field questions. Cheers
Towards RJB’s point above:
I don’t think the metatheory the authors write about is a metatheory; its a domain of inquiry. Psychology arose 140 years ago as the study of the individual’s psyche. The study of social action already existed as a separate domain, and would around the same time split into ”economics” and ”sociology” (roughly as we know them today). So aren’t they just saying that (some) psychologists should switch to sociology? That’s fine – but I think that debate is best approached from a theory of science perspective, not the critical theory approach the authors seem to take.
Also:
1. Mats Alvesson is not a psychologist.
2. ”Maslow’s view should be a non-starter.” But… it is?
Perhaps Frain and Jans are right that the individual receives more scholarly attention than is warranted. But to strengthen their case, they should at least engage with perspectives from evolutionary biology, particularly kin selection and cultural evolutionary theory. Kin selection predicts that individuals are more inclined to prioritize their own welfare, and that of their close relatives, over the welfare of unrelated others or strangers. This presents an obvious challenge for building an effective military: it requires overcoming the individual’s psychological preference for self-preservation (and family preservation) in order to foster collective commitment.
Frain and Jans attempt to sidestep these preferences by critiquing Maslow’s hierarchy of needs, which places safety and physiological needs at the base of the pyramid. They argue that according to Laslow-inspired theory, “enculturation and belonging that drives employee commitment and wellbeing are seen as fringe factors” and that Laslow’s model “flies in the face of everyday observation (adventure sports is but one obvious example).” This critique is not persuasive. How much money or time does the average person devote to rock-climbing, bungee jumping, or hang-gliding? Now compare that with what they spend on food, water, housing, and health care. Everyday behavior overwhelmingly affirms the priority of basic survival needs. To dismiss this is to misread the evidence, and to ignore a key challenge of organizational design in militaries.
I also think that it is important to acknowledge the unique challenges of the military and military leadership. Soldiers should not be thought of as ordinary employees, and I find it weird that this article doesn’t highlight the extraordinary differences between the soldiers job and that of a typical employee. The military is a special case because it places people in environments where their strongest individualistic instincts of self-preservation and protection of kin become potential liabilities. Isn’t a key challenge of military culture to override these default human instincts through training, socialization, and enculturation? You make a better case for that enculturation by acknowledging the self-preservation and safety of the individual than by pretending it doesn’t exist. If the military’s goal is to train individuals to downplay their own safety for the sake of others, then “knowing your enemy” means grappling seriously with Maslow and with the biological underpinnings of self-preservation, not dismissing them.
I know I’m a broken record on this, but the presumption of individualism in economics is encoded in the curvature properties of production and utility functions. They are assumed for convenience, but they have social theoretic content. And in evolutionary biology, the unit of selection is a trait, but the fitness landscape of traits is nonconvex because traits are not individuals; they interact through structure. Multiple local equilibria abound.
Peter what do you say to some stuff ive read from Steve Keen saying that in surveys of factories and things the operations people always report downward sloping marginal cost of production, that the design of production lines is always for much more production than typical values (excess capacity) as people want to be able to grow without spending a lot to revamp their facility, and that a core assumption in economics, that marginal cost increases and that pricing equilibrium occurs where marginal cost of production equals customer willingness to pay so marginal profit is zero… Is complete horseshit.
It seems to me this and several other criticisms make economics as taught at least to undergrads more like a religion than anything else… We are asked to believe things on faith because the empirics clearly show something else and thats not convenient.
These are parallel observations. There’s a large literature on increasing returns in production, most prominently Brian Arthur. My point is different. A, it’s theoretical, although there are some empirical applications,e.g. spacial econ. B, the “violations” I’m referring to are off-diagonal in the matrix of cross partials, not the principal diagonal as with increasing returns. Additionally, excess capacity is most likely due to the asymmetry between insufficient and excess supply in most competitive contexts. I recall blogging about this many years ago.
While waiting for a response from Peter:
1. I think the typical assumption is a vaguely u-shaped marginal cost curve, not an upwards sloping one, because of economy of scale.
