Computational Constraints Mean Sociology Is Bullshit Too

By Graham Peterson

As a undergraduate I wast thrilled by Gerd Gigerenzer’s criticisms of economics in Rationality for Mortals.  Gigerenzer is Herb Simon’s old buddy, and at least as smart.  You should buy his book.  The takeaway is an old criticism of economics: humans are so computationally constrained in terms of the CPU evolution has given us, that it is impossible to imagine people actually rank their preferences, transform them on a monotone map to some function, and then maximize that function relative to a battery of cost vector information.

The criticism is extremely appealing for a few reasons.  First: it nips at the heels of economics while staying within the technological metaphors of computers.  Criticisms torn from the pages of the humanities never fare well.  Second: it’s intuitively plausible to anyone who’s solved even a rudimentary constrained optimization problem.  The idea that we’re doing so subconsciously bajillions of times a day, when it takes days of instruction for a 19 year old to find the top of a parabola, is silly.

Moreover, there are some great anecdotes suggesting another mechanism: fast and frugal heuristics.  It turns out that baseball players do not observe the parabola a ball flies in, calculate its first and second derivatives, and then run to the spot they’ve predicted the ball will land — they just run around and keep the angle between their eye and the ball constant — eventually the ball lands where they are.  Generalize from anecdote to a big toolkit of heuristics, and you’ve got a plausible decision mechanism.

But how many heuristics does one need to really get through the day, or an argument with one’s lover, say?  The list is potentially infinite.  And then we’re right back in the snare of the computational problem.

In sociology (and originally in the separate study of Organizations, which economists have forgotten that Simon embarked on with sociologists, and that people like Ken Arrow were enthusiastic about), we call such collections of heuristics institutions.  Institutions are made up of embedded routines, scripts, sets of rules, that are used for decision making.

This kind of thinking got lapped up by sociologists in mid century and into the 70s and 80s.  They called it Role Theory.  The language is absolutely colloquial by now: who were your “role models” growing up?  Our first theories of organizations, going back to our very special Max Weber, included some version of the Occupational Role — a social construct that exists independent of the people occupying it, accruing its own rights and duties (or expectations).*

The problem with Role Theory is that it’s equally as implausible as Rational Choice is, for the exact same reason.

How many roles do you imagine there are that anyone fills in one day?  On a lazy day, I visit three social media websites, go to the laundromat and do my sheets, stop and buy a cup of coffee, read for school, and write a cocky blog post.  In the intuition of role theory, all of these activities constitute a social stage for which I read off a script of lines and stage direction (how long to make eye contact with the Mexican ladies watching telenovelas at the laundromat; which version of thank you to use for the 22 year-old hipster girls at the coffee shop; which tone to use while I respond to an argument I haven’t heard on the internet before).

This theory of behavior seems to put even greater demands on the computational resources of the agent than rational choice.  If you thought solving calculus problems was tough; hark back to your experience in a school play, trying to learn all of your lines, stage blocking, and then of course making it all come off naturally.

I think the problem we have here is that in conjuring both theories we are starting from the various intuitions that modern people have come up with in the street to talk about their behaviors–that military people have been called corporals and that traders have traded right up until they made a loss–and tried to generalize these ex post rationalizations, or stories about our behaviors, into scientific ex ante predictions, about how social systems, or individual psychology, compel behavior in the first place.

The problem may not be the evidence supporting or refuting either of these theories (or others), but rather that we are so whipped up in a game of telling one another stories about ourselves after the story has already been written by our actions, that we are relatively helpless to do anything but turn these ex post rationalizations into theories of the determinants, causes, antecedents, motivations, etc. of action.

*This theory of occupational roles is I think one of the major reasons that such a fantastic misunderstanding of labor markets and wages remains — people literally think of jobs as slots.  Jobs are not slots.  Jobs come and go as a function of the exigencies of technological innovation.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s