Came across post on Gigerenzer's "The Bias Bias in Behavioral Economics." Know Gigerenzer's work well - teach it in my classes. While agree on some points, my sense is he 1) creates strawman of BE/JDM research that is strange/nonexistent 2) overclaims statmodeling.stat.columbia.edu/2019/07/14/gig… 1/n
The behavioral story that people are dumb is missing the fact that we are quite good at learning.
And deciphering causality.
(When that’s *actually* the goal. Unlike, eg, w/ political discussions.)
C’mon Moshe, what behavioral story that people are dumb? We have whole fields on when learning converges and when it doesn’t. It’s really not fair to keep attacking a straw man of behavioral Econ that does not exist. I wrote a thread about it: twitter.com/alexoimas/stat…
Not every behavioral researcher. Not every behavioral story.
But yes many explain, eg, people’s political misperceptions using cognitive and informational constraints.
Imo one nice piece of evidence against that premise (which again, is not always shared by all behavioral scientist, but is common) is that we are quite good at deciphering causality when not driven by motivated reasoning.
The toddler evidence imo is pertinent.
(As is evidence that we figure stuff out in non-political domains just fine.)
I don’t think the alternative that, eg, political views are driven by ignorance or stupidity is an inherently bad take. It’s a priori plausible. And many believe it.
I don’t think it’s straw manning to summarize it and describe evidence against.
* my apologies if made it sound like *all* behavioral scientists advocate this alternative.
Thanks Moshe. A few things. I think in general when offering critiques of specific arguments it's helpful to specify "researchers who hold this view" or, even better, cite specific work. Here, this would have been useful because the view you target (fairly, in my opinion) is...
*not* held by the majority of BE researchers. So to spur constructive conversation, it's useful to specify exactly what you're targeting rather than getting reactive responses from those in the field you highlight but who don't recognize what you are targeting.
Second point on learning. There is a healthy, growing literature in BE on exactly what you are describing: people holding entrenched beliefs due to incentives. This work on motivated reasoning, or belief-based utility, argues that such phenomena are due to incentives.
There are great empirical and theoretical papers on this, this recent paper by @schwardmann & @JoelvdWeele being one example: docs.google.com/viewer?a=v&pid…. Whole conferences are devoted to it (at least two that I know of this summer).
Yup. Well aware of that paper (and like it).
FWIW though I think that the approach therein (and the one I take) differs from that of belief based utility approach.
And what I (and the theory behind that paper) call incentives, isn’t quite the same.
Namely, we don’t mean what feels good. We mean the primary rewards—like food sex and social esteem— that we evolved to pursue, and evolved to shape learning processes.
Which leads to very different types of explanations (and is the cornerstone of much of my criticism of b.e.).
(But I admit that the particular evidence I cited at start of this convo is consistent w/ both approaches. Just rules out cognitive constraints.)
But I think we've talked about this before. There is a reason economics has a sufficient statistics approach, and is distinct from other sciences. Your points are on *why* something feels good. Which is important!
But for purposes that economists use these models for, it may be sufficient to just assume X feels good and abstract from underlying motives. Researchers can argue, no, it's not sufficient to assume this for the stated purposes. And then more structure is imposed on reduced form.
What I hope is clear from this is that I'm not saying doing research on motives is not super important. It is! What I'm saying is that if bar for BE research is to model every single motive, the model will not be usable for the types of things economists need these models for.
Hmm, that’s fair. And sometimes that’s fine.
But imo it’s not always separable into different “levels of analysis.”
Often the “what feels good” story has implicit assumptions about the cause.
Eg that “good feelings” can be had by hacking a flawed hedonic system.
??? I cannot agree with the claim that we are quite good at determining causality absent politics. This is a situational claim, and there are plenty of situations where we find all sorts of ad hoc explanations. See Gilovic’s How We Know What Isn’t So.
Stephen E Weinberg
I have seen gilovich’s book. And think it very much fits my (intended) argument: he is (typically) talking bout domains where we have strong persuasion motives. Eg we want others to think we are smart and attractive.
(Admittedly such persuasion motives aren’t just in political domain. But that’s a context where such motives are HUGE. AND accuracy motives are absent. So persuasion motives rule the day. But agree, e.g., beliefs related to self presentation almost as bad)