2. Businesses are, of course, not ”always” planning for excess capacity. E.g. ”just in time”.
I agree that introductory courses in economics can make things sound very definitive.
But these ”criticisms” of ”economics” are typically targeting a vague conception of the first few chapters in an introductory textbook.
I can reveal that memory doesn’t work in the way described in an introductory psychology textbook. That might be a problem for how we teach psychology, but not for the science an sich.
I wish simplistic assumptions were confined to the intro chapters of econ textbooks; unfortunately they are used almost everywhere. Welfare econ is *completely* based on them, and, to pick one egregious example I analyzed in my climate book, simple CES functions are central to integrated assessment models. In this sense individualism in economics is deep, pervasive and nearly invisible.
Mattias.
Just In Time is basically a way to deal with the fact that if you just ran flat-out you’d produce more than you could sell. The idea is rather than running flat out, finding out that you’ve overproduced and having to warehouse stuff, furlough workers, have production problems on start-up, etc you instead pipeline specific quantities of stuff that relate to your needs to produce some explicit quantity of stuff per unit time, for reasons other than price equilibrium.
We might think that a company that could produce say 1000 widgets a day and it’s cheaper to produce per each than 500, but chooses to only produce 500 is “leaving money on the table” but my impression is that lots of people face a situation where they can’t just lower prices and get people to buy more of something. Maybe on longer timescales, but on short enough timescales (months, or a year ish) not necessarily. For example, you’re producing widgets for a hardware store, if you drop prices, the hardware stores will order the same amount, sell the same amount, and pocket the extra profit as middlemen. So the factory has no incentive to produce more and sell at lower prices. The consumer may not really need more, and dropping retail prices just leaves the hardware merchant with lower profits. If the retail price could be reduced, perhaps new uses for the product would be found, but this takes significant time, and we sit at non-optimal situation for a long time, long enough to put people out of business.
My impression is Keen claims these kinds of situations are way way way more common than anything at the foundation of Econ theories supposes.
Myself, for a book I’m writing, I created a dynamic model in which a manufacturer sells bolts to a wholesale warehouse, who sells to a chair maker… A counterfeit producer enters the market and undercuts the real manufacturer with bolts that fail more easily. The chair maker buys a random mix of bolts from the warehouse, and begins to have failures, so increases their cost of QC, which makes their chairs less profitable and they shift production to other products and raise prices, which reduces their sales, the feedback between all these things clearly shows what you expect, but if all the parties are adjusting their actions there isn’t necessarily any kind of equilibrium and it clearly has potential oscillations through time and soforth. For example the counterfeiter can adjust their prices through time to make the counterfeit bolts be sold more during high bolt price periods and less when bolt prices decline (ie. when people stop buying bolts because they suck and to incentivize more sales the warehouser drops prices)
One thing that’s assumed almost universally in Econ is that there’s “an equilibrium” and imho this core assumption is wrong in vast stretches of real econ… For example, where is the equilibrium in the copper price graph:
https://tradingeconomics.com/commodity/copper
I’m afraid I may be getting off topic, but there is probably a connection here to simplistic individualistic models of human behavior… people behave the way they do because they’re embedded in large systems that react to their behavior. If you rock the boat in some ways you get fired, in other ways you get massively rewarded… if you can find a rule to guide your behavior (such as the feedback rule for the bolt counterfeiter) you can make money, but only in the presence of the other people not behaving too differently… etc.
I’m not familiar with Keen, but I doubt there are any general conclusions regarding the shape of production functions and whether excess capacity is chosen. I know that most economic theories assume that marginal costs eventually rise with output and that producing beyond capacity (not the actual limit of capacity but the economic definition of capacity – where average cost is minimized) is optimal in such cases. I also know that in particular industries (telecommunications networks as an example) that installing excess capacity is a better choice rather than having to add capacity later (think of wireline technologies where lines are buried). So, there are cases where both are true. Neither case causes economic theories to explode – indeed, the declining marginal cost case is the textbook case for “natural monopolies.”
I believe that Peter (he can correct this if I am wrong) is referring more to cases where production costs include social costs (externalities) rather than solely private costs to the producer. In such cases, nonconvexities can arise and they do cause major problems for economic theory. But that is different than the traditional theories of production dealing with private production costs. And, in the full social cost case, I do believe that the assumptions are convenient for economic theory, despite the fact that they may not be realistic at all.
Yes, there’s a literature going back to the 70s on nonconvexity and externalities, but the issue I have in mind arises with no externalities in the Coasian sense. Each worker’s performance affects the others’ performance etc.
So Peter, I am not steeped in econ jargon, so it helps me to think about concrete examples, so for example suppose we’re producing some clothing… and there are 20 people cutting and sewing… they produce quantity Q1 per day. But there are frequent jams and issues with the machinery. So we hire a new person who is a maintenance person who looks for ways to reduce problems with the machinery etc. Let’s say for ease of calculation this person is paid the same as the others. So now we have cost 21W, but we produce quantity Q2. Whether the average cost per garment goes up or down depends on whether 21W/Q2 is greater than or less than 20W/(Q1), so whether Q2/Q1 is greater than or less than 21/20 = 1.05. If the new person makes everyone 15% more productive the cost of production goes down… if the new person makes everyone 3% more productive the cost goes up…
now, whether people get 15% or 3% more productive may depend on whether there’s a particular worker who overuses the maintenance person’s time because they don’t understand the changes the maintenance person makes… or if the maintenance person’s large improvement in productivity makes the manager look bad and he prefers to fire the maintenance person rather than lose his own job or etc etc
Is this the kind of stuff you’re talking about? you can’t think of the cost of production as a function of the number of workers, but rather as a function of the *set* of workers and their particular skills and attitudes and soforth.
Getting there. But now imagine there exists a set of tasks that each worker might perform, either by choice or assignment. W21 has an effect on the Q effect of each worker-task pair in combination with the formation of the other pairs. Solving this for a global optimum is theoretically possible if you have a ton of data. What isn’t very helpful is trying to identify individual MPs for individual workers based on a production function that assumes no interaction effects. Sorry this is non-numerical, but it should be obvious the effects have to be nonlinear if the cross-partials aren’t going to zero out. (This was a confusion in the initial literature on nonconvexity and externalities.)
Peter, put another way people exist in a system and their behavior affects other people’s behavior and they have to kind of directed random-walk through an ever changing landscape of what might benefit them… trying things, and seeing what others do about it, and without these *systems dynamics* ideas at least considered, you’re hopelessly lost.
I’ve been an advocate of dynamical models for a long time. You can see some discussion elsewhere in the thread of a simple one I built for a book:
https://statmodeling.stat.columbia.edu/2025/09/10/going-beyond-naive-individualistic-models-of-social-science/#comment-2403410
I’ll be interested to look in a few days when I get home from a camping trip. It’s a bother to peck away at a cell phone….
This is in response to Daniel’s response to my comment:
Sorry, I think I see what you are driving at now.
I don’t think the criticism you attribute to Keen is very thoughtful. JIT is a planned strategy – it sets production capacity. But regardless of JIT, ”always” is an exaggeration (at best). That adaptation is not instantaneous is also not news. Fluctuating prices is perfectly consistent with a trend towards an equilibrium, but one which changes all the time due to constant exogenous shocks.
Now, your point is that unless that evolution is modelled explicitly we aren’t capturing important aspects of the process? (This is the point where I start to see where you’re coming from). I agree that this can be the case. But every model is built for a purpose, and different kinds of models are differently well-suited to different purposes.
Now I finally get on topic in relation to the original post again: I think we need a plethora of models (and sciences) to fulfill *different* purposes, because there isn’t one privileged purpose that trumps the rest. I think ”naive individualism” is fine. As is ”naive collectivism”. Or a mix. It depends on what the purpose is.
As a final point, I’ve enjoyed this discussion. Thank you